Financial and socio-economic viability of irrigated agricultural development in the Roper catchment Australia’s National Science Agency A technical report from the CSIRO Roper River Water Resource Assessment for the National Water Grid Chris Stokes1, Diane Jarvis2, Tony Webster1, Ian Watson1, Shokhrukh Jalilov1, Yvette Oliver1, Allan Peake1, Alex Peachey3, Steve Yeates1, Caroline Bruce1, Seonaid Philip1, Di Prestwidge1, Adam Liedloff1, Perry Poulton1, Ben Price1, Stephen McFallan1 1CSIRO, 2James Cook University, 3Northern Territory Government ISBN 978-1-4863-1897-1 (print) ISBN 978-1-4863-1898-8 (online) Citation Stokes C, Jarvis D, Webster A, Watson I, Jalilov S, Oliver Y, Peake A, Peachey A, Yeates S, Bruce C, Philip S, Prestwidge D, Liedloff A, Poulton P, Price B and McFallan S (2023) Financial and socio-economic viability of irrigated agricultural development in the Roper catchment. A technical report from the CSIRO Roper River Water Resource Assessment for the National Water Grid. CSIRO, Australia. Copyright © Commonwealth Scientific and Industrial Research Organisation 2023. To the extent permitted by law, all rights are reserved and no part of this publication covered by copyright may be reproduced or copied in any form or by any means except with the written permission of CSIRO. Important disclaimer CSIRO advises that the information contained in this publication comprises general statements based on scientific research. The reader is advised and needs to be aware that such information may be incomplete or unable to be used in any specific situation. No reliance or actions must therefore be made on that information without seeking prior expert professional, scientific and technical advice. To the extent permitted by law, CSIRO (including its employees and consultants) excludes all liability to any person for any consequences, including but not limited to all losses, damages, costs, expenses and any other compensation, arising directly or indirectly from using this publication (in part or in whole) and any information or material contained in it. CSIRO is committed to providing web accessible content wherever possible. If you are having difficulties with accessing this document please contact Email CSIRO Enquiries . CSIRO Roper River Water Resource Assessment acknowledgements This report was funded through the National Water Grid’s Science Program, which sits within the Australian Government’s Department of Climate Change, Energy, the Environment and Water. Aspects of the Assessment have been undertaken in conjunction with the Northern Territory Government. The Assessment was guided by two committees: i.The Assessment’s Governance Committee: CRC for Northern Australia/James Cook University; CSIRO; National Water Grid (Department of Climate Change, Energy, the Environment and Water); NT Department of Environment, Parks and Water Security; NT Department of Industry, Tourism and Trade; Office of Northern Australia; Qld Department of Agriculture and Fisheries; Qld Department of Regional Development, Manufacturing and Water ii.The Assessment’s joint Roper and Victoria River catchments Steering Committee: Amateur Fishermen’s Association of the NT; Austrade; Centrefarm; CSIRO, National Water Grid (Department of Climate Change, Energy, the Environment and Water); Northern Land Council; NT Cattlemen’s Association; NT Department of Environment, Parks Australia; Parks and Water Security; NT Department of Industry, Tourism and Trade; Regional Development Australia; NT Farmers; NT Seafood Council; Office of Northern Australia; Roper Gulf Regional Council Shire Responsibility for the Assessment’s content lies with CSIRO. The Assessment’s committees did not have an opportunity to review the Assessment results or outputs prior to its release. Numerous people were generous in assisting with information on cropping input prices and agronomy used in the crop gross margin analyses: Chris Howie (Bindaroo Pastures), Don Telfer (DPIRD, WA), Frank Miller (African Mahogany Australia), Vin Lange (Centrefarm), Alex Lindsay (Forsite Forestry), Scott Fedrici (Quintis), George Revell (Ag Econ), Arthur Cameron (NT Ag), Muhammad Sohail Mazhar (NT Ag), David McNeil (DPIRD, WA), Sarah Ryan (Tiwiplantations), Alireza Houshmandfar (NT DITT), NT Farmers. Steve McFallan (CSIRO) provided estimates of freight costs from the TraNSIT model. The Northern Territory Government assisted with gathering information, including Robyn Cowley (who provided GRASP parameter files for modelling native pastures) and subsequently provided much advice on the Banjo landsystem and cattle carrying capacities of the Roper catchment, as well as James Christian (who provided information on local producer practices). Acknowledgement is made to the Queensland Government who are the custodians for the GRASP code and provide ongoing model testing and improvement. GRASP reference: Rickert KG, Stuth JW and McKeon GM (2000) Modelling pasture and animal production. In: Mannetje L’t and Jones RM (eds) Field and laboratory methods for grassland and animal production research. CABI Publishing, New York, 29–66. This report was improved based on helpful review comments from Dr Natalie Stoeckl, Dr Andrew Ash, Tom Vanderbyl and Kev Devlin. Acknowledgement of Country CSIRO acknowledges the Traditional Owners of the lands, seas and waters, of the area that we live and work on across Australia. We acknowledge their continuing connection to their culture and pay our respects to their Elders past and present. Photo: Melon production, Roper catchment. Source: Nathan Dyer - CSIRO Director’s foreword Sustainable regional development is a priority for the Australian and Northern Territory governments. Across northern Australia, however, there is a scarcity of scientific information on land and water resources to complement local information held by Indigenous owners and landholders. Sustainable regional development requires knowledge of the scale, nature, location and distribution of the likely environmental, social and economic opportunities and the risks of any proposed development. Especially where resource use is contested, this knowledge informs the consultation and planning that underpins the resource security required to unlock investment. In 2019 the Australian Government commissioned CSIRO to complete the Roper River Water Resource Assessment. In response, CSIRO accessed expertise and collaborations from across Australia to provide data and insight to support consideration of the use of land and water resources for development in the Roper catchment. While the Assessment focuses mainly on the potential for agriculture, the detailed information provided on land and water resources, their potential uses and the impacts of those uses are relevant to a wider range of regional-scale planning considerations by Indigenous owners, landholders, citizens, investors, local government, the Northern Territory and federal governments. Importantly the Assessment will not recommend one development over another, nor assume any particular development pathway. It provides a range of possibilities and the information required to interpret them - including risks that may attend any opportunities - consistent with regional values and aspirations. All data and reports produced by the Assessment will be publicly available. Chris Chilcott Project Director C:\Users\bru119\AppData\Local\Microsoft\Windows\Temporary Internet Files\Content.Word\C_Chilcott_high.jpg The Roper River Water Resource Assessment Team Project Director Chris Chilcott Project Leaders Cuan Petheram, Ian Watson Project Support Caroline Bruce Communications Chanel Koeleman/Kate Cranney, Siobhan Duffy, Amy Edwards Activities Agriculture and socio- economics Chris Stokes, Caroline Bruce, Shokhrukh Jalilov, Diane Jarvis1, Adam Liedloff, Yvette Oliver, Alex Peachey2, Allan Peake, Maxine Piggott, Perry Poulton, Di Prestwidge, Thomas Vanderbyl7, Tony Webster, Steve Yeates Climate David McJannet, Lynn Seo Ecology Groundwater hydrology Indigenous water values, rights, interests and development goals Danial Stratford, Laura Blamey, Rik Buckworth, Pascal Castellazzi, Bayley Costin, Roy Aijun Deng, Ruan Gannon, Sophie Gilbey, Rob Kenyon, Darran King, Keller Kopf3, Stacey Kopf3, Simon Linke, Heather McGinness, Linda Merrin, Colton Perna3, Eva Plaganyi, Rocio Ponce Reyes, Jodie Pritchard, Nathan Waltham9 Andrew R. Taylor, Karen Barry, Russell Crosbie, Phil Davies, Alec Deslandes, Katelyn Dooley, Clement Duvert8, Geoff Hodgson, Lindsay Hutley8, Anthony Knapton4, Sebastien Lamontagne, Steven Tickell5, Sarah Marshall, Axel Suckow, Chris Turnadge Pethie Lyons, Marcus Barber, Peta Braedon, Kristina Fisher, Petina Pert Land suitability Ian Watson, Jenet Austin, Elisabeth Bui, Bart Edmeades5, John Gallant, Linda Gregory, Jason Hill5, Seonaid Philip, Ross Searle, Uta Stockmann, Mark Thomas, Francis Wait5, Peter L. Wilson, Peter R. Wilson Surface water hydrology Justin Hughes, Shaun Kim, Steve Marvanek, Catherine Ticehurst, Biao Wang Surface water storage Cuan Petheram, Fred Baynes6, Kevin Devlin7, Arthur Read, Lee Rogers, Ang Yang, Note: Assessment team as at June 15, 2023. All contributors are affiliated with CSIRO unless indicated otherwise. Activity Leaders are underlined. 1James Cook University; 2NT Department of Industry, Tourism and Trade; 3 Research Institute for the Environment and Livelihoods. College of Engineering, IT & Environment. Charles Darwin University; 4CloudGMS; 5NT Department of Environment, Parks and Water Security; 6Baynes Geologic; 7independent consultant; 8Charles Darwin University; 9Centre for Tropical Water and Aquatic Ecosystem Research. James Cook University. ii | Financial and socio-economic viability of irrigated agricultural development in the Roper catchment Shortened forms SHORT FORM FULL FORM AACo Australian Agricultural Company ABARES Australian Bureau of Agricultural and Resource Economics and Sciences ABS Australian Bureau of Statistics ANCOLD Australian National Committee on Large Dams APSIM Agricultural Production Systems sIMulator BCR benefit–cost ratio (present value of benefits/present value of costs for a project) CBA cost–benefit analysis CBD central business district CCS commercial cane sugar (percentage of extractable raw sugar in harvested cane) CGE Computable General Equilibrium CO2CRC a Cooperative Research Centre (CRC) investigating carbon capture and storage technologies CPI consumer price index CSIRO Commonwealth Scientific and Industrial Research Organisation DCF discounted cashflow DIDO drive-in drive-out (applied to type of workforce) DKIS Darwin-Katherine Interconnected System (electricity distribution) DOI Document Object Identifier DRBWCD Daly Roper Beetaloo Water Control District DS dry season EBITDA earnings before interest, taxes, depreciation and amortisation EI–O environmental input–output EIS environmental impact statement EPRI Electric Power Research Institute ET0 reference evapotranspiration FACE Free Air CO2 Enrichment FIFO fly-in fly-out (applied to type of workforce) GCM global climate model GDP Gross Domestic Product GFA gross floor area GM gross margin GSP Gross State Product GVAP Gross value of agricultural production (an ABARES statistic) GVIAP Gross value of irrigated agricultural production (an ABARES statistic) SHORT FORM FULL FORM GWWAP Georgina Wiso Water Allocation Plan HSD Health Service District HV high voltage (electricity transmission lines) ILUA Indigenous land use agreement I–O input–output IRR internal rate of return LCOE Least Cost of Energy MTLAWAP Mataranka Tindall Limestone Aquifer Water Allocation Plan NAWRA Northern Australia Water Resource Assessment NPF Northern Prawn Fishery NPV net present value NSW New South Wales NT Northern Territory O&M operation and maintenance (type of recurring cost) PAW plant available water PAWC plant available water capacity PCR post-completion review PE potential evaporation PHN primary health network PVR plant variety rights Qld Queensland RH relative humidity SA South Australia SA1 to SA4 ABS Statistical Area (spatial boundary for data collection), number indicates hierarchy level SA1s are the smallest unit for general release of census data (and are aggregated into larger units) SEIFA Socio-Economic Indexes for Areas (published by ABS) SGG soil generic group TDH total dynamic head (1 m TDH = 9.8 kPa) TraNSIT Transport Network Strategic Investment Tool USA United States of America VPD vapour pressure deficit WA Western Australia WS wet season Units UNITS DESCRIPTION $ Australian dollars, at constant June 2021 value $ CapEx per ML/y at dam Capital expenditure per ML/y of the dam’s supply capacity at the dam wall % percent °C degree Celsius AE animal equivalent (cattle) AE/km2 adult equivalent per kilometre square d day bale bale of processed cotton lint (227 kg) FTE full time equivalent GL gigalitre GW⋅h gigawatt hour h hour ha hectare (= 10,000 m2) ha/AE hectares per adult equivalent hPa hectopascal (1 hPa = 100 pascals) km kilometre km/h kilometre per hour kPa kilopascal kV kilovolt kVA 1000-volt amps kW kilowatt kW⋅h kilowatt hour L litre m metre m3/ha metres-cubed per hectare mm millimetre MJ/m2/day megajoules per metre square per day ML megalitre ML/ha megalitres per hectare (equivalent to 100 mm) ML/year megalitres per year (ML/y) MVA megavolt amp (1 MVA = 1 MW) MW megawatt MW⋅h megawatt hour t metric tonne UNITS DESCRIPTION t/ha metric tonnes per hectare TEU twenty-foot equivalent unit (standard shipping container) y year Preface Sustainable regional development is a priority for the Australian and Northern Territory governments. For example, in 2023 the Northern Territory Government committed to the implementation of a new Territory Water Plan. One of the priority actions announced by the government was the acceleration of the existing water science program ‘to support best practice water resource management and sustainable development’. The efficient use of Australia’s natural resources by food producers and processors requires a good understanding of soil, water and energy resources so they can be managed sustainably. Finely tuned strategic planning will be required to ensure that investment and government expenditure on development are soundly targeted and designed. Northern Australia presents a globally unique opportunity (a greenfield development opportunity in a first-world country) to strategically consider and plan development. Northern Australia also contains ecological and cultural assets of high value and decisions about development will need to be made within that context. Good information is critical to these decisions. Most of northern Australia’s land and water resources, however, have not been mapped in sufficient detail to provide for reliable resource allocation, mitigate investment or environmental risks, or build policy settings that can support decisions. Better data are required to inform decisions on private investment and government expenditure, to account for intersections between existing and potential resource users, and to ensure that net development benefits are maximised. In consultation with the Northern Territory Government, the Australian Government prioritised the catchment of the Roper River for investigation (Preface Figure 1-1) and establishment of baseline information on soil, water and the environment. Northern Australia is defined as the part of Australia north of the Tropic of Capricorn. The Murray– Darling Basin and major irrigation areas and major dams (greater than 500 GL capacity) in Australia are shown for context. The Roper River Water Resource Assessment (the Assessment) provides a comprehensive and integrated evaluation of the feasibility, economic viability and sustainability of water and agricultural development. While agricultural developments are the primary focus of the Assessment, it also considers opportunities for and intersections between other types of water-dependent development. For example, the Assessment explores the nature, scale, location and impacts of developments relating to industrial and urban development and aquaculture, in relevant locations. The Assessment was designed to inform consideration of development, not to enable any particular development to occur. As such, the Assessment informs – but does not seek to replace – existing planning, regulatory or approval processes. Importantly, the Assessment does not assume a given policy or regulatory environment. As policy and regulations can change, this enables the results to be applied to the widest range of uses for the longest possible time frame. Preface Figure 1-1 Map of Australia showing Assessment area It was not the intention – and nor was it possible – for the Assessment to generate new information on all topics related to water and irrigation development in northern Australia. Topics not directly examined in the Assessment are discussed with reference to and in the context of the existing literature. Functionally, the Assessment adopted an activities-based approach (reflected in the content and structure of the outputs and products), comprising eight activity groups; each contributes its part to create a cohesive picture of regional development opportunities, costs and benefits. Preface Figure 1-2 illustrates the high-level links between the eight activities and the general flow of information in the Assessment. For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au Preface Figure 1-2 Schematic diagram of the high-level linkages between the eight activities and the general flow of information in the Assessment. Assessment reporting structure Development opportunities and their impacts are frequently highly interdependent and consequently, so is the research undertaken through this Assessment. While each report may be read as a stand-alone document, the suite of reports most reliably informs discussion and decisions concerning regional development when read as a whole. The Assessment has produced a series of cascading reports and information products: • Technical reports; that present scientific work at a level of detail sufficient for technical and scientific experts to reproduce the work. Each of the eight activities has one or more corresponding technical report. • A Catchment report; that for the Roper catchment synthesises key material from the technical reports, providing well-informed (but not necessarily-scientifically trained) readers with the information required to make decisions about the opportunities, costs and benefits associated with irrigated agriculture and other development options. • A Summary report; that for the Roper catchment provides a summary and narrative for a general public audience in plain English. • A Summary factsheet; that for the Roper catchment provides key findings for a general public audience in the shortest possible format. The Assessment has also developed online information products to enable the reader to better access information that is not readily available in a static form. All of these reports, information tools and data products are available online at https://www.csiro.au/roperriver. The website provides readers with a communications suite including factsheets, multimedia content, FAQs, reports and links to other related sites, particularly about other research in northern Australia. For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au Executive summary The purpose of this report is to provide information on the costs, risks and benefits of new irrigated development in the catchment of the Roper River, at farm to scheme and regional scales, and supply chains beyond. The overall conclusion is that large public dams would be marginal, but on-farm water sources, suitably sited, could provide good prospects for viable new farms. Farming options for the Roper catchment • Amongst the range of irrigated cropping options suited to Roper catchment environments, those that are most likely to be profitable (if development costs can be kept low enough) are annual horticulture, cotton, forages, and peanut. Most broadacre cropping is best suited to dry season planting (late March to August), but this requires more irrigation. Wet season planting (December to early March) would be possible on well drained soils, particularly for annual horticulture (targeting harvests for winter gaps in supply in southern markets). The amount of irrigation required depends on a number of factors, but as an example, cotton planted at the end of the wet season would need about 6–9 ML/ha while the perennial forage crop Rhodes grass would require up to 20 ML/ha each year. • Sequential cropping systems present opportunities for generating additional net revenue from the same capital investment. There are many potential cropping sequences (more than one crop per year in the same field) that show agronomic potential for matching back-to-back crop requirements with Roper catchment growing conditions, particularly on well-draining loamy soils (like those in the western Roper catchment). However, these farming systems would need to be developed and proven locally, and the challenges involved should not be underestimated. • Trafficability constraints on finer-textured clay soils (mainly scattered patches of alluvial soils) would make scheduling crop sequences in the same year more difficult, and so would restrict the choice of crops to those with shorter growing seasons. • The farm-scale performance of cropping systems will be determined by: (i) finding markets and supply chains that can provide a sufficient price, scale and reliability of demand; (ii) the skill of farmers in managing the operational and financial complexity of adapting crop mixes and production systems to Roper catchment environments; (iii) the nature of water resources in terms of their costs to develop, the volume and reliability of supply, and the timing of when water is available relative to optimal planting windows; and (iv) the nature of the soil resources in terms of their scale and distribution, proximity to water sources and supply chains, farming constraints, the crops they can support with viable yields, and costs to develop. • There are natural synergies in growing irrigated forages to integrate with existing beef enterprises in the Roper catchment. Annual beef revenue would need to be increased by the order of $2000—$3000 per hectare of irrigated forages to provide an acceptable return on the initial capital costs of development. Few options were able to generate such returns, with Rhodes grass being the most likely forage to be viable (but only under scenarios where development costs could be kept low and beef prices remained at current (2023) high levels). Economic considerations beyond the farm gate • A review of recent public dams built in Australia highlighted some areas where cost–benefit analyses (CBAs) for water infrastructure projects could be improved, particularly regarding more realistic forecasting of demand for water. This report provides information for benchmarking a range of assumptions commonly used in such CBAs, including demand forecasting, that can be used to check when proposals for new dams are being unrealistically optimistic (or pessimistic). • Financial analyses indicated that large dams in the Roper catchment are unlikely to be viable (if governments required full cost recovery at a 7% internal rate of return and provided no assistance). Irrigators could afford to contribute at most $20,000 to $30,000/ha towards the cost of new off-farm water infrastructure (about half, before accounting for risks), whereas the most cost-effective potential large dam developments would likely cost about $50,000 per hectare of new irrigated farmland (e.g. capital cost of $500 million to build a dam and supporting infrastructure that could irrigate about 10,000 ha). • Dams could be marginally viable if public investors accepted a 3% internal rate of return or partial contributions to water infrastructure costs similar to established irrigation schemes in other parts of Australia. • On-farm water sources provide better prospects and, where sufficiently cheap water development opportunities can be found, these could likely support viable broadacre farms and horticulture with low development costs. Horticulture with high development costs (like fruit orchards with modern packing facilities) in the Roper catchment would be more challenging unless farm financial performance could be boosted by finding niche opportunities for premium produce prices, or savings in production and marketing costs. • For broadacre crops, gross margins of the order of $4000/ha/y (before accounting for the costs of water or risks) are required to provide a sufficient return on investment. Those crops likely to achieve such a return (under current conditions, in 2023) include Rhodes grass hay and wet- season cotton. • Horticultural gross margins would have to be higher (of the order of $7,000—$11,000/ha/y) to provide an adequate return on the higher capital costs of developing this more intensive type of farming (relative to broadacre). Profitability of horticulture is extremely sensitive to prices received, so the locational advantage of supplying ‘out of season’ (winter) produce to southern markets is critical to viability. Wet season planted annual horticultural row crops would be the most likely to achieve these returns in the Roper catchment. • Farm performance can be affected by a range of risks, including water reliability, climate variability, price fluctuations, and learning to adapt farming practices to new locations. Setbacks that occur early on after an irrigation scheme is established have the largest effect on scheme viability. There is a strong incentive to start any new irrigation development with well-proven crops and technologies, and to be thoroughly prepared for the anticipatable agronomic risks of establishing new farmland. Risks that cannot be avoided need to be managed, mitigated where possible, and accounted for in determining the realistic returns that may be expected from a farm/scheme and the capital buffers that would be required. • Any development of new irrigated agriculture and supporting infrastructure would have knock- on benefits to the regional economy beyond the direct economic growth from the new farms and construction. During the ongoing production phase of a new irrigation development, there could be an additional $0.46 to $1.82 million of indirect regional benefits for each million dollars of direct benefits from increased agricultural activity (gross farm revenue net of any payments for water), depending on the type of agricultural industry. Each net $100 million increase in agricultural activity could create about 100 to 852 jobs, depending on the agricultural industry. Identifying investment opportunities As market, regulatory, infrastructure and other conditions in the Roper catchment change from those prevailing at the time this report was written, investors/farmers would be expected to adapt and respond to opportunities and challenges accordingly. Ultimately, to establish and sustain new irrigated developments in the Roper catchment, investors will need to identify opportunities that simultaneously solve all three of the following questions: • Markets: Where is the investor going to sell their produce and how are they going to set up the supply chains to get their products, at low-enough cost, from the Roper catchment to those who want to buy them? • Production systems: What is the investor going to grow and do they understand how this needs to be grown differently in tropical Australia (and the soils, water resources and climates of Roper catchment environments specifically) to where they have gained their previous experience? • Competition: Why is it better to grow the chosen product(s) in tropical Australia, relative to alternative options of growing the same product elsewhere, or growing different products in the chosen location? There is a wide variety of potential investors in northern Australia agriculture, each of whom will come with different strengths regarding the above three criteria but will also likely have blind spots where they are not initially completely aware of the full scale of the challenges involved. Successful investments have typically been able to find comprehensively planned answers to all three of these questions, while failures have not. This Assessment (including companion reports) has focused primarily on ‘production system’ challenges by filling knowledge gaps on the land and water resources in the Roper catchment. This report evaluated the farming options that could be sustainably and profitably developed on that resource base, and it provides additional supporting information for overcoming the competitive disadvantages and market constraints for northern Australia. Widespread expansion of agriculture in the Roper catchment is unlikely to occur in the near term. However, smaller scale opportunities will continue to emerge for those able to find niches for cost savings and suitable markets, and who have the capital and capacity to persist through the challenging establishment years. Contents Director’s foreword .......................................................................................................................... i The Roper River Water Resource Assessment Team ...................................................................... ii Shortened forms .............................................................................................................................iii Units ............................................................................................................................... v Preface .............................................................................................................................. vii Executive summary .......................................................................................................................... x Part I Background context 1 1 Introduction ........................................................................................................................ 2 1.1 Rationale and approach ........................................................................................ 2 1.2 Structure of this report .......................................................................................... 3 2 Socio-economic context ..................................................................................................... 6 2.1 Agricultural industries of the Northern Territory .................................................. 6 2.2 Market opportunities and challenges ................................................................. 14 2.3 Demography and economy of the Roper catchment .......................................... 22 Part II Agricultural development options 45 3 Biophysical factors affecting agricultural performance ................................................... 46 3.1 Climate ................................................................................................................. 46 3.2 Soils and land suitability ...................................................................................... 54 3.3 Irrigation systems ................................................................................................ 60 3.4 Crop types ............................................................................................................ 64 3.5 Crop and forage management ............................................................................ 73 3.6 Cattle and beef production ................................................................................. 77 4 Approach for evaluating agricultural options ................................................................... 80 4.1 Multi-scale framework for evaluating agricultural viability ................................ 80 4.2 Modelling crop yields and water use .................................................................. 81 4.3 Greenfield crop gross margin tool ....................................................................... 88 4.4 Modelling the integration of forage and hay crops within existing beef cattle enterprises ........................................................................................................................ 91 5 Performance of agricultural development options .......................................................... 93 5.1 Principles of dryland and irrigated cropping ....................................................... 93 5.2 Performance of irrigated crop options ................................................................ 97 5.3 Cropping systems .............................................................................................. 109 5.4 Integrating forages into livestock systems ........................................................ 114 Part III Economics 125 6 Lessons learned from recent Australian dam-building experiences .............................. 126 6.1 Introduction ....................................................................................................... 126 6.2 Methods and case study selection .................................................................... 128 6.3 Proposed and realised outcomes for each case study development ............... 130 6.4 Key lessons......................................................................................................... 136 7 New infrastructure demand and costs ........................................................................... 142 7.1 Introduction ....................................................................................................... 142 7.2 Agricultural growth and water demand trajectories ........................................ 143 7.3 Development costs for land and water resources ............................................ 146 7.4 Processing costs ................................................................................................. 150 7.5 Transport costs .................................................................................................. 152 7.6 Energy infrastructure costs ............................................................................... 156 7.7 Community infrastructure costs ........................................................................ 158 8 Financial viability of new irrigated development ........................................................... 160 8.1 Introduction ....................................................................................................... 160 8.2 Balancing scheme-scale costs and benefits ...................................................... 162 8.3 Price irrigators can afford to pay for a new water source ................................ 168 8.4 Financial targets required to cover costs of large, off-farm dams .................... 169 8.5 Financial targets required to cover costs of on-farm dams and bores ............. 172 8.6 Risks associated with variability in farm performance ..................................... 174 8.7 Achieving financial viability in a new irrigation development .......................... 178 9 Regional economics ........................................................................................................ 181 9.1 Multiplier and input–output (I–O) approach .................................................... 181 9.2 Regional economic benefits .............................................................................. 190 9.3 Water multipliers and environmental input–output (EI–O) analysis ................ 198 Part IV Concluding comment 211 10 The ‘sweet spot’ for northern development .................................................................. 212 References ........................................................................................................................... 217 Part V Appendices 235 Aquaculture opportunities and viability ............................................................ 236 Figures Preface Figure 1-1 Map of Australia showing Assessment area ................................................... viii Preface Figure 1-2 Schematic diagram of the high-level linkages between the 8 activities and the general flow of information in the Assessment. ............................................................................. ix Figure 1-1 Map of the Assessment area showing the Roper catchment and catchments from previous related assessments of land and water resources in northern Australia ........................ 4 Figure 2-1 Trends in gross value of agricultural production for crops and livestock in the Northern Territory over 40 years (1981–2021) .............................................................................. 6 Figure 2-2 Production of major horticultural crops in NT in 2019–20 showing (a) quantity (tonnes) and (b) economic value of production ($ million) .......................................................... 10 Figure 2-3 Changes in agricultural subsectors’ relative values (GVAP) in (a) Australia and (b) the Northern Territory over 40 years (1981–2021) ............................................................................ 10 Figure 2-4 Trend in horticultural crops production across Australian states and territories over 40 years (1981–2021) ................................................................................................................... 11 Figure 2-5 Mango production in Australia and the NT in 2019–20 expressed in terms of (a) the number of trees and (b) the number of businesses ..................................................................... 12 Figure 2-6 NT cattle movement over the last decade (2012–2021) ............................................. 14 Figure 2-7 Comparison of marketing costs, across three categories of agriculture, relative to GVAP for (a) Australia and (b) the Northern Territory (average 2011–12 to 2020–21) ............... 18 Figure 2-8 Adaptability of Australia’s exports of broadacre commodities, as demonstrated by year-to-year variations in export volumes and market mixes before and after the disruptions associated with the Covid pandemic ............................................................................................ 20 Figure 2-9 Farmers’ terms of trade in Australia for (a) input prices and (b) prices received for commodities.................................................................................................................................. 21 Figure 2-10 Boundaries of the Australian Bureau of Statistics Statistical Area Level 4 (SA4) and Statistical Area Level 2 (SA2) regions used for demographic data in this Assessment ................ 24 Figure 2-11 Land use classification for the Roper catchment ...................................................... 27 Figure 2-12 Map of regions in the Northern Prawn Fishery ......................................................... 30 Figure 2-13 Road rankings and conditions for the Roper catchment ........................................... 32 Figure 2-14 Type 2 vehicle access across the Roper catchment................................................... 33 Figure 2-15 Common configurations of heavy freight vehicles used for transporting agricultural goods in Australia .......................................................................................................................... 34 Figure 2-16 Road speed restrictions for the Roper catchment .................................................... 35 Figure 2-17 Agricultural enterprises in the Roper catchment and amount of annual trucking to/from them ................................................................................................................................ 37 Figure 2-18 Electricity generation and transmission network and natural gas pipelines in the Roper catchment ........................................................................................................................... 39 Figure 2-19 Location, type and volume of annual licenced surface water and groundwater entitlements .................................................................................................................................. 42 Figure 3-1 Long-term fortnightly climate variation in (a) rainfall, (b) maximum and (c) minimum temperatures for the historical climate (1890 to 2015) at Mataranka ........................................ 47 Figure 3-2 Long-term fortnightly climate variation in (a) solar radiation, (b) relative humidity (RH), and (c) vapour pressure deficit (VPD) under the historical climate (1890 to 2015) at Mataranka ..................................................................................................................................... 48 Figure 3-3 Historical potential evaporation (PE) in the Roper catchment at Mataranka for (a) monthly PE (range is the 20th to 80th percentile monthly PE) and (b) time series of annual PE (line is the 10-year running mean) ................................................................................................ 50 Figure 3-4 Monthly mean daily (a) solar radiation and (b) vapour pressure deficit for three locations in the Roper catchment (Bulman, Mataranka and Ngukurr: latitude 13.7–14.9° S) and Griffith (subtropical: latitude 34.3° S) ........................................................................................... 51 Figure 3-5 Monthly mean daily (a) maximum and (b) minimum daily temperatures for three locations in the Roper catchment (Bulman, Mataranka, and Ngukurr: latitude 13.7—14.9° S) and Griffith (subtropical: latitude 34.3° S) ........................................................................................... 51 Figure 3-6 Mean annual number of tropical cyclones in Australian for (a) El Niño years and (b) La Niña years ...................................................................................................................................... 53 Figure 3-7 The soil generic groups (SGGs) of the Roper catchment produced by digital soil mapping ........................................................................................................................................ 55 Figure 3-8 Agricultural versatility index map for the Roper catchment ....................................... 59 Figure 3-9 Annual cropping calendar for cropping options in the Roper catchment ................... 75 Figure 3-10 Soil wetness indices that indicate when seasonal trafficability constraints are likely to occur on Kandosols (sandy) and Vertosols (high clay) with a Bulman climate ........................ 76 Figure 4-1 Overview of multi-scale approach for evaluating the viability of agricultural development options .................................................................................................................... 80 Figure 4-2 Climate comparisons of Roper catchment sites versus established irrigation areas at Katherine and Ord River (WA) ...................................................................................................... 83 Figure 4-3 Farm gross margin tool used for consistent comparative analysis of different greenfield farming options ........................................................................................................... 88 Figure 5-1 Influence of planting date on dryland grain sorghum yield at Bulman for (a) Kandosol and (b) Vertosol ............................................................................................................................ 95 Figure 5-2 Influence of available irrigation water on grain sorghum yields for planting dates (a) on 1 February and (b) 1 August, for Kandosols with a Bulman climate ....................................... 97 Figure 5-3 Fluctuations in seedless watermelon prices at Melbourne wholesale markets from April 2020 to February 2023 ....................................................................................................... 104 Figure 5-4 Probability of exceedance graphs for (a) simulated irrigation requirement (mm) and (b)grain yield (t/ha), for a grain sorghum crop grown under current climate conditions and forboth a drier and wetter future climate scenario at Mataranka in the Roper catchment .......... 108 Figure 5-5 Average liveweights for each scenario for male animals born at the end of November ..................................................................................................................................................... 122 Figure 6-1 Map showing locations of the five case study dams used in this review .................. 129 Figure 7-1 Trends in gross value of agricultural production (GVAP) in (a) Australia and (b) the NT over 40 years (1981–2021) ......................................................................................................... 144 Figure 7-2 Trends for increasing gross value of irrigated agricultural production (GVIAP) as available water supplies have increased for (a) fruits, (b) vegetables, (c) fruits and vegetables combined, and (d) total agriculture ............................................................................................ 145 Figure 7-3 Mean annual water application rate by horticultural type across Australian states and territories .................................................................................................................................... 146 Figure 7-4 Road layer used in TraNSIT, showing road ranks and heavy vehicle restrictions ..... 153 Figure 7-5 Freight paths from Mataranka to key ports and southern markets ......................... 155 Figure 8-1 Financial structure of irrigation scheme used in accounting for costs, revenue and use of land and water resources ....................................................................................................... 165 Figure 9-1 Regions used in the input–output (I–O) analyses relative to the Roper catchment Assessment area ......................................................................................................................... 184 Figure 9-2 Water consumption in the NT analysed by ‘household’, ‘agricultural’, ‘mining and manufacturing’ and ‘other industries’ consumption from 2000–01 to 2016–17 ...................... 202 Figure 9-3 Water consumption in Queensland analysed by ‘household’, ‘agriculture’, ‘mining and manufacturing’ and ‘other industries’ consumption from 2001–01 to 2016–17 ............... 203 Figure 9-4 Water consumption for the NT and for Queensland (Qld), in total and agricultural consumption alone, illustrating percentage of total consumption arising from agriculture from 2000–01 to 2016–17 ................................................................................................................... 204 Figure 9-5 Water multipliers estimated for Roper catchment, demonstrating direct and indirect (production and consumption induced) demand for additional ML of water resulting from additional $1 million stimulus applied to expanding a particular sector of the economy ......... 207 Figure 10-1 Viable irrigated agriculture investments in the Roper catchment require a combination of capturing opportunities and mitigating risks in three critical areas: markets, production systems and competition ......................................................................................... 214 Tables Table 2-1 Major demographic indicators for the Roper catchment ............................................. 23 Table 2-2 SEIFA scores of relative socio-economic advantage for the Roper catchment ............ 25 Table 2-3 Key employment data for the Roper catchment .......................................................... 25 Table 2-4 Value of agricultural production within the wider SA4 region and estimates of the value of agricultural production for the Roper catchment .......................................................... 28 Table 2-5 Overview of commodities annually transported into and out of the Roper catchment ....................................................................................................................................................... 36 Table 2-6 Schools servicing the Roper catchment ........................................................................ 44 Table 2-7 Number and percentage of unoccupied dwellings and population for the Roper catchment ..................................................................................................................................... 44 Table 3-1 Soil generic groups (SGGs), descriptions, management considerations and correlations to Australian Soil Classification (ASC) for the Roper catchment .................................................. 55 Table 3-2 Area and proportions covered by each soil generic group (SGG) for the Roper catchment ..................................................................................................................................... 57 Table 3-3 Qualitative land evaluation observations for locations in the Roper catchment shown in Figure 3-8 .................................................................................................................................. 60 Table 3-4 Details of irrigation systems applicable for use in the Roper catchment ..................... 61 Table 3-5 Pumping costs by irrigation operation .......................................................................... 62 Table 4-1 Crop options where performance was evaluated in terms of water use, yields and gross margins ................................................................................................................................ 82 Table 4-2 Crop yields and median irrigation water requirement delivered to the field .............. 84 Table 5-1 Soil water content at sowing and rainfall for the 90-day period following sowing for three sowing dates, based on a Bulman climate on Vertosol ...................................................... 94 Table 5-2 Performance metrics for broadacre cropping options in the Roper catchment: applied irrigation water, crop yield and gross margin (GM) for three environments ............................... 99 Table 5-3 Breakdown of variable costs relative to revenue for broadacre crop options ........... 101 Table 5-4 Sensitivity of cotton crop gross margins (GMs) to variation in yield, lint prices, and distance to gin ............................................................................................................................. 102 Table 5-5 Sensitivity of forage (Rhodes grass) crop gross margins (GMs) to variation in yield and hay price ...................................................................................................................................... 102 Table 5-6 Performance metrics for horticultural options in the Roper catchment: annual applied irrigation water, crop yield and gross margin (GM) ................................................................... 103 Table 5-7 Breakdown of variable costs relative to revenue for horticultural crop options ....... 105 Table 5-8 Sensitivity of watermelon crop gross margins (GMs) to variation in melon prices and freight costs ................................................................................................................................. 105 Table 5-9 Performance metrics for plantation tree crop options in the Roper catchment: annual applied irrigation water, crop yield and gross margin (GM)....................................................... 107 Table 5-10 Breakdown of variable costs relative to revenue for plantation tree crop options . 107 Table 5-11 Likely annual irrigated crop planting windows, suitability and viability in the Roper catchment ................................................................................................................................... 113 Table 5-12 Sequential cropping options for Kandosols .............................................................. 114 Table 5-13 Production and financial outcomes from the different irrigated forage and beef production scenarios for a representative property on the Sturt Plateau ................................. 119 Table 5-14 Net present values for forage development options ............................................... 120 Table 6-1 Summary characteristics of the five dams used in this review .................................. 129 Table 6-2 Summary of the expectations and reported outcomes for each dam reviewed ....... 131 Table 6-3 Benefits (and disbenefits) included in proposals justifying the five dams reviewed . 137 Table 6-4 Summary of key issues and potential improvements arising from a review of recent dam developments ..................................................................................................................... 140 Table 7-1 Indicative development costs for different types of irrigated farms ......................... 148 Table 7-2 Indicative capital costs for developing on-farm water sources (including distribution from source to cropped fields) ................................................................................................... 149 Table 7-3 Indicative capital costs for developing two irrigation schemes based on the most cost- effective dam sites in the Roper catchment ............................................................................... 149 Table 7-4 Indicative capital and operating (fixed and variable) costs for a cotton gin from two sources ........................................................................................................................................ 151 Table 7-5 Indicative capital and operating costs for a basic sugar mill capable of processing 1000 t cane per hour ............................................................................................................................ 152 Table 7-6 Indicative road transport costs between the Roper catchment and key markets and ports ............................................................................................................................................ 154 Table 7-7 Indicative costs of transmission and distribution lines, for sizes relevant to this Assessment ................................................................................................................................. 157 Table 7-8 Indicative costs of transformer, for sizes likely to be relevant to developments in the Assessment area ......................................................................................................................... 157 Table 7-9 Indicative construction costs for different types of community facilities in Darwin . 158 Table 7-10 Indicative construction costs for new schools .......................................................... 159 Table 8-1 Types of questions that users can answer using the tools in this chapter ................. 161 Table 8-2 Assumed indicative capital and operating costs for new off- and on-farm irrigation infrastructure .............................................................................................................................. 167 Table 8-3 Price irrigators can afford to pay for water based on the type of farm, the farm water use, and annual gross margin (GM) of the farm ......................................................................... 168 Table 8-4 Farm gross margins (GMs) required to cover the costs of off-farm water infrastructure (at the suppliers’ target internal rate of return (IRR)) ................................................................ 170 Table 8-5 Water pricing required to cover costs of off-farm irrigation scheme development (dam, water distribution, and supporting infrastructure) at the investors target internal rate of return (IRR) .................................................................................................................................. 171 Table 8-6 Farm gross margins (GMs) required to achieve target internal rate of return (IRR) given different capital costs of farm development (including an on-farm water source) ......... 172 Table 8-7 Equivalent costs of water per megalitre for on-farm water sources with different capital costs of development, at the internal rate of return (IRR) targeted by the investor ..... 173 Table 8-8 Risk adjustment factors for target farm gross margins (GMs), accounting for the effects of reliability and severity (level of farm performance in ‘failed’ years) of periodic risks ..................................................................................................................................................... 176 Table 8-9 Risk adjustment factors for target farm gross margins (GMs), accounting for the effects of reliability and timing of periodic risks......................................................................... 176 Table 8-10 Risk adjustment factors for target farm gross margins (GMs), accounting for the effects of learning risks ............................................................................................................... 178 Table 9-1 Key 2016 data comparing the Roper catchment with the related I–O analysis regions ..................................................................................................................................................... 185 Table 9-2 Regional economic impact estimated by I–O analysis for the total construction phase of an irrigated agricultural development based on estimated Type ll multipliers determined from two independent I–O models ............................................................................................ 192 Table 9-3 Estimated full time equivalent (FTE) number of jobs created for the construction phase of an irrigated agricultural development ......................................................................... 193 Table 9-4 Estimated regional economic impact per year resulting from four scales of direct increase in agricultural output (rows) in the Roper catchment, for the different categories of agricultural activity for two I–O models (columns) .................................................................... 195 Table 9-5 Type II regional economic multipliers applicable to the ongoing agricultural production phase of the Roper catchment development .......................................................... 195 Table 9-6 Estimated impact on annual household incomes and full time equivalent (FTE) jobs within the Roper catchment resulting from four scales of direct increase in agricultural output (rows) for the different categories of agricultural activity (columns) ........................................ 196 Table 9-7 Re-analysis of corrected water use data by industry in the NT for 2000–01 ............. 206 Table 9-8 Estimated increase in demand for water (ML) based on increased demand for construction activity, for the construction phase of an irrigated agricultural development ..... 208 Table 9-9 Estimated increase in total demand for water (ML) based on increased demand for agricultural output during the operational phase of an irrigated agricultural development .... 209 Table 10-1 Opportunities and risks across three key criteria for the success of irrigated development in the Roper catchment ........................................................................................ 215 Part I Background context Part I provides the background information and context to support the analyses in Parts II and III. Chapter 1 summarises the main principles governing successful irrigated development in northern Australia and describes the structure of this report. Chapter 2 describes the current social and economic characteristics of the Roper catchment and the existing agriculture and infrastructure base, as background context for the chapters that follow, and the foundation from which any new development would build. Part II analyses the farm-scale performance of potential irrigated agricultural development options and covers the agronomic principles that determine the types of cropping systems that could be sustainably and profitably implemented. Part III analyses the scheme-scale viability of irrigated development options and economic considerations beyond the farm gate that would be required for those developments to succeed in the Roper catchment. 1 Introduction 1.1 Rationale and approach Large infrastructure projects, such as new irrigation developments, can deliver substantial social and economic benefits to the regions in which they are built, but are complex and costly investments. The difficulty in accurately estimating costs and the chance of incurring unanticipated expenses during construction, or not achieving projected benefits when completed, mean that there are risks to the viability of developments if they are not thoroughly planned. For example, large water (and other) infrastructure projects routinely cost more and deliver less benefit than originally planned (see review in Chapter 6). In recent decades there has been growing emphasis in Australia on greater accountability and transparency in how water resources are managed and priced (e.g. Infrastructure Australia, 2021a, 2021b; NWGA, 2022, 2023), and part of this shift has involved greater scrutiny of the viability of potential new water infrastructure. Similar issues arise, at smaller scale, for on-farm water sources for irrigated development. Past work has examined the factors that contribute most to whether greenfield (mainly irrigated) agricultural developments succeed or fail; this includes lessons from historical farming experiences in northern Australia (Ash et al., 2014; Ash and Watson, 2018), analyses of potential new development options in other northern Australian catchments (Stokes et al., 2017), and a financial evaluation of the Bradfield Scheme and more cost-effective water infrastructure alternatives (Stokes and Jarvis, 2021). The broad principles emerging from that work highlight the most important determinants of success for greenfield agricultural development in locations like the Roper catchment: • Although northern Australian environments are challenging for agriculture, the main historical factors determining the success of farming ventures have been management, planning and finances. • By their nature, greenfield developments in new farming locations lack the strong support networks of peers for sharing experiences and learning together, which makes overcoming the initial challenges of adapting farming systems to local conditions more difficult. • It is inevitable that greenfield farms in locations without an established history of farming will initially perform below their long-term potential and allowance needs to be made for a period of learning-by-doing. Staging developments, where possible, allows making mistakes at a small scale, where risks are contained and rectified, before expanding. • Blind overoptimism is unhelpful; it ignores anticipatable risks that otherwise could have been mitigated or avoided (including unrealistic assumptions about productivity, sizes of markets and prices), and wastes time and resources pursuing options long after they should be abandoned. • The rate at which water resources are developed (especially large public water infrastructure investments) needs to be scaled to realistic expansion rates for agriculture and associated trajectories of demand for new water. Building oversized infrastructure, that can’t be fully utilised shortly after development, is very cost inefficient. • Long supply chains and distant processing and phytosanitary facilities often put northern agriculture at a competitive disadvantage. Economies of scale are required to support viable local processing and shortened supply chain routes, which often creates a chicken-and-egg dilemma. • Given the competitive disadvantages of farming in northern Australia (versus established southern farming areas) there is a greater imperative for finding the most cost-effective development opportunities, that is, good soils in close proximity to good water resources, both of which can be developed at affordable expense. • Agricultural industries that have succeeded in northern Australia have often done so by finding niche opportunities for cost savings and markets (such as out-of-season production), but these usually come at the expense of scalability, limiting the rate at which expansion can occur. All the above themes are strongly echoed and reinforced throughout this report. The report aims to provide information that can assist in planning and evaluating the viability of investments in irrigated development, and quantifying the costs, benefits and risks involved. The intention is to provide a generic information resource that is broadly applicable to a wide range of irrigated agriculture development options, rather than being prescriptive about how future development (if any) in the Roper catchment (Figure 1-1) should proceed, or examining specific proposals in detail. 1.2 Structure of this report This report complements the overall Assessment of potential opportunities and constraints for new irrigated agriculture in the Roper catchment, by conducting a multi-scale analysis (from farm to scheme, region and markets) that identifies the agronomic, social and economic conditions required for potential new developments to succeed. The chapters of this report are structured into three main parts as follows: Part I: provides the background information for the analyses in the later parts of the report. Chapter 1 is this introduction. Chapter 2 describes the current social and economic characteristics of the Roper catchment and the existing agriculture and infrastructure base, as background context for the chapters that follow, and the foundation from which any new development would build. Part II analyses the farm-scale performance of potential irrigated agricultural development. Chapter 3 provides background information on tropical agronomy including the environmental factors affecting crop performance (climate, soils, land suitability and water resources), the range of potential crop options, and crop management considerations. Chapter 4 describes the approach used for crop modelling and other quantitative analyses of a set of 19 possible crop options for the Roper catchment and the methods used to estimate their potential performance (in terms of yields, water use and farm gross margins). Chapter 5 presents the results of the farm-scale analyses, uses narrative risk analyses to illustrate opportunities and challenges for establishing viable new farms, and interprets the practical implications of the farm-scale information provided for the types of cropping systems that could be fine-tuned to Roper catchment environments. Figure 1-1 Map of the Assessment area showing the Roper catchment and catchments from previous related assessments of land and water resources in northern Australia Se-R-500_Prawn_fishery_boundaries_v2 For more information on this figure or equation please contact CSIRO on enquiries@csiro.au Part III analyses the scheme-scale viability of irrigated development options and economic considerations beyond the farm gate. Chapter 6 reviews recent large dam projects in Australia for how well proposed benefits were realised in practice to elicit lessons for future developments and to provide context for the financial analyses that follow. Chapter 7 provides indicators of the demand trajectories for new water (and other) infrastructure from growth in agriculture in the NT and describes the types of infrastructure that would be required to support large-scale irrigated development, together with indicative costs and options for building that infrastructure. Chapter 8 uses a generic financial analysis approach, to demonstrate the key determinants of irrigation scheme viability that investors need to balance. Chapter 9 quantifies the regional costs and benefits of irrigated development using regional input–output analysis. It also includes estimates of the proportions of those benefits that are likely to flow to Indigenous Australians, and an environmental input–output analysis of how increased agricultural water use would stimulate additional demand from other water users. Part IV concludes by summarising key principles for identifying agricultural investment opportunities in the Roper catchment. Part V is an appendix on aquaculture opportunities and their potential viability (mainly summarising previous work). 2 Socio-economic context This chapter begins with a general overview of agricultural industries in the NT (Section 2.1) and the market opportunities and challenges involved in expanding agriculture in the NT (Section 2.2), before providing more specific details on the demography, economy and existing infrastructure in the Roper catchment (Section 2.3). 2.1 Agricultural industries of the Northern Territory Over the past decade the economy of the NT has experienced significant growth, reaching a Gross State Product (GSP) of $26.2 billion in 2020–21, and is forecast to continue growing at an average 2.9% over the 5 years to 2025–26 (Northern Territory Economy, 2022). Having allocated more than 45% of the territory’s land to agriculture, the NT agricultural sector produced a gross commodity value of $1.14 billion in 2019 (NT Farmers, 2019). The key value contributions were $800 million from beef cattle, and $340 million from plant-based agricultural and horticultural crops (NT Farmers, 2019). However, the sector’s production varies significantly from year to year due to seasonal conditions, and changes in global and domestic demand for NT commodities (NT Department of Treasury and Finance, 2022) (Figure 2-1). While the NT’s agricultural industry is diversified, the main subsectors are livestock (mainly cattle grazing) and cropping (mainly horticulture). Figure 2-1 Trends in gross value of agricultural production for crops and livestock in the Northern Territory over 40 years (1981–2021) Source: ABS (2022) The agriculture, forestry and fishing sector represented 3.6% of the NT’s GSP compared with 2.3% nationally (Northern Territory Economy, 2022). Despite the relatively small contribution to GSP, the industry is vital in generating economic activity and employment in regional areas. For more information on this figure please contact CSIRO on enquiries@csiro.au 03006009001981–821982–831983–841984–851985–861986–871987–881988–891989–901990–911991–921992–931993–941994–951995–961996–972001–022002–032003–042004–052005–062006–072007–082008–092009–102010–112011–122012–132013–142014–152015–162016–172017–182018–192019–20Gross value of production ($ million) YearCropsLivestock The agriculture, forestry and fishing industry of the NT provides a significant number of jobs and supports economic activity in rural remote areas of the NT (NT Department of Treasury and Finance, 2022). The agriculture, forestry and fishing sector accounted for 1.8% of the NT’s workforce in 2020–21, compared with the national rate of 2.6%. It has major connections with other sectors of the NT’s economy such as wholesale and retail sale, manufacturing and transport. The NT’s agricultural sector consists of four major subsectors: broadacre crops, horticultural crops, livestock (cattle grazing), and forestry. Crops grown in the NT include mangoes (Mangifera indica) (accounting for half of Australia’s total production), melons, Asian vegetables, grapes (Vitis spp.), tropical fruit, ornamental plants, pastures and fodder crops. Current NT crop production is focused predominantly on domestic markets (with limited export). There is growing interest in broadacre crops for food (e.g. rice (Oryza sativa) and peanuts (Arachis hypogaea)) or fibre (e.g. cotton (Gossypium spp.) and industrial hemp (Cannabis sativa ssp. Sativa)), as well as a maturing plantation-forestry estate. After cattle grazing, forestry is the second largest user of land, with stands of hickory wattle (Acacia mangium) for wood chip, African mahogany (Khaya spp.) for cabinet timber, and Indian sandalwood (Santalum album) for a range of wood and oil products (Northern Territory Government, 2022). The extensive coastline provides many sites of pristine water suitable for fishing and aquaculture with substantial scope to participate directly in wild catch fisheries, aquaculture, fish processing and value-adding opportunities. 2.1.1 Broadacre crops Most value for broadacre cropping in northern Australia is realised through its integration with existing agribusiness enterprises, particularly where cropping supports, boosts and diversifies beef production (CRCNA, 2020). The capacity to produce a variety of crops in northern Australia is well- established. However, due to the barriers, costs and risks of tropical broadacre cropping, farmers in the north are seeking to leverage crop production as part of a broader effort to promote the profitability and sustainability of their businesses (CRCNA, 2020). Farms growing a single broadacre crop may be marginal, so it may be necessary to develop farming systems that are able to exploit the synergies of multiple crops to establish viable new businesses. Thus, production of each crop needs to consider where it fits and could derive added value within a wider business model. For instance, a single irrigated farm could comprise such crops as melons, hay, cotton, mungbean (Vigna radiata) and maize (Zea mays) that could be delivered to domestic as well as export markets (CRCNA, 2020). Broadacre cropping is not yet well-established in the NT, but investments in crop trials and processing industries, such a cotton ginning, could facilitate new farming opportunities in crops such as cotton, peanuts, sorghum (Sorghum spp.), hemp and rice. Cotton Australian cotton growers produce some of the best quality fibre cotton in the world with additional market opportunities for lint and seed by-products. In addition to lint, cotton seed can be used as a feed supplement for cattle or processed into oil, soaps, cosmetics and meal. Cotton is a highly scalable crop that is well suited to the soils and climates of the NT. The NT Farmers Association identified and proposed the Katherine–Daly Basin as one of the promising regions for further agricultural development. For example, the development of a 62,000 ha cotton industry in the NT with four gins was estimated to cost $732 million and could support up to 424 jobs (PwC, 2019). As of 2023, a cotton gin is currently being built north of Katherine and will require 55,000 bales per year initially to be financially viable. Peanut Peanuts could be an important crop in the NT, given their high gross margin and suitability to the region (NT Farmers, 2019). Despite peanuts often being classified as drought tolerant, the highest quality and yields are often obtained from areas with reliable rainfall and with access to irrigation. As such, irrigation management is critical to obtaining a high-yielding, economically viable crop. There are about 120 manufacturers in Australia using peanuts for snack food, confectionery and peanut butter. Seven processors account for about 80% of the market (GRDC, 2017). Domestic peanut consumption is about 50,000 t of nut-in-shell, increasing at 2 to 3% per year (GRDC, 2017). Domestic production can satisfy less than 50% of the local demand with the rest being imported (Bega Peanut Butter, 2022). There is a potential opportunity to expand the production of peanuts in the NT. Australian growers only supply a fraction of the local domestic demand and prices remain relatively stable throughout the year. Currently, there are no shellers/processors in the NT, so all produce would have to be freighted to processors in Queensland, which has historically been a difficult market for NT producers to penetrate. Sorghum Dryland grain sorghum has been the only crop grown consistently in the NT over the past two decades. It is used in the local stockfeed market and has been grown successfully between Daly Waters and Darwin, with the main growing areas being Douglas–Daly and Katherine. Dryland grain sorghum can be grown in regions where the average wet-season rainfall is between 700 and 1400 mm. Although yields of 4.5 to 5.0 t/ha have been achieved in small areas, more realistic commercial yields would be 2.5 to 3.5 t/ha. As well as grain, sorghum provides a very useful stubble for grazing. Weight gains for cattle on sorghum stubble have been recorded at up to 1 kg per head per day in the early to mid-dry season (Hausler et al., 2002). ABARES forecast the value of sorghum production in 2022–23 to be $929 million (ABARES, 2022). China is currently the dominant market for Australian sorghum (83% of exports were destined for this market in 2021–22) (ABARES, 2022). Hemp In the last decade, regulations regarding cultivation of industrial hemp have eased in all states of Australia making it possible to grow industrial hemp once again. The NT Government made it legal to grow industrial hemp in August 2019 (if a licence is granted). The development of industrial hemp presents farmers with a possible opportunity, following the momentum and traction it has gained in the past decade. Industrial hemp is a fast growing, annual herbaceous plant with a deep tap root. To achieve a viable gross margin, the production of industrial hemp would need to be grown at scale. As such, the farming operations would likely need to be comparable to that of other broadacre crops, such as sorghum, cotton and soybean (Glycine max). The plant is largely intolerant to wet, flooded or waterlogged soil, thus would need to be sown at a 4–5 cm depth in non-crusting soils to achieve rapid germination. Furthermore, adequate moisture is required during active growth to obtain an economically viable yield (NT Farmers, 2019). The NT climate is potentially very suitable for hemp production, particularly if farmers are able to harvest two crops a year. Broadacre mechanised production of non-medicinal hemp products would require tailoring to northern Australian conditions and developing suitable harvesting, transport and processing technology. One of the most critical issues with the northern Australian hemp industry is a lack of market for products. There are, however, several options for marketing hemp, including processing, packing and selling products online or joining a company or cooperative that processes and markets grain and products on growers’ behalf (Kumar and Telfer, 2022). 2.1.2 Horticulture industry Australia’s horticulture industry is the second largest industry in terms of rural production, after wheat (Triticum aestivum). Broadly, horticultural production includes four major subindustries: fruits, vegetables, nuts, and nursery products (including cut flowers, turf) (DAFF, 2022). Fruit and nuts account for 52% of total horticultural production, vegetables account for 31%, and nursery goods and ornamental crops account for 17% (Entegra Signature Structures, 2022). In 2019–20, horticultural production was valued at more than $15 billion (DAFF, 2022). Horticultural produce is typically perishable and expensive to store and transport, with stringent phytosanitary standards for export, so most Australian horticultural production is sold domestically, and growth in horticultural produce is constrained by growth in demand from local consumers (and establishment of new supply chains for export). For individual products, fresh exports seldom represent more than 15% of Australian production. Tropical and subtropical fruits grown in Australia include bananas (Musa spp.), citrus (Citrus spp.), macadamias (Macadamia integrifolia) and mangoes. Plantings of other fruits include some of Asian origin such as rambutan (Nephelium lappaceum), mangosteen (Garcinia mangostana) and durian (Durio spp.). Most production of tropical and subtropical fruits occurs in Queensland. Vegetables grown in Australia include asparagus (Asparagus officinalis), zucchini (Cucurbita pepo), squash (Cucurbita spp.), potatoes (Solanum tuberosum), tomatoes (Solanum lycopersicum), carrots (Daucus carota), mushrooms (e.g. Agaricus bisporus), onions (Allium cepa) and lettuce (Lactuca sativa). Greenhouses are also increasingly being used for horticulture, allowing produce to can be grown throughout the year. In 2022–23 the value of horticultural exports is projected to be $3.4 billion (ABARES, 2022). The NT horticultural industry is represented by three subindustries: fruit, vegetables, and nurseries (including cut flowers and cultivated turf). Almost all production is sold interstate. According to the NT Farmers Association, the value of horticultural production in the NT (excluding forestry) was $341 million in 2019–20. Together, mangoes and melons accounted for most horticulture in the NT by both produce volume and value (Figure 2-2). The total value of horticultural production for 2019–20 season comprised (Figure 2-2): • $128.8 million for mango production • $69.4 million for melon production • $61 million for vegetables • $37.1 million for field crops and fodder • $22.4 million for other fruits (including citrus, grapes, dates, tropical and exotic fruits) • $15 million for nursery and turf production. (a) Quantity (b) Value Figure 2-2 Production of major horticultural crops in NT in 2019–20 showing (a) quantity (tonnes) and (b) economic value of production ($ million) Source: Sangha et al. (2022) The market price of horticultural crops is highly elastic, thus any large increase in seasonal supply of horticultural crops, relative to seasonal market demand, may create an oversupply and subsequently reduce the market price and farm gate return for farmers (NT Farmers, 2019). Figure 2-3 illustrates trends in gross value of agricultural production (GVAP) for horticulture relative to other agricultural industries for Australia and the NT for the last 40 years. Over this period horticulture has increased in value at a faster rate than other agricultural industries and has come to represent a higher share of total agricultural production in both the NT and Australia overall. (a) Australia (b) Northern Territory Figure 2-3 Changes in agricultural subsectors’ relative values (GVAP) in (a) Australia and (b) the Northern Territory over 40 years (1981–2021) Data points are decade averages of annual values. Source: ABS (2022) For more information on this figure please contact CSIRO on enquiries@csiro.au 020406080MelonsMangoesVegetablesGrapes andother fruitsOther cropsQuantity (thousand tonnes) For more information on this figure please contact CSIRO on enquiries@csiro.au 04080120160MangoesMelonsVegetablesGrapes andother fruitsOther cropsValue of production ($ million) For more information on this figure please contact CSIRO on enquiries@csiro.au 0% 50% 100% 1981-901991-002001-102011-21GVAP (%) DecadeCrops (horticulture)Crops (other)Livestock For more information on this figure please contact CSIRO on enquiries@csiro.au 0% 50% 100% 1981-901991-002001-102011-21GVAP (%) DecadeCrops (horticulture)Crops (other)Livestock Australia’s horticultural industry produces most of its vegetables for domestic consumption and exports only a small number of vegetables, with the major export product being asparagus. The major horticultural growing regions are located in Victoria, NSW and Queensland. Although the NT only accounts for a small fraction of national horticultural production, it has raised its share threefold from 0.3% in 1981–90 to 1.0% in the last decade (2011–21) (Figure 2-4). Figure 2-4 Trend in horticultural crops production across Australian states and territories over 40 years (1981–2021) Data points are decade averages of annual values. The share of the ACT is too small to be visible in the bars above. Numbers above columns show the NT share of total Australian horticultural production. Source: ABS (2022) Mango Nationally, the mango industry is worth more than $110 million, and is heavily concentrated in the NT and Queensland (Plant Health Australia, 2021). Northern Australia is the epicentre of Australian mango production, with over 4 million trays produced each year in the NT alone (Brann, 2021). The NT mango industry produces roughly half of Australia’s total mango crop annually, selling to both domestic and international markets. The Australian mango export landscape is exhibiting some strong growth prospects, particularly into north Asian markets, supported by the recent (2021) completion of a phytosanitary treatment facility as part of a new export hub at Darwin International Airport. There are potential opportunities to grow markets for Australian mangoes in South-East Asia, China and the USA. The mango industry has a proactive peak industry body that has been working with governments to secure and improve technical market access and support fledgling trade in several protocol markets. In 2020, Australia exported 6574 t, or around 9% of its fresh mango production, valued at $35 million. For more information on this figure please contact CSIRO on enquiries@csiro.au 04,0008,00012,0001981–901991–002001–102011–21GVAP ($ million) DecadeQueenslandVictoriaSouth AustraliaNew South WalesWestern AustraliaTasmaniaAustralian Capital TerritoryNorthern TerritoryNT 0.3% NT 1.0% NT 0.9% NT 0.6% Mango flowering, fruiting and harvesting in the NT occur earlier in the season than in other mango production regions in Australia. Therefore, mangoes from the NT are the earliest fruit to reach the Australian market (Clonan et al., 2020). More than half of mango-producing businesses in the NT are located in the Darwin area (Figure 2-5). (a) Number of mango trees (b) Number of mango businesses Figure 2-5 Mango production in Australia and the NT in 2019–20 expressed in terms of (a) the number of trees and (b) the number of businesses Source: Clonan et al. (2020) There are existing challenges in the supply of fruit to market that occur when extreme or unusual conditions have an impact on flower induction. These events have an impact on the volume and timing of fruit supplied to the market. Storage and transport of mangoes is highly restricted due to the fruit’s short shelf life and sensitivity to storage conditions. Similarly, transport of mangoes overseas to export markets is difficult due to biosecurity and quality issues. Mangoes are therefore more susceptible to changes in harvest timing than many other commodities. These confounding effects make it challenging to manage the supply of mangoes to reach markets at times of high demand and high prices, particularly since late harvesting in the NT will overlap with production from regions like Queensland that typically produce most of their fruit slightly later in the season (Clonan et al., 2020). Such regional interactions in seasonal supply are typical for horticultural products, and are exacerbated by highly perishable produce, expensive storage and inelastic consumer demand to create high volatility in the prices paid to growers. But the high value of horticultural crops, particularly during seasonal windows of unmet demand, means they can be very profitable for growers who are able to effectively manage this volatility. 2.1.3 Plantation forestry Plantation forestry is becoming an increasingly important industry and is currently the second largest agricultural land user in the NT after cattle grazing, with more than 42,000 ha of the NT currently used to produce forestry products in managed plantations. There are currently three plantation forestry projects in the NT: • Hickory wattle plantations are being grown for woodchip exports on the Tiwi Islands, managed by Midway Limited on behalf of the Tiwi Plantations Corporation on Melville Island. A total of 23,000 t of acacia woodchips were sold from the Tiwi Islands in 2020–21. For more information on this figure please contact CSIRO on enquiries@csiro.au 05001,0001,500AustraliaNTDarwinNumber of trees (thousands) For more information on this figure please contact CSIRO on enquiries@csiro.au 0150300450AustraliaNTDarwinNumber of businesses • African mahogany is being grown in the Douglas–Daly and Katherine regions by African Mahogany Australia. The company manages the largest plantation estate of this species in the world. It is being grown for a high‑value, sawn-timber market, which includes veneer boards, floorboards and feature-grade timber. These plantings are currently in mid‑rotation, with a predicted rotation of 18 to 25 years. • Indian sandalwood is also grown in the Douglas–Daly and Katherine regions for oil and pharmaceuticals. These plantations are currently in mid-rotation and will not realise the bulk of their value for another 3 to 4 years. The NT’s forestry industry supplies timber products to many markets, including timber for manufacturing luxury furniture, oil for fragrance and beauty products, and logs and woodchips for domestic timber mills and overseas markets such as China. There are potential opportunities to expand the forestry industry and existing plantations. Currently, there are existing African mahogany plantations in the Douglas–Daly region, an expanding woodchip industry on the Tiwi Islands and several Indigenous-owned forests across Arnhem Land. Despite the fragmented nature of the region’s plantations, several opportunities for industry development have been identified. These include plantation expansions in the Douglas–Daly region, eucalyptus plantation growing advantages on the Tiwi Islands, and possible opportunities for branded products for international markets (NT Farmers, 2019). Such opportunities, if developed, could significantly increase the value of forestry and forest products in the NT (NT Farmers, 2019). 2.1.4 Beef cattle In 2022, the NT supplied 582,300 head of live beef cattle to both overseas and domestic markets (Figure 2 6). The major live export markets for the NT are the Philippines, Vietnam, and Malaysia. The current reliance on beef cattle farming significantly increases the financial and social risk to farm owners and their employees, leaving them vulnerable to any change in trade relationships with major export nations (NT Farmers, 2019). While the Australian cattle industry has experienced a significant drop in prices due to herd restocking, with a 20% decrease in 2023 (AFR, 2023), analysts remain optimistic about the future, predicting more stable market conditions and a positive outlook for exports. The Australian industry has successfully rebuilt its cattle herd, aided by favourable weather conditions, leading to increased supply and a larger national cattle herd. However, the recent high prices for Australian cattle, leading up to 2023, have prompted Indonesia, a major trading partner, to seek alternative supplies, including Brazilian beef. Despite this challenge, experts predict a positive outlook for the industry, with Rabobank forecasting stable prices and strong beef producer margins (AFR, 2023). The anticipated lower US cattle production, increased demand from Japan, South Korea, and China's economic recovery further contribute to the positive market environment. Figure 2-6 NT cattle movement over the last decade (2012–2021) Source: NT Department of Treasury and Finance (2022) There are also natural synergies of new cropping with the established beef industry. For instance, forages are well suited as a first crop to grow in greenfield locations (they are more forgiving and have a ready local market), there are opportunities for vertical integration of forages with beef production (both on-farm consumption and in pelleted form for live export cattle), and cotton seed (separated from lint during ginning) is a good dietary supplement for cattle. 2.2 Market opportunities and challenges 2.2.1 Overview The Northern Territory's strategic location near major export markets in Asia, well-established industrial and export infrastructure, rich natural resources, and status as Australia's northernmost jurisdiction and base for defence activities in the Indo-Pacific region have all historically influenced the Territory's investment landscape. As a result, the mining, defence, and agriculture sectors have played a significant role in the NT's economic output, as evidenced by the major projects currently in progress in those areas (Northern Territory Economy, 2022). In this section, we will provide an overview of the market outlook for increased agricultural production in the NT, along with an assessment of the potential opportunities and challenges that could arise for future economic development and investment. 2.2.2 Key advantages Having abundant land resources, suitable water and soil resources, and being close to growing Asian markets, the NT has significant potential for further development of agriculture and For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au agricultural-related businesses and industries. Several characteristics in the NT make it conducive to new agricultural investment: • stable growth trends in production of agricultural products, in particular horticulture • stringent controls over agrichemical use and management of biosecurity risk in agri-food industries to assure a reputation of clean, green and safe agricultural produce • proximity to large and growing Asian markets • counter-seasonal production, where fresh produce is harvested to fill seasonal gaps in supply in southern (and export) markets, which is often associated with higher produce prices. The NT’s favourable climate presents opportunities for horticultural development. Regional differences in temperature, rainfall and humidity are important determinants for crop and land selection. Many crops that grow in the NT are harvested earlier than those grown in southern Australia. This gives the NT a distinct market advantage in Australia and also provides counter- seasonal opportunities in overseas markets. 2.2.3 Risks, challenges, and constraints One of the issues routinely raised by farmers in northern Australia relates to the regulatory hurdles that make it challenging to secure sufficient land and water resources to not only make their own on-farm investments viable, but also to reach the scale of production to make investments in processing and the other steps in new supply chains viable (for example, cotton gin and pulse packaging facilities) (CRCNA, 2020). Underdeveloped pastoral areas are also often difficult to convert to alternative land uses such as horticulture due to the high level of investment required for developing road infrastructure and other facilities on farms. Furthermore, any investment for farm expansion requires a consistent and secure water supply to make land productive. The process to negotiate land tenure and access to land and water resources can be complex and expensive (Sangha et al., 2022). Increasing input costs, mainly for fertilisers, pesticides and transport, further challenge the sustainability and profitability of the horticultural industries (Sangha et al., 2022). Other key challenges for new agricultural development in the NT are covered below. Water Access to water is a major limitation to the expansion of crop production in northern Australia. In the NT, growers rely primarily on groundwater for irrigation, but current allocations of groundwater are already well utilised, restricting further horticultural development. A lack of understanding of the impacts of long-term surface water and groundwater extraction in the NT (Sangha et al., 2022) could impede making additional water available for agriculture. Even if new water becomes available, there are uncertainties about the feasibility of cropping options and whether they could sustainably and profitably make use of that water (CRCNA, 2020). Biosecurity A range of exotic plant pests and diseases can have an impact on horticultural production for commercial fruit and vegetable growers in the NT. Emerging pests and diseases remain a constant threat to horticultural industries and can have a range of impacts, from production loss to food safety risks, downgrades in produce quality and temporary loss of access to markets. Beehives are used extensively for pollination services across Australia with many apiarists providing mobile services throughout key pollination periods. The recent varroa mite detection around the port of Newcastle in NSW has led to hives being destroyed within the eradication zone and restrictions on the movement of beehives from NSW into other states and territories. This is particularly an issue for almond production as almonds rely exclusively on pollination services from beehives. Other plant pests and diseases such as banana freckle, blueberry rust and fruit fly also have the potential to have an impact on production and produce quality. Outbreaks of fruit fly remain a challenge to horticultural exports (ABARES, 2022). Labour supply Horticultural farms rely more heavily on casual and contract labour compared to most other agricultural activities. This is mainly due to the intensive seasonal labour requirement for picking, sorting and packing produce. Many workers are employed for short, intensive time periods over the harvest period before moving on to work in other regions. As a result, this work is often suited to workers who are more mobile (ABARES, 2022). Many of the occupations that are projected to see increasing demand in 2021–2025 are related to agriculture, including ‘farmers and farm managers’, ‘crop farmers’, ‘mixed crop and livestock farmers’, and ‘mixed crop and livestock farm workers’ (Australian Industry and Skills Committee, 2022). The NT plant-based agricultural and horticultural industry is forecast to employ over 3500 full time equivalents (FTEs) by 2030, which would create demand for an additional 1803 FTEs in these industries (NT Farmers, 2019). Market and supply chain The strong influence of retailers and supermarkets in Australian markets means horticultural commodity prices are typically determined by retailers rather than growers (Sangha et al., 2022). As a result of market mechanisms (including supply chain costs and losses), growers receive a much lower price (i.e. 9–10 fold less) than what consumers pay. Many family-farm owners have expressed concerns that declining marginal profitability is exerting increasing pressure on small- and medium-scale farms, which may be forced to sell to larger enterprises (Sangha et al., 2022). There are inflexibilities in supply chain processes that create challenges for growers. For example, major retail chains require 6 weeks’ notice of supply, and transport bookings needing to be made well in advance of harvest (Clonan et al., 2020). Australia’s horticultural supply chains face the following key issues (L.E.K. Consulting, 2021): • horticultural products are generally perishable, meaning freight is time-sensitive and long delays, such as roadblocks, can make the product unsalable • pests, diseases and contaminants, potentially from overseas, can enter the supply chain and damage the produce • the supply chain is dynamic, as particular farms can change the produce they grow or the use of the land. While this adds flexibility, it increases the reliance on flexible road freight, which is often less efficient than rail • a high sensitivity to variability in weather makes fruit and vegetable production volatile. Weather events can affect the production, harvest timing and quality (including size and weight) of the produce. Variation in volumes can make decisions around infrastructure investment difficult. The timing of harvest for horticultural produce has significant flow-on effects for the entire supply chain, including influencing market prices. Currently, national industry crop forecasts are used to convey any changes to volume or supply to the whole industry. The national industry body also facilitates information supply between growers, marketers, wholesalers and retailers. Horticultural industries could be assisted by new tools to better forecast these impacts from the farm level through to the effects on national supply chains (Clonan et al., 2020). Climate change In considering future development, it is highly likely temperatures will increase under climate change, as will rainfall variability (Mathew et al., 2018). Whether rainfall will increase or decrease in the NT is uncertain (Timbal et al., 2015; Section 3.1.8). Growers have expressed concerns about whether the gene pools in existing crop varieties are sufficiently diverse for industries to be able to adapt to threats of disease outbreaks and climate variability in the future. There is a lack of long- term data on how an increase in temperature will have an impact on mango production, which is a dominant, high-value commodity in the region (Sangha et al., 2022). Less information is available for small-scale emerging crops such as Asian vegetables and fruits. 2.2.4 Markets and infrastructure The low density of the population in northern Australia means that there is little demand for locally grown produce, so producers have to focus on markets and supply chains in the southern part of the country (for domestic consumption of horticultural produce) and export destinations (for bulk broadacre commodities) (CRCNA, 2020). The NT has a geographic advantage, relative to other parts of Australia, in its proximity to Asian and other markets. But it also subject to a range of disadvantages and constraints. Export markets are heavily dependent on supporting facilities. For example, small volumes or ‘non-bulk’ produce, such as cotton lint, require container packing, which is currently only available in Darwin, Gladstone, Mackay and Townsville, while bulk export of grains is possible in Wyndham, Darwin and possibly Townsville, but with different levels of infrastructure, mostly built for non- grains (CRCNA, 2020). Despite the closer geographic proximity of northern Australia to many key markets, supply chains are extended because most agricultural exports leave through southern ports. For example, currently no bulk food-grade containers are handled by Darwin Port (either import or export). Darwin Port has a low volume of container traffic. For example, the Port of Fremantle, which is the primary container port in WA, handles over three times more TEU (twenty-foot equivalent unit container) annually than Darwin Port. Approximately 20 t of grain can be loaded in a single TEU (depending on its density). Efficient supply chains require integrated circuits that minimise underutilisation of transport capacity on all legs. The challenge is to not just develop transport and handling capacity for exports, but to balance that with the compatible imports required to avoid the added cost of dead freighting empty containers (CRCNA, 2020). Another constraint for NT growers is competition with suppliers from other locations (domestically and internationally) with lower costs of production. For example, Figure 2-7 shows that marketing costs (including transport costs) are higher for growers in NT than other parts of Australia. Australian Bureau of Statistics (ABS) marketing costs include a range of expenses incurred for agricultural produce between farm gate and markets including freight, cost of containers, commission and other charges incurred in marketing. While these data are not collected on a completely comparable basis between jurisdictions, they are still useful for highlighting supply chain challenges for the NT relative to the rest of Australia. For all three categories of agriculture, marketing costs (farm gate to market) are higher for the NT than the national average, putting NT producers at a competitive disadvantage. (a) Australia (b) Northern Territory Figure 2-7 Comparison of marketing costs, across three categories of agriculture, relative to GVAP for (a) Australia and (b) the Northern Territory (average 2011–12 to 2020–21) Source: ABS (2022) 2.2.5 Export opportunities Asia accounts for well over half of Australia’s agricultural exports (Sefton and Associates, 2013) and strong income growth in the region could see demand continue to rise (Lockie, 2015). Asian agrifood demand is expected to double by 2050. This creates growth opportunities for Australia’s food industry in supplying safe, premium meat, dairy, wine, vegetables and processed, branded product to China’s growing middle class (KPMG and The University of Sydney China Studies Centre, 2013). An estimated 300 million people in middle-class households, projected to double in the next 10 years, are rapidly developing westernised consumption habits and diets. The value of red meat imports is expected to grow from a current $3 billion to $150 billion by 2050 (Lockie, 2015). For more information on this figure please contact CSIRO on enquiries@csiro.au 051015Percent (%) Crops (horticulture)Crops (other)Livestock Market access is critical for the Australian horticultural industry. Australian growers export approximately $2.92 billion worth of horticultural commodities annually, accounting for about 25% of Australia’s total production (Hort Innovation, 2021). Over the last 3–5 years in Australia there has been significant expansion in the area planted for horticultural commodities such as oranges, mandarins, avocados (Persea americana), almonds (Prunus dulcis) and macadamias. This has allowed horticultural exports to increase, contributing to a higher GVAP for the subsector. As the Australian domestic market is relatively small, expansion of international market access for horticultural commodities will be essential for future growth. If horticultural industries are unable to develop additional markets, there is potential for the pipeline of expanded production to deflate prices on the domestic market, impacting growers’ profitability. This could in turn lead to industry consolidation as less profitable growers exit the market (ABARES, 2022). Products with high labour and transport costs are not ‘natural’ candidates for large-scale Australian exports. However, returns for Australian fresh produce can be profitable in market niches (due to product seasonality, uniqueness, or Australia’s reputation for ‘clean and green’ produce) or quality lines. Fresh pear (Pyrus spp.), kiwi fruit (Actinidia deliciosa), grape and onion exports are a significant proportion of local production. The real value of fresh and processed horticultural exports has increased at an average annual rate of 5% (in Australian dollars). Australia benefits from being located fairly close to the developing markets in Asia. The country also benefits from being a supplier to northern hemisphere markets when products are out of season in those markets (Entegra Signature Structures, 2022). Total Australian horticultural exports for 2021–22 were estimated at $3.4 billion (ABARES, 2022). Exports are dominated by fruits and nuts. The top five horticultural exports by volume in 2019–20 were oranges, grapes, carrots, mandarins and almonds, while by value they were almonds, grapes, oranges, macadamias and mandarins (L.E.K. Consulting, 2021). The most important vegetable exports are asparagus, carrots and cauliflower (Brassica oleracea var. botrytis). There is also a significant export market in cut flowers, especially to Japan (Entegra Signature Structures, 2022). Much of Australia’s produce is seasonal, with a few exceptions such as potatoes, carrots, cauliflower and broccoli (Brassica oleracea var. italica) (L.E.K. Consulting, 2021). The value of horticultural imports to Australia also shows steady growth. In the last decade (2010– 11 to 2019–20) horticultural imports increased by 71.2%; imports of fruits, nuts and vegetables increased by 62.1%, 99.6% and 71.9%, respectively (DAFF, 2022). The largest suppliers of imported vegetables to Australia are South Korea (mushrooms), Mexico (asparagus), and China (garlic (Allium sativum) and onions). The largest suppliers of imported fruits to Australia are New Zealand (kiwi fruit and avocados) and the USA (oranges and table grapes). The value of fresh fruit exports amounted to more than three times the value of fresh fruit imports in 2019–20. Unlike horticultural crops, bulk broadacre commodities are traded on large global markets, with multiple competing international buyers. Australia already exports the vast majority of its broadacre commodities (e.g. 82% of cereals, 92% of pulses and 98% of oilseeds by value (ABARES, 2022)), so export markets could easily absorb the scale of increases in production that would be possible from the Roper catchment. Figure 2-8 demonstrates the adaptability of broadacre export markets that readily accommodate changes from year to year in product volumes and market access. The period in Figure 2-8 covers substantial disruptions to supply chains and market access restrictions, including the Covid pandemic, yet broadacre commodity exporters readily adapted to available markets to sell all commodities produced each year. Figure 2-8 Adaptability of Australia’s exports of broadacre commodities, as demonstrated by year-to-year variations in export volumes and market mixes before and after the disruptions associated with the Covid pandemic Only the ten largest export destinations for each year are shown. SAR = special administrative region Source: ABARES Trade dashboard (beta) (2022) 2.2.6 Recent volatility in costs and prices All costs and prices in this Assessment are standardised in real June 2021 Australian dollars (with inflation adjustments made to older sources when necessary). For agricultural commodity prices, that can fluctuate substantially from year-to-year, the decade mean (2011—2021) was used instead (so as not to be influenced by short-term dips and spikes in prices when comparing alternative cropping options). Historically, many agricultural inputs have experienced more moderate year-to-year price volatility than individual food and fibre commodities. However, over the duration of the Assessment, major global events such as the Covid pandemic, the war in Ukraine and other substantial disruptions to market access and supply chains have introduced a period of higher-than-normal volatility in agricultural input prices. Recent changes in agricultural terms of trade are therefore presented below as context for interpreting the June 2021 pricing used in this report (Figure 2-9). The interrelated inputs of fertiliser and fuel have both more than doubled in price since 2020–21 and materials costs have risen by more than 50% (Figure 2-9). Relative changes in the costs of many inputs over the past 2 years exceed those over the previous decade. In contrast, the prices that farmers receive for farm produce have not kept pace with increasing input costs (yet), leading to declining terms of trade. Until increases in the costs of farming inputs flow through to increases For more information on this figure please contact CSIRO on enquiries@csiro.au 020,00040,00060,0002016–172017–182018–192019–202020–212021–222022–23 Jul–DecAgricultural export ($ million) YearChina (excludes SARs and Taiwan)GermanyHong Kong (SAR of China) IndiaIndonesiaJapanKorea, Republic of (South)MalaysiaNew ZealandOtherPhilippinesSaudi ArabiaSingaporeThailandUnited Arab EmiratesUnited States of AmericaVietnam (a) Prices of farm inputs (relative to 2020–21) (b) Prices farms received for farm commodities (relative to 2020–21) Figure 2-9 Farmers’ terms of trade in Australia for (a) input prices and (b) prices received for commodities ABARES (2022) terms of trade indices have been rebased to 2020–21. Indices for 2021–22 are preliminary, and for 2022–23 are forecasts. Axes are on the same scale for both panels to aid comparison. Price volatility for individual commodities can be much greater than for the aggregated categories displayed (e.g. see Figure 5-3). For more information on this figure please contact CSIRO on enquiries@csiro.au 0.00.51.01.52.02.53.02010–11 2011–12 2012–13 2013–14 2014–15 2015–16 2016–17 2017–18 2018–19 2019–20 2020–21 2021–222022–23Price index (relative to 2020–21) MaterialsFuelFertiliserChemicalsMarketingVehicle & machinery maintenanceStructure maintenanceElectricityWaterInsuranceLabour For more information on this figure please contact CSIRO on enquiries@csiro.au 0.00.51.01.52.02.53.02010–11 2011–12 2012–13 2013–14 2014–15 2015–16 2016–17 2017–18 2018–19 2019–20 2020–21 2021–222022–23Price index (relative to 2020–21) GrainsOilseedsPulsesIndustrial cropsHayHorticultureLive cattle for export in the prices of agricultural commodities, balancing farm finances will be more difficult. This includes the irrigated farming options evaluated in this report (using 2021 prices). Until it is clear how recent disruptions to farming terms of trade balance out in the longer term, there will be added risk for investors in the agricultural sector. These risks should be borne in mind when using information in this report, particularly the financial analyses. 2.3 Demography and economy of the Roper catchment This section describes the current social and economic characteristics of the Roper catchment in terms of the demographics of local communities (Section 2.3.1), the current industries and land use (Section 2.3.2), and the existing infrastructure of transport networks, supply chains, utilities and community infrastructure (Section 2.3.3). Together these characteristics describe the built and human resources that would serve as the foundation upon which any new development in the Roper catchment would be built. 2.3.1 Demographics The Roper catchment comprises around half of the Roper Gulf Regional Council local government area together with small parts of a number of other adjacent local government areas, including Katherine Town Council, West Arnhem Regional Council, East Arnhem Regional Council and Victoria Daly Regional Council. At the state/territory level, the catchment includes the majority of the electoral division of Arnhem and a small part of a number of other electoral divisions, including Katherine, Arafura, Mulka and Gwoja. At the federal level, the catchment forms a part of the Division of Lingiari (which encompasses the majority of the NT, excluding the Division of Solomon that covers an area near Darwin). The population density of the Roper catchment is extremely low at one person per 32.6 km2, which is about five times lower than the NT, and 100 times lower than Australia as a whole. The region contains no significant urban areas (population >10,000 people), however, there are a number of small towns and communities within the catchment including Barunga, Beswick, Bulman, Daly Waters, Larrimah, Mataranka (the regional centre), Minyerri and Ngukurr. The only one of these settlements with a population greater than 1000 is Ngukurr (population 1149 as at the 2016 Census). Katherine (population 6303 in 2016) is the closest urban service centre and is located about 100 km north-west of Mataranka, just outside the catchment. The nearest major city and population centre is the NT capital of Darwin (population of Greater Darwin area was 136,828 in 2016), approximately 420 km from Mataranka. The demographic profile of the catchment, based on data from the 2016, 2011 and 2006 censuses is shown in Table 2-1. The ABS reports statistics by defined statistical geographic regions (such as the nested hierarchy of statistical areas), but none of those regions closely approximate the Roper catchment. Instead, data are shown for: (i) Elsey (ABS Statistical Area Level 2 (SA2) region 702051065), being the single region which most closely approximates the catchment boundary (Figure 2-10); and (ii) estimated data based on combining the appropriate portions of a number of ABS regions to best match the actual spatial coverage of the catchment (62.2% of Elsey SA2 region, 19.0% of Gulf SA2 region, plus small proportions (each less than 2%) of the SA2 regions of East Arnhem, Katherine, Victoria River and West Arnhem). Table 2-1 Major demographic indicators for the Roper catchment For more information on this figure or table please contact CSIRO on enquiries@csiro.au †Weighted averages of scores for SA2 regions falling wholly or partially within the catchment boundary. Source: ABS (2006), ABS (2011) and ABS (2016) census data The typical resident of the region is younger, poorer and more likely to identify as Indigenous than the typical resident of the NT and of Australia as a whole. The population is predominantly younger (median age less than 30) than is typical compared to the NT and to the country as a whole (median age more than 30), however, the trend from 2011 to 2016 suggests that the median age is increasing. The population contains a much larger proportion of Indigenous people (more than 70%), compared to the NT (25.5%) and the country overall (less than 3%), and the median household incomes were considerably below the average for the NT and for the country as a whole in 2016. Furthermore, the proportion of households on low incomes (less than $650/week) was far higher, and the proportion on high incomes (more than $3000/week) far lower than the proportion for the NT and for the country as a whole. Figure 2-10 Boundaries of the Australian Bureau of Statistics Statistical Area Level 4 (SA4) and Statistical Area Level 2 (SA2) regions used for demographic data in this Assessment The Roper catchment falls within the 1st decile for each of the Socio-economic Indexes for Areas (SEIFA) metrics (Table 2-2), indicating the region is scoring below 90% of the rest of the country on each of the measures. When considering the various SA2 regions that fall within the catchment boundary, virtually all (West Arnhem, Elsey, Gulf, Victoria River) individually rank within the 1st decile for each of the four measures. Only the Katherine region (less than 2% of which falls within the Roper catchment border) avoids this lowest decile for all measures (ranging from 3rd [Index of Economic Resources (IER)] to 6th decile) while the East Arnhem region (less than 1% of which falls within the Roper catchment border) ranks in the 2nd decile for the Index of Education and Occupation (IEO). Se-R-505_Map_Australia_Roper_tourism_SA2_v3 For more information on this figure or equation please contact CSIRO on enquiries@csiro.au Table 2-2 SEIFA scores of relative socio-economic advantage for the Roper catchment Scores are relativised to a national mean of 1000, with higher scores indicating greater advantage. For more information on this figure or table please contact CSIRO on enquiries@csiro.au †Weighted averages of scores for SA2 regions falling wholly or partially within the catchment boundary. ‡Accessibility and Remoteness Index of Australia Score. 1Based on both the incidence of advantage and disadvantage. 2Based purely on indicators of disadvantage. Source: ABS (2018a) 2.3.2 Current industries and land use Employment The economic structure of the Roper catchment differs substantially from that of the NT and Australia as a whole. The proportion of the adult population (aged 15 and older) within the labour force is far smaller (see participation rates in Table 2-3), indicating that a large proportion of the potential workforce is unable or unwilling to seek work. Furthermore, the unemployment rates are far higher than the NT and national averages (see unemployment rates in Table 2-3), indicating that of those who are willing and able to seek work a larger proportion have been unable to find work. Table 2-3 Key employment data for the Roper catchment For more information on this figure or table please contact CSIRO on enquiries@csiro.au For more information on this figure or table please contact CSIRO on enquiries@csiro.au †Weighted averages of scores for SA2 regions falling wholly or partially within the catchment boundary. Source: ABS (2016), ABS (2011) and ABS (2006) census data There are also noticeable differences in the industries providing the most jobs within the region (Table 2-3). While ‘Education and training’ and ‘Healthcare and social assistance’ are important employers in the region and nationally, ‘Retail trade’, ‘Construction’ and ‘Professional, scientific and technical services’ feature within the top five industries by employment across the nation on average but are far less significant in the Roper catchment. Similar to the NT as a whole, ‘Public administration and safety’ and ‘Other services’ are relatively more important to the employment prospects of workers in this region compared to the national average. However (and of particular relevance to this Assessment), ‘Agriculture, forestry and fishing’ features strongly within the top five industries for the Roper catchment, and furthermore, the importance of the sector has been growing over time when results of the previous censuses are considered. Over the last three censuses (2006, 2011 and 2016), the percentage of employment from the agricultural sector nationally has been reported as 3.1%, 2.5% and 2.5%, respectively, and for the NT, 2.4%, 1.9% and 2.0%, respectively, over the same years. That is, the industry proportion of employment has been small and fairly flat. In contrast, the importance of agricultural employment within the Roper catchment is large and growing, having provided 12.2% of employment in 2006, 13.5% in 2011 and 14.0% in 2016. The structural differences in this region compared to elsewhere can have a significant impact on the regional economic benefits that can result from development projects initiated within the region compared to development projects that may be initiated elsewhere. Land use The Roper catchment covers an area of about 77,400 km2, much of which is conservation and protected land (48.78%) (Figure 2-11). A further 5.03% is classified as water and wetlands, most of which is in several large areas classified as marsh and wetlands (4.22%) throughout the northern parts of the Assessment area. Most of the remaining area (45.74%) is used for grazing natural vegetation. Intensive agriculture and cropping make up a very small portion of the catchment: dryland and irrigated agriculture and intensive animal production together comprise just 0.14% of the land area. The other intensive localised land uses are transport, communications, services, utilities and urban infrastructure (0.31%), and mining (less than 0.01% of the catchment area). Figure 2-11 Land use classification for the Roper catchment Source: Northern Territory Land Use Mapping Project 2016–current, Department of Environment, Parks and Water Security, Northern Territory Government, http://www.ntlis.nt.gov.au/metadata/export_data?type=html&metadata_id=5779F987695AE0FAE050CD9B21447ADC Se-R-514_Map_landuse_Roper_v2 For more information on this figure or equation please contact CSIRO on enquiries@csiro.au Agriculture and fisheries The value of agricultural production for the ABS SA4 region that covers the Roper catchment is given in Table 2-4, together with the estimated proportion of that production that occurs within the catchment. The value of agricultural production in the SA4 region was about 30% higher in 2019–20 than 2015–16, mainly due to increased gross revenue from the beef industry. The value of crops from the region has remained relatively stable over the same period (a 3% decline: Table 2-4). Table 2-4 Value of agricultural production within the wider SA4 region and estimates of the value of agricultural production for the Roper catchment For more information on this figure or table please contact CSIRO on enquiries@csiro.au Estimate for Roper catchment based on SA2 data apportioned using weighted averages of scores for SA2 regions falling wholly or partially within the catchment boundary (ABS, 2017). 2019–20 estimate for Roper catchment based on applying percentage of ‘total crops’ and ‘total livestock’ by SA4 that fall within the catchment based on the ratio of 2019–20 data to the 2015–16 SA4 data (ABS, 2021e). Sources: ABS (2017) Value of agricultural commodities in 2015–16; ABS (2021e) Value of agricultural commodities in 2019–20 The most recent annual survey data from the ABS describing the value of agriculture by different types of industries (2019–20 survey), are only available at a much larger scale (SA4 level; see Figure 2-10) than the Roper catchment, making it difficult to accurately estimate the value of agriculture products within the region. Hence estimates have been made using 2015–16 agricultural census data, which were published by ABS at finer spatial scales (SA2 level), and then adjusting these by the ratio of the SA4 value for 2019–20 relative to 2015–16 (Table 2-4). Agriculture is a major source of employment in the Roper catchment, featuring within the top three industries by employment levels, as shown in Table 2-3. This is very different to the importance of agriculture to employment on a national basis. Agricultural production in the Roper catchment is dominated by extensive grazing of beef cattle, valued at $55.5 million in 2019–20 (Table 2-4). The first cattle were brought to Elsey Station, near Mataranka, in 1882 (Gleeson and Richards, 1985). The life of European settlers at the Elsey Station and the surrounding region have become well known through the account given in the autobiographical novel We of the never-never (Gunn, 1908). Subsequent attempts at expanding livestock production along the Roper River initially met with mixed success. The Mataranka Horse and Sheep Experimental Station was established in 1913 as part of plans for closer settlement in the upper Roper catchment, which envisioned agricultural-led growth in the region by which Mataranka would replace Darwin as the NT capital (Gleeson and Richards, 1985). However, sheep production proved unsuitable for the region because of the harsh climate, rough terrain and blowfly incidence, so was abandoned in favour of beef cattle, which proved more successful and persists as the dominant agricultural activity in the catchment to this day. Present-day cattle grazing occurs on dryland native and naturalised pastures where productivity is constrained by the variable climate and low-fertility soils. These constraints have shaped the types of beef production systems currently in the Roper catchment, which typically have low stocking rates and target live exports to South-East Asia through Darwin Port. Cropping in the Roper catchment has an annual value of only about $18 million (Table 2-4), mainly from melons and mangoes grown near Mataranka. Mataranka complements Katherine as a mango growing area since the climate is slightly cooler, which means that flowering and fruit ripening occur later and thus extends the overall duration of the harvest season for the region. Despite more than a century of attempts at establishing crop industries in the NT, there is still very little irrigated or dryland cropping in the Roper catchment. After the agricultural experiments around the time of the First World War, the Second World War prompted another wave of interest in facilitating northern agricultural development, which included a set of agricultural experimental stations. In 1942, approval was given to establish army farms at Katherine and Mataranka with the aim of more efficiently supplying the fruit and vegetables needed to maintain the nutrition of troops. The army experimental farm at Katherine was initially established to test what fruit and vegetables were suitable for the area. After the war this became the Katherine Experimental Station, where a wider range of crops were explored (run by the Commonwealth until it was handed over to the Northern Territory Government in the 1980s). Several crops, such as peanuts in the 1950s, initially proved to be agronomically suitable for the local environment but were unable to be established as competitive local industries, partly because of difficulties with market access and high transport costs. There is currently no active aquaculture in the Roper catchment. A freehold area of approximately 12,000 ha about 25 km upstream from the mouth of the Roper River was developed by Carpentaria Aquafarm Pty Ltd with about 40 ha of grow-out ponds in the 1980s. While the ponds remain, the business stopped operating in the early 1990s (Australasia Aquaculture, 2005). Offshore, the Roper River drains into one of the most valuable fisheries in the country. The Northern Prawn Fishery (NPF) spans the northern Australian coast between Cape Londonderry in WA to Cape York in Queensland (Figure 2-12), with most of the catch being landed at the ports of Darwin, Karumba and Cairns. Over the 10-year period from 2010–11 to 2019–20, the annual value of the catch from the NPF has varied between $65 million and $124 million, with a mean of $100 million (Steven et al., 2021). The Roper catchment flows into the South Groote NPF region (Figure 2-12), one of the smallest regions by annual prawn catch. Like many tropical fisheries, the target species exhibit an inshore–offshore larval life cycle and are dependent on inshore habitats, including estuaries, during the postlarval and juvenile phases (Vance et al., 1998). Monsoon-driven freshwater flood flows cue juvenile prawns to emigrate from estuaries to the fishing grounds and flood magnitude explains 30 to 70% of annual catch variation, depending on catchment region (Buckworth et al., 2014; Vance et al., 2003). Fishing activity for banana prawns (Fenneropenaeus spp.) and tiger prawns (Penaeus spp.), which constitute 80% of the catch, is limited to two seasons: a shorter banana prawn season from April to June, and a longer tiger prawn season from August to November. The specific dates of each season are adjusted depending on catch rates. Banana prawns generally form the majority of the annual prawn catch by volume. Key target and by-product species are detailed by Woodhams et al. (2011). The catch is often frozen on-board and sold in domestic and export markets. Figure 2-12 Map of regions in the Northern Prawn Fishery The regions in alphabetical order are Arnhem-Wessels (AW), Cobourg-Melville (CM), Fog Bay (FB), Joseph-Bonaparte Gulf (JB), Karumba (KA), Mitchell (ML), North Groote (NG), South Groote (SG), Vanderlins (VL), Weipa (WA), West- Mornington (WM). Source: Dambacher et al. (2015) The NPF is managed by the Australian Government (via the Australian Fisheries Management Authority) through input controls, such as gear restrictions (number of boats and nets, length of nets) and restricted entry. Initially comprising over 200 vessels in the late 1960s, the number of vessels in the NPF has reduced to 52 trawlers and 19 licensed operators after management initiatives including effort reductions and vessel buy-back programs (Dichmont et al., 2008). Given recent efforts to alleviate fishing pressure in the NPF, there is little opportunity for further expansion of the industry. However, any development of water resources in the Roper catchment would need to consider the downstream impacts on prawn breeding grounds and the NPF. 2.3.3 Current infrastructure Transport The most significant road in the Roper catchment is the Stuart Highway, which runs from Darwin to Port Augusta in SA, about 300 km north of Adelaide. The Stuart Highway is formally designated Route A1 from Darwin to Daly Waters and Route A87 from Daly Waters to Port Augusta. The road passes through Mataranka and Larrimah at the top of the catchment in the west and is the main link northwards to Katherine and Darwin and southwards to the south-eastern states via Alice Springs. Figure 2-13 shows the network of roads within the Roper catchment together with rankings according to the types of road surface. All road network information in this section is Se-R-501_Portrait_map_Australia_NPF_regions_v3 For more information on this figure or equation please contact CSIRO on enquiries@csiro.au from spatial data layers in the Transport Network Strategic Investment Tool (TraNSIT: Higgins et al., 2015). Aside from the Stuart Highway, the Roper catchment is served by a sparse network of mainly unsealed roads. The most important roads branching off the Stuart Highway into the catchment are the Roper Highway (Route B20), linking Ngukurr near the mouth of the river (in the east) to Mataranka at the top of the catchment (in the west), and the Central Arnhem Road (Route C24), which runs across the north of the catchment from the Stuart Highway through Bulman/Gulin Gulin. Figure 2-14 shows the heavy vehicle access restrictions for roads within the Roper catchment, as determined by the National Heavy Vehicle Regulator. All non-residential roads in the study area permit Type 2 road trains, which are vehicles up to 53 m in length, typically a prime mover pulling three 40-foot trailers (Figure 2-15). Despite the poorer road conditions of many of the local unsealed roads, large (Type 2) road trains are permitted due to minimal safety issues from low traffic volumes and minimal road infrastructure restrictions (e.g. bridge limits, intersection turning safety). Drivers would regularly use smaller vehicle configurations on the minor roads due to the difficult terrain and single lane access, particularly during wet conditions. Figure 2-16 shows the speed limits for the road network within the Roper catchment. These speed limits are usually higher than the average speed achieved for freight vehicles, particularly on unsealed Rank 3 roads. Heavy vehicles using such unsealed roads would usually achieve average speeds of no more than 60 km/hour, and often as low as 20 km/hour when transporting livestock. A good quality standard gauge rail line passes through the western edge of the Roper catchment. This provides freight access to Darwin Port (East Arm Wharf) to the north, and to major southern markets via Alice Springs. The rail line is primarily used for bulk commodity transport (mostly minerals) to Darwin Port. There are no branch lines in the Roper catchment so goods would have to be transported to and from loading points by roads. Figure 2-13 Road rankings and conditions for the Roper catchment Rank 1 = well-maintained highways or other major roads, usually sealed; Rank 2 = secondary ‘state’ roads; Rank 3 = minor routes, usually unsealed local roads. The ‘Rank 1’ road is the Stuart Highway, which runs from Darwin (in the north) to Port August (in South Australia, about 300 km north of Adelaide). Se-R-508_Roper_TraNSIT_road rankings_v2 For more information on this figure or equation please contact CSIRO on enquiries@csiro.au Figure 2-14 Type 2 vehicle access across the Roper catchment Truck classes referred to in the legend are illustrated in Figure 2-15. For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au Figure 2-15 Common configurations of heavy freight vehicles used for transporting agricultural goods in Australia For more information on this figure, please contact CSIRO on enquiries@csiro.au Figure 2-16 Road speed restrictions for the Roper catchment Supply chains and processing Table 2-5 provides volumes of agricultural commodities transported into and out of the Roper catchment and Figure 2-17 shows the location of existing agricultural enterprises in the catchment. As previously noted, agricultural production in the Roper catchment is currently dominated by horticulture and beef, particularly melons and live cattle export. This is also For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au reflected in annual volumes of commodities transported into and out of the catchment through the road network. About 31,000 t of melons were transported out of the catchment, predominantly to domestic distribution centres in 2021, according to TraNSIT records of truck movements. There are also large volumes of freight transporting cattle into (~13,000 head in 2021) and out of (~36,000 head combined) the Roper catchment, mainly via the Stuart Highway. Live export of cattle via Darwin Port accounts for the majority of cattle movements, but there are also substantial transfers of cattle between properties and smaller volumes directed to domestic markets via abattoirs and feedlots. Table 2-5 Overview of commodities annually transported into and out of the Roper catchment Indicative transport costs are means for each commodity and include differences in distances between source and destinations. For more information on this figure or table please contact CSIRO on enquiries@csiro.au Source: 2021 data from TraNSIT (Higgins et al., 2015) There are currently no processing facilities for agricultural produce within the Roper catchment, but there are (or soon will be) facilities nearby that could support producers in the catchment. The closest meatworks was run by AACo (Australian Agricultural Company) at Livingstone, about 40 km south of Darwin but has not been operational since 2018. When operating, it was accessible by large (Type 2) road trains along the entire route from the Roper catchment. The first cotton gin in the NT commenced construction in 2021 (scheduled for completion in 2023) about 30 km north of Katherine as part of growing interest in establishing a new cotton industry in the region. The closest port for bulk export of agricultural produce from the Roper catchment is in Darwin. Darwin Port, operated by Landbridge Group, handles about 20,000 to 30,000 TEU each year split roughly evenly between imports and exports. The main exports are dry bulk commodities (mainly manganese) and livestock, but there are also annual exports of about 100 TEU of refrigerated containers. Exports of new bulk agricultural produce would require construction of a new storage facility. Figure 2-17 Agricultural enterprises in the Roper catchment and amount of annual trucking to/from them Smaller horticultural enterprises, mainly melon and mango farms near Mataranka, are too small to show on the map at this scale. The thickness of purple lines indicates how much traffic (as number of tailers) there is on regional roads connecting local enterprises. For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au Energy The Darwin-Katherine Interconnected System (DKIS) is the largest electricity grid in the NT (Figure 2-18). The DKIS is electrically isolated from other grids in Australia (but see below for how electricity and natural gas transmission systems are interconnected). The DKIS transmission network reaches the western edge of the Roper catchment, passing through Mataranka and reaching as far south as Larrimah. A small branch off this main transmission line serves Barunga (Bamyili) and Beswick, and a distribution line links Jilkminggan to nearby Mataranka. Generation on the DKIS is primarily by gas turbine power stations at Channel Island (279 MW), Weddell (129 MW), Katherine (36.5 MW) and Pine Creek (26.9 MW, privately owned). The closest generator to the Roper catchment is at Katherine, where there is an additional back-up diesel generator. Figure 2-18 Electricity generation and transmission network and natural gas pipelines in the Roper catchment Distribution networks are not shown, but communities marked with red lightning symbols are connected to nearby generation or transmission sources of electricity. The Amadeus Gas Pipeline runs north–south (bi-directional) through Katherine; the McArthur River Pipeline branches off eastwards from Daly Waters to the McArthur River zinc-lead mine. Se-R-507_Roper_energy_generation_distribution_v2 For more information on this figure please contact CSIRO on enquiries@csiro.au Most of the Roper catchment, however, is too remote to be covered by the DKIS. The three largest off-grid remote communities rely on hybrid systems powered by diesel generators supplemented with solar: Ngukurr (400 kW solar system), Minyerri (275 kW) and Bulman (100 kW). Distribution lines link nearby smaller settlements to these off-grid sources of electricity: Rittarangu is connected to Ngukurr and Weemol is connected to Bulman. Historically, gas pipelines have been a cheaper way of transporting energy than electrical transmission lines (DeSantis et al., 2021; GPA, 2021). So, a network of natural gas pipelines has been a cost-effective way of linking energy supplies across the NT by connecting sources of gas to electricity generators and other demand centres. The Amadeus Gas Pipeline is a bi-directional pipeline running from the gas fields of the Amadeus Basin near Alice Springs in the south northwards through the western edge of the Roper catchment (near the Stuart Highway) towards Darwin. The McArthur River Pipeline connects to the Amadeus Gas Pipeline at Daly Waters and runs across the southern edge of the Roper catchment to the generator at the McArthur River zinc-lead mine. The Northern Gas Pipeline, which runs 622 km between Tennant Creek and Mount Isa in Queensland (south of the Roper catchment), provides a connection between the energy systems of the NT and the eastern states. Water There are no major dams or water transmission pipelines in the Roper catchment. Urban water for domestic consumption therefore depends mainly on treated groundwater (from bores) as the preferred source for larger settlements. Surface water is also used in some cases: river pumping supplements the water supply for Ngukurr, while outstations may also use river water, springs and lagoons (Zaar, 2009). Surface water entitlements Licenced surface water entitlements are sparse across the Roper catchment which occupies the eastern part of the Daly Roper Beetaloo Water Control District (DRBWCD). The Northern Territory government prepares Water Allocation Plans to sustainably manage and allocate water resources, the Georgina Wiso Water Allocation Plan 2023 - 2031 (GWWAP) is currently in place with the Mataranka Tindall Limestone Aquifer Water Allocation Plan (MTLAWAP) (Figure 2-19) and the Surface Water Take – Wet Season Flows Policy currently being developed. There have been four licences granted for a combination of public water supplies and cultural and industrial uses, all of which fall outside of proposed water allocation planning areas. The largest entitlement of 80 ML/year is for public water supply at Barunga (Figure 2-19). The water is sourced from Beswick Creek which is fed by water discharging from Bamyili Spring. The next largest entitlement of 26 ML/year is from the Roper River for industrial purposes. Minor entitlements of <8 ML/year from the Roper River have been granted for cultural and industrial purposes. Groundwater entitlements Licenced groundwater entitlements have been granted across the central and south-western parts of the Roper catchment, most prominently around Mataranka. Most of the groundwater entitlements are for water sourced from the regional-scale Tindall Limestone Aquifer (TLA) in the south-west of the catchment. These licenced entitlements all occur within the proposed MTLAWAP with the exception of one licence for public water supply at Daly Waters (Figure 2-19). Only very minor entitlements are sourced elsewhere from localised fractured and weathered rock aquifers hosted in the Roper Group. The purpose of the majority of these entitlements is for irrigated agriculture (31.1 GL/year). Licenced entitlements totalling 354 ML/year have also been granted for public water supplies at Mataranka, Jilkminggan, Larrimah and Daly Waters. The remainder of entitlements (309 ML/year) have been granted for industrial purposes including tourist accommodation, and council and cement operations. Other licenced entitlements come from aquifers hosting intermediate to local-scale groundwater systems inside the DRBWCD but outside of the proposed MTLAWAP and GWWAP areas (Figure 2-19). Figure 2-19 Location, type and volume of annual licenced surface water and groundwater entitlements Data sources: Daly Roper Beetaloo Water Control District sourced from NT Department of Environment, Parks and Water Security (2019); Water Allocation Plan areas sourced from NT Department of Environment, Parks and Water Security (2018) Water allocation map \\FS1-CBR.nexus.csiro.au\{lw-rowra}\work\11_Groundwater\3_Roper\1_GIS\1_Map_docs\Gr-R-509_water_allocation_CR_NI_v04_NotInProgress.mxd For more information on this figure please contact CSIRO on enquiries@csiro.au Licences have been granted for public water supply at Beswick (190 ML/year), Barunga (280 ML/year) and Minyerri (150 ML/year). Groundwater for Barunga is sourced from localised aquifers hosted in Cretaceous sandstone. Groundwater for Beswick is sourced from the intermediate-scale dolostone aquifer hosted in the Dook Creek Formation. Groundwater for Minyerri is sourced from a localised aquifer hosted in the Bessie Creek Sandstone. Community infrastructure The availability of community services and facilities in remote areas can play an important role in attracting or deterring people from living in those areas. Development of remote areas therefore also needs to consider whether housing, education and healthcare are sufficient to support the anticipated growth in population and demand, or to what extent these would need to be expanded. There are no hospitals in the Roper catchment but, like most remote parts of Australia, the area is serviced by a primary health network (PHN). Australia is divided into 31 PHNs and one of these covers the whole of the NT. General practitioners and allied health professionals provide most primary healthcare in Darwin and the regional centres within the Northern Territory PHN, while smaller communities are supported by remote health clinics (NT PHN, 2020). The Roper catchment falls within the Katherine Health Service District (HSD) (also known as the Big Rivers Region) of the Northern Territory PHN where the Sunrise Health Service Aboriginal Corporation and Katherine West Health Board provide remote health services. PHNs work closely with local hospital networks, and for the Katherine/Big Rivers Region the associated hospital is Katherine Hospital, which is located just outside the western border of the Roper catchment. This hospital has 60 beds and provides emergency services, surgical and medical care, paediatrics and obstetrics (NT PHN, 2020). A network of eight schools cover the small communities throughout the Roper catchment. A total of 807 FTE students are enrolled in these schools with 77.2 teachers (FTE) in 2021 (Table 2-6). The largest school in the catchment is at Ngukurr. There are a further six schools in Katherine, just outside the Roper catchment and about 100 km north-west of Mataranka, and there is also a school of the air in Katherine that serves 167.5 students (FTE) in the region. At the time of the 2016 census, only about 11% of private dwellings were unoccupied, representing a similar proportion to the national average although slightly lower than the NT (Table 2-7). This suggests that the current pool of housing may have some capacity to absorb small future increases in population. Table 2-6 Schools servicing the Roper catchment For more information on this figure or table please contact CSIRO on enquiries@csiro.au FTE = full time equivalent Source: ACARA (2022) (data presented with permission) Table 2-7 Number and percentage of unoccupied dwellings and population for the Roper catchment For more information on this figure or table please contact CSIRO on enquiries@csiro.au †Weighted averages of scores for SA2 regions falling wholly or partially within the catchment boundary. Source: ABS (2016) census data Part II Agricultural development options Part II analyses the farm-scale performance of potential irrigated agricultural development options and covers the agronomic principles involved in implementing them. Chapter 3 provides background information on tropical agronomy including the environmental factors affecting crop performance (climate, soils, land suitability, water resources), the range of potential crop options and crop management considerations. Chapter 4 describes the approach used for crop modelling and other quantitative analyses of a set of 19 possible crop options for the Roper catchment and the methods used to estimate their potential performance (in terms of yields, water use and farm gross margins). Chapter 5 presents the results of the farm-scale analyses, uses narrative risk analyses to illustrate opportunities and challenges for establishing viable new farms, and interprets the practical implications of the information provided in Part II for the types of cropping systems that could be fine-tuned to Roper catchment environments. Part III analyses the scheme-scale viability of irrigated development options and economic considerations beyond the farm gate that would be required for those developments to succeed. 3 Biophysical factors affecting agricultural performance 3.1 Climate Climate is a key factor in determining the productivity of agricultural and pastoral production systems. While temperature, radiation and rainfall influence the rate of crop growth, extreme weather events such as floods, hail, drought or heat waves have additional episodic, and sometimes catastrophic, effects on agricultural production systems. Crop water use is determined by the interaction between atmospheric evaporative demand (controlled by air temperature, vapour pressure deficit (VPD) and windspeed), crop canopy and root system capacity, and the amount of water stored in the soil. The climate of the Roper catchment is discussed in detail in the companion technical report on climate (McJannet et al., 2023), and briefly summarised below (Figure 3-1; Figure 3-2). The Roper catchment has a hot and arid climate that is highly seasonal with an extended dry season between May and October. The Roper catchment receives, on average, 792 mm of rain per year, 96% of which falls during the summer wet season (1 November to 30 April). Mean daily temperatures and potential evaporation are high relative to other parts of Australia. On average, annual potential evaporation is approximately 1900 mm; however, the annual net evaporative loss (annual evaporation minus rainfall) ranges between 350 and 1770 mm. Overall, the climate of the Roper catchment generally suits the growing of a wide range of crops, though in most years rainfall would need to be supplemented with irrigation. The variation in rainfall from one year to the next is moderate compared to elsewhere in northern Australia yet is high compared to other parts of the world with similar mean annual rainfall. The length of consecutive dry years is not unusual in the Roper catchment and the intensity of the dry years is similar to many centres in the Murray–Darling Basin and east coast of Australia. Since 1969–70, one tropical cyclone in 40% of cyclone seasons and two tropical cyclones in 8% of seasons have affected parts of the Roper catchment. Future climate projections for the Roper catchment suggest little change in rainfall: approximately a third of the global climate models (GCMs) project an increase in mean annual rainfall by more than 5%, a fifth project a decrease in mean annual rainfall by more than 5% and about half indicate ‘little change’. Each of the key climate parameters that control plant growth and crop productivity are discussed in turn under the subheadings below, although it should be noted that they are interrelated and never act in isolation. Throughout this section, the tropical monsoonal climate of the Roper catchment is contrasted against that of more temperate southern agricultural areas (using Griffith, NSW, as an example), to highlight how different cropping systems in northern Australia are to those where most of the country’s farming expertise resides. 3.1.1 Rainfall While rainfall in the Roper catchment largely occurs during the summer wet season, variability in rainfall is high, with long-term rainfall totals over a 14-day period varying by over 140 mm between seasons at Mataranka (Figure 3-1). Irrigation can be used to supplement rainfall in the wet season when below average rainfall is experienced, and also facilitate cropping during the dry season (winter months) when sufficient irrigation water is available. (a) Rainfall, and number of days per fortnight daily rainfall exceeds 5 mm (b) Maximum temperature, and number of days per fortnight minimum temperatures are above 35 °C and 40 °C thresholds (c) Minimum temperature, and number of days per fortnight minimum temperatures are below 10 °C and 5 °C thresholds Figure 3-1 Long-term fortnightly climate variation in (a) rainfall, (b) maximum and (c) minimum temperatures for the historical climate (1890 to 2015) at Mataranka Whiskers on box plots show 10% and 90% exceedance values. Source: Data sourced from SILO website https://www.longpaddock.qld.gov.au/silo/ (Jeffrey et al., 2001) For more information on this figure please contact CSIRO on enquiries@csiro.au For more information on this figure please contact CSIRO on enquiries@csiro.au For more information on this figure please contact CSIRO on enquiries@csiro.au (a) Solar radiation, and number of days per fortnight radiation is below 20 and 15 MJ/m2/day thresholds (b) Relative humidity (RH), and number of days per fortnight RH is below 40% while temperatures exceed 35 °C (c) Vapour pressure deficit, and number of days per fortnight RH is above 40% while temperatures exceed 35 °C Figure 3-2 Long-term fortnightly climate variation in (a) solar radiation, (b) relative humidity (RH), and (c) vapour pressure deficit (VPD) under the historical climate (1890 to 2015) at Mataranka Whiskers on box plots show 10% and 90% exceedance values. Source: Data sourced from SILO website https://www.longpaddock.qld.gov.au/silo/ (Jeffrey et al., 2001) For more information on this figure please contact CSIRO on enquiries@csiro.au For more information on this figure please contact CSIRO on enquiries@csiro.au For more information on this figure please contact CSIRO on enquiries@csiro.au Wet-season rainfall is associated with the monsoon trough, tropical lows or intense storms, which also have implications for crop growth and management. The former can reduce crop yield potential through warm night temperatures and lower solar radiation (due to prolonged cloud cover) as shown for the wet season (December to March) in the Mataranka example (Figure 3-2). Intense storm events on the other hand produce strong winds, which have the potential to physically damage crops. Excessive rainfall can also complicate the management of agricultural land, for example in delaying farm operations, or the loss of soil nutrients such as nitrogen through leaching, runoff and denitrification. Waterlogging can also reduce crop growth on clay soils and reduce machine access to fields on heavier soils found in floodplains of the Roper catchment. The mean annual rainfall, averaged over the Roper catchment is 792 mm (McJannet et al., 2023). Annual rainfall is highest in the northern part of the catchment that receives more active monsoon episodes during the wet season. Rainfall is lowest in the most southerly part the catchment. Mean annual rainfall is about 900 mm at Bulman in the north-east, 770 mm at Ngukurr in the east, and 780 mm at Larrimah in the south-west. The Roper catchment is relatively flat, and consequently there is no noticeable topographic influence on climate parameters such as rainfall or temperature. The highest monthly rainfall totals typically occur during January, February and March. While daily wet-season rainfall is strongly correlated with the Australian Monsoon Index, seasonal rainfall variability experienced in the Roper catchment is strongly influenced by Indonesian sea surface temperatures and El Niño-Southern Oscillation indices (Rogers and Beringer, 2017). Year- to-year variation in the timing and amount of rainfall affects the amount of water available for irrigation due to fluctuations in stream flows and the consequent opportunities for water harvesting. Irrigated cropping options need to consider the timing and amount of water available. 3.1.2 Evaporation Evaporation is the ‘drying’ process by which water is lost from open water, plants and soils to the atmosphere. It has become common usage to also refer to this as evapotranspiration. Transpiration is ‘that part of the total evaporation that enters the atmosphere from the soil through the plants’ (Shuttleworth, 1993). The rate and amount of water evaporated from the soil surface is influenced by surface shading by the crop canopy or surface stubble residues and soil water in the top soil layers. Crop transpiration is the product of not only solar radiation but also air temperature, air humidity and wind that affect the vapour pressure gradient between plant leaf stomata and the atmosphere (see Section 3.1.5), along with crop factors such as the height and leaf area of the crop, the extent of the root system, and the amount of water in the soil. Evaporation losses from water storages (dams and ringtanks) and delivery systems (diversion streams and channels) need to be considered in determining the overall water availability to meet crop water demand. The mean annual potential evaporation (PE) for the Roper catchment is 1883 mm (McJannet et al., 2023). Seasonal and inter-annual variation in PE is illustrated for Mataranka (Figure 3-3). The mean annual rainfall deficit (mean annual net evaporative water loss from potential open storages) at Mataranka is about 1065 mm (McJannet et al., 2023). (a) Monthly potential evaporation (b) Annual potential evaporation Figure 3-3 Historical potential evaporation (PE) in the Roper catchment at Mataranka for (a) monthly PE (range is the 20th to 80th percentile monthly PE) and (b) time series of annual PE (line is the 10-year running mean) Source: Data sourced from SILO website https://www.longpaddock.qld.gov.au/silo/ (Jeffrey et al., 2001) 3.1.3 Radiation Shortwave radiation from sunlight influences plant growth through the process of photosynthesis converting atmospheric carbon dioxide into carbohydrates within the plant. The potential amount of solar radiation intercepted by the crop is determined by latitude (which influences day length), time of year, cloudiness, atmospheric transparency and scattering, and crop canopy characteristics for the growth stage. Solar radiation during the summer months (December to March) is supressed in the Roper catchment due to increased cloud cover associated with the monsoon trough over northern Australia (Figure 3-4a). While long-term mean radiation during the wet season is reduced to less than 18 MJ/m2/day from mid-January, radiation levels during the dry season remain high compared to agricultural regions in southern Australia. Figure 3-4a demonstrates how differences in latitude between the Roper catchment (tropical latitude, about 14° S) and Griffith in southern NSW (subtropical latitude 34.3° S) affect monthly solar radiation. For the Mataranka example, solar radiation from April to October remained above 18 MJ/m2/day, much higher than the radiation experienced during the same period at Griffith (Figure 3-4a), indicative of the subtropical and temperate patterns of radiation in the southern parts of Australia where most crop production occurs. Farmers in the Roper catchment can maximise crop yields by successfully managing the time of sowing and growing season length to maximise peak radiation intercepted by the crop (March–April and August–September) while avoiding the temperature extremes experienced in October and November. For more information on this figure please contact CSIRO on enquiries@csiro.au 050100150200250300JFMAMJJASONDPotential evapration (mm) RangeMedianMean For more information on this figure please contact CSIRO on enquiries@csiro.au 1500175020002250250027503000196519751985199520052015Annual evaporation (mm) Year (a) Mean daily solar radiation (b) Mean daily vapour pressure deficit Figure 3-4 Monthly mean daily (a) solar radiation and (b) vapour pressure deficit for three locations in the Roper catchment (Bulman, Mataranka and Ngukurr: latitude 13.7–14.9° S) and Griffith (subtropical: latitude 34.3° S) 3.1.4 Temperature Temperature influences all plant physiological processes and plays a role in determining the length of crop development phases. The optimal temperature for plant growth and therefore maximum individual crop productivity varies between crop species. Temperature extremes at sensitive phenological stages can adversely affect crop productivity. Plant species have differing temperature thresholds for optimum growth and differing responses during periods of extreme high or low temperature. High plant canopy temperatures reduce the efficiency of photosynthesis via increased respiration (particularly at night) and photorespiration, the latter affecting C3 crops (e.g. rice, soybean, mungbean, sesame (Sesamum indicum), cotton, forage legumes). For northern Australia, the highest temperatures generally occur during the months of October to December as shown for the Roper catchment (Figure 3-5), where the long-term mean daily maximum temperature can exceed 36 °C and night temperature (i.e. minima) exceed 24 °C. High temperature effects (both day and night) on plant photosynthesis are exacerbated by high humidity and low solar radiation. (a) Mean daily maximum temperature (b) Mean daily minimum temperature Figure 3-5 Monthly mean daily (a) maximum and (b) minimum daily temperatures for three locations in the Roper catchment (Bulman, Mataranka, and Ngukurr: latitude 13.7—14.9° S) and Griffith (subtropical: latitude 34.3° S) For more information on this figure please contact CSIRO on enquiries@csiro.au 51015202530JanFebMarAprMayJunJulAugSepOctNovDecMJ/m2/dayBulmanMatarankaNgukurrGriffith For more information on this figure please contact CSIRO on enquiries@csiro.au 05101520253035JanFebMarAprMayJunJulAugSepOctNovDechPaBulmanMatarankaNgukurrGriffith For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au 10152025303540JanFebMarAprMayJunJulAugSepOctNovDec°CBulmanMatarankaNgukurrGriffith For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au 051015202530JanFebMarAprMayJunJulAugSepOctNovDec°CBulmanMatarankaNgukurrGriffith Figure 3-5 shows that while the amplitude of annual mean monthly temperatures experienced in the Roper catchment are smaller than those further south in Griffith, the differences in mean monthly maximum temperatures between the two locations are greatest between August and October. The onset of the wet season (December to March) generally coincides with periods of hot temperatures (slightly cooler than the pre-monsoonal build-up), lower solar radiation and higher humidity/lower VPD (Figure 3-4). When high temperatures occur at times that crops are growing rapidly and soil water profiles are depleted, the cooling effects of transpiration are diminished and crop canopy temperatures rise. Under such stress conditions, photosynthesis is reduced and plant tissue damage can occur. Collectively these physiological effects are often referred to as ‘water stress’. Prior to the onset of summer rains, low soil water, higher air temperatures and high solar radiation combine to heat soils, particularly those low in vegetative cover. High soil temperatures can reduce seedling emergence and crop establishment. For an irrigated crop, higher temperatures induce higher evaporative demand and increase evapotranspiration, resulting in a higher irrigation requirement to achieve maximum production. 3.1.5 Vapour pressure deficit Relative humidity (RH), the amount of water vapour in the air as a proportion of the potential amount of water the air can hold for a given air temperature and altitude, is well understood. But VPD is a more accurate measurement of how plants respond to changes in humidity and temperature. VPD is the difference between the current partial pressure of water vapour in the atmosphere and the amount of water vapour that could be held at saturation (at 100% RH at the current temperature). At higher VPDs, the vapour pressure gradient between plants and the atmosphere is stronger, which drives higher rates of transpiration and water use by crops (Rashed, 2016). It is the combination of VPD and high air temperature that reduces the ability of plants to transpire and regulate temperature. High temperatures and low VPD (particularly at night) are as detrimental to canopy temperature regulation as high temperatures and high VPD. During periods of high temperature, supplementary irrigation may assist in reducing plant stress but is of limited value during periods of high VPD. The long-term mean RH for Mataranka fluctuates between 25% in the dry season and slightly above 50% in the wet season (Figure 3-2). The occurrence of periods of high humidity also influences the development of many plant diseases. Irrigated crops can be exposed to high levels of humidity that can favour disease infection during the wet season and during cooler nights in the dry season. Lower RH in the spring build-up to the monsoonal season (September to November) correlates with an increase in VPD and higher maximum and minimum temperatures that would require additional irrigation resources to meet higher surface evaporation and transpiration loss (Figure 3-2; Figure 3-4b). 3.1.6 Windspeed Wind can be both beneficial and harmful to crop productivity. It can aid the process of pollination and is particularly important in the development of fruit and seed from wind-pollinated flowers. However, strong winds can cause excessive water loss through transpiration which can cause crops and trees to wilt. In strong winds, tall crops, particularly crops that are covered with water from rain or spray irrigations, may lodge (fall over), leading to lower photosynthetic potential and making crops more difficult to harvest. Combined with other factors, winds can be particularly harmful; for example, wind-blown sand particles can damage vegetative surfaces. Destructive winds and potential flooding associated with tropical cyclones pose a significant threat, particularly to tree crops. 3.1.7 Cyclones Cyclones are a significant risk to any above-ground infrastructure (sheds, irrigation pivots etc.) and to tree crops with long life cycles. Tropical cyclones and tropical lows also contribute a considerable proportion of total annual rainfall in the Roper catchment, but the actual amount is highly variable from one year to the next (see companion technical report on climate (McJannet et al., 2023)). There is a reasonably high risk of cyclones in the Roper catchment from November to April, predominantly in the eastern coastal part of the district and particularly in La Niña years (Figure 3-6). For the 53 tropical cyclone seasons from 1969–70 to 2021–22, 53% of seasons experienced no tropical cyclones, 40% experienced one tropical cyclone, and 8% experienced two cyclones in part of the Roper catchment (McJannet et al., 2023). Figure 3-6 Mean annual number of tropical cyclones in Australian for (a) El Niño years and (b) La Niña years Source: Bureau of Meteorology cyclone maps http://www.bom.gov.au/climate/maps/averages/tropical-cyclones/. 3.1.8 Future climate Australia’s climate has been progressively warming since the early 1900s (CSIRO and BoM, 2015). Mean overnight minimum temperatures have increased by 1.1 °C and mean daily maximum temperatures by 0.8 °C. Northern Australia, including the Roper catchment, has experienced a mean temperature increase of between 0.5 °C and 1.0 °C since 1910. Temperatures are expected to increase in the future, resulting in an increased number of extremely hot days. While winter rainfall has declined by 19% in the south-west of the country, parts of northern Australia have experienced above average increases in rainfall since the 1970s. Future climate projections of rainfall for northern Australia do not show a clear trend, with some models suggesting decreases and others projecting increases in rainfall. An analysis of 21 downscaled GCMs for the Roper catchment gave a consensus result that mean annual rainfall in the Roper catchment could change For more information on this figure please contact CSIRO on enquiries@csiro.au by less than 5% under a 2.2 °C warming scenario, with slightly more models projecting >5% wetting (29%) than >5% drying (19%) (McJannet et al., 2023). The same analysis projected mean annual PE to increases by about 3 to 10% in the Roper catchments. In addition to changes in temperature, evaporation and rainfall as a consequence of increased greenhouse gas emissions, agricultural production will also be affected directly by elevated atmospheric CO2 concentrations. The direct impacts of elevated atmospheric CO2 concentrations on crop physiological processes of photosynthesis and leaf stomatal conductance are well- documented from Free Air CO2 Enrichment (FACE) experiments (e.g. Hendrey et al., 1993; Tubiello et al., 2007). In the absence of temperature stress, elevated CO2 improves water use efficiency of crops and grasses by regulating a stomatal closure response in the plant to increase intercellular CO2 (Parry et al., 2004) and by the passive effects of increasing CO2 relative to vapour gradients between substomatal spaces and the atmosphere. One anomaly of projected increases in mean temperature associated with elevated greenhouse gases is temperature-induced acceleration of crop development as a result of an increase in the rate of thermal time accumulation. While overall crop yields may decrease in response to increased daily temperature, the rate of decline may be mitigated due to a shortening of the vegetative and grain-filling periods, which result in phenological development and maturation occurring earlier and possibly within a more favourable climate period. The timing and use of supplementary irrigation will also have a role in reducing the severity of temperature-induced stress in crops. 3.2 Soils and land suitability 3.2.1 Soils Soils play a vital role in enabling crop production by providing a medium for physical support, nutrient supply and cycling (including associated soil organic matter and soil biota), and water storage and supply. The companion technical report on digital soil mapping and land suitability (Thomas et al., 2022) classified soils of the Roper catchment into soil generic groups (SGGs) (Figure 3-7; Table 3-1). The ten SGG groupings provide a means of aggregating soils with broadly similar properties and management considerations. Each of the SGGs has a different potential for agriculture, some with almost no potential, such as the shallow and/or rocky soils (e.g. SGG 7, Table 3-1) and some with moderate to high potential (e.g. SGG 9, Table 3-1) depending on other factors such as flooding and the amount of salt in the profile. Figure 3-7 The soil generic groups (SGGs) of the Roper catchment produced by digital soil mapping The inset map shows the data reliability, which for SGG mapping is based on the confusion index as described in the companion technical report on land suitability (Thomas et al., 2022). Table 3-1 Soil generic groups (SGGs), descriptions, management considerations and correlations to Australian Soil Classification (ASC) for the Roper catchment Figure 3-7 shows the distribution of the SGGs within the Roper catchment while Table 3-2 provides the areas, in hectares, within the catchment. For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au For more information on this figure please contact CSIRO on enquiries@csiro.au For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au Source: Companion technical report on land suitability (Thomas et al., 2022) The Roper catchment contains soils from all ten soil generic groups with the exception of peaty soils (SGG 5). Of the nine SGGs found in the catchment, only three occupy more than 10% of the area and together these soils represent 81% of the catchment (Table 3-2). The dominant soils of the Roper catchment are the red loamy soils, principally of the Sturt Plateau (SGG 4.1, making up 35.21%), the shallow and/or rocky soils principally found throughout the central parts of the catchment (SGG 7, 35.33%) and the cracking clay soils typically found along the rivers and other alluvium (SGG 9, 10.09%). Table 3-2 Area and proportions covered by each soil generic group (SGG) for the Roper catchment For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au na = not applicable, not found in the Roper catchment Source: Companion technical report on digital soil mapping and land suitability (Thomas et al., 2022) 3.2.2 Land suitability The overall suitability of a location for a particular land use is determined by a range of attributes. Examples of these attributes include climate at a given location, slope, drainage, permeability, plant available water capacity (PAWC), pH, soil depth, surface condition and texture. From these attributes a set of limitations are derived, which are then considered against each potential land use. Note that the use of the term suitability in the Assessment refers to the potential of the land for a specific land use such as furrow-irrigated cotton. The companion technical report on digital soil mapping and land suitability (Thomas et al., 2022) provides a complete description of the land suitability assessment framework and the material presented below is summarised from that report. The framework aggregated individual crops into a set of 21 crop groups that have shared land suitability constraints. Land suitability was then determined for 58 land use combinations of crop group × season × irrigation type (including dryland cropping). Thomas et al. (2022) calculated the overall suitability for a particular land use by considering the set of relevant attributes at each location and determining the most limiting attribute among them. This most limiting attribute then determined the overall land suitability classification on a scale from Class 1 (‘Suitable with negligible limitations’) to Class 5 (‘Unsuitable with extreme limitations’) for that particular combination of crop group × season × irrigation type. Note that this classification explicitly excludes consideration of flooding, risk of secondary salinisation, or availability of water. The intention is that such risks would be considered separately, along with further detailed soil physical, chemical and nutrient analyses before planning any developments at scheme, enterprise or property scale. Caution should therefore be employed when using these data and maps at fine scales. In order to provide an aggregated summary of the land suitability products, an index of agricultural versatility was derived for the Roper catchment (Figure 3-8). Versatile agricultural land was calculated by identifying where the highest number of 14 selected land use options were mapped as being suitable (i.e. suitability classes 1 to 3). Qualitative observations on each of the areas mapped as ‘A’ to ‘E’ in Figure 3-8 are provided in Table 3-3. Figure 3-8 Agricultural versatility index map for the Roper catchment High index values denote land that is likely to be suitable for more of the 14 selected land use options. The map also shows areas of interest (A to E) from a land suitability perspective, discussed in Table 3-3. Note that this map does not take into consideration flooding, risk of secondary salinisation, or availability of water. Source: Companion technical report on digital soil mapping and land suitability (Thomas et al., 2022) For more information on this figure please contact CSIRO on enquiries@csiro.au Table 3-3 Qualitative land evaluation observations for locations in the Roper catchment shown in Figure 3-8 Further information on each soil generic group (SGG) and a map showing spatial distribution can be found in Thomas et al. (2022). For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au Source: Companion technical report on digital soil mapping and land suitability (Thomas et al., 2022) 3.3 Irrigation systems 3.3.1 Irrigation efficiency and pumping costs Water that is captured and stored from rivers must be transported to and applied in the field where it is needed. This conveyance of water can result in losses from leakage, seepage, evaporation, outfall, unrecorded usage and system filling. Water extracted from groundwater is usually extracted locally, and transport losses are reduced, but losses can still occur during application. Losses can occur at all points along the delivery system depending on system design, and across Australia the mean water conveyance efficiency from the river to the farm gate has been estimated to be 71% (Marsden Jacobs Associates, 2003). On-farm losses occur between the farm gate and delivery to the field and usually take the form of evaporation and seepage from on-farm storages and delivery systems. Even in irrigation developments where water is delivered to the farm gate via a channel or in groundwater systems, many farms still have small on-farm storages. These on-farm storages enable the farmer to have a reliable supply of irrigation water with a higher flow rate than might otherwise be possible from channels and may also be used to recycle tailwater. Several studies have been undertaken in southern Australia of on-farm distribution losses. Meyer (2005) estimated an on-farm distribution efficiency of 78% in the Murray and Murrumbidgee regions, while Pratt Water (2004) estimated on-farm efficiency to be 94% and 88% in the Coleambally Irrigation Area and the Murrumbidgee Irrigation Area, respectively. In these irrigation areas, measured channel seepage losses in both supply channels and on-farm channels were generally less than 5% (Akbar et al., 2013). Estimates of channel seepage losses in the Burdekin Irrigation Area range from 2 to 22% (Williams, 2009). Once water is delivered to the field, it needs to be applied to the crop using an irrigation system. In-field application efficiency is the percent of water applied that is available for crop uptake. Efficiency losses occur when applied water evaporates, runs off the field or drains below the root zone. The application efficiency of irrigation systems typically varies between 60 and 90%, with more efficient pressurised systems being more expensive. There are three types of irrigation systems that can potentially be applied in the Roper catchment: surface irrigation, spray irrigation and micro irrigation (Table 3-4). Irrigation systems need to be tailored to the soil, climate and crops that may be grown, and matched to the availability and source of water for irrigation. System design also needs to consider investment risk in irrigation systems as well as likely returns, degree of automation, labour availability, and maintenance and operation costs, including pumping costs (Table 3-5). Typically spray and micro irrigation systems are more suitable for permeable or well-drained soils, whereas less expensive surface systems are suitable predominantly on clay soils. Surface irrigation systems have the lowest pumping costs, particularly where they can mainly rely on gravity to distribute water. Table 3-4 Details of irrigation systems applicable for use in the Roper catchment Adapted and updated from Ash et al. (2018), Hoffman et al. (2007), Raine and Bakker (1996) and Wood et al. (2007). IRRIGATION SYSTEM TYPE APPLICATION EFFICIENCY (%) CAPITAL COST ($/ha) LIMITATIONS Surface Basin, border and furrow 60 to 85% $800 to $3600 For most crops; topography, sandy soils, and surface levelling costs may be limiting factors Spray Centre pivot 75 to 90% $2700 to $5900 Not suitable for tree crops; high energy requirements for operation Lateral move 75 to 90% $2700 to $5400 Not suitable for tree crops; high energy requirements for operation Micro Drip 80 to 90% $6400 to $9600 High energy requirement for operation; high level of skills needed for successful operation Table 3-5 Pumping costs by irrigation operation Adapted and expanded from Culpitt (2011) with costs calculated from first principles based on assumptions of $1.20/L for diesel ($1.62/L less $0.42/L rebate), $0.30/kW⋅h for electricity, and diesel consumption of 0.25 L/kW⋅h equivalent. Bore pumping is the cost to lift water to the surface per m TDH (total dynamic head) required, where the TDH and maximum flow rate depend on the nature of the aquifer. 1 m TDH = 9.8 kPa. SURFACE SPRAY MICRO BORE ITEM UNITS FLOOD HARVESTING SURFACE IRRIGATION TAILWATER RETURN CENTRE PIVOTS LATERAL MOVES SUBSURFACE DRIP LOW PRESSURE DRIPPERS PER m TDH Total dynamic head (TDH) m 7 6 5.5 50 35 50 17 per m Pumping plant efficiency % 50% 50% 50% 66% 66% 75% 66% 40% Energy required kWh/ML 38.2 32.7 30.0 206.4 144.5 181.7 70.2 6.8 Equivalent diesel requirement L/ML 9.5 8.2 7.5 51.6 36.1 45.4 17.5 1.7 Pumping cost, electricity $/ML $11.40 $9.80 $9.00 $61.90 $43.40 $54.50 $21.10 $2.00 Pumping cost, diesel $/ML $11.40 $9.80 $9.00 $61.80 $43.20 $54.40 $21.00 $2.00 3.3.2 Surface irrigation systems Surface irrigation encompasses basin, border strip, and furrow irrigation, as well as variations such as bankless channel systems. In surface irrigation, water is applied directly to the soil surface with structures used to direct water across a field. These structures are often individual crop rows (furrows) but can be up to tens of metres wide (basins). Gravity is used to propel the water across the paddock, with levelling often required to increase the uniformity and efficiency of application. Generally, fields are laser levelled to increase the uniformity of applied water and allow adequate surface drainage from the field. The uniformity and efficiency of surface systems are highly dependent on the system design and soil properties, timing of the application of irrigation water, and the skill of the individual irrigator in operating the system. Mismanagement can severely degrade system performance and lead to systems that operate at poor efficiencies. Surface irrigation can generally be adapted to almost any crop and has a lower capital cost compared with alternative systems (Table 3-4), therefore it is well suited to broadacre crops that have lower gross margins and larger cropped areas. Surface irrigation systems perform better when soils are of uniform texture because infiltration characteristics of the soil play an important part in the efficiency of these systems. They are not so well suited to sandy soils due to losses along the furrows. Therefore, surface irrigation systems should be designed into uniform soil management units and layouts (run lengths, basin sizes) tailored to match soil characteristics and water supply volumes. Australian agriculture is increasingly employing water inflow controls to automate surface irrigation systems. High application efficiencies are possible with surface irrigation systems that are well designed and managed, and sited on appropriate clay soils. On ideal soil types and with systems capable of high flow rates, efficiencies can be as high as 85%. On poorly designed and managed systems on soil types with high variability, efficiencies may be below 60%. The major cost in setting up a surface irrigation system is generally land grading and levelling, and construction of structures to enable storage, water capture and recycling of runoff water. Costs are directly associated with the volume of soil that must be moved. Typical earthworks volumes are in the order of 800 m3/ha but can exceed 2500 m3/ha. Volumes greater than 1500 m3/ha are generally considered excessive due to costs (Hoffman et al., 2007). Surface irrigation systems are the dominant irrigation system used throughout the world. With surface irrigation, little or no energy is required to distribute water throughout the field and this gravity-fed approach reduces energy requirements of these systems (Table 3-5). 3.3.3 Spray irrigation systems Spray irrigation systems discussed here refer specifically to lateral move and centre pivot irrigation systems. Centre pivot systems consist of multiple sprinklers spaced laterally along a series of irrigation spans, supported by a series of towers. The towers are self-propelled and rotate around a central pivot point, forming an irrigation circle of generally less than 500 m radius with areas less than 80 ha. Output volumes of individual sprinkler heads are set based on proximity to the centre of the circle so that water is applied at a constant rate per hectare across the arc covered by the pivot. The time taken for the pivot to complete a full circle can range from as little as half a day to multiple days depending on crop water demands and application rate of the system. The rotation speed of the centre pivot and flow rate of sprinklers used determine the irrigation application rate. Lateral or linear move systems are similar to centre pivot systems in construction but instead of moving in a circle around a central point, an entire row of sprinklers moves laterally down a rectangular-shaped field. Water is supplied by a channel or flexible hose running the length of the field. Lateral system lengths are generally in the range of 800 to 1000 m. Spray irrigation systems offer the advantage over surface systems that they can be more easily utilised on rolling topography and generally require less land forming. Furthermore, fertiliser can be applied through fertigation where crop nutrients are injected through the irrigation system rather than applied to the field. Both centre pivot and lateral move irrigation systems have been extensively used for irrigating a range of annual broadacre crops and are capable of irrigating most field crops. They are generally not suitable for tree crops or vine crops. Saline irrigation water applications in arid environments would rapidly rust standard components of the system and can lead to foliage damage (since water is sprayed from above the crop). Centre pivot and lateral move systems usually have higher capital costs but are capable of very high efficiencies of water application. Generally, application efficiencies for these systems range from 75 to 90% (Table 3-4). A key factor for deciding whether spray systems are suitable is sourcing the energy needed to operate these systems, which are usually powered by electricity or diesel depending on costs and infrastructure available. Under high groundwater pressure, centre pivots and lateral moves may be propelled using water pressure (without the need for additional energy from pumping). 3.3.4 Micro irrigation systems Micro irrigation systems use thin-walled polyethylene pipe to apply water to the root zone via small emitters spaced along the drip tube. These systems are capable of precisely applying water to the plant root zone, thereby maintaining a high level of irrigation control and water use efficiency. Historically, micro irrigation systems have been extensively used in tree, vine and row crops, with limited applications in complete-cover crops such as grains and pastures due to the expense of these systems. Micro irrigation is suitable for most soil types and can be practiced on steep slopes. There are two main types of micro irrigation systems: above-ground and below- ground (where drip tape is buried beneath the soil surface). Below-ground micro irrigation systems offer advantages in reducing evaporative losses and improving trafficability. However, below-ground systems are more expensive and require higher levels of expertise to manage. With pressurised irrigation systems such as micro irrigation, water application can be more easily controlled, and fertigation can be used to precisely apply nutrients during irrigation. For high-value crops, such as horticultural crops, where crop yield and quality parameters dictate profitability, micro irrigation systems should be considered suitable across the range of soil types and climate conditions. Properly designed and operated micro irrigation systems are capable of very high application efficiencies, with field efficiencies of 80 to 90% (Table 3-4). In some situations, micro irrigation systems also offer labour savings and improved crop quality (i.e. more marketable fruit through better water control and precision application of crop nutrients). Intensive management of micro irrigation systems, however, is critical; to achieve these benefits requires a much greater level of expertise than other traditional systems such as surface irrigation systems. Micro irrigation systems also have high energy requirements, with most systems operating at pressures of about 15–500 kPa (about 15–50 m total dynamic head (TDH)) with diesel or electric pumps most often used (Table 3-5). 3.4 Crop types 3.4.1 Broadacre crops Cereal crops Cereal production is well-established in Australia. The area of land devoted to production of grass grains (e.g. wheat, barley (Hordeum vulgare), grain sorghum, maize, oats (Avena sativa) and triticale (× Triticosecale)) each year has stayed relatively consistent at about 20 million ha over the decade from 2012–13 to 2021–22, yielding over 55 Mt with a value of $19 billion in 2021–22 (ABARES, 2022). Production of cereals greatly exceeds domestic demand, and the majority (82% by value) was exported in 2021–22 (ABARES, 2022). Significant export markets exist for wheat, barley and grain sorghum, with combined exports valued at $15 billion in 2021–22. There are additional niche export markets for grains such as maize and oats. Amongst the cereals, summer crops such as grain sorghum and maize have the highest potential in the Roper catchments. These could be grown opportunistically using dryland production, utilising stored soil water from the wet season, or in the dry season using irrigation. To grow cereal crops, farmers would require access to tillage, fertilising, planting, spraying and harvesting equipment. Harvesting is often a contract operation, and in larger growing regions other activities can also be performed under contract. Pulse crops Pulse production is well-established in Australia. The area of land devoted to production of pulses (mainly chickpea (Cicer arietinum), lupin (Lupinus spp.) and field pea) each year has varied from 1.1 to 2.0 million ha over the decade from 2012–13 to 2021–22, yielding over 3.8 Mt with a value of $2.5 billion in 2021–22 (ABARES, 2022). The vast majority of pulses (93% by value) were exported in 2021–22 (ABARES, 2022). Pulses produced in the Roper catchments would most likely be exported, although there is presently no cleaning or bulk handling facility. Many pulse crops have a relatively short growing season, meaning they are well-suited to opportunistic dryland production, as well as irrigated production either as a single crop or in rotation with cereals or other non-legume crops. In the Roper catchment, pulse crops would most likely be suited to a production system where harvesting is in the dry season to avoid the negative impacts of rain on seed quality. Pulses are often advantageous in rotation with other crops because they provide a disease break and, being legumes, are able to fix atmospheric nitrogen into the soil, often providing carry-over nitrogen for subsequent crops. Even where this is not the case, their ability to meet their own nitrogen needs can be beneficial in reducing costs of fertiliser and associated freight. Pulses are a high-value broadacre crop (chickpeas and mungbeans have in recent years achieved prices over $1000/t) yet produce modest yields (e.g. 1 to 3 t/ha), which means freight costs represent a smaller percentage of the value of the crop compared with higher yielding, lower value cereal crops. This becomes of great importance as the distance from processing facilities and ports increases. To grow pulse crops, farmers would require access to tillage, fertilising, planting, spraying and harvesting equipment. Harvesting is generally a contract operation, and in larger growing regions other activities can also be performed under contract. The equipment required for pulse crops is the same as that required for cereal crops, so farmers intending a pulse and cereal rotation would not need to purchase extra equipment. Oilseed crops Soybean, canola (Brassica napus) and sunflowers (Helianthus annus) are oilseed crops used to produce vegetable oils, biodiesel, and high protein meals for intensive animal production. Soybean is also used in processed foods such as tofu; and it can provide both green manure and soil benefits in crop rotations, with symbiotic nitrogen fixation adding to soil fertility and sustainability in an overall cropping system. Soybean is used commonly as a rotation crop with sugarcane (Saccharum officinarum) in northern Queensland. Summer oilseed crops such as soybean and sunflower are more suited to tropical environments than winter-grown oilseed crops such as canola. Cottonseed, a by-product of cotton farming separated from the lint during ginning, is also classified as an oilseed. Cottonseed is used for animal feed and oil extraction. The area of land in Australia devoted to production of oilseeds (predominantly canola) each year has varied between 2.1 and 3.4 million ha over the decade from 2012–13 to 2021–22, yielding over 8.4 Mt with a value of $6.1 billion in 2021–22 (ABARES, 2022). The majority of oilseed production (98% by value) was exported in 2021–22 (ABARES, 2022). Canola dominates Australian oilseed production accounting for 98% of the gross value of oilseeds in 2021–22, while soybeans, sunflower and other oilseeds (including peanuts) each accounted for less than 1%. There is growing interest in soybean production in the NT, particularly from overseas companies looking to export oil to Asia. Soybean is generally grown for grain but is a useful forage crop (cut green or baled) for livestock. Soybean is sensitive to photoperiod (day length) and requires careful consideration in selection of the appropriate variety for a particular sowing window. Newer varieties will need suitability testing in the Roper to ensure they match the local climate. Sunflowers are widely grown in central Queensland and in recent years they have been grown in some areas of the Ord Valley. Crop yields are known to decline from southern Australia to northern Australia due to a less suitable climate in the north. There has been little evaluation of sunflowers in the NT. With no oilseed processing facility in the NT, soybean and sunflowers would need to be transported a significant distance until sufficient scales of production are achieved to justify the investment in processing facilities. Given both their modest yield and price, transport costs are likely to be a major constraint on profitability unless there is a well-developed supply chain into Asia. Root crops, including peanuts Root crops including peanut, sweet potatoes (Ipomoea batatas) and cassava (Manihot esculenta), are potentially well-suited to the lighter soils found across much of the Roper catchment. Root crops such as these are not suited to growing on heavier clay soils because they need to be pulled from the ground for harvest, and the heavy clay soils, such as cracking clays, are not conducive to mechanical pulling. While peanut is technically an oilseed crop, it has been included in the root crop category due to its similar land suitability requirements (i.e. the need for it to be ‘pulled’ from the ground as part of the harvest operation). The most widely grown root crop in Australia, peanut, is a legume crop that requires little or no nitrogen fertiliser and is very well-suited to growing in rotation with cereal crops, as it is frequently able to fix atmospheric nitrogen in soil for following crops. The Australian peanut industry currently produces approximately 15,000 to 20,000 t/year from around 11,000 ha, which is too small an industry to be reported separately in Australian Bureau of Agricultural and Resource Economics and Sciences statistics (ABARES 2022). The Australian peanut industry is concentrated in Queensland. In northern Australia a production area is present on the Atherton Tablelands, and peanuts could likely be grown in the Roper catchment. The Peanut Company of Australia established a peanut-growing operation at Katherine in 2007 and examined the potential of both wet- and dry-season peanut crops, mostly in rotation with maize. Due to changing priorities within the company, coupled with some agronomic challenges (Jakku et al., 2016), the company sold its land holdings in Katherine in 2012 (and Bega bought the rest of the company in 2018). For peanuts to be successful, considerable planning would be needed in determining the best season for production and practical options for crop rotations. The nearest peanut processing facilities to the Roper catchment are Tolga on the Atherton Tablelands or Kingaroy in southern Queensland. The stubble remaining after peanut harvest can be used as a high-quality supplementary feed for cattle. Most of the equipment suitable for cereal production (for planting, fertilising, spraying and harvesting) can be used for root crop production; however, specialised equipment is required to remove the roots from the ground prior to harvest. Such harvesting considerations mean that heavy clay soils are not suitable for peanut production. The residue makes good-quality hay that can be sold locally to the cattle industry, if farms have the required hay-making equipment. Industrial crops Industrial crops require post-harvest processing, usually soon after harvest in a nearby facility. Examples of industrial crops that are grown in the Australian tropics are cotton and sugarcane. Cotton Dryland and irrigated cotton production are well-established in Australia. The area of land devoted to cotton production varies widely from year to year, largely in response to availability of water, varying from 70,000 to 600,000 ha between 2012–13 and 2021–22, with an average of 400,000 ha grown per year for the decade (ABARES, 2022). Likewise, the gross value of cotton lint production varied greatly over the past decade, from $0.3 billion in 2019–20 to $5.2 billion in 2021–22. Genetically modified cotton varieties were introduced in 1996 and now account for almost all cotton produced in Australia (over 99%). Australia was the fourth largest exporter of cotton in 2022, behind the USA, India and Brazil. Cottonseed is a by-product of cotton processing and is a valuable cattle feed. Mean lint production in 2015–16 was 2.0 t/ha (ABARES, 2022). Cotton has a chequered history in northern Australia: the crop has shown vulnerability to insect pests (particularly in the mid-1970s in the Ord River region); moratoria on genetically modified crops by the WA and NT governments prevented commercial investment early this century (Yeates et al., 2013; Yeates, 2001); and production constraints have restricted the scale of production required to encourage investment in local processing facilities. Although many of these issues have been resolved, some negative public perceptions and misconceptions remain. Growers of genetically modified cotton are required to comply with the approved practices for growing those varieties, including preventative resistance management. Research and commercial test farming have demonstrated that the biophysical challenges are manageable if cotton growing is tailored to the climate and biotic conditions of northern Australia (Moulden et al., 2006; Grundy et al., 2012; Yeates et al., 2013). Specialised harvesting and baling equipment is required for cotton production. In recent years irrigated cotton crops achieving 10– 12 bales/ha have been grown successfully in the Ord River Irrigation Area. Cotton trials were also conducted at Katherine in the early 2000s but, due to the length of the wet season, poorly drained soils and the economic area required to support a cotton gin, no cotton industry developed at the time. The need to grow cotton in the dry season to avoid insect pests was historically a limiting factor in regions with a long wet season. More recently however, the lifting of restrictions on genetically modified cotton in 2018 and the development of new varieties have permitted wet- season production. Optimism for developing a local cotton industry, including investment in a gin near Katherine, has followed. Recent studies suggest that dryland cotton may be viable in the Katherine–Daly and Tipperary regions (Yeates and Poulton, 2019). Other industrial crops Other industrial crops, such as tea (Camellia sinensis) and coffee (Coffea spp.), are unlikely to yield well in the Roper catchment due to climate constraints. Sugarcane requires a large area (possibly greater than 25,000 ha) with reliable annual water, as well as a central sugar milling facility. There has been interest in hemp production. Hemp is a photoperiod-sensitive summer annual with a growing season of 70–120 days, depending on variety and temperature. Hemp is well suited to growing in rotation with legumes as hemp can use the nitrogen fixed by the legume crop. Industrial hemp can be harvested for grain with modifications to conventional headers, otherwise all other standard farming machinery for ground preparation, fertilising and spraying can be used. There are legislative restrictions to growing hemp in Australia and northern Australian jurisdictions (the NT, WA and Queensland) have all implemented legislation to license growing of industrial hemp to facilitate development of the industry. 3.4.2 Forage crops Forage, hay and silage are crops that are grown for consumption by animals. Forage is consumed in the paddock in which it is grown and is often referred to as ‘stand and graze’. Hay is cut, dried, baled and stored before being fed to animals, usually in yards for weaning or when animals are being held for sale. Silage production resembles that for hay, but harvested forage is stored wet in wrapped bales or covered ground pits, where anaerobic fermentation occurs to preserve the feed’s nutritional value. Silage is often used as a production feed to grow animals to meet the specifications of premium markets. Dryland and irrigated production of fodder is well-established in Australia, with over 20,000 producers, most of whom are not specialist producers. Fodder is grown on approximately 30% of all commercial Australian farms each year, and 70% of fodder is consumed on the farms on which it was produced. Approximately 85% of production is consumed domestically. The largest consumers are the horse, dairy and beef feedlot industries. Fodder is also widely used in horticulture for mulches and for erosion control (RIRDC, 2013). There is a significant fodder trade in support of the northern beef industry, with further room for expansion since fodder costs comprise less than 5% of beef production costs (Gleeson et al., 2012). The Roper catchment is suited to dryland or irrigated production of forage, hay and silage. Significant amounts of dryland hay production occur in the Douglas–Daly region, to the north-east of the Roper catchment. Most of that hay is either used for feeding cattle destined for live export or used as part of a feed pellet used on boats carrying live export cattle. Forage crops, both annual and perennial, include sorghum, Rhodes grass (Chloris gayana), maize and Jarra grass (Digitaria milanjiana ‘Jarra’), with specific forage cultivars. If irrigated, these grass forages require considerable amounts of water and nitrogen as they can be high-yielding (20 to 40 t dry matter per ha per year). Given their rapid growth, crude protein levels can drop quickly to less than 7%, reducing their value as a feed. To maintain high nutritive value (10 to 15% crude protein), high levels of nitrogen fertiliser need to be applied and in the case of hay, the crop needs to be cut every 45–60 days. Forage legumes are desirable because of their high protein content and their ability to fix atmospheric nitrogen in the soil. The nitrogen fixed during a forage legume phase is often in excess of requirements and remains in the soil as additional nitrogen available to subsequent crops. Annual production of legumes tends to be much lower than grasses (10 to 15 t dry matter per ha per year) but their input costs are usually much lower due to reduced nitrogen fertiliser requirements and, because they are shorter cycle crops, their total water use is often lower. Cavalcade (Centrosema pascuorum ‘Cavalcade’) and lablab (Lablab purpureus) are currently grown in northern Australia. The high crude protein content of forage legumes means that growth rates of cattle can be high. Apart from irrigation infrastructure, the equipment needed for forage production is machinery for planting; fertilising and spraying equipment is also desirable but not necessary. Cutting crops for hay or silage requires more specialised harvesting, cutting, baling and storage equipment. Hay is best stored dry, and silage requires either bunkers or large tarpaulins for covering silage above ground to maintain anaerobic conditions. Grass crops usually make better silage than legume crops because they have higher levels of sugars to aid with fermentation. Forage crops such as maize can be grown until the head just reaches the ‘milk stage’ to provide high levels of digestible energy while the leaves and stems are still green and high in protein. 3.4.3 Horticultural crops Intensive horticulture is an important and widespread industry in Australia, occurring in every state, particularly close to capital city markets. Horticultural production varied between 2.9 and 3.3 Mt per year between 2012–13 and 2021–22, of which 65 to 70% was vegetables (ABARES, 2022). Unlike broadacre crops, most horticultural production in Australia is consumed domestically. The total gross value of horticultural production was $13.2 billion in 2021–22, up from $9.3 billion in 2012–13, of which 24% was from exports (ABARES, 2022). Horticulture is also an important source of jobs, employing approximately a third of all people who work in agriculture. Horticultural production is more intensive than broadacre production and has a higher degree of risk, such as a short season of supply and highly volatile prices as a result of highly inelastic supply and demand. Managing these issues requires a heightened understanding of risks, markets, transport and supply chain issues (including associated interactions with other horticultural production regions). Production is highly seasonal and can involve multiple crops produced on individual farms to manage labour resources. The importance of freshness in many horticultural products means seasonality of supply is important in the market. Farms in the NT have the advantage of being able to produce out-of-season supplies to southern markets. However, they must also compete with production regions in northern Queensland and northern WA, which are already established production areas with associated infrastructure and are geographically closer to most of the urban centres of southern Australia. Horticultural row crops Horticulture row crops are generally short lived, annual crops, grown in the ground such as watermelon and rockmelon. Almost all produce is shipped to major markets (cities) where central markets are located. Row crops such as watermelon and rockmelon use staggered plantings over a season (for example every two to three weeks) so that the period over which harvested produce is sold can be extended. This strategy allows better use of labour and allows better management for risks of price fluctuations. Often only a short period of time with very high prices is enough to make melon production a profitable enterprise. Horticultural row crops are well-established throughout the NT. The NT melon industry, consisting of watermelon (seedless) (Citrullus lanatus), rockmelon (Cucumis melo var. cantalupensis) and honeydew (Cucumis melo (Inodorus Group) ‘Honey Dew’), produces approximately 25% of Australia’s melons. Melon production is well suited across many parts of the NT and would be well suited to the Roper catchment. Intensive production of vegetables is widely practiced close to Darwin, with $61 million in production annually. Asian vegetables consist mainly of okra (Abelmoschus esculentus), snake bean (Vigna unguiculata ssp. sesquipedalis), cucumber (Cucumis sativus), and Asian melons. Asian melons (e.g. bitter (Momordica charantia), hairy (Benincasa hispida), and luffa (Luffa ssp.)) are from the cucurbit family and are climbing annual plants. They are consumed as immature fruits either in stir fries, soups or curries. While these crops may also be grown in the Roper catchment, the high cost of transportation to Darwin would be a disadvantage in comparison to farms already situated in the Darwin area. Horticulture typically requires specialised equipment and a large labour force. Therefore, a system for attracting, managing and retaining sufficient staff is also required. Harvesting is often by hand, but packing equipment is highly specialised. Irrigation is generally with micro equipment, but overhead spray is also feasible. Leaf fungal diseases need to be carefully managed when using spray irrigation. Micro spray equipment has the advantage of being able to deliver fertiliser along with irrigation. Horticultural tree crops Some fruit and tree crops, such as mangoes and citrus, are well-suited to the climate of the Roper catchment. Other species such as avocado and lychee (Litchi chinensis) are not likely to be as well- adapted to the climate and soils. Tree crops are generally not well-suited to cracking clays, which make up some of the arable soils for irrigated agriculture in the Roper catchment. Horticultural tree production is more feasible on the lighter, well-drained soils in the west of the Roper catchment. A feature of fruit tree production is the time lag between planting and production, meaning decisions to plant need to be made with a long time frame for production and return in mind. Mango production in the NT is buffered somewhat against large-scale competition as its crop matures earlier than the main production areas in Queensland and it can achieve high returns. Mango production in the NT had a gross value of $129 million in 2020, accounting for 38% of the $341 million total value of horticultural production in the NT and half of mangoes produced (by mass) in Australia (Sangha et al., 2022). Other niche tropical fruit trees such as jackfruit (Artocarpus heterophyllus), rambutan and dragon fruit (Selenicereus undatus) are being commercially grown closer to Darwin. Prices received for niche crops can be very high when demand outstrips production, but the market is very sensitive to oversupply, particularly from cheaper overseas imports. The perennial nature of tree crops makes a reliable year-round supply of water essential. Some species, such as mango and cashew (Anacardium occidentale), can survive well under mild water stress until flowering. It is critical for optimum fruit and nut production that trees are not water stressed from flowering through to harvest, approximately from June to between November and February, depending on plant species and variety. This is a period in the Roper catchment when very little rain falls, and farmers would need to have a system in place to access reliable irrigation water during this time. High night-time minimum temperatures can reduce flowering in mangoes, and this has been a problem in orchards in the NT that may be exacerbated in the Roper catchment areas that experience high minimum temperatures. Specialised equipment is required for fruit and nut tree production. The requirement for a timely and significant labour force necessitates a system for attracting, managing and retaining sufficient staff. In a remote location the cost of providing accommodation to such staff may be significant. Tree pruning and packing equipment is highly specialised for the fruit industry, as are the micro irrigation systems typically used in horticulture (see Section 3.3.4). 3.4.4 Plantation tree crops Of the potential plantation tree crops that could be grown in the Roper catchment, Indian sandalwood and African mahogany are likely to be the most economically feasible. Many other plantation species could be grown but returns are much lower than for sandalwood or African mahogany. African mahogany is well-established in plantations near Katherine and in north Queensland. Indian sandalwood is grown in the Ord River Irrigation Scheme, around Katherine and in north Queensland. The first commercial crops of Indian sandalwood grown in Australia are now being harvested in the Ord River Irrigation Scheme and over the 16-year period of their cultivation to date many agronomic challenges have been solved. Although they are fertile, the cracking clay soils found in parts of the Roper catchments are not well-suited to tree crops due to increased potential for root shearing without very careful and ongoing irrigation management, and their susceptibility to seasonal inundation. Plantation species require greater soil depth than most other crop species so deeper loams and sands can be well- suited where irrigation is available. Plantation tree crops require over 15 years to mature, but once established they can tolerate prolonged dry periods. Irrigation water is critical in the establishment and in the first 2 years of a plantation for a number of species. In the case of Indian sandalwood (which is a hemi root parasite), the provision of water is not just for the trees themselves but the leguminous host plant. Some plantation tree crops can be grown under entirely dryland conditions (e.g. African mahogany). After harvest, trees are prepared for milling or processing, which does not need to occur locally. For example, given its high value, sandalwood is transported from northern Australia to Albany in southern WA, where the oil is extracted. 3.4.5 Niche crops Niche crops such as guar (Cyamopsis tetragonoloba), chia (Salvia hispanica), quinoa (Chenopodium quinoa), bush products and others may be feasible in the Roper catchment, but there is limited verified agronomic and market data available for these crops. Niche crops are niche due to the limited demand for their products. As a result, small-scale production can lead to very attractive prices, but only a small increase in productive area can flood the market, leading to greatly reduced prices and making production unsustainable. There is growing interest in bush products, but insufficient publicly available information for inclusion with the analyses of irrigated crop options in this report. Bush product production systems could take many forms, from culturally appropriate wild harvesting targeting Indigenous cultural and environmental co-benefits to intensive mechanised farming and processing, resembling something like macadamia farming, with multiple possible combinations and variants in between. The choice of production system would have implications for the extent of Indigenous participation in each stage of the supply chain (farming, processing, marketing and/or consumption), the co-benefits that could be achieved, the scale of the markets that could be accessed (in turn affecting the scale of the industry for that bush product), the price premiums that produce may be able to attract, and the viability of those industries. The current publicly available information on bush products is mainly focused on eliciting Indigenous aspirations, biochemical analysis (for safety, nutrition and efficacy of potential health benefits of botanicals), and considerations of safeguarding Indigenous intellectual property (e.g. Woodward et al., 2019). Analysing bush products in a comparable way to other crop options in this report would first require these issues to be resolved, for communities to agree on the preferred type of production systems (and pathways for development), and for agronomic information on yields, production practices and costs to be publicly available. Past research on guar has been conducted in the NT and current trials are underway in north Queensland, which could prove future feasibility. There is increasing interest in non-leguminous, small-seeded crops such as chia and quinoa, which have high nutritive value. The market size for these niche crops is quite small compared with cereals and pulses and so the scale of production is likely to be small in the short-to-medium term. There is a small, established chia industry in the Ord River region of WA, but its production and marketing statistics are largely commercial-in-confidence. Nearly all Australian production of chia is contracted to The Chia Company of Australia or is exported to China. In Australia, The Chia Company produces whole chia seeds, chia bran, ground chia seed and chia oil for wholesale and retail sale and exports these products to 36 countries. The growing popularity of quinoa in recent years is attached to its marketing as a super food. It is genetically diverse and has not been the subject of long-term breeding programs. This diversity means it is well-suited to a range of environments, including northern Australia, where its greatest opportunity is as a short-season crop in the dry season under irrigation. It is a high-value crop with farm gate prices of about $1000/tonne. Trials of quinoa production have been conducted at the Katherine Research Station (approximately 50 km from the western edge of the Roper catchment), with reasonable yields being returned. More testing is required in the northern environments of the Roper catchment, before quinoa could be recommended for commercial production. 3.4.6 Aquaculture Aquaculture opportunities were not evaluated in this Assessment but were covered as part of a previous resource assessment for the Darwin catchments (Irvin et al., 2018). Appendix A draws on that report to summarise: the three most likely candidate species for new aquaculture industries in the Roper catchment; an overview of the different types of intensive and extensive production systems that could be employed; and the financial viability of different options for aquaculture development, presenting an updated financial analysis that follows the same approach used previously in Irvin et al. (2018). 3.5 Crop and forage management 3.5.1 Irrigation Irrigated agriculture in the Roper catchment will be limited by the amount of irrigation water that can be reliably supplied. The companion technical report on river system simulation (Hughes et al., 2023), the companion technical report on surface water storage (Petheram et al., 2022), and the companion technical report on hydrogeological characterisation (Taylor et al., 2023) provide an overview of reliable water yields. Irrigation is required to allow reliable establishment and production of most crops at the time of year they are optimally grown. The Roper catchment exhibits a strong wet-season/dry-season rainfall pattern (Figure 3-1). Short-duration crops sown during the wet season (November to April) may require little or no supplementary irrigation, while those sown during the drier winter months may require full irrigation during the growing period to meet crop transpiration demand. Perennial crops also require irrigation through the dry season. The primary determinants of the amount of irrigation water required are the time of year the crop is grown, the duration of the growing season, how much water can be stored in the soil (particularly what is available at the time of sowing), the amount of in-crop rainfall received, and PE (especially during periods when the canopy is fully developed). The amount of irrigation required per hectare is also determined by the crop grown and crop management, such as the irrigation system used. Section 3.3 covered the various types of irrigation systems that could be used in the Roper catchment, together with the implications of each for crop management, including irrigation efficiency and pumping costs. When irrigation water is limited, farmers need to consider a range of factors to determine the best way to make use of the limited water. Where multiple crops of different value are grown on the farm, the decision would be straightforward to give priority to irrigating a high-value horticultural plantation crop such as mangoes over planting a low-value broadacre crop. In other situations, a decision would need to be made about whether to grow a small area of fully irrigated crop, or a large area of partially irrigated crop. The choice of strategy in this situation can be heavily dependent on the amount of rain likely to fall during the cropping period, the degree to which water stress affects yields and the farmers attitude to risk. For example, one study showed that deficit irrigating wheat could be a viable strategy for managing water limitations in subtropical areas of south-eastern Australia (Peake et al., 2016), while yields of crops like cotton can be very sensitive to water deficits. Ultimately this would be an economic decision about trading off the high irrigated water use efficiency that can be achieved with deficit irrigation against the impact on crop yield, harvest quality and revenue. Opportunities for dryland cropping in the Roper catchment may be limited. Dryland crops, grown without any applied irrigation water, rely on rainfall (either stored in the soil or received during crop growth) for all of their water requirements. The more rainfall that is received, the greater a dryland crop will typically yield. Dryland yields are usually lower than irrigated yields, but in years receiving above average rainfall during the growing season dryland yields may be similar to irrigated yields with careful management. Short-duration crops such as mungbean and sorghum established during the wet season are able to utilise in-crop rainfall during early stages of crop development, and then rely on stored water in the soil to minimise water stress during the later grain-filling period (although there are only limited areas of soils with high water-holding capacity in the Roper catchment). Harvesting would occur at the end of the wet season. To achieve increased dryland yields in seasons with above average in-crop rainfall, additional fertiliser inputs and pest management are also required. The inter-annual variability of rainfall means that continuous year-on-year dryland cropping is unlikely to be feasible in the Roper catchment. Opportunistic cropping, pursued when conditions are favourable, particularly in the higher rainfall areas of the catchment in combination with soils that possess high water-holding capacity, is likely to provide the most profitable and sustainable approach to dryland cropping. 3.5.2 Sowing time and cropping calendar Time of sowing can have a significant effect on achieving economical crop and forage yields, and on the availability and amount of water for irrigation required to meet crop demand. Cropping calendars identify optimum sowing times and compare the growing seasons of different crops. A cropping calendar is an essential crop management planning tool that is used to schedule farm operations for a given crop, from land preparation and sowing/planting times through the growing season to harvest (Figure 3-9) so that crops can be reliably and profitably grown. The calendar assumes best agronomic management in establishment, weed and insect control, and nutrient and water inputs to minimise stress during crop and grain development. Sowing windows vary in both timing and length among crops and regions, and consider the likely suitability and constraints of weather conditions (e.g. heat and cold stress, radiation and conditions for flowering, pollination and fruit development) during each subsequent growth stage of the crop. Limited field experience currently exists in the Roper catchment for most of crops and forages evaluated. The cropping calendar in Figure 3-9 is therefore based on knowledge of crops derived from past and current agricultural experience in the Ord River Irrigation Area (WA), Katherine and Douglas–Daly (NT) and the Burdekin region (Queensland), combined with an understanding of plant physiology, which enables crop response to differences in local climates to be anticipated. The optimum planting window and growing season for each crop were further refined through local experience and use of the APSIM (Agricultural Production Systems sIMulator) cropping systems model. Some annual crops have both wet season (WS) and dry season (DS) cropping options. Perennial crops are grown throughout the year, so growing seasons and planting windows are less well defined. Generally, perennial tree crops are transplanted as small plants (not seeds), and in northern Australia this is usually timed towards the beginning of the wet season to take advantage of wet-season rainfall. Sometimes crops can be successfully sown outside of the identified sowing windows without incurring a substantial yield penalty. In this analysis, sowing dates between September and November have generally been avoided because high evaporative demand and low water availability (see Section 3.1) are not conducive to seedling establishment; however, it is possible to sow at this time for many crops. Figure 3-9 Annual cropping calendar for cropping options in the Roper catchment WS = wet season; DS = dry season Figure 3-9 considers only the optimal climatic conditions for crop growth and is intended to be used to together with considerations of other location-specific operational constraints. Such constraints would include wet-season difficulties in access and trafficability and limitations on the number of hectares per trafficable day that available farm equipment can sow/plant. For example, For more information on this figure please contact CSIRO on enquiries@csiro.au CROP TYPECROPDECJANFEBMARAPRMAYJUNJULAUGSEPOCTNOVCROP DURATION(days) Cereal cropsSorghum (WS)ssssssgggg110—140Sorghum (DS)gssssssssssggg110—140Maize (WS)ssssssssgggg110—140Maize (DS)ssssssgggg110—140Rice (WS)ssssgggg120—160+ Rice (DS)ssssgggg90—135 Pulse crops (food legumes)Mungbean (WS)ssssssggg70—85Mungbean (DS)ssssggg70—85Chickpeassssgggg100—120OilseedsSoybean (WS)ssssssgggg110—130Sesamessssgggg110—130Root cropsPeanut (WS)ssssssggggg100—140Peanut (DS)gssssgggg100—140Cassavassssssssssssssggggg180—210Industrial cropsCotton (WS)ssssssgggg100—120Cotton (DS)ssssgggg100—120Hemp (fibre)ssssssssgggg110—150Forage, hay, silageRhodes grassggspspspgggspspspspPerennial (regrows) Forage sorghumssssssssgggssssssgg60—80 (regrows) Forage milletssssssssgggssssssgg60—80 (regrows) Forage maizegssssssgggssssssgg75—90Forage legumesCavalcadessggggggssss150—180Lablabssssssssssggggg130—160Horticulture (row crops)Melonsssssssgggg70—110Oniongssssssssssgggg130—160Capsicum, chilli, tomatossssggggg70—90 from transplantPineapplespspspgggggggPerennialHorticulture (vine)Table grapesspspspgggggggggPerennialHorticulture (tree crops)MangospspspgggggggggPerennialAvocadospspspgggggggggPerennialBananaspspspspggggggggPerennialLimespspspgggggggggPerennialLemonspspspgggggggggPerennialOrangespspspgggggggggPerennialCashewspspspgggggggggPerennialMacadamiaspspspgggggggggPerennialPlantation trees (silviculture)Africian mahoganyspspspgggggggggPerennialIndian sandalwoodspspspgggggggggPerennialSowing window forannual cropsGrowingperiodFallowSowing window for perennial cropsLikely sowing period clay-rich alluvial Vertosols are likely to present severe trafficability constraints throughout much of the wet season in the Roper catchment, while sandier Kandosols would present far fewer trafficability restrictions in scheduling farming operations (Figure 3-10). Figure 3-10 Soil wetness indices that indicate when seasonal trafficability constraints are likely to occur on Kandosols (sandy) and Vertosols (high clay) with a Bulman climate The indices show the proportion of years (for dates at weekly intervals) when plant available water (PAW) in the top 30 cm of the soil is below two threshold proportions (70% and 80%) of the maximum PAW value. Lower values indicate there would be fewer days at that time of year when fields would be accessible and trafficable. Estimates are from 100-year APSIM simulations without a crop. In actual farming situations, once a crop canopy is established later in the season, crop water extraction from the soil would assist in alleviating these constraints. Many suitable annual crops can be grown at any time of the year with irrigation in the Roper catchment. Optimising crop yield alone is not the only consideration. Ultimately, sowing date selection must balance the need for the best growing environment (optimising solar radiation and temperature) with water availability, pest avoidance, trafficability during the season and at harvest, crop rotation, supply chain requirements, infrastructure development costs, market access considerations, and potential commodity price. For example, for annual horticultural crops growing season selection is based on meeting market windows outside of when southern production areas can supply product, or to coincide with optimal growing conditions for yield and quality; while cotton is most reliable when flowering occurs in the sunny warm months of April– May and picking follows in the dry months of July–August. Many summer crops from temperate regions are suited to the tropical dry season (winter) because temperatures are closer to their optima and/or there is more consistent solar radiation (e.g. maize, chickpea and rice). For sequential cropping systems (that grow more than a single crop in a year in the same field), growing at least one crop partially outside its optimal growing season can be justified if total farm profit per year is increased and there are no adverse biophysical consequences (e.g. pest build-up). Growers also manage time of sowing to optimally use stored soil water and in-season rainfall and to avoid rain damage at maturity. Access to irrigation provides flexibility in sowing date and in the choice and timing of crop or forage systems in response to seasonal climate conditions. Depending on the rooting depth of a particular species and the length of growing season, crops established at the end of the wet season may access a full profile of soil water (e.g. 200+ mm PAWC for some For more information on this figure please contact CSIRO on enquiries@csiro.au 0% 20% 40% 60% 80% 100% 1–Jan1–Feb1–Mar1–Apr1–May1–Jun1–Jul1–Aug1–Sep1–Oct1–Nov1–Dec% of years PAW is below thresholdKandosol 80% thresholdKandosol 70% thresholdVertosol 80% thresholdVertosol 70% threshold Vertosols). While timing of sowing to maximise available water can reduce the overall irrigation requirement, it may expose crops to periods of lower solar radiation and extreme temperatures during plant development and flowering. It may also prevent the implementation of a sequential cropping system. 3.5.3 Nutrition Adequate crop nutrition is essential for achieving economic yields. Tropical soils are typically highly weathered and are usually low in the water-soluble nutrients nitrogen, phosphorus, potassium and sulfur and require their addition as fertiliser. Soil organic carbon is typically also low. Hence, nutrient management systems in the Roper catchment will require practices that can maximise organic carbon inputs via cover crops, stubble retention and mulch farming while minimising the loss of water-soluble nutrients, particularly during the wet season. Synchronising nutrient availability with crop demand is key to achieving adequate and cost-effective crop nutrition. Managers can mitigate nutritional risks by conducting thorough soil testing of paddocks. Because soil can be variable over relatively short distances, it may be necessary to sample soil for testing in a number of locations. 3.5.4 Weed and pest management Weeds can be a major contributor to economic loss in agricultural production systems and may also provide a mechanism for disease transmission. Management of weeds, particularly in irrigated systems, is important for reducing competition for resources (particularly water and nitrogen) and for maximising water and nitrogen use efficiencies in production. The cropping systems modelled in this report assume best practice in managing weed and pest infestation. 3.6 Cattle and beef production 3.6.1 Characteristics of the beef production system About 46% of the Roper catchment is used for grazing of natural vegetation by beef cattle and this is the dominant land use by area. The typical beef production system is a cow-calf operation with sale animals turned off at weights to suit the live export market (see below). The industry had an estimated annual gross value of $55.5 million in 2019–20 (Table 2-4). The within-year variation produced by the wet-dry climate is the main determinant for cattle production. Native pasture growth is dependent on rainfall, therefore, pasture growth is highest during the December to March period. During the dry season, the total standing biomass and the nutritive value of the vegetation declines. Changes in cattle liveweight closely follow this pattern, with higher growth rates over the wet season compared to the dry season. Indeed, in many cases cattle lose liveweight and body condition throughout the dry season until the next pulse of growth initiated by wet-season rains. A large area of land is needed to maintain one unit of cattle (typically termed an AE, or adult equivalent). This carrying capacity of land is determined primarily by the soil (and landscape) type, the average annual rainfall and its seasonality, and the consequent native vegetation type. NT Government estimates of carrying capacity on the Sturt Plateau range from a maximum of 15 to 21 AE/km2 (i.e. 4.8 to 6.7 ha/AE) on the relic floodplains of the Larrimah land system in ‘A’ condition (from a four point scale where ‘A’ is highest and ‘D’ is lowest) to a low of 4.5 to 5.0 AE/km2 (i.e. 20 to 22.2 ha/AE) on ‘C’ condition pastures of the Elsey and Bulwaddy land systems, noting that ‘D’ condition lands across the region have a recommended carrying capacity of zero AE/km2 (Pettit, undated). Carrying capacity estimates for the alluvial plains of the Gulf Fall are not as readily available as for the Sturt Plateau but these types of landscapes are typically considered of ‘moderate to high’ or ‘high’ pastoral value. A whole of industry survey (Cowley, 2014) provides a snapshot of the industry as it was in 2010. While some of the survey results below have inevitably changed since then, the general enterprise type has not changed significantly in the last decade and the following can be considered still current. Cowley (2014) presents data for the whole of the Katherine region, broken into five districts; Roper, Sturt Plateau, Katherine/Daly, Victoria River and Gulf. The information below comes from either the Roper or the Sturt Plateau districts unless noted to be from the Katherine region as a whole (i.e. across all five districts). Note that the Roper and Sturt Plateau districts do not follow Roper catchment boundaries but can be considered broadly representative of those properties within the catchment. The majority of properties in the Roper and Sturt Plateau districts were less than 1000 km2 in size, with a minimum of 20 km2 in the Roper. In both these districts, about 10% of properties were greater than 4000 km2. The average property size was 1133 km2 (Roper) and 1308 km2 (Sturt Plateau). At the time of survey, only about 65% of the area of each property in the Roper and 83% in the Sturt Plateau districts were considered grazed, with the remainder ‘yet to be developed’. Since 2010, property development in both districts has continued, so these percentages will now be higher. About 25% of properties in the Roper district and 70% of properties in the Sturt Plateau district had attempted the introduction of non-native species to improve the pastures, although the area of seeding was only 0.2% or 7.2%, respectively. Typically, this involved low input operations where seed was broadcast into an uncultivated seedbed. Improved pasture operations were much more common in the Sturt Plateau district than in the Roper district, with the most common species introduced being the legume, Stylosanthes spp. Nearly one-third of producers planned to increase their use of improved pastures. Only two properties across the Katherine region irrigated any pastures. Across the Katherine region, 75% of properties did some sort of prescribed burning. Cattle were typically run in paddocks (median of 9 per property in the Roper and 16 in the Sturt Plateau districts) with cattle predominantly relying on man-made water points (median of 12 in the Roper and 36 in the Sturt Plateau districts) with a low number of permanent natural water points (median of 3 and 2, respectively). Watering point development and paddock subdivision were common in the 2010 survey and have continued since. The most common grazing strategy was a combination of continuous grazing and wet-season spelling. Rotational grazing, or cell grazing, were not typically used. Across the Katherine region as a whole, nearly 40% of properties were ‘Owner-Manager’. Typically, these were smaller enterprises with less cattle than ‘Company-Manager’ properties, with 48% of the cattle and 43% of the land under this latter category. Company-Manager properties were often part of an integrated enterprise, involving transfers of cattle between properties in the company and sharing of staff and resources. Owner-Manager properties were more likely to consist only one property and run as a stand-alone enterprise. About 78% of all cattle across the Katherine region were Brahman, with about another 17% being Brahman derived. The majority of surveyed properties in both the Roper and Sturt Plateau districts ran between 2000 and 5000 head of cattle. Owner-Manager properties typically ran fewer cattle than Company-Manager properties. The majority of properties in both the Roper and Sturt Plateau districts bred cattle for the live export market, although a significant percentage (38% and 24%, respectively) bred cattle to transfer and grow-out elsewhere. Across the Katherine region, 83% of cattle turned-off made their way to live export either directly or indirectly through inter-company transfers, backgrounding or floodplain agistment, closer to Darwin. The most common live export destination was South-East Asia. Across the Katherine region most of the cattle were sold off-property early in the dry season, at the time of the first round of mustering. The most common sales months were May to-July, with a secondary peak in September–October. This corresponds to the common practice of two rounds of mustering, with the first early in the dry season and the second late in the dry season. Liveweight gain data are not available, but the Katherine region survey results reported by Cowley (2014. Table 14) provide the following information. Light steers were turned off at an average weight of 243 kg at an average age of 1 year and an average weight of 246 kg at an average age of 1.5 years. Heifers and steers for live export were turned off at an average weight of 299 kg and 308 kg respectively, at an average age of 1.8 years. Heavier steers for live export were turned off at an average weight of 405 kg at an average age of 2.7 years. Estimated average annual mortality rates across the Katherine region ranged from 3.8% for weaner heifers to 5.5% for old cows. Weaning rates in the Roper district averaged 55% and in the Sturt Plateau district 62%. While the cattle typically graze on native pastures, many properties supplementary feed hay to the weaner cohort, partly to train them to be comfortable around humans for management purposes and partly to add to their growth rates during the dry season when the nutritive value and total standing biomass of native pastures is falling. Urea-based supplements and supplements containing phosphorus were fed to a range of age and sex classes of the cattle. The urea-based supplements are to provide a source of nitrogen for cattle grazing dry season vegetation while the phosphorus supplements, mostly provided over the wet season, are used because phosphorus is deficient in many areas yet it is required for many of the body’s functions such as building bones, metabolising food and producing milk (Jackson et al., 2015). Winter (1988) working in the Katherine region, found substantial benefits to phosphorus fertilisation and supplementation, particularly in early and late wet season periods and when grazing pastures which had been oversown with legumes. Supplements were fed in 88% of the Roper district and 100% of the Sturt Plateau district. The most common animal health treatment was vaccination against botulism. 4 Approach for evaluating agricultural options 4.1 Multi-scale framework for evaluating agricultural viability The approach used to analyse the viability of agricultural development options draws on similar past technical assessments of new irrigated farming (Ash et al., 2014, 2018; Petheram et al., 2013a, 2013b; Stokes et al., 2017; Stokes and Jarvis, 2021) and an historical analysis of the successes and challenges of agricultural developments across northern Australia (Ash et al., 2014). The Assessment takes a multi-scale approach, from farm to regional scales (Figure 4-1): • The farm-scale performance component is a bottom-up analysis of farm performance, working from the biophysical and management determinants of crop yields and water use to indicative farm gross margins (GMs) that could be achieved for a range of cropping and fodder options (methods covered here in Chapter 4, with results presented in Chapter 5). • The scheme-scale viability component takes a generic top-down approach, working backwards from the costs of developing new farms and water resources (Chapter 7) to the water pricing and farm GMs and that would have to be sustained in the long term to cover those costs (Chapter 8). • The regional-scale component looks at the knock-on economic effects that could occur if new agricultural areas were developed in the Roper catchment (Chapter 9), and the market opportunities and constraints for the supply chains required for new farm produce (Section 2.2). Figure 4-1 Overview of multi-scale approach for evaluating the viability of agricultural development options For more information on this figure please contact CSIRO on enquiries@csiro.au StakeholdersOpportunitiesFarming options to be evaluatedFarm-scale: bottom-upCrop agronomy; Farm performanceScheme-scale: top-downPerformance required to cover costsRegional-scale: input–outputKnock-on regional economic impactSoilsWaterClimateCostsManagementVIABLE ? InfrastructureMarketsHistoricaldam benefits The combined analytical framework also allows fully integrated cost–benefit analysis of specific case studies, based on farm-scale analyses and information from assessments of land and water resources and associated water storage options. The added effort of rigorously adhering to such an integrating framework is more than offset by the advantages it provides: biophysical and financial resources are all accounted for in a consistent and coherent manner; the design of all analyses remains focused on the ultimate goal of identifying the most suitable development options; and interpretation of results is focused on maximising the viability of those opportunities in the context of Roper catchment environments and mitigating the risks and challenges involved. It also avoids getting distracted by sub- disciplinary ‘optimisations’ of intermediate metrics, such as maximising crop yields, maximising water use efficiency, or minimising unit costs of water and farm infrastructure, which can lead to suboptimal outcomes for configuring greenfield irrigation developments). The aim of the farm-scale analyses was to determine: the level of farm ‘performance’ that can be achieved in the Roper catchment (specifically quantified here in terms of crop yields and water use (Section 4.2), and GMs (Section 4.3)); the relative ranking of crop options that show the most potential; the management practices that can maximise those opportunities, while dealing with local challenges; and the cropping systems that could combine that understanding into possible working configurations of farming options and crop sequences on profitable commercial farms. Ultimate financial viability would depend on additional capital and overhead costs and associated considerations for developing water resources and establishing new farms (which are covered in chapters 6 to 8). 4.2 Modelling crop yields and water use 4.2.1 Analysis approach Nineteen irrigated crop options were selected to evaluate their potential performance in new irrigated farms in the Roper catchment (Table 4-1). The crops were selected to ensure that there was at least one option for each of the 13 ‘major crop groupings’ used in the companion technical report on land suitability (Thomas et al., 2022), provided that they had the potential to be viable in the Roper catchment (based on knowledge of how well these crops grow in other parts of Australia), were of commercial interest for possible development in the region, and there was sufficient information on their agronomy and farming costs/prices for quantitative analysis. Some of the 13 major land suitability crop groupings were subdivided to give a total of 21 ‘crop groups’, based on shared sensitivities to soil constraints (Thomas et al., 2022). Of these 21 ‘crop groups’, some were excluded because they were not suitable for Roper environments: for example, leafy green vegetables were unsuited to the often dry and desiccating conditions in the Roper catchment (land suitability crop group 5); there was insufficient suitable land and water for rice (crop group 8); sugarcane (crop group 11), while suited to the Roper catchment climate, was excluded because there was insufficient scale of contiguous clay soils to support a local mill (see Section 7.4.4); and only the most suitable type of forage for hay production was evaluated (perennial grass, in crop group 14, not annual forages (crop group 12) or legume forages (crop group 13)). Note that the typology of crop groupings used in the land suitability assessments (Thomas et al., 2022) was based on crop responses to soil constraints, and does not correspond to the standard agronomic classification of crops according to the types of commodities they produce (as used in Table 4-1 and the rest of this report, following the ABARES typology). The analyses used a combination of APSIM crop modelling and climatically informed extrapolation to estimate potential yield and water use for each of the 19 crop options (Table 4-1). Table 4-1 Crop options where performance was evaluated in terms of water use, yields and gross margins The methods used for estimating crop yield and irrigation water requirements are coded as: A = APSIM; E = climatically informed extrapolation. Where two letters are used, the first is the primary method, and the second is used for sensibility testing (A, E) or applying adjustments (E, A; with adjustment multipliers shown in parentheses where the APSIM median was more than 10% outside the range of sensibility testing estimates). Mango (KP) is Kensington Pride, and Mango (PVR) is an indicative new high-yielding variety, likely to have plant variety rights (e.g. Calypso). CROP TYPE CROP IRRIGATION WATER ESTIMATE METHOD YIELD ESTIMATE METHOD Broadacre crops Cereal Sorghum (grain) A, E A, E Pulse Mungbean E, A (1.30) A, E Chickpea E, A (1.19) E, A (1.14) Soybean A, E A, E Oilseed Sesame E E Peanut A, E A, E Industrial Cotton (dry season) E, A (1.69) A, E Cotton (wet season) E, A (1.34) E, A (1.24) Hemp E E Forage Rhodes grass A, E A, E Horticulture (row) Rockmelon E E Watermelon E E Onion E E Capsicum E E Horticulture (tree) Mango (PVR) E E Mango (KP) E E Lime E E Plantation tree African mahogany E E Sandalwood E E Agronomic climate analogues for the Roper catchment The nature of evaluating greenfield farming options in locations like the Roper catchment, where little irrigated commercial farming currently occurs, is that there is very limited agronomic data available of the type that is required for quantitative analyses. However, there are good analogues for Roper catchment climate and soils in agronomically similar environments at similar latitudes where irrigated cropping is well-established: the Katherine–Daly Basin (NT) is indicative of irrigated farming systems and potential crops grown on well-drained loamy soils, and the Ord River Irrigation Area (WA) is indicative of furrow irrigation on heavy clay soils. Figure 4-2 shows the close similarities in climate between possible cropping locations in the Roper catchment and Katherine (NT) and the Ord River (Kununurra, WA): •There are very small differences in solar radiation and maximum temperature among locations. •At Larrimah and Daly Waters, slightly lower minimum temperatures from May to August canextend growing season length. However, lower rainfall prior and after this period permitsflexibility in planting date to avoid the lower temperatures if required. •The climate from October to mid-December at all locations is characterised by extreme hightemperatures and high evaporative demand. The hot conditions during this period are notoptimal for active growth for the majority of crops and are only suitable for crop maturation anddesiccation. Risks of pre-harvest weathering and poor trafficability increases significantly aftermid-November. (a)Mean monthly rainfall (b)Mean daily maximum temperature (c)Mean daily solar radiation (d)Mean daily minimum temperature Figure 4-2 Climate comparisons of Roper catchment sites versus established irrigation areas at Katherine and Ord River (WA) Roper catchment locations are Bulman, Mataranka, Ngukurr, Larrimah and Daly Waters For more information on this figure please contact CSIRO on enquiries@csiro.au 050100150200250SepOctNovDecJanFebMarAprMayJunJulAugmm For more information on this figure please contact CSIRO on enquiries@csiro.au 010203040JanFebMarAprMayJunJulAugSepOctNovDec°C For more information on this figure please contact CSIRO on enquiries@csiro.au 0510152025JanFebMarAprMayJunJulAugSepOctNovDecMJ/m2/day For more information on this figure please contact CSIRO on enquiries@csiro.au 010203040JanFebMarAprMayJunJulAugSepOctNovDec°C KatherineOrdBulmanMatarankaNgukurrLarrimahDaly Waters The approach here was therefore to initially estimate likely ranges of crop yields and water use based on cropping knowledge from these climatically analogous regions, data sourced from past research and farming experience at nearby locations, and consideration of biophysical differences between Roper catchment environments and those of source data (Figure 4-2). For example, crop yields of 7 to 10 t/ha (sorghum), 2.2 t/ha (mungbean) and 5 t/ha (peanut) have been achieved under irrigation in Queensland (GRDC, 2017; QDAF, 2017), and irrigated broadacre crops such as cotton, mungbean, niche grains, peanuts, sesame and forages have produced excellent yields when grown on these soils in Katherine–Daly and Ord Valley (Beach, 1995; O’Gara, 2010; Yeates and Martin, 2006; Yeates et al., 2022). Table 4-2 shows the extrapolated estimates made this way for ranges of likely yields, irrigation water requirements, and growing seasons for the broadacre crops that were simulated in APSIM. These estimates were used for sensibility testing and calibration of modelled outputs. For other crops where there was no APSIM model, yield and water use were estimated in a similar manner, using expert experience and climatically informed extrapolation from the most similar analogue locations in northern Australia where commercial production currently occurs (those estimates are provided with the results in Section 5.2). Given the lack of direct cropping data available from within the Roper catchment, a 20% margin of error should be allowed for all these estimates at the indicative catchment level (with further allowance for variation between farms and fields). Optimum planting windows within the growing season for each species were shown in Figure 3-9 (Section 3.5.2). Table 4-2 Crop yields and median irrigation water requirement delivered to the field WS = wet-season planted (December to early March); DS = dry-season planted (late March to August); Y = year (for perennial crop). Overhead spay irrigation usually requires 10% more irrigation water than subsurface tape. CROP (YIELD UNIT) IRRIGATION WATER REQUIREMENT (ML/ha) LIKELY YIELDS (t/ha or bales/ha) Subsurface tape Furrow (clay) Sorghum (t) 4.2 DS 6.0 DS 7.0–8.0 Mungbean (t) 2.3 WS, 3.0 DS 3.0 WS, 4.3 DS 1.7–2.2 Chickpea (t) 3.4 DS 5.0 DS 2.5–3.0 Soybean (t) 5.4 DS 7.7 DS 3.5–4.0 Peanut (t) 4.0–5.0 WS Cannot farm on clay 3.5–4.5 Cotton (bales) 4.0 WS, 6.0 DS 6.0 WS, 8.0 DS 10–12 WS, 9–11 DS Rhodes grass (t) 12–14 Y, 4.2 WS 18 Y, 6.0 WS 35 Y, 9 WS Sources: Climatically extrapolated from data in Beach (1995), O’Gara (2010), Yeates and Martin (2006) and Yeates et al. (2022) APSIM APSIM was used to the estimate the crop water use and yield for those crops where modules were available (Table 4-1). All crops were sown between mid-March and the end of April (end of the wet season), except for the wet-season cotton which was sown in early February. Yield estimates from APSIM should be considered the maximum potential under ideal management and growing conditions (e.g. before allowing for pest damage or imperfect management). Crop water use in APSIM outputs was estimated from fully irrigating the crop assuming 100% efficiency. Irrigation was triggered in simulations whenever PAW (plant available water) for the top 800 mm of the soil profile fell below a crop-specific threshold proportion (65 to 80%) of PAWC. Adjustments were then made to this 100%-efficiency estimate of crop water use to estimate the amount of irrigation water that would be applied on-farm (including application losses), based on the type of irrigation system used, as described in Section 3.3.1. APSIM is a modelling framework that simulates biophysical process in farming systems (Holzworth et al., 2014) and has been used for a broad range of applications, including on-farm decision making, seasonal climate forecasting, risk assessment for government policy making and evaluating changes to agronomic practices (Keating et al., 2003; Verburg et al., 2003). It has demonstrated utility in predicting performance of commercial crops, provided that soil properties are well characterised (Carberry et al., 2009). Some crop modules have been validated for environments similar to the Roper catchment and used in previous assessments of cropping potential (Ash et al., 2014; Carberry et al., 1991; Pearson and Langridge, 2008; Webster et al., 2013; Yeates, 2001). Many APSIM crop modules use a deterministic modelling approach to simulating crop processes of development, growth and partitioning, and hence require detailed measurements from field observations to parametrise and validate the model for each location. Such field observation data do not exist for the Roper catchment and were beyond the scope of this Assessment to acquire. In particular, some APSIM crop models under estimate crop water use in inland northern Australian environments (where crop water use is elevated by windy conditions with high VPDs that APSIM has not been calibrated to). Detailed cotton trials with accurate measurements of water use, windspeeds and VPDs at Katherine were able to simulate these high levels of water use with a locally calibrated APSIM cotton model (Yeates and Martin, 2006). However, such well-calibrated models are not available for the Roper catchment, and the meteorological data required are not widely available (even outside the Roper catchment APSIM is not typically configured and calibrated to use windspeed data). To address this problem, if APSIM estimates of crop water use (after allowing for application losses) were outside the range estimated for sensibility testing by more than 10%, then an adjustment multiplier was applied to bring it into the estimated range (Table 4-2). On this basis, water use estimates for four cropping options were adjusted: cotton (wet season and dry season), mungbean and chickpea. An alternative ‘crop factor’ method (Allen et al., 1998) was also trialled but proved inferior (as discussed below). Yield estimates from APSIM generally performed well for most crop options in sensibility testing against available data, and adjustment multipliers were only required for wet-season cotton and chickpea (Table 4-2). More detailed sensibility testing of simulated metrics of crop growth processes within each season confirmed that APSIM was reasonably representing crop phenology and biomass production, so the yield errors were mainly associated with the physiological processes that ultimately determine the partitioning between harvested and non-harvested plant components (e.g. cotton bolls versus leaf and stem). This is not surprising where detailed parametrisation of these processes had not been conducted specifically for the Roper catchment. The adjustment multipliers can therefore be considered as adjusting the harvest index of these crops (defined as the ratio of the mass of harvested crop product, such as cotton bolls, to the total shoot biomass). Variation in Roper catchment environments For the main APSIM simulations of farm performance (crop yield and water use for each crop in Table 4-2) three locations were selected to represent some of the best potential farming conditions across the varied environments available in the Roper catchment. Each location consisted of a soil type and the climate associated with those areas of soils: • A sandy red Kandosol (locally called Blains) with a Mataranka (14.92 °S, 133.07 °E) climate. This soil represents some of the better farming conditions amongst the loamy soils of the Sturt Plateau (SGG 4.1 and 4.2, marked ‘A’ in Figure 3-8) that provide the largest arable areas in the Roper catchment. Using grain sorghum as an indicator crop, the PAWC of the modelled soil was 79 mm (noting that PAWC differs between crops with different rooting patterns and physiologies). Daily historical meteorological data used for these simulations was from the Mataranka weather station, which has a mean annual rainfall of about 950 mm. • A Vertosol with a Ngukurr (14.73 °S, 134.73 °E) climate. This soil represents some of the better farming conditions among the cracking clays on the alluvial plains of the major rivers (SGG 2 and 9, marked ‘B’ in Figure 3-8). The PAWC of this soil for sorghum was 212 mm, and the mean annual rainfall for Ngukurr is about 850 mm. Only small, dissected patches of this soil are suitable for cropping because of limitations from floodplain inundation, workability and the complex distribution of flood channels (which both break up patches that would be large enough to crop and cut off wet-season access to some larger pockets of otherwise suitable soil). • A Dermosol with a Bulman (13.66 °S, 134.33 °E) climate. This soil represents some of the better farming conditions among the brown non-cracking clay soils along the Central Arnhem Road and red friable loamy clay soils in the Hodgson River area (SGG 2, marked ‘C’ and ‘D’ in Figure 3-8). The PAWC of this soil for sorghum was 156 mm, and the mean annual rainfall for Ngukurr is about 1000 mm. Additional APSIM simulations were conducted to demonstrate agronomy principles, such as seasonal patterns of stored PAW and crop responses across a range of different levels of irrigation. To isolate the effects of a single factor at a time in these models (e.g. comparing Kandosol to Vertosol), all other factors were kept the same (e.g. the same climate for two different soils), which could result in additional combinations of soils and climate beyond the three listed above (used in the main simulations of crop performance). The locations of the three meteorological stations used for the simulations are also shown in Figure 3-8. The availability of meteorological data is very poor for the Roper catchment in terms of density of weather stations, gaps in historical records and the range of agronomically relevant measurements made (particularly the absence of vapour pressure and windspeed data). This limited the choice of the locations that could be modelled and the accuracy with which crop water demand can be modelled (i.e. before making calibration adjustments of the type used in Table 4-1). Crop factor and reference evapotranspiration approach The crop analyses needed to estimate not only median annual crop water use (and annual exceedance values), but also long-term (about 100 year) monthly estimates of crop demand to be used in the companion technical report on river system simulation (Hughes et al., 2023). Calculating the volumes of surface water that could safely be extracted, and the reliability with which they can be supplied to irrigators, requires data on long-term patterns of crop water demand that are coherent with simulated estimates of surface water yields (i.e. both estimates reflecting the influence of the same historical meteorological data). This is so that inter-annual and intra-seasonal variation in the timing, duration and volume of water demand by irrigators can be matched to synchronous variation in surface water supplies, to determine the volumes and timing of water than can reliably be extracted (noting that drier years, which yield less surface water, tend also to be associated with less in-crop rainfall and hence higher crop demand). The ‘crop factor’ method (Allen et al., 1998) was tested for situations where no APSIM module was available, or where the APSIM model performed poorly. The crop factor method uses a standardised index of evapotranspiration (ET0) that is adjusted by a multiplier determined by the type of crop and its stage of development (the ‘crop factor’) to estimate crop irrigation requirements. The standard application of this method is for near real-time irrigation decisions on farms (where the stage of crop development and amount of stored water in the soil profile are known at the time and used together with crop-factor-based values to decide how much irrigation water to apply). However, there are challenges when trying to apply this approach to simulating long-term patterns of inter-annual and intra-seasonal variation in crop water demand. Variation in crop demand is not only affected by meteorological factors (ET0), but is affected at least as much by variation in crop growth (e.g. how fast the canopy closes, and the duration of the growing season until irrigation is withdrawn before harvest) and antecedent conditions that affect soil water stores. Using ET0 to simulate long-term patterns of crop water demand would therefore require replicating many of the components of full crop models, such as dynamic crop canopy development and phenology (that vary within seasons and between years), and dynamic soil water profiles. Developing such models for each crop option was beyond the scope of this Assessment, but models with simple dynamic rules for crop phenology and dynamic crop factors were developed and tested for three major potential crops in the Roper catchment: cotton, watermelons and mangoes. Testing of the cotton dynamic crop factor model showed it performed worse than APSIM (under estimated annual water demand by a greater margin) and required high crop factor adjustments (>1 at full canopy) to calibrate against field trial data. If such calibration adjustments had to be applied to crop factor estimates anyway (and likely for the same reason: poor quality vapour pressure data and lack of windspeed data required to accurately represent periods of high crop demand), it was deemed that it was preferable to use the water use outputs from the more thorough APSIM model and apply a correction factor to those where the model substantially (>10% margin) under estimated crop water demand (Table 4-1). 4.2.2 Cropping systems New agricultural developments that focus on annual field crops may require sequential cropping (more than a single crop in a year) to generate sufficient revenue to cover the substantial costs of developing new farms. Annual broadacre crops have been grown sequentially for many decades in tropical Australia (e.g. in the Burdekin, Ord and Atherton Tablelands). The approach used was to explore what cropping systems could be practically implemented in the Roper catchment environments, as a way of synthesising and interpreting the results from the other farm-scale analyses. The aim was not to be prescriptive about cropping systems, but rather to provide insights on the issues and opportunities associated with developing integrated cropping systems relative to farming individual crops. 4.2.3 Dryland cropping Although the focus of this Assessment was on irrigated crop and forage production, some limited analysis was also undertaken for opportunistic dryland cropping. The APSIM simulations for the dryland analyses were used to illustrate general agronomic principles across the contrasting environments in the Roper catchment (rather than for the full analyses of farm performance done for irrigated cropping options). 4.3 Greenfield crop gross margin tool The annual farm GM is the difference between the revenue received for harvested produce and the variable costs incurred in growing, harvesting and marketing the crop each year. It is a key, but partial, metric of farm financial performance. GMs here are calculated and expressed per hectare of cropped farmland, without explicitly specifying the total area farmed other than that it would be of sufficient scale to be cost efficient in the Roper catchment context (notionally about 500 ha for broadacre farms and 200 ha for horticulture). Undertaking a comparative analysis of farm GMs for multiple greenfield development options in a region lacking established commercial farms creates unique challenges that required a bespoke ‘greenfield farm GM tool’ to be developed (Figure 4-3). Figure 4-3 Farm gross margin tool used for consistent comparative analysis of different greenfield farming options For more information on this figure please contact CSIRO on enquiries@csiro.au Base lookup data tablesCrop yieldWater useInputs $Operations $Labour $ (Variable) Farming systems (crop, operations, inputs, labour: scheduled to climate) Crop 1OperationsInputsCrop 2OperationsInputsCrop 3OperationsInputsCrop …nOperationsInputsFarming scenarios (farming system ×location ×management ×…) Farm Option 1Farm Option 2Farm Option 3Farm Option …x$ Outputs The challenges faced, and the approach taken to address them, are summarised below: • Mix of GM templates of different historical provenance: Previous similar assessments have built on-farm GM tools from multiple different sources that used different approaches for farm financial accounting. A consistent accounting approach was required in this Assessment both so that cropping options could be compared on a like-for-like basis, and so that accounting was compatible with how these GMs were combined with capital and overhead costs in subsequent full discounted cashflow (DCF) analyses (Chapter 8). For example, if a particular farming operation is treated and costed as being undertaken by an outsourced contractor, then capital costs of the associated equipment should not be included in subsequent capital costs of farm establishment used in full financial analyses. In addition, a consistent GM analysis framework provides for a smoother and more automated workflow, including the input of required data from the farm agronomy parts of the evaluation, and output of data from the subsequent scheme financial analyses. • Inappropriate translation from existing to greenfield farming location: When using a farm GM template for an established farming region (mainly southern and coastal areas) there are implicit assumptions about what farming operations are conducted and when they are scheduled, based on proven local practices. But when those templates are extrapolated to a new location without proven commercial farming, those implicit assumptions can break down and the GM accounting can become disconnected from the farming practices that would actually be required locally. In greenfield situations there is a need to tightly couple GM accounting with how farm operations would be scheduled and conducted in those locations (e.g. scheduling of farm operations and the equipment required needs to consider seasonal trafficability and other climate constraints, as does the choice of which fertilisers and pest/weed management methods are used and how they are applied). Building GMs ‘bottom-up’ from an explicit, locally adapted schedule of farm operations ensures this is the case. • Arbitrary inconsistencies in assumptions: When using GM templates from multiple different sources there are inevitably arbitrary differences in assumptions and costings (or, at least, it is laborious to keep these rigorously synchronised across templates). For example, such discrepancies would include the choice of fertilisers, which micronutrients are being applied (on the same soils), and which markets are being used for pricing transport costs of farm inputs and produce (including the point in the supply chain to which goods are delivered, freight is costed and payments to farmers are priced). These issues can be addressed with a standardised set of data tables, and rigorously logging the assumptions and the basis used for estimating each cost. An added advantage of this approach is that once a GM is developed for a farming system in one scenario, it is easier to rigorously adapt it to other applications (because it is obvious precisely how assumptions have changed and the exact cost basis on which new values need to be adjusted). The farm GM tool consisted of three main types of components, as illustrated in Figure 4-3. The foundation of analyses was a set of data tables with all the farm agronomic performance data generated in Section 4.2 (crop yields and water use) and a standard set of costs for inputs, farm operations and labour requirements to be applied consistently to all farming scenarios. Each farming system to be evaluated then had its own template that drew on the standard data tables. The farming system templates consisted primarily of a schedule of farm operations that linked to the machinery operating costs in the data tables, together with associated costs of inputs and labour requirements. Each farming operation allowed specifying up to three simultaneous compatible activities, for example, using a 166 kW tractor with airseeder and harrows to (i) plant 13 kg/ha cotton seed, (ii) with a BollGuard fee, and (iii) 100 kg/ha Granulock Z fertiliser, all in a single operation. Each operation also had a date associated with it, used to display a calendar of the farming operations, so that sensibility testing could ensure the farming system being costed was operationally viable and agronomically sound (relative to local climatic and trafficability constraints and optima, as discussed in Section 3.5.2). Farming templates also included other parameters specific to that farming system, such as the prices received for produce (which allowed splitting yield into different products/produce classes and specifying different prices for the same crop grown under different conditions in variant farming systems). The final component of the GM tool consisted of farming scenarios, which are parameter sets for each scenario specifying details of the farming system, crop performance data, and crop management required to calculate the final of set of GMs to be compared. The scenario parameters included specifying the type of irrigation, in order to automatically account for associated irrigation application losses and pumping costs. An adjustment could also be made to the total calculated cost of labour required for all farm operations to account for the portion that would be performed by permanent staff (accounted for separately in the overhead costs: noting labour costs have both variable (mainly seasonal workers) and fixed/overhead (mainly permanent staff) components). Because this Assessment focuses on the viability of greenfield irrigated development (i.e. including a new water source), the cost of water is not included in GMs as a variable cost, but is accounted for the in the capital and operating costs of the new water source. The costs of the water sources are treated on a consistent like-for-like basis, so that alternate water sources can be substituted for each other in any arbitrary pairing with different farming options in the later scheme-level analyses of financial viability (see Chapter 8). All costs were specified in real constant June 2021 Australian dollars (as is the standard throughout this Assessment, adjusted for inflation from older sources where necessary). Costs of farming inputs were based on prices from suppliers in Darwin, and freight costs assumed that this is where they would be purchased. Since agricultural commodity prices (versus inputs) typically fluctuate more over time, they were notionally set at the average for the past decade (e.g. as documented in ABARES (2022) data series). Commodity prices do not represent a forecast, just a long-term historical precedent (to reduce the effects of temporary spikes and dips in current prices). Investors would need to make their own decisions about long-term future trends in input costs and commodity prices (see also Section 2.2.6 covering recent volatility in farmers’ terms of trade). Farm GMs were calculated, together with breakdown summaries of variable costs and revenue, for each farming option listed in Table 4-1. Given the uncertainties in estimating farm performance in greenfield situations, narrative risk analyses were also undertaken to illustrate how different challenges and opportunities could affect farm GMs. 4.4 Modelling the integration of forage and hay crops within existing beef cattle enterprises A commonly held view within the northern cattle industry is that the development of water resources would allow irrigated forages and hay to be integrated into existing beef cattle enterprises, thereby improving their production and potentially, their profitability. Currently, cattle graze on native pastures, which rely solely on rainfall and any consequent overland flow. The quality of these pastures is typically low, and it declines throughout the dry season, so that cattle either gain little weight, or even lose weight, during this period. Theoretically, the use of on-farm irrigated forage and hay production would allow graziers greater options for marketing cattle: meeting market liveweight specifications for cattle at a younger age; meeting the specifications required for different markets than those typically targeted by cattle enterprises in the Roper catchment; and providing cattle which meet market specification at a different time of the year. Forages and hay may also allow graziers to implement management strategies, such as early weaning or weaner feeding, which should lead to flow-on benefits throughout the herd, including increased reproductive rates. Some of these strategies are already practiced within the Roper catchment but are reliant on hay or other supplements purchased on the open market. By growing hay on-farm, the scale of these management interventions might be increased, at reduced net cost. Furthermore, the addition of irrigated feeds may also allow graziers to increase the total number of cattle which can be sustainably carried on the property. The use of irrigated hay or forage production to feed cattle on-farm in the Roper catchment is very little used, if at all (Cowley, 2014). In fact, there are very few cattle enterprises in northern Australia which are set up to integrate on-farm irrigation, notwithstanding the theoretical benefits. Despite its apparent simplicity, fundamentally altering an existing cattle enterprise in this way brings in considerable complexity, with a range of unknowns about how best to increase productivity and profitability. The most comprehensive guide to what might be possible to achieve by integrating forages into cattle enterprises can be found in Moore et al. (2021) who have used a combination of industry knowledge, new research and modelling to consider the costs, returns and benefits. Because there are so few on-ground examples, modelling has been used in a number of studies to consider the integration of forages and hay into cattle enterprises, summarised in Watson et al. (2021b). This study in the Roper catchment used CLEM (Crop Livestock Enterprise Model; Version 2023.3.7172.0; Crop Livestock Enterprise Model ) to model the impact of on-farm irrigated forages and hay for a representative property on the Sturt Plateau. CLEM is a whole-of-farm model. Native pasture (modelled with GRASP; McKeon et al., 2000) and several irrigated forage and hay options (modelled with APSIM; Holzworth et al., 2014) were prepared as input into CLEM on a monthly time step. Animal production, herd dynamics, financial parameters (overhead and variable costs and cattle and hay prices) and management actions within CLEM were then parameterised with information from a number of sources (Section 5.4.1). CLEM’s output then included information on cattle production and hence herd dynamics as well as financial metrics, which were used to compare across the baseline, forage and hay scenarios. Central to CLEM is a set of animal production equations which calculate reproductive rates, milk production, liveweight changes, mortality and other key functions. It is a relatively new model which builds on the North Australian Beef Systems Analyser (NABSA; Ash et al., 2015) but models the performance of all individual animals within the herd, rather than calculating outputs for each cohort of livestock (typically age and sex class). 5 Performance of agricultural development options This chapter presents the results and interpretation of the farm-scale analyses detailed in Chapter 4. It begins with a discussion of agronomic principles of dryland and irrigated cropping in the Roper catchment (Section 5.1). Those principles provide context for the results on the 19 individual crop options that were analysed in terms of crop yields, the amount of irrigation water used, and GMs (the three metrics referred to collectively as farm ‘performance’ in this and following chapters) (Section 5.2). The irrigated crop options are grouped into broadacre, horticulture and plantation tree crops. The viability of these options is then discussed in a section on cropping systems, that considers the mix of farming practices that could most profitably be integrated within local Roper catchment environments, using both single and sequential cropping systems (Section 5.3). The final section evaluates the viability of integrating irrigated forages into existing beef production systems (the dominant current agricultural activity in the Roper catchment) (Section 5.4). This chapter aims to determine: the level of farm performance that can be achieved in the Roper catchment; the relative ranking of crop options that show the most potential; the management practices that can maximise those opportunities, while dealing with the local constraints; and the cropping system configurations that might conceivably use that understanding to implement mixes of these crop options on profitable commercial farms. Ultimate financial viability would depend on additional capital and overhead costs and associated considerations for developing water resources and establishing new farms (covered in chapters 6 to 8 that follow). 5.1 Principles of dryland and irrigated cropping 5.1.1 Dryland broadacre cropping Dryland cropping (crops grown without irrigation, relying only on rain) has been practiced by farmers in the NT for almost 100 years, yet only small areas of dryland crop production currently occur each year. This indicates that despite the theoretical possibility that dryland crops could be produced using the significant rainfall that occurs during the wet season in the Roper catchment, in practice there are significant agronomic and market-related challenges to dryland crop production that have prevented its expansion to date. As dryland farming depends on stored soil water and in-crop rainfall, the timing of crop establishment to maximise both production and economic yield is critical. Without the certainty provided by irrigation, dryland cropping is opportunistic in nature, relying on favourable conditions in which to establish, grow and harvest a crop. The annual cropping calendar in Figure 3-9 shows that, for many crops, the sowing window includes the month of February. For relatively short-season crops such as sorghum and mungbean, this coincides with both the sowing time that provides close-to-maximum crop yield and the time at which the season’s water supply can be most reliably assessed with a high degree of confidence. Table 5-1 shows how plant available soil water content at sowing and subsequent rainfall in the 90 days after each sowing date varies over three different sowing dates for a Vertosol in the Roper catchment at Bulman. As sowing is delayed from February to April, the amount of stored soil water increases. However, there is a significant decrease in rainfall in the subsequent 3 months after sowing. Combining the median PAW in the soil profile at sowing, and the median rainfall received in the 90 days following sowing, provides totals of 581, 464 and 311 mm for the February, March and April sowing dates, respectively. For ‘drier than average years’ (80% probability of exceedance), the soil water stored at sowing and the expected rainfall in the ensuing 90 days (<460 mm) would result in water stress and comparatively reduced crop yields. In ‘wetter than average years’ (20% probability of exceedance) the amount of soil water at the end of February combined with the rainfall in the following 90 days (764 mm) is sufficient to grow a good short-season crop (noting that the timing of rainfall is also important since some rain is ‘lost’ to runoff, evaporation, and deep drainage between rainfall events). Opportunistic dryland cropping would target those wetter years where PAW at the time of sowing indicated a higher chance of harvesting a profitable crop. Table 5-1 Soil water content at sowing and rainfall for the 90-day period following sowing for three sowing dates, based on a Bulman climate on Vertosol PAW = plant available water stored in soil profile. The 80%, 50% (median) and 20% probability of exceedance values are reported, for the 100 years between 1920 and 2020. The lower-bound values (80% exceedance) occur in most years, while the upper bound values only occur in the most exceptional upper 20% of years. SOWING DATE PAW AT SOWING DATE (mm) RAINFALL IN 90 DAYS FOLLOWING SOWING DATE (mm) TOTAL STORED SOIL WATER + RAINFALL IN SUBSEQUENT 90 DAYS (mm) 80% 50% 20% 80% 50% 20% 80% 50% 20% 1 February 80 151 212 299 424 614 457 581 764 1 March 143 228 305 146 250 405 354 464 617 1 April 193 274 299 16 53 128 269 311 393 The success of dryland cropping is clearly dependent on wet-season rainfall, but also the ability of the soil to store water for the crop to use as it finishes growing into the dry season. Figure 5-1 highlights the effects of diminishing water availability and increasing evapotranspiration likely to be encountered when sowing a dryland crop at the start of April or later. This constraint is much more severe for sandier soils that have less capacity to store PAW (like Kandosols in the Roper catchment, Figure 5-1a), compared to finer textured soils (like the alluvial Vertosols in the Roper catchment, Figure 5-1b). (a) Bulman Kandosol (sandy, PAWC 79 mm) (b) Bulman Vertosol (high clay, PAWC 212 mm) Figure 5-1 Influence of planting date on dryland grain sorghum yield at Bulman for (a) Kandosol and (b) Vertosol Estimates are from APSIM simulations with planting dates on the 1st and 15th of each month. PAWC values give the plant available water capacity that each soil profile can store. The shaded band around the median line indicates the 80% to 20% exceedance probability range in year-to-year variation. Well-drained, but infertile, Kandosols predominate throughout much of the Roper catchment and across northern Australia generally (Williams et al., 1985). Such soils also tend to be susceptible to erosion and hard-setting, which can decrease the infiltration of intense monsoon rainfall into the soil for storage and increase the difficulty of establishing crops. The low water-holding capacity of Kandosols, in combination with the extreme heat that often occurs in the Roper catchment between rain events, can quickly induce water stress at any stage during the crop life cycle. This contrasts with cropping systems in southern Australia where crops on similar soils are grown during winter when temperatures are cooler and rainfall is more regular and less intense, so crops experience less water stress. Heavier clay soils, such as Vertosols in alluvial areas of the Roper catchment, hold more PAW, so dryland crops grown on these soils would likely experience less water stress (Figure 5-1). However, alluvial soils are subject to frequent inundation and waterlogging during the wet season due to their location in the landscape and particularly poor drainage in some Roper catchment Vertosols (Figure 3-10). This means that crops cannot always be sown at optimum times, fertiliser can be lost to runoff, drainage and denitrification, and in-crop management (e.g. for weed, disease and insect control) cannot be undertaken cost-effectively with ground-based equipment in a timely manner, a critical requirement for dryland crop production to succeed (Robertson et al., 2016). Soil is rarely uniform within a single paddock, let alone across entire districts. Without the homogenising input of irrigation to alleviate water limitations (and associated high inputs of fertilisers to alleviate nutrient limitations), yields from low-input dryland cropping are typically much more variable (both across years and locations) than yields for irrigated agriculture. Furthermore, the capacity of the soil to supply stored water varies not only with soil type, but also depends on crop type and variety because each crop’s root system has differing ability to access water, particularly deep in the profile. This makes it harder to make generalisations about the viability of dryland cropping in the Roper catchment as farm performance (e.g. yield and GM) is much more sensitive to slight variations in local conditions. Rigorous estimates of dryland crop For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au 024681–Jan15–Jan1–Feb15–Feb1–Mar15–Mar1–Apr15–AprYield (t/ha) Sow dateRangeMedian For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au 024681–Jan15–Jan1–Feb15–Feb1–Mar15–Mar1–Apr15–AprYield (t/ha) Sow dateRangeMedian performance would require detailed localised soil mapping and crop trials before investment decisions could be confidently made. Socio-economic factors have also been identified as limitations to the development of dryland farming in the NT (Chapman et al., 1996). Lack of significant local markets for broadacre commodities mean that transport costs to markets are much higher than costs incurred by alternative production regions across southern Australia, and that GMs for low value, small- grained commodity crops (such as sorghum and maize) are too low to justify significant expansion of dryland cropping for these crops. Producers also experience difficulties in attracting and retaining a trained labour force to hot, remote locations. These challenges have combined to prevent expansion of dryland cropping in the NT. These socio-economic constraints affect irrigated agriculture too, in an interrelated way, since the two types of farming typically complement each other in achieving sufficient combined economies of scale to overcome many of these constraints. A core of irrigated farming often provides the impetus to attract an expansion of dryland farming around it (and, conversely, the limited scale of irrigated broadacre farming in the NT has impeded development of dryland cropping). Despite the challenges described above, recent efforts in the NT have identified potential opportunities for dryland farming using higher value crops, such as pulses or cotton. A preliminary APSIM assessment of the potential for dryland cotton in the region suggested that mean lint yields of 2.5–3.5 bales/ha may be possible at a range of locations in the vicinity of the Roper catchment (Yeates and Poulton, 2019). However, there was very high variability in median yields between farms (1–5 bales/ha), depending on management and soil type. 5.1.2 Irrigated cropping responses and options Crops that are fully irrigated can yield substantially more than dryland crops. Figure 5-2 shows how yields for sorghum grown on Kandosols in the Roper catchment increase as more water becomes available to alleviate water limitations and meet increasing proportions of crop demand. With sufficient irrigation, yields are highest for (wet-season sown) crops grown over the dry season when radiation tends to be less limiting (plateau of Figure 5-2a versus Figure 5-2b). For wet-season sowing, unirrigated yields can approach fully irrigated yields in good years (yields exceeded in the top 20% of years, marked by the upper shaded range in Figure 5-2a). However, irrigation allows greater flexibility in sowing dates, allows sowing in the dry season too (for crops that would then grow through the wet season), and generates more reliable (and higher median) yields. (a) 1 February sowing (wet season) (b) 1 August sowing (dry season) Figure 5-2 Influence of available irrigation water on grain sorghum yields for planting dates (a) on 1 February and (b) 1 August, for Kandosols with a Bulman climate Estimates are from 100-year APSIM simulations. The shaded band around the median line indicates the 80% to 20% exceedance probability range in year-to-year variation. Dryland production is indicated by the zero point where no allocation is available for irrigating. The different amounts of irrigation water available (Figure 5-2) also indicate the range of options for growing crops from dryland (zero ML/ha available irrigation water), to supplemental irrigation (where less water is available than required to maximise yield, but sufficient to achieve higher and more reliable yields than from purely dryland cropping), to full irrigation (where there is sufficient water to achieve close to the maximum yield). Increasing amounts of ‘available’ water do not mean that those volumes were applied in Figure 5-2, only that it was available to apply to crops when needed; so, the yield curves plateau once crop demand is fully met. The simulations did not seek to ‘optimise’ supplemental irrigation strategies in years where available water was insufficient to attain maximum crop yields; irrigators would need to make those decisions in years where available water was lower than total crop demand. A key advantage of irrigated dry-season cropping in northern Australia is that the availability of water in the soil profile and surface water storages is largely known at the time of planting (in the early wet season: Table 5-1). This means irrigators have good advance knowledge for planning how much area to plant, which crops to grow, and what irrigation strategies to use, particularly in years where they have insufficient water to fully irrigate all fields. A mix of irrigation approaches could be used, such as expanding the scale of a core irrigated cropping area with other less intensively farmed areas, opportunistic dryland cropping, opportunistic supplemental irrigation, opportunistic sequential cropping, and/or adjusting the area of fully irrigated crops grown to match available water supplies that year. 5.2 Performance of irrigated crop options Measures of farm performance (in terms of yields, water use and GMs) are presented for the 19 cropping options that were evaluated (Table 4-1). As noted in Chapter 4, given the limited commercial irrigated farming that currently occurs in the Roper catchment to provide real world data, estimates of crop water use and yields should be considered as indicative, and to have a possible 20% margin of error at the catchment scale (with further variation expected between farms and fields). For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au 02468012345Yield (t/ha) Available irrigation water (ML/ha) RangeMedian For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au 02468012345Yield (t/ha) Available irrigation water (ML/ha) RangeMedian GMs are a key partial metric of farm performance but should not be treated as fixed constants determined by the cropping system alone. They are a product of the farming and business management decisions made by individual farmers, input prices, commodity prices and market opportunities. As such, the GMs presented below should be treated as indicative of what might be attained for each cropping option once their sustainable agronomic potential has been achieved. Any divergence from assumptions about yields and costs would flow through to GM values, as would the consequences of any underperformance or overperformance in farm management. It is unrealistic to assume that the levels of performance in the results below would be achieved in the early years of newly established farms, and allowance should be made for an initial period of learning when yields and GMs are below their potential (see Chapter 8). Collectively however, the GMs and other performance metrics presented here provide an objective and consistent comparison across a suite of likely cropping options for the Roper catchment and an indicative maximum performance that could be achievable for greenfield irrigated development for each of the groupings of crops below. 5.2.1 Irrigated broadacre crops Table 5-2 shows the farm performance (yields, water use and GMs) for the ten broadacre cropping options that were evaluated. For crops that were simulated with APSIM, estimates are provided for locations with three different soil types associated with climates in the Roper catchment (Kandosol at Mataranka, Vertosol at Ngukurr and Dermosol at Bulman) and include measures of variability (expressed in terms of years with yield exceedance probabilities of 80%, 50% (median) and 20%). For other crops, yield and water use estimates (and resulting GMs) were estimated based on expert experience and climatically informed extrapolation from the most similar analogue locations in northern Australia where commercial production currently occurs. The broadacre cropping options with the best GMs (>$1500/ha) were cotton (both wet-season and dry-season cropping), forages (Rhodes grass) and peanuts. These suggest GMs of $4000 to $5000 might be achievable for broadacre cropping in the Roper catchment, although not necessarily at scale. Mungbean, chickpea and industrial hemp had intermediate GMs (about $1000/ha). The GMs for sorghum, soybean and sesame were low (<$800/ha in most cases). Simulated yields (and consequent GMs) were generally lowest on the Kandosol and highest on the Vertosol because of the increased buffering capacity that a high PAWC clay soil provides against hot weather that triggers water stress even in irrigated crops. The Dermosol yields and GMs were slightly lower than the Vertosol due to its lower PAWC. It was not possible to model cotton on the Vertosol in APSIM because of the difficulty in replicating the nuances of managing waterlogging on inter-furrow mounds on these heavy clay soils, and the sensitivity of cotton roots to waterlogging. Estimates of cotton yield (used in place of cotton simulations) for Vertosols assume that this waterlogging could be managed if fields were carefully sited and furrows were skilfully managed. However, as illustrated before, some Vertosols in the Roper catchment present particularly severe drainage challenges (Figure 3-10) that could limit the suitable area for farming, and may require more careful management than Vertosols that are currently used for cotton farming in other parts of Australia. Table 5-2 Performance metrics for broadacre cropping options in the Roper catchment: applied irrigation water, crop yield and gross margin (GM) for three environments Performance metrics are an indication of the upper bound that could be achieved after best management practices for Roper catchment environments had been identified and implemented. All options are for dry season (DS) irrigated crops sown between mid-March and the end of April (end of the wet season), except for the wet season (WS) cotton, sown in early February. Variance in yield estimates from APSIM simulations is indicated by providing 80%, 50% (median) and 20% probability of exceedance values (Y80%, Y50% and Y20%, respectively), together with associated applied irrigation water (including on-farm losses) and GMs in those years. The lower-range yields (Y80% exceedance) occur in most years, while the upper-range Y20% yields only occur in the most exceptional upper 20% of years. Note that applied irrigation water is not always higher in years with higher yields (Y20%). ‘na’ indicates 20% and 80% exceedance estimates that were not applicable because APSIM outputs were not available and expert estimates of just the median yield and water use were used instead. Peanut is omitted for the Vertosol location because of the practical constraints of harvesting root crops on clay soils. Freights costs assume processing near Katherine for cotton and peanut, and that hay is sold locally. No crop model was available for sesame or hemp, so indicative estimates for the catchment were used. Cotton yields and prices are for lint bales (227 kg after ginning), not tonnes (t). PAWC = plant available water capacity. CROP APPLIED IRRIGATION WATER CROP YIELD YIELD UNIT PRICE VARIABLE COSTS TOTAL REVENUE GROSS MARGIN (ML/ha/y) (Yield units) ($/unit) ($/ha/y) ($/ha/y) ($/ha/y) Y80% Y50% Y20% Y80% Y50% Y20% Y80% Y50% Y20% Dermosol (156 mm PAWC), Bulman climate (~1000 mm annual rainfall) Cotton WS 6.9 6.1 3.8 10.4 11.2 12.0 bales/ha 580 3604 7448 3439 3844 4366 Cotton DS 8.0 7.2 8.5 8.4 9.1 9.8 bales/ha 580 3291 6073 2400 2782 3110 Sorghum (grain) 4.0 5.6 4.9 7.5 7.9 8.2 t/ha 310 1734 2449 674 715 801 Mungbean 3.2 3.2 3.2 1.7 1.9 2.0 t/ha 1100 940 1919 797 979 1068 Chickpea 4.8 4.3 5.4 2.5 2.7 3.0 t/ha 750 1119 2052 772 933 1053 Soybean 7.6 6.6 7.3 3.3 3.6 3.8 t/ha 570 1342 2052 540 710 784 Peanut 4.5 5.3 5.5 4.5 4.8 5.1 t/ha 1000 3126 4800 1508 1674 1888 Rhodes grass (hay) 13.3 10.8 12.4 34.2 35.1 36.1 t/ha 42 4189 8775 4266 4586 4671 Kandosol (79 mm PAWC), Mataranka climate (~950 mm annual rainfall) Cotton WS 3.6 5.8 7.1 6.9 10.9 12.5 bales/ha 580 3544 7283 1760 3738 4510 Cotton DS 8.9 8.7 10.5 8.3 9.1 10.0 bales/ha 580 3530 6073 2091 2543 2890 Sorghum (grain) 6.8 6.5 5.9 6.1 6.4 6.8 t/ha 310 1730 1984 142 254 416 Mungbean 4.1 4.6 5.0 1.6 1.7 2.0 t/ha 1100 1156 1717 507 561 809 Chickpea 6.2 5.3 4.5 2.3 2.4 2.6 t/ha 750 1340 1796 302 455 692 CROP APPLIED IRRIGATION WATER CROP YIELD YIELD UNIT PRICE VARIABLE COSTS TOTAL REVENUE GROSS MARGIN (ML/ha/y) (Yield units) ($/unit) ($/ha/y) ($/ha/y) ($/ha/y) Y80% Y50% Y20% Y80% Y50% Y20% Y80% Y50% Y20% Soybean 7.1 7.4 7.3 2.3 2.5 2.6 t/ha 570 1637 1425 –283 –212 –147 Peanut 4.6 6.1 6.5 3.6 3.8 3.9 t/ha 1000 2850 3800 936 950 1001 Rhodes grass (hay) 19.3 19.9 18.6 32.0 32.9 33.4 t/ha 42 4694 8225 3407 3531 3709 Vertosol (212 mm PAWC), Ngukurr climate (~850 mm) Cotton WS na 6.0 na na 11.0 na bales/ha 580 3807 7341 na 3535 na Cotton DS na 8.0 na na 9.5 na bales/ha 580 3599 6340 na 2741 na Sorghum (grain) 4.6 5.7 5.7 7.6 8.1 8.3 t/ha 310 1714 2511 715 797 839 Mungbean 2.6 5.2 3.9 2.1 2.3 2.4 t/ha 1100 1000 2323 1192 1323 1438 Chickpea 5.4 5.4 5.4 2.7 3.0 3.2 t/ha 750 1137 2223 939 1086 1234 Soybean 9.0 8.0 9.1 3.8 4.1 4.4 t/ha 570 1372 2337 803 965 1083 Rhodes grass (hay) 13.2 12.1 15.1 35.6 36.8 38.5 t/ha 42 4377 9200 4538 4823 4947 General estimate for Roper catchment (not soil specific) Sesame na 6.2 na na 0.9 na t/ha 1300 1737 1170 na –567 na Hemp (grain seed) na 5.9 na na 1.1 na t/ha 3150 2149 3465 na 1316 na A breakdown of the variable costs for growing broadacre crops shows that the largest two costs are the costs of inputs (31%) and farm operations (35%) (Table 5-3). Both of these cost categories would have similar dollar values when growing the same crop in southern parts of Australia, but the cost category that puts northern growers at a disadvantage is the higher market costs (23%, for freight and other costs involved in selling the crop – also see Section 2.2.4). Total variable costs consume 58% of the gross revenue generated, which leaves sufficient margin for profitable farms to be able to temporarily absorb small declines in commodity prices or yields without creating severe cashflow problems. Table 5-3 Breakdown of variable costs relative to revenue for broadacre crop options The first eight crops (Cotton WS to Rhodes grass) are for the Dermosol (intermediate performance), and the last two crops are for general catchment estimates. ‘Input’ costs are mainly for fertilisers, herbicides, and pesticides; the cost of farm ‘operations’ includes harvesting; ‘labour’ costs are the variable component (mainly seasonal workers) not covered in fixed costs (mainly permanent staff); ‘market’ costs include levies, commission and transport to the point of sale. WS = wet season; DS = dry season CROP TOTAL REVENUE TOTAL VARIABLE COSTS PERCENTAGE BREAKDOWN OF VARIABLE COSTS VARIABLE COSTS VS REVENUE ($/ha/y) ($/ha/y) INPUTS (%) OPERATIONS (%) LABOUR (%) MARKET (%) (%) Cotton WS 7448 3604 33% 35% 7% 26% 48% Cotton DS 6073 3291 36% 34% 7% 23% 54% Sorghum (grain) 2449 1734 23% 19% 12% 45% 71% Mungbean 1919 940 36% 27% 16% 22% 49% Chickpea 2052 1119 33% 26% 15% 25% 55% Soybean 2052 1342 34% 23% 16% 27% 65% Peanut 4800 3126 32% 44% 9% 15% 65% Rhodes grass (hay) 8775 4189 17% 62% 5% 17% 48% Sesame 1170 1737 30% 43% 12% 15% 148% Hemp (grain seed) 3465 2149 38% 37% 12% 14% 62% Mean 4020 2323 31% 35% 11% 23% 58% Narrative risk analyses were conducted for the two broadacre crops with the highest GMs: cotton and forages. The cotton analysis explored the sensitivity of GMs to opportunities and challenges created by changes in cotton lint prices, crop yields and distance to the nearest gin (Table 5-4). Results show that high recent cotton prices (about $750/bale) have created a unique opportunity for those looking to establish new cotton farms in NT locations like the Roper catchment, since growers could transport cotton to distant gins or produce suboptimal yields and still generate GMs above $3000/ha. At lower cotton lint prices, a local gin becomes more important for farms to remain viable. Recent high cotton prices have reduced some of the risk involved in learning to grow cotton to its full sustainable potential in the region and while awaiting the commissioning of the new gin 30 km north of Katherine (due in 2023). At high yields and prices, the returns per ML of irrigation water may favour growing a single cotton crop per year, instead of committing limited water supplies to sequential cropping with a dry-season crop (that would likely provide lower returns per ML and be operationally difficult/risky to sequence). Table 5-4 Sensitivity of cotton crop gross margins (GMs) to variation in yield, lint prices, and distance to gin The base case is the Ngukurr Vertosol (Table 5-2) and is highlighted for comparison. The gin locations considered are a local gin near a new cotton farming region in the Roper catchment, the new gin in Katherine, and two other potential gins in the NT (Adelaide River) and north-west Queensland (Richmond). Cotton lint prices are for the average over the past decade ($580/bale), recent high prices ($750/bale), and lower prices from about 10 years ago ($450/bale). Effects of a lower yield are also tested (the 9.5 bales/ha estimated as the dry-season yield for this location versus the base case of 11 bales/ha for wet-season cropping). FREIGHT COST $/t (DISTANCE TO GIN) LINT PRICE = $450/bale LINT PRICE = $580/bale LINT PRICE = $750/bale YIELD YIELD YIELD 9.5 bales/ha 11 bales/ha 9.5 bales/ha 11 bales/ha 9.5 bales/ha 11 bales/ha $13 (50 km to local gin) 1881 2517 3116 3947 4731 5817 $79 (300 km to Katherine gin) 1526 2105 2761 3535 4376 5405 $113 (500 km to Adelaide River gin) 1342 1892 2577 3322 4192 5192 $317 (1700 km to Richmond gin) 242 619 1477 2049 3092 3919 The narrative risk analysis for irrigated forages also looked at the sensitivity of farm GMs to variations in hay price and distance to markets, but here focuses on the issues of local supply and demand (Table 5-5). Forages, such as Rhodes grass, are a forgiving first crop to grow on greenfield farms as new farmers gain experience of local cropping conditions and ameliorate virgin soils while producing a crop with a ready local market in cattle. While there are limited supplies of hay in the region, growers may be able to sell hay at a reasonable price, given the large amount of beef production in the region and challenges of maintaining livestock condition through the dry season when the quality of native pastures is low. This would particularly be the case in dry years, when the quantity and quality of native pasture is low and demand for livestock dietary supplements increases. But the scale of unmet local demand for hay limits opportunities to scale hay production without depressing local prices and/or having to sell hay further away, both of which lead to rapid declines in GMs (to below zero in many cases, Table 5-5). Another opportunity for hay is for feeding to cattle during live export which could be integrated into an existing beef enterprise to supply their own live export livestock; this would require the hay to be pelleted. Section 5.4 considers how forages could be integrated into local beef production systems for direct consumption by livestock within the same enterprise. Table 5-5 Sensitivity of forage (Rhodes grass) crop gross margins (GMs) to variation in yield and hay price The base case is the Ngukurr Vertosol (Table 5-2) and is highlighted for comparison. Transporting the hay further distances would increase opportunities for finding counter-seasonal markets paying higher prices, but this would be rapidly offset by higher freight costs. FREIGHT COST (DISTANCE TO DELIVER) HAY PRICE $150/t $250/t $350/t $20 (Local) 1142 4823 8502 $79 (300 km to Katherine) –1028 2651 6331 $317 (1700 km to Richmond) –9787 –6107 –2427 5.2.2 Horticultural crops Table 5-6 shows estimates of potential performance for a range of horticultural crop options in the Roper catchment. Upper potential GMs for annual horticulture (about $9,000 per ha per year) were less than upper potential GMs for farming perennial fruit trees (about $11,000 per ha per year). Capital costs of farm establishment and operating costs increase as the intensify of farming increases, so ultimate farm financial viability is not necessarily better for horticulture compared to broadacre crops with lower GMs (see Chapter 8). Note also that perennial horticultural crops typically require more water than annual crops because irrigation occurs for a longer period each year (mean of 9.0 versus 4.8 ML per ha per year, respectively, in Table 5-6); this also, indirectly, affects capital costs of development since perennial crops require a larger investment in water infrastructure compared to annual crops to support the same cropped area. Table 5-6 Performance metrics for horticultural options in the Roper catchment: annual applied irrigation water, crop yield and gross margin (GM) Applied irrigation water includes losses of water during application. Horticulture is most likely to occur on well-drained Kandosols. KP = Kensington Pride mangoes; PVR = new high-yielding mangoes varieties with plant variety rights (e.g. Calypso). Product unit prices listed are for the dominant top grade of produce, but total yield was apportioned among lower graded/priced categories of produce as well in calculating total revenue. Transport costs assume sales of total produce are a split among southern capital-size markets in proportion to their size. Applied irrigation water accounts for application losses assuming efficient pressurised micro irrigation systems. CROP APPLIED IRRIGATION WATER CROP YIELD PRICE PRICING UNIT VARIABLE COSTS TOTAL REVENUE GROSS MARGIN (ML/ha/y) (t/ha/y) ($/unit) (unit) ($/ha/y) ($/ha/y) ($/ha/y) Row crop fruit and vegetables, annual horticulture (less capital intensive) Rockmelon 5.3 25.0 28 15 kg tray 40,819 44,000 3,181 Watermelon 6.0 47.0 450 500 kg box 42,756 42,300 –456 Capsicum 3.2 32.0 19 8 kg carton 66,757 76,000 9,243 Onion 4.7 30.0 15 10 kg bag 35,661 41,850 6,189 Fruit trees, perennial horticulture (more capital intensive) Mango (KP) 7.8 9.3 24 7 kg tray 20,751 28,398 7,648 Mango (PVR) 7.8 17.5 21 7 kg tray 40,386 47,250 6,864 Lime 11.4 28.5 18 5 kg carton 89,451 100,890 11,439 Crop yields and GMs can vary substantially amongst varieties, as is demonstrated here for mangoes. Mango production is well-established in multiple regions of northern Australia, including in the Darwin, Douglas–Daly and Katherine regions of the NT, with a smaller area of orchards at Mataranka in the Roper catchment. For example, the well-established Kensington Pride (KP) mangoes typically produce 5 to 10 t/ha while newer varieties can produce 15 to 20 t/ha. These new varieties (such as Calypso) are likely to be released with plant variety rights (PVR) accreditation. Selection of varieties also needs to consider consumer preferences and timing of harvest relative to seasonal gaps in market supply that can offer premium prices. Prices paid for fresh fruit and vegetables can be extremely volatile (Figure 5-3) because produce is perishable and expensive to store, and regional weather patterns can disrupt target timing of supply that can result in unintended overlaps or gaps in combined supply between regions. This creates regular fluctuations between oversupply and undersupply, against inelastic consumer demand, to the extent that prices can fall so low at times that it would cost more to pick, pack and transport produce than farms receive in payment. Amongst this volatility are some counter- seasonal windows in southern markets (where prices are typically higher) that northern Australian growers can target. Figure 5-3 Fluctuations in seedless watermelon prices at Melbourne wholesale markets from April 2020 to February 2023 Source: ABARES (2023) Horticultural enterprises typically run on very narrow margins, where about 90% of gross revenue would be required just to cover variable costs of growing and marketing a crop grown in the Roper catchment (Table 5-7). This makes crop GMs extremely sensitive to fluctuations in variable costs, yield and produce prices, amplifying the effect of already volatile prices for fresh fruit and vegetables. The majority of the variable costs of horticultural production occur from harvest onwards, mainly in freight. This affords the opportunity to mitigate losses if market conditions are unfavourable at the time of harvest, since most costs can be avoided (at the expense of forgone revenue) by not picking the crop. The narrative risk analysis for horticulture used the crop with the lowest GM (watermelons: Table 5-7) to illustrate how opportunities for reducing freight costs and targeting periods of higher produce prices could improve GMs to find niches for profitable farms (Table 5-8). Reducing freight costs by finding backloading opportunities or concentrating on just the smaller closest southern capital city market of Adelaide would substantially improve GMs (to $6547 and $4039 per ha per year, respectively). The base case already assumed that growers in the Roper catchment would target the predictable seasonal component of watermelon price fluctuations (Figure 5-3), but any further opportunity to attain premiums in pricing could help convert an unprofitable baseline case into a profitable one. This example also highlights the issue that while there may be niche opportunities that allow an otherwise unprofitable enterprise to be viable, the scale of those niche For more information on this figure please contact CSIRO on enquiries@csiro.au opportunities also then limits the scale to which the industry in that location could expand, for example, there is a limit to the volume of backloading capacity at cheaper rates; only supplying produce to the closest market excludes the largest markets (e.g. accessing the larger Sydney and Melbourne markets remains nonviable except when prices are high, Table 5-8); and chasing price premiums restricts the seasonal windows into which produce is sold or restricts markets to smaller niches that target specialised product specifications. Niche opportunities are seldom scalable, particularly in horticulture, which is a contributing factor to why horticulture in any region usually involves a range of different crops (often on the same farm). Table 5-7 Breakdown of variable costs relative to revenue for horticultural crop options ‘Input’ costs are mainly for fertilisers, herbicides, and pesticides; the cost of farm ‘operations’ includes harvesting; ‘labour’ costs are the variable component (mainly seasonal workers) not covered in fixed costs (mainly permanent staff); ‘market’ costs include levies, commission and transport to the point of sale. WS = wet season; DS = dry season CROP TOTAL REVENUE TOTAL VARIABLE COSTS PERCENTAGE BREAKDOWN OF VARIABLE COSTS VARIABLE COSTS VS REVENUE ($/ha/y) ($/ha/y) INPUTS (%) OPERATIONS (%) LABOUR (%) MARKET (%) (%) Row crop fruit and vegetables, annual horticulture (less capital intensive) Rockmelon 44,000 40,819 25% 15% 12% 48% 93% Watermelon 42,300 42,756 13% 17% 15% 54% 101% Capsicum 76,000 66,757 33% 13% 12% 42% 88% Onion 44,000 35,661 10% 17% 11% 61% 81% Fruit trees, perennial horticulture (more capital intensive) Mango (KP) 28,398 20,751 21% 20% 15% 44% 73% Mango (PVR) 47,250 40,386 18% 23% 19% 40% 85% Lime 100,890 89,451 21% 21% 21% 38% 89% Mean 54,691 48,083 20% 18% 15% 47% 88% Table 5-8 Sensitivity of watermelon crop gross margins (GMs) to variation in melon prices and freight costs The base case (Table 5-2) is highlighted for comparison. FREIGHT COST MELON PRICE (PERCENTAGE DIFFERENCE FROM BASE PRICE) (MARKET LOCATION) $225 (–50%) $337 (–25%) $450 (BASE PRICE) $675 (+50%) $900 (+100%) $210/T $210/t (backloading to Adelaide) –11,642 –2,588 6,547 24,736 42,925 $263/t (close market: Adelaide) –14,150 –5,056 4,039 22,228 40,417 $359/t (all capital cities) –18,662 –9,568 –456 17,716 35,905 $387/t (Sydney) –19,978 –10,884 –1,789 16,400 34,589 $391/t (Melbourne) –20,166 –11,072 –1,977 16,212 34,401 The risk analysis also illustrates just how much farm financial metrics like GMs amplify fluctuations to input costs and commodity prices to which they are exposed. For horticulture, far more than broadacre agriculture, it is very misleading to look just at a single ‘median’ GM for the crop, because that is a poor reflection of what is going on within an enterprise. For example, a –50% to +100% variation in watermelon prices would result in theoretical annual GMs fluctuating between–$18,662/ha and $35,905/ha (Table 5-8). While, in practice, potentially negative GMs could begreatly mitigated (by not harvesting the crop), this still creates cashflow challenges in managingyears of negative returns between years of windfall profits. This amplified volatility is anothercontributor to horticultural farms often growing a mix of produce (as a means of spreading risk). For row crop production, another common way of mitigating risk is using staggered plantingthrough the season, so that subsequent harvesting and marketing are spread out over a longertarget window to smooth out some of the price volatility. 5.2.3 Plantation tree crops (silviculture) Estimates of annual performance for African mahogany and sandalwood are provided in Table 5-9. The best available estimates were used in the analyses, but information on plantation tree production in northern Australia is often commercially sensitive and/or not independently verified. The measures of performance presented, therefore, have a low degree of confidence and should be treated as broadly indicative noting that actual commercial performance could be either lower or higher. Plantation forestry has long life cycles with low-intensity management during most of the growth cycle, so variable costs typically consume less of the gross revenue (27%) than broadacre or horticultural farming (Table 5-10). However, long life cycle production systems have additional risks over annual cropping in that there is a much longer period between planting and harvest for adverse events to affect the yield quantity and/or quality, prices of inputs and harvested products could change substantially over that period, and market access and arrangements with buyers could also change. The long lags from planting to harvest also mean that potential investors need to consider other similar competing pipeline developments (that may not be obvious because they are not yet selling product) and long-term future projections of supply and demand (for when their own plantation will start to be harvested and enter supply chains). The cashflow challenges are also significant, given the long-term outlay of capital and operating costs before any revenue is generated. Carbon credits might be able to assist with some early cash flow (if the ‘average’ state of the plantation, from planting to harvest, stores more carbon than the vegetation it replaced). Table 5-9 Performance metrics for plantation tree crop options in the Roper catchment: annual applied irrigation water, crop yield and gross margin (GM) Yields are values at final harvest and for sandalwood are just for the heartwood component. Other values are annual averages assuming a 20-year life cycle of the crop (representing the idealised ultimate steady state of an operating farm that was set up with staggered plantings for a steady stream of harvests). No discounting is applied to account for the substantial timing offset between when costs are incurred and revenue is received: any investment decision would need to take that into account. African mahogany performance is for unirrigated production. CROP CROP LIFE CYCLE APPLIED IRRIGATION WATER CROP YIELD AT HARVEST PRICE PRICING UNIT VARIABLE COSTS TOTAL REVENUE GROSS MARGIN (y) (ML/ha/y) (t/ha) ($/unit) ($/ha/y) ($/ha/y) ($/ha/y) African mahogany 20 unirrigated 160 4,000 t 682 4,000 3,318 Sandalwood 20 4.7 4 8,800 t heartwood 901 1,760 859 Table 5-10 Breakdown of variable costs relative to revenue for plantation tree crop options ‘Input’ costs are mainly for fertilisers, herbicides and pesticides; the cost of farm ‘operations’ includes harvesting and labour; ‘market’ costs include levies, commission and transport to the point of sale. CROP TOTAL REVENUE TOTAL VARIABLE COSTS PERCENTAGE BREAKDOWN OF VARIABLE COSTS VARIABLE COSTS VS REVENUE ($/ha/y) ($/ha/y) INPUTS (%) OPERATIONS (%) MARKET (%) (%) African mahogany 4000 682 22% 47% 31% 17% Sandalwood 1760 901 5% 79% 16% 51% Mean 2,880 792 14% 63% 24% 27% 5.2.4 Climate change and crop production As noted previously (Section 3.1.8), mean annual rainfall in the Roper catchment is projected by most GCMs to change by less than 5%, and slightly more models project >5% wetting (29%) than >5% drying (19%). As an illustrative example of the possible impacts of climate change on futurecropping in the Roper catchment, APSIM was used to simulate grain sorghum yield and water usefor Mataranka under current (historical) climate and two contrasting 2070 scenarios from GCMprojections: a hotter drier future (based on GFDL-CM3, 3.4 °C warmer and 52 mm/year drier thancurrent), and a hotter wetter future (based on CCSM4, 2.7 °C warmer and 81 mm/year wetter thancurrent). Simulations of both climate change scenarios used CO2 levels of 725 ppm, as projectedfor a future climate under RCP 8.5 (Riahi et al., 2011). APSIM simulation results for irrigated sorghum, sown in mid-January, indicated that irrigation requirement was higher under the drier future climate scenario (Figure 5-4a), representing a median increase in annual demand for irrigation water of 40 mm (0.4 ML/ha) above the baseline scenario in a median year. Little change occurred between the irrigation requirement for the baseline and wetter future climate scenarios. Median sorghum grain yields of both wet and dry warming scenarios were lower than baseline yields, due to the detrimental effect of extreme temperatures on crop growth and development which are worse in the drier climate scenario (about 1.5 t/ha lower) than the wetter scenario (about 0.5 t/ha lower) (Figure 5-4b). (a) Irrigation water requirement (b) Yield For more information on this figure please contact CSIRO on enquiries@csiro.au 02550751000100200300400500600Probability (%) Irrigation (mm of equivalent rainfall) Current climateDry futureWet future For more information on this figure please contact CSIRO on enquiries@csiro.au 0255075100012345678910Probability (%) Yield (t/ha) Current climateDry futureWet future Figure 5-4 Probability of exceedance graphs for (a) simulated irrigation requirement (mm) and (b) grain yield (t/ha), for a grain sorghum crop grown under current climate conditions and for both a drier and wetter future climate scenario at Mataranka in the Roper catchment Note that the APSIM model results provide an estimate of crop responses to alternative climate change scenarios while holding farming practices constant. Projections of real-world impacts are constrained by incomplete knowledge of crop and farming system response to alternative environmental conditions. The effect of extreme temperatures on sensitive crop growth processes (particularly flowering) in northern Australia is not well understood, and crop responses in reality may differ from those presented here. Additionally, adaptive management changes are available to farmers that may mitigate the negative effect of climate change on crop growth (e.g. using alternative sowing times to avoid heat stress during critical growth periods, and sowing longer duration varieties (including new climate-adapted varieties that may be developed) to counteract the reduced growth periods caused by higher temperatures). Nonetheless, it is prudent for any potential developer to consider the risks that future lower yields and higher water use could have on new farm developments, and the implications of this for recovering the costs of investments. For some crops, climate change impacts could involve more than just incremental changes in yields. This is particularly the case for crops that are already at the edge of their distributional ranges for phenological triggers (such as cold triggers for flower initiation in mangoes, e.g. NESP Earth Systems and Climate Change Hub (2019)). At the lower end of impacts, phenological changes may primarily change just the timing of harvest. Depending on how the new seasonal supply coincides with altered phenology and production windows from other regions, price premiums for out-of-season production could be affected. In worse cases, flowering, pollination and/or fruit set (or other phenological progression) may be curtailed in an increasing number of years, until crop production may no longer be viable without new climate-adapted varieties. 5.3 Cropping systems This section evaluates the types of cropping systems (crop species x growing season x resource availability x management options) that are most likely to be profitable in the Roper catchment based on the analyses of farm performance above (Section 5.2), information from companion technical reports in this Assessment, and cropping knowledge from climatically analogous regions (relative to local biophysical conditions). Cropping system choices could include growing a single crop during a 12-month period, or growing greater than one, commonly referred to as sequential, double, or rotational cropping. This section covers the principles for implementing both types of cropping systems (beyond the issues for individual crops already dealt with in sections 3.4 and 5.2), with an emphasis on sequential cropping systems and the mix of cropping options that might make up a new farming area in the Roper catchment. 5.3.1 Cropping system considerations Selecting two or more crops to grow in sequence brings additional complexity, beyond the issues already discussed in finding and adapting individual cropping options for the Roper catchment. The rewards from successfully growing crops in sequence (versus single cropping) can be substantial if additional net annual revenue can be generated from the same initial capital investment (to establish the farm). To find viable mixes of cropping options for the Roper catchment, developers will need to consider each of the four key factors below. Markets Whether growing a single crop or sequential cropping, the choice of crop(s) to grow is driven by the markets and supply chains that can provide a sufficient price and reliability of demand, while being able to supply those markets at sufficient scale and affordable cost. As the price received (and scale of markets) for different crops fluctuates, so too will the crops grown. In the Roper catchment freight costs, determined by the distance to selected markets and processing facilities, will also need to be considered. A critical scale of production may be needed for a new market opportunity or supply chain to be viable (e.g. exporting grains from Darwin would require sufficient economies of scale for the required supporting port infrastructure and shipping routes to be viable). Crops such as cotton, peanut and sugarcane require a nearby processing facility. A consistent and critical scale of production is required for processing facilities to be viable (see Section 7.4). From 2023, cotton will have the advantage of local processing when a gin will be operational 30 km north of Katherine. Transport of raw cotton from the Roper catchment to this gin would go a long way to improving the viability of cotton production (Table 5-4), particularly for the western part of the Roper catchment (where it is possible to get to the gin and back in a day, without the added expense of an overnight stop). Most horticultural production from the Roper catchment would be sent to capital city markets, often using refrigerated transport. Roper catchment horticultural production would have to accept a high freight cost relative to producers in southern parts of Australia. The competitive advantage of horticultural production in the Roper catchment is that higher market prices can be achieved from ‘out of season’ production compared to large horticultural production areas in southern Australia. Annual horticultural row crops such as melons would use staggered plantings, for example, planting at fortnightly intervals over a 3-to-4-month period, to reduce risk of exposure to low market prices and to make it more likely that very high market prices would be achieved for at least some of the produce. Market considerations are covered in more detail in Section 2.2. Operations Farmers need to be skilled at managing the operational complexity of adapting crop mixes and production systems to Roper catchment environments (including soils, water resources and climates), particularly in ‘learning’ through the challenging establishment years. Sequential cropping can require a trade-off in sowing times to allow crops to be grown within a back-to-back schedule. This trade-off could lead to slightly lower yields from planting at suboptimal times. For annual horticultural crops there would be additional trade-offs in the seasonal window over which produce can be sent to market (affecting opportunities to target seasonal peaks in prices and to use staggered planting dates to mitigate risks from price fluctuations). Growing crops sequentially depends on timely transitions between the crops and selecting crops that are agronomically and operationally compatible with each other, including growing seasons that reliably fit together in the available cropping windows. In the Roper catchment’s variable and often intense wet season, rainfall increases operational risk via reduced trafficability and the subsequent limited ability to conduct timely operations. A large machinery investment (either multiple or larger machines) could increase the area that could be planted per day when fields are trafficable within a planting window. With sequential cropping, additional farm machinery and equipment may be required where there are crop-specific machinery requirements, or to help complete operations on time where there is tight scheduling between crops. Any additional capital expenditure on farm equipment would need to be balanced against the extra net farm revenue generated. Sequential cropping can also lead to a range of cumulative issues that need careful management, for example, build-up of pests, diseases and weeds; pesticide resistance, often exacerbated by sequential cropping; increased watertable depth; and soil chemical and structural decline (e.g. Piaui, 2010; Chauhan et al., 2012; Lopes et al., 2012; Lopes and Guilherme, 2016). Many of these challenges can be anticipated prior to commencement of sequential cropping. Integrated pest, weed and disease management would be essential when multiple crop species are grown in close proximity (adjacent fields or farms). Many of these pests and controls are common to several crop species where pests move between fields (e.g. aphids). Such situations are exacerbated when the growing seasons of nearby crops partially overlap or when sequential crops are grown, because both scenarios create ‘green bridges’, facilitating the continuation of pest life cycles. When herbicides are required, it is critical to avoid products that could damage a susceptible crop the following season or sequentially. Water Cropping systems are strongly influenced by the nature of water resources in terms of their costs to develop, the volume and reliability of supply, and the timing of when water is available relative to optimal planting windows (see companion technical reports on river modelling (Hughes et al., 2023), surface water storage (Petheram et al., 2022), and groundwater modelling (Knapton et al., 2023)). Sequential cropping leads to a higher annual crop water demand (versus single cropping) because: the combined period of cropping is longer; it includes growing during the Roper catchment dry season; and PAW at planting will have been depleted by the previous crop. Typically, an additional 1 ML/ha on well-drained soils, and 1.5 ML/ha on clays, is required for sequential cropping relative to the combined water requirements of growing each of those crops individually (with the same sowing times). This additional water demand needs to be accounted for in initial farm planning, particularly where on-farm water storage or dry season water extraction is required. Irrigating using surface water in the Roper catchment would face issues with the reliability and the timing of water supplies. River flows are unlikely to be sufficient to trigger pumping into on-farm storages for irrigation (i.e. to meet environmental flow and river height requirements) before mid- to late wet season (mid-February to March) in the mid-Roper catchment (Hughes et al., 2023). The timing of water availability is therefore not well suited to crops that would need to be reliably sown by March (e.g. wet-season sorghum, soybean and sesame) and would push cotton planting to the later part of the wet-season window (Figure 3-9). Late availability of water for extraction each wet season reduces the options for sequencing a second crop. The cost of developing water sources (or the price at which water is supplied to irrigators) is also critical in determining what crops are grown, because only high-value cropping options will be able to afford to pay for more expensive water (see Chapter 8). For example, in other parts of Australia that use ‘deep’ bore water (>50 m TDH) for irrigation, farming is restricted to high-value horticulture because of the high capital and pumping costs involved in accessing and distributing that water. Soils Farming systems are governed by the nature of the soil resources in terms of their scale and distribution, their proximity to water sources and supply chains, their farming constraints, the crops they can support with viable yields, and their costs to develop (see companion technical reports on digital soils mapping and land suitability (Thomas et al., 2022), flood modelling (Kim et al., 2023), and Part III of this report). The largest arable areas in the Roper catchment are loamy Kandosols of the Sturt Plateau (SGG 4.1 and 4.2, marked ‘A’ in Figure 3-8), with smaller, scattered patches of cracking clay Vertosols on the alluvial plains of the major rivers (SGG 2 and 9, marked ‘B’ in Figure 3-8). There are good analogues of these Roper catchment environments in successful irrigated farming areas in other parts of northern Australia. Katherine is indicative of farming systems and potential crops grown on well-drained loamy soils irrigated by pressurised systems, and the Ord River Irrigation Area is indicative of furrow irrigation on heavy clay soils. The good wet-season trafficability of the well-drained loamy Kandosols permits timely cropping operations and would enhance the implementation of sequential cropping systems. However, Kandosols also present some constraints for farming. Kandosols are inherently low in organic carbon, nitrogen, phosphorus, sulfur, zinc and potassium with other micronutrients often requiring supplementation (molybdenum, boron and copper). Very high fertiliser inputs are therefore required when first cultivated. Due to the high risk of leaching of soluble nutrients (e.g. nitrogen and sulfur) during the wet season, in-crop application (multiple times) of the majority of crop requirement for these nutrients is necessary (Yeates, 2001). In addition, high soil temperatures and surface crusting combined with rapid drying of the soil at seed depth reduce crop establishment and seedling vigour for many broadacre species sown during the wet season and early dry season, for example, maize, soybean and cotton (Abrecht and Bristow, 1996; Arndt et al., 1963). In contrast, the cracking clay Vertosols have poor trafficability following rainfall (Figure 3-10), inundation or irrigation, disrupting cropping operations. These constraints are compounded by complex braiding of river channels, which means that there are no large contiguous areas of suitable clay soils in the Roper catchment, and they occur instead as dissected and dispersed patches (limiting those crops where scale of production is important). Farm design is a major factor on cracking clay soils and needs to minimise flooding of fields from nearby waterways, ensure prompt runoff from fields after irrigation or rain events, and ensure that farm roads maintain access to fields. Timely in-field bed preparation can reduce delays in planting. Clay soils also have some advantages, particularly in costs of farm development by allowing lower cost surface irrigation (versus pressurised systems) and on-farm storages (where expensive dam lining can be avoided if soils contain sufficient clay) (see companion technical report on surface water storage by Petheram et al. (2022)). Clay soils also typically have greater inherent fertility than Kandosols (but initial sorption by clay means that phosphorus requirements can be high for virgin soils in the first 2 years of farming). 5.3.2 Potentially suitable cropping systems Potential crop species that could be grown as a single crop per year were identified and rated for the Roper catchment (Table 5-11) based on indicators of farm performance presented above (yields, water use and GMs: Section 5.2), together with considerations of growing season, experiences at climate-analogous locations, past research, and known market and resource limitations and opportunities. Many of these crops currently have small to medium-sized high- value markets, hence they are sensitive to Australian and international supply. Annual horticulture, cotton, peanut and forages are the most likely to generate returns that could exceed farm development and growing costs (Table 5-11). Table 5-11 Likely annual irrigated crop planting windows, suitability and viability in the Roper catchment Crops are rated as to how likely they are to be financially viable: *** = likely at low-enough development costs; ** = less likely for single cropping (at current produce prices); * S = marginal but possible in a sequential cropping system. Rating qualifiers are codes as L development limitation, M market constraint, P depends on sufficient scale and distance to local processor, and B depends on distance to and type of beef (livestock production) activity it is supporting. Farm viability is dependent on the cost at which land and water can be developed and supplied (Chapter 8). na = not applicable. CROP RATING CROP RATING Wet season (planted December to early March) Dry season (planted late March to August) Cotton *** P Annual horticulture *** M Forages *** B Cotton *** P Sugarcane *** LP Niche grains (e.g. chia, quinoa) *** SM Peanut (not on clay) *** LMP na na Mungbean ** Mungbean ** Maize ** na na Chickpea ** na na Rice ** L na na Sorghum (grain) * S Sorghum (grain) * S Soybean * S Soybean * S Sesame * S Sesame * S Due to good wet-season trafficability on loamy soils, there are many possible sequential cropping options for the Roper catchment Kandosols (Table 5-12). Given the predominance of broadleaf and legume species in many of the sequences (Table 5-12), a grass species is desirable as an early wet-season cover crop. Although annual horticulture and cotton could individually be profitable (Table 5-11), an annual sequence of the two would be very tight operationally. Cotton would be best grown from late January with the need to pick the crop by early August, then destroy cotton stubble, prepare land and remove volunteer cottons seedlings. That scheduling would make it challenging to fit in a late-season melon crop that would need to be sown by late August to early September. Similar challenges would occur with cotton followed by mungbean or grain sorghum. Fully irrigated sequential cropping on the Roper catchment Vertosols would likely be opportunistic and favour combinations of short-duration crops that can be grown when irrigation water reliability is greatest (March to October), for example, annual horticulture (melons), mungbean, chickpea, and grass forages (2 to 4 months growing-season length). Following an unirrigated (dryland) wet-season grain crop with an irrigated dry-season crop could also be possible. However, seasonally dependent soil wetting and drying would limit timely planting and the area planted, which means that farm yields between years would be very variable. Sorghum, mungbean and sesame are the species most adapted to dryland cropping due to favourable growing-season length and their tolerance to water stress and higher soil and air temperatures. The scattered distribution of suitable pockets of clay soils would limit the scale of farming at any location within the Roper catchment (which would restrict opportunities for establishing local processors). Table 5-12 Sequential cropping options for Kandosols E = early in month; L = late in month; M = middle of month SPECIES GROWING SEASON SPECIES GROWING SEASON Wet season (planted December to early March) Dry season (planted late March to August) Mungbean E-February to L-April Annual horticulture M-May to L-October Sorghum (grain) January to April Peanut (not on clay) January to April or February to May Cotton L-January to E-August Mungbean M-August to L-October Sorghum (grain) M-August to M-November Forage/silage M-August to E-November; cut then retained as wet-season cover crop Mungbean E-February to L-April Cotton E-May to E-November Mungbean Peanut Sesame Soybean E-February to L-April E-January to L-April E-January to L-April E-January to L-April Maize May to October Sesame or Sorghum (grain) January to L-April Chickpea May to August Mungbean Sesame Soybean E-February to L-April January to L-April January to L-April Grass forage/silage May to E-November; cut then retained as wet-season cover crop 5.4 Integrating forages into livestock systems 5.4.1 Baseline enterprise A baseline beef cattle enterprise was developed for the Sturt Plateau. The nominal soil type was a red earth on the Banjo land system (Pettit undated) in B–C condition. The rainfall location used was Larrimah, because of the length and quality of record. Output from the model was used for the 54 year period 1965 to 2018. The parameter estimates used to set up the model were derived from a number of published sources (Ash et al., 2018; Chudleigh et al., 2019; Cowley, 2014; Jackson et al., 2015; McLean and Holmes, 2015; Meat and Livestock Australia, 2006; Moore et al., 2021; Pettit, undated; Tyler et al., 2012) as well as local knowledge and online sources (e.g. Australian Fodder Industry Association ; Feed Central ; Feed Test ; Nutrien Ag Solutions ; Meat and Livestock Australia ). The baseline enterprise was set up in CLEM as a self-replacing cow-calf operation, focused on selling into the live export market, with castrate males sold at a minimum 300 kg liveweight (but noting that actual sale weights of individuals were typically in excess of this because there were only two sale dates per year). Any remaining castrate males were sold at the first selling month after the animals reached 25 months age. The baseline enterprise was set up with two rounds of mustering. The main muster was in May with the second muster in September. The mating system was ‘controlled’ (i.e. bulls were introduced to the cows in January and removed at the end of May). Females were retained for breeding purposes with excess breeders sold by the CLEM model to maintain maximum breeder numbers set within the model (2070 breeders in the baseline). Animal numbers were managed in order to keep the average annual utilisation rate of pasture at about 15% (i.e. cattle offtake of native pasture equal to 15% of native pasture growth). Calves were weaned relatively early at 150 kg liveweight minimum or 5 months old in May and a second weaning in September at 100 kg minimum or 5 months old. Calves were naturally weaned once they had reached an age of 8 months. Animals marked for sale were sold in May and either August or September (depending on the irrigation scenario). Remaining breeders were sold when they reached a maximum age of 120 months. All animals were fed a supplement containing nitrogen and phosphorus between May and November and a phosphorus supplement between December and April. In its current configuration, CLEM assumes that phosphorus is not limiting so that the addition of a phosphorus supplement in the model is for the purposes of accurate costing rather than altering production outputs. Broadly speaking, these enterprise characteristics can be thought of as a typical cattle enterprise within the Roper catchment with a size of about 100,000 ha and an Owner-Manager. The exception to this is the use of controlled breeding. While not unknown in the Roper catchment, it is not commonly practiced for the whole herd, although Cowley (2014) reports that 56% of producers on the Sturt Plateau carried out some controlled mating on specific cohorts within their herd. However, the concentration of calving in the CLEM model due to the controlled mating made it much easier to track cohorts of animals for comparisons across the forage and hay scenarios. Variable and overhead costs were drawn from a number of sources (see above) and then indexed from either the date of publication, or the period of collection, through to June 2021, recognising that there has been high volatility, and general increases in costs, since that time. Similarly, livestock prices in recent years have been highly volatile with Meat and Livestock Australia’s National Feeder Steer Indicator for ‘Qld Yearling Steer 280–330 kg liveweight’ reaching a maximum in January 2022 of 661 c/kg, but with an average over the 10-year period between February 2013 and February 2023 of 332 c/kg (Meat and Livestock Australia ). Clearly, such volatile livestock prices will have a big impact on enterprise profitability, with or without irrigated forages. In CLEM, liveweight prices can be set for different age and sex classes. The Sturt Plateau baseline model was set up to test the sensitivity of beef prices based on the following: • LOW beef price. Beef prices were set to 275 c/kg for males between 12 months and 24 months old, declining across age and sex classes to 134 c/kg for cows older than 108 months. For the modelled baseline scenario, this gave a price of 219 c/kg averaged across the herd and across years. • MED beef price. Beef prices were set to 350 c/kg for males between 12 months and 24 months old, declining across age and sex classes to 170 c/kg for cows older than 108 months. For the modelled baseline scenario, this gave a price of 278 c/kg averaged across the herd and across years. • HIGH beef price. Beef prices were set to 425 c/kg for males between 12 months and 24 months old, declining across age and sex classes to 206 c/kg for cows older than 108 months. For the modelled baseline scenario, this gave a price of 338 c/kg averaged across the herd and across years. A GM per AE was calculated as the total revenue from cattle sales minus total variable costs (Table 5-13). A profit metric, earnings before interest, taxes, depreciation and amortisation (EBITDA) was also calculated as total revenue minus variable and overhead costs, which allows performance to be compared independently of financing and ownership structure (McLean and Holmes, 2015) and is used in the analysis of net present value (NPV). 5.4.2 Irrigated forage and hay scenarios As outlined in Section 4.4, the use of forages and hay grown on-farm to supplement cattle is uncommon in northern Australia. Across the entire Katherine region (which includes the Roper catchment) Cowley (2014) reports ten producers making hay from improved pastures in 2010 with a median of 270 tonnes per property, although it is unclear how many producers irrigated. This is relatively small scale, at 27 t/ha this would require only 10 ha of production. There is still much to be learned about the most appropriate forage and hay species to grow, how best to manage the forages and hay to ensure high-quality feed, which cohort(s) of cattle to feed, how the feeding should be managed and which market specifications should be targeted to obtain maximum return. The number of possible combinations of options is large, making it difficult to compare scenarios. The modelling outlined in this section took a conservative approach, using three species of forage and hay crops, feeding young cattle only and keeping a constant market specification based on a minimum sale weight of 300 kg, noting that the average sale weight was greater than this because sales occurred only twice per year, in May and August (for lablab scenario, see below) or May and September (baseline and all other scenarios). The primary market was considered to be live export, either directly or through sales to backgrounders or agistment, closer to Darwin. Ideally, production would increase by allowing male animals to reach minimum selling weight at a younger age and allowing for greater weight gain during the dry season when animals on native pasture alone either lose weight, or gain very little weight. There are also potential benefits to the reproductive capacity of the herd by providing better nutrition to young females. Finally, the addition of forages and hay allows more cattle to be carried, while still maintaining a utilisation rate of native pastures at around 15%. • The approach considered three different forage and/or crop options, which were modelled in APSIM and used as an input to the CLEM modelling: • Rhodes grass, which is a perennial grass, capable of high biomass values but requiring careful management to optimise biomass and nutritive content. At high biomass levels, the nitrogen content is diluted. It also requires frequent cutting in order to maintain sufficiently high Dry Matter Digestibility. Rhodes grass is probably the most common crop grown on irrigation on cattle enterprises in northern Australia, and while there are some data available regarding its management and production in an environment broadly comparable with the Roper catchment (e.g. Giovi Agriculture, 2018) readily published data for comparison are scarce. • forage sorghum, an annual grass crop, grown over a period of 7 months. Careful management of forage sorghum is required if cattle are put on to the crop to graze it directly (i.e. stand and graze) due to the risk of prussic acid poisoning (O’Gara, 2010). • Lablab, an annual legume crop which typically provides a higher quality of feed compared to the two grasses but over a shorter period, and at lower biomass yields. A fourth option was also included, that of buying hay for feeding to weaners for the 2 months following weaning, which is a common practice in the northern grazing industry, including in the Roper catchment (Cowley 2014; Tyler 2012). The costs for producing the irrigated forages and hay were based on those which sat behind the NABSA modelling found in Ash et al (2018) and indexed to consumer price index (CPI). These costs were treated as variable costs and were on a per hectare basis. The area of forages and hay grown was determined by matching the monthly availability from the irrigated forages and hay with the nutritional demands of the cattle being fed, accepting small shortfalls from time to time. Such an approach over estimates the amount of land required for irrigation because in practice a manager can move livestock from the irrigated area to native pasture within time steps of a day and can be more flexible in approach than the model allows. A total of six scenarios were tested (all included nitrogen and phosphorus supplementation) with summarised results shown in Table 5-13: 1. Baseline. No supplementary hay or forage feeding. 2. Baseline plus hay. That is, baseline but with the addition of reasonable quality hay bought off- farm to supplement the weaners in the first 2 months after weaning. 3. Irrigated forage sorghum fed as stand and graze from June to September for all animals which were weaned and less than 24 months. In the model, the animals did not have access to native pasture (although in practice the animals would be moved between the irrigated forage sorghum and native pasture as required). In the model, the aim was to reduce irrigated forage shortfalls in any month to a minimum and balance that with the number of hectares irrigated, noting that any additional hectares incurred a cost. The irrigated area was set to 150 ha and the average of the dry matter yield in each month in the months of feeding ranged between 2338 kg/ha and 2970 kg/ha. 4. Irrigated forage sorghum grown for hay which was fed from June to September for all animals which were weaned and less than 24 months old at the time of access to the irrigated forage or hay. The animals remained in a paddock with access to native pasture and the amount of hay provided was set to 80% of their potential intake. About 20% of the hay was considered to be wasted by trampling, etc. Excess hay was sold into the market. The irrigated area was set to 160 ha and the average annual dry matter yield of hay was 12,773 kg/ha. 5. Irrigated lablab fed as stand and graze from June to August for all animals which were weaned and less than 24 months old at the time of access to the irrigated forage or hay. In the model, the animals did not have access to native pasture (although in practice the animals would be moved between the irrigated lablab and native pasture as required). In the model, the aim was to reduce irrigated forage shortfalls in any month to a minimum and balance that with the number of hectares irrigated, noting that any additional hectares incurred a cost. The irrigated area was set to 200 ha and the average of the dry matter yield in each month in the months of feeding ranged from 670 kg/ha in August to 2304 kg/ha in June. 6. Irrigated Rhodes grass grown for hay which was fed from June to September for all animals which were weaned and less than 24 months old at the time of access to the irrigated forage or hay. The animals remained in a paddock with access to native pasture and the amount of hay provided was set to 80% of their potential intake. About 20% of the hay was considered to be wasted by trampling, etc. Excess hay was sold into the market. The irrigated area was set to 60 ha and the average annual dry matter yield of hay was 31,278 kg/ha. 5.4.3 Herd and financial impacts GMs at MED beef prices for the baseline and feeding scenarios ranged between $181/AE and $226/AE (Table 5-13). This is broadly consistent with GMs found in similar studies (Ash et al., 2018; Moore et al., 2021) given the beef prices used here. A noteworthy difference in this study compared with Ash et al. (2018) and Moore et al. (2021) is that at MED beef price the GMs for the baseline ($226/AE) and the baseline plus hay ($220/AE) scenarios (i.e. the current practice in the Roper catchment) were similar to the highest of the irrigated scenarios ($224/AE for Rhodes grass hay). In the two previous studies the baseline GMs were lower than for all irrigated scenarios. Forage sorghum, either as stand and graze or fed as hay, provided the lowest GMs ($181/AE and $188/AE, respectively) in this study while lablab stand and graze ($200/AE) was between these two groupings. Considering GMs only, the decision to irrigate becomes less attractive at LOW beef prices and more attractive at HIGH beef prices. Increasing the period of forage and/or hay feeding would have increased the gross margins, since many of the significant variable costs (e.g. sowing, fertilising) would already have been incurred, while the amount of beef produced would increase. However, this may not be possible in practice without very active matching of forage availability to animal numbers. The main aim in the model was to keep forage or hay shortfalls to a minimum while trying to minimise the area of irrigation needed. At all three beef prices, total revenue was highest for the four irrigated forage or hay scenarios compared to the two baseline scenarios but the higher costs for the irrigated scenarios led to similar or lower GMs. Table 5-13 Production and financial outcomes from the different irrigated forage and beef production scenarios for a representative property on the Sturt Plateau Details for LOW, MED and HIGH beef prices are found in the text in Section 5.4.1. Scenario descriptions are found in Section 5.4.2. AE = adult equivalent; EBITDA = earnings before interest, taxes, depreciation and amortisation BASELINE BASELINE PLUS HAY FORAGE SORGHUM – STAND AND GRAZE FORAGE SORGHUM – HAY LABLAB – STAND AND GRAZE RHODES GRASS – HAY Forage/hay None Bought hay Forage sorghum Forage sorghum Lablab Rhodes grass Maximum number of breeders 2030 2070 2400 2300 2250 2290 Herd size (AE) averaged across calendar year 2752 2760 3316 3215 3167 3170 Pasture utilisation (%) 15.1 15.0 15.2 15.0 15.1 15.0 Weaning rate (%) 64 63 63 64 64 63 Mortality rate (%) 6.9 6.9 6.4 6.4 6.5 6.5 Average weight of all castrate males sold in May (kg/animal) 343 331 355 356 356 357 Average weight of 18 month old (i.e. end- November-born) castrate males sold in May (kg/animal) 307 311 354 352 350 352 Average weight of 30 month old (i.e. end- November-born) castrate males sold in May (kg/animal) 378 387 n/a n/a n/a n/a Average age of castrate males sold in May (months) 24 20 18 18 18 18 Percent of castrate male cohort aged 15 months to 19 months (compared with 27 to 31 month cohort) sold in May (%) 51 87 100 100 100 100 Beef produced per year (kg) 380,119 390,161 478,419 465,534 455,166 460,597 Gross margin ($/AE) (LOW BEEF PRICE) 142 133 95 100 113 136 Profit (EBITDA) ($) (LOW BEEF PRICE) 128,073 103,770 52,157 58,223 93,104 166,263 Gross margin ($/AE) (MED BEEF PRICE) 226 220 181 188 200 224 Profit (EBITDA) ($) (MED BEEF PRICE) 359,466 342,142 337,246 339,901 369,890 445,448 Gross margin ($/AE) (HIGH BEEF PRICE) 310 306 267 275 288 312 Profit (EBITDA) ($) (HIGH BEEF PRICE) 590,860 580,513 622,335 621,580 646,676 724,633 At MED beef prices, EBITDA was highest for Rhodes grass hay ($445,448/year). The EBITDA for all other scenarios was between $337,246/year (forage sorghum stand and graze) and $369,890/year (lablab stand and graze). While production (measured as beef sold per financial year) is clearly boosted by the introduction of irrigated forages or hay, the profitability is highly sensitive to the cost of the irrigated scenarios. An NPV analysis allows consideration of the capital costs involved in development, which is not captured in the gross margin or EBITDA. The analysis used two costings ($15,000/ha and $25,000/ha) for the capital costs of development used in the NPV analysis (Table 5-14). Table 5-14 Net present values for forage development options Details for LOW, MED and HIGH beef prices are found in Section 5.4.1. SCENARIO CAPITAL COSTS BEEF PRICE NET PRESENT VALUE Baseline plus hay for weaners $15,000 LOW -330,014 MED -341,093 HIGH -352,183 $25,000 LOW -330,014 MED -341,093 HIGH -352,183 Forage sorghum - stand and graze $15,000 LOW -3,626,168 MED -3,180,400 HIGH -2,734,632 $25,000 LOW -5,487,121 MED -5,041,352 HIGH -4,595,584 Forage sorghum - hay $15,000 LOW -3,752,944 MED -3,340,532 HIGH -2.928,110 $25,000 LOW -5,737,960 MED -5,325,548 HIGH -4,913,126 Lablab - stand and graze $15,000 LOW -4,156,222 MED -3,791,649 HIGH -3,427,076 $25,000 LOW -6,637,491 MED -6,272,919 HIGH -5,908,346 Rhodes grass - hay $15,000 LOW -463,272 MED -75,240 HIGH 312,793 $25,000 LOW -959,526 MED -571,494 HIGH -183,461 NPVs were calculated using the same assumptions as elsewhere in this Assessment (i.e. over a 40- year evaluation period at a 10% discount rate and assumed a 50:50 breakdown of assets with a 40- year lifespan and a 15-year lifespan. The NPV analyses showed that only one scenario had a positive NPV, that of Rhodes grass hay at HIGH beef price and the lower of the two development costs per ha. All other scenarios gave a negative NPV and even the one positive NPV was low ($312,793), suggesting that a decision to irrigate would need to assume beef prices well above their 10-year average in order to be viable. The EBITDAs would need to increase by about $2,000 per year per irrigated ha at the $15,000/ha development cost in order to meet the costs of development or about $3,000 per year per irrigated ha at the $25,000/ha development cost. Much of the animal production and EBITDA increases due to the irrigated forage scenarios came from the increased number of breeders which could be carried, while still keeping the utilisation rate of native pastures at about 15%. The two irrigated hay scenarios allowed the highest number of breeders to be carried, an average of 2295, compared with 2030 and 2070 for the two baseline scenarios. This flowed through to the total number of AE carried being about 15% to 20% higher than the two baseline scenarios averaged across all years. The amount of beef produced each year was about 20% to 24% higher, using the same scenario comparison. For the two baseline and two stand and graze scenarios, 100% of the revenue was from sales of beef (noting all livestock were sold on a per kg basis, including ‘cast for age’ herd bulls). For the two irrigated hay scenarios, excess hay was sold into the market. However, this contributed to less than 12% of total revenue, (i.e. the scenarios were not set up for hay sales to be a significant part of the enterprise structure). The most obvious biophysical impact of the various feeding strategies was the increase in liveweight, compared to the baseline (Figure 5-5). Feeding hay to weaners for the 2 months following weaning led to about a 24 kg increase in liveweight compared to the baseline. For the November-born male animals (i.e. born at the end of November in the model and therefore 1 month old at the end of December and 18 months old in their second May) shown in Figure 5-5 this weight differential decreased to about 20 kg at 23 months old. Both stand and graze options (forage sorghum and lablab) provided a significant increase in liveweight compared with the baseline (about 70 to 72 kg/head) for the 4 months of sorghum or the 3 months of lablab feeding. The two irrigated hay scenarios provided the highest liveweight gains, being about 84 kg heavier than the baseline animals by September of their first year, with 100% being sold in May at 18 months (Table 5-13). The average sale weight and average age of all castrate males sold at the May sales (Table 5-13) requires some explanation. The average weight for cattle in the baseline and baseline plus hay scenarios (343 kg and 331 kg) is similar to those in the four forage or hay scenarios (between 355 kg and 357 kg). The reason for this is that only 51% (baseline) and 87% (baseline plus hay) of animals were sold in their second May (15 to 19 months old) with the remainder sold in their third May (27 to 31 months old). By contrast, 100% of the animals in all four forage and hay scenarios were sold in their second May (15 to 19 months old) and their average at the May sale reflects this. For 30 month old steers in the two baseline scenarios, the average sale weights were 378 kg and 387 kg. Figure 5-5 Average liveweights for each scenario for male animals born at the end of November For the purposes of this graph, all sales were switched off, in order to show growth rates over the full period of feeding, without the removal of sale animals having an impact on the average weights of the remainder of the cohort. For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au 0100200300400500OctNovDecJanFebMarAprMayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepLiveweight (kg) Month Baseline Baseline plus weaner hay Forage sorghum graze Forage sorghum hay Lablab graze Rhodes grass hay Weaning rates were all either 63% or 64%. Given that the age or weight at weaning was identical for all scenarios there was no impact on the reproductive rate due to any different strategy for the male calves. For the female calves, their minimum age for first mating was set to 24 months (i.e. the various feeding scenarios had finished by the time nearly all females had reached that age). Those which had been fed either forages or hay increased their weights in comparison to the baseline. However, the average weight of (November-born) female cattle in all scenarios at 25 months (i.e. the month before bulls were introduced) was above 270 kg, the threshold set in the model for the minimum weight at first mating. Furthermore, the utilisation rate of around 15% suggests a conservative approach to stocking rate, with a relatively low number of animals given the carrying capacity of the vegetation. Taken together, this suggests that the weaning rate found was at the upper end of what might be expected (as an average taken over a range of years) in this environment and the addition of the feeding scenarios had no material impact on the weaning rate even though female liveweights were higher when fed hay or forages. Those heifers weighing more than 270 kg at first mating may have achieved a lifetime reproductive benefit but this did not show in the model results. While there are advantages to some form of irrigated forage or hay production, the introduction of irrigation to an existing cattle enterprise is not for the faint-hearted. The scenarios here range from an area which would require 1.5 pivots of 40 ha each to an area which would require five 40 ha pivots. A water allocation of about 0.8 GL to 1.2 GL would be required to provide sufficient irrigation water. The capital cost of development would range between $900,000 for 60 ha of Rhodes grass hay at a development cost of $15,000/ha to $5,000,000 for 200 ha of lablab at a development cost of $25,000/ha. In addition, the grazing enterprise would need to develop the expertise and knowledge required to run a successful irrigation enterprise of that scale, which is quite a different enterprise to one of grazing only. This is a constraint recognised by graziers elsewhere in northern Australia (McKellar et al 2015) and almost certainly contributes to the lack of uptake of irrigation in the Roper catchment. Part III Economics Part III analyses the scheme-scale viability of irrigated development options and economic considerations beyond the farm gate required to succeed. Chapter 6 reviews recent large dam projects in Australia for how well proposed benefits were realised in practice to elicit lessons for future developments and to provide context for the subsequent economic analysis chapters that follow. Chapter 7 provides indicators of the agricultural demand trajectories for new water in the NT and describes the types and costs of the enabling infrastructure required to support large-scale irrigated development. Chapter 8 uses a generic financial analysis approach to demonstrate the key determinants of irrigation scheme viability that investors need to balance and provides tools that allow users to estimate the viability of different development configurations. Chapter 9 quantifies the regional benefits of irrigated development using regional input–output analysis and presents an environmental input–output analysis showing how increased agricultural water use would stimulate additional demand from other water users. Part IV concludes by summarising key principles for identifying agricultural investment opportunities in the Roper catchment. 6 Lessons learned from recent Australian dam- building experiences 6.1 Introduction Large public infrastructure projects are complex investments, where it is difficult to decide in advance whether sufficient benefits will be derived to justify the costs involved. This is exacerbated by the fact that many costs are not readily apparent until after construction has begun, and it can take many years after construction is complete before it becomes clear whether the planned growth trajectory and ultimate scale of benefits is achieved. Cost–benefit analysis (CBA) has been widely used to assist decision makers in evaluating the likely net benefits from proposed projects and prioritising investments, including for transport developments (roads, railways, bridges etc.) and water resource developments (including dams, pipelines etc.). The economics part of this Assessment, therefore, begins by looking at the lessons that can be learned from past use of CBAs in large infrastructure projects. Lessons from these experiences provide context for the indicative infrastructure costs (Chapter 7), scheme financial analyses (Chapter 8) and regional benefits (Chapter 9) in the following chapters, and an opportunity to better plan and evaluate future water infrastructure projects. Despite CBA having been very widely used for a long period of time, there are far fewer examples where the estimated costs and benefits (used to justify the project) have been revisited at a later date, after the development has been constructed and in operation for a number of years. Ex-post evaluation of CBAs is important to highlight (i) whether estimates for both the scale and timing of flows of costs and benefits are achieved in practice, and (ii) opportunities for learning to improve evaluations of future projects. Such insights could improve forecasting and decision making in the future. In a review of Australian dam CBA costings estimates, Petheram and McMahon (2019) observed a strong likelihood of cost overruns compared to CBA estimates. Such biases have implications for the quality of decisions for prioritising investments in projects. The benefits of ex-post evaluation are increasingly being recognised in Australia. For example, ex- post evaluations have been completed on a sample of national road investment projects since 2005, with findings and lessons learned published to inform future ex-ante and ex-post project evaluations (BITRE, 2018). Infrastructure Australia1 has provided guidance on developing and appraising high-quality infrastructure project proposals and have encouraged wider application of post-completion reviews, that is, using ex-post comparisons between actual outcomes and the forecasts identified within the business case.2 This guidance emphasised that the 1 Infrastructure Australia is an independent statutory body established to advise governments, industry and the community on the investments, processes and reforms required to deliver better infrastructure for all Australians (for more information see https://www.infrastructureaustralia.gov.au/). 2 The most recently updated guidance, published 2021, includes information on defining problems and opportunities, identifying and analysing options, developing the business case, and preparing an economic appraisal including a CBA (for more information see https://www.infrastructureaustralia.gov.au/publications/assessment-framework). … overarching objective … is not to find fault in the implementation of the project, but to capture lessons that can improve future planning, delivery and risk mitigation… (Infrastructure Australia, 2021a:8) While there are some examples of ex-post evaluations of the costing data from public infrastructure CBAs, such cases are much more common for road and transport related developments than for water infrastructure CBAs. Of the limited examples where water resource development CBAs have been evaluated, the focus has been on exploring the accuracy of the forecast capital costs (rather than on the benefits/demand component of the CBA). Such research has shown a history of cost overruns in dam construction projects, in Australia and internationally, where a capital cost overrun is defined as the percentage difference between the actual cost of constructing the dam and the publicly stated or contracted cost immediately prior to construction. Examples include an international study that found mean cost overruns of 96% for mega-dam construction projects (Ansar et al., 2014), and an Australian-focused study that found mean cost overruns of 120% (Petheram and McMahon, 2019). Systematic biases in costings of large infrastructure projects occur both from under estimating unit costs of individual components and by omitting essential enabling infrastructure components altogether (Ansar et al., 2014; Auditor General Western Australia, 2016; Flyvbjerg et al., 2002; Odeck and Skjeseth, 1995; Wachs, 1990). For example, a review of the Ord-East Kimberley Development Plan (for expansion of the Ord irrigation system by about 15,000 ha) found that there were additional costs of $114 million to the WA Government, beyond the planned $220 million state investment in infrastructure to directly support the expansion (Auditor General Western Australia, 2016). Literature on ex-post evaluations of the forecast benefits from public infrastructure developments is scarce, particularly for water infrastructure. Only one such study was available, an international evaluation that found a sample of large dams from 52 different countries had underperformed with regards to the anticipated benefits and service delivery (World Commission on Dams, 2000a). This study noted that ‘Large dams designed to deliver irrigation services have typically fallen short of physical targets, did not recover their costs and have been less profitable in economic terms than expected’ (World Commission on Dams, 2000b:xxxi). This study’s findings included that the forecasting of future demand for water from dam developments around the world was frequently inaccurate, and, with regards to irrigation dams in particular, that the estimates of demand tended to be overstated. Given that (i) there is limited research exploring the accuracy of benefits/demand forecasting for CBA compared to evaluations of the costing component, and (ii) there are indications that demand forecasts are often poorly related to real water needs, this report focuses on the less researched element of CBAs: the demand for increased water supplies and their associated benefits. Within Australia, ex-post evaluations of the accuracy of water demand and benefit forecasting in CBA supporting water resource developments have not historically been prepared. However, the importance of such evaluations is increasingly being recognised. For example, the 2021 update to the Infrastructure Australia Assessment Framework recommends post-completion reviews (PCRs) for all major infrastructure projects and requires PCRs for all projects where Infrastructure Australia assessed the original business case (Infrastructure Australia, 2021a, p. 8). Further, the recently published National Water Grid Investment Framework (DCCEEW, 2022) specifies that agreement to conduct a post-completion project evaluation, in consultation with the National Water Grid, will be an Australian Government condition for investment in future water infrastructure projects. The review in this chapter used a sample of large and recently constructed Australian dams based on publicly available information and reports. This review provides baseline information regarding the ex-ante and ex-post information available for recent water resource developments, and highlights lessons for possible ways of improving future water infrastructure planning and assessments. The review also provides context for interpreting CBAs from independent analyses (such as those presented in Chapter 8 and those that adhere to the Infrastructure Australia technical guidelines for economic appraisal (Infrastructure Australia, 2021b)) relative to those from project proponents (where there may be selection biases and incentives to present scenarios where benefits exceed costs). Methods for this review are set out in Section 6.2, the summary of the case studies is described in Section 6.3, and key findings are set out in Section 6.4. 6.2 Methods and case study selection The Australian National Committee on Large Dams (ANCOLD) websiteFigure 6-1, and summary information on each dam is provided in Table 6-1. 3 lists 570 dams, ranging in capacity from 11 ML to 12,400 GL and constructed between 1857 and 2012 to provide water for domestic, industrial and agricultural use, in addition to hydro-electricity generation and flood mitigation. Based on criteria of having completed construction in 2000 or later, and having a capacity in excess of 40 GL, five developments were selected for review. The geographic locations of the five dams are show in 3 https://www.ancold.org.au For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au Figure 6-1 Map showing locations of the five case study dams used in this review The case study dams are numbered in blue as 1: New Harvey Dam, 2: Paradise Dam, 3: Meander Dam, 4: Wyaralong Dam, and 5: Enlarged Cotter Dam. Table 6-1 Summary characteristics of the five dams used in this review Dam completion date and capacity sourced from ANCOLD website (https://www.ancold.org.au). Documents reviewed for each dam are listed in Table 6-2 NEW HARVEY DAM† PARADISE DAM MEANDER DAM WYARALONG DAM ENLARGED COTTER DAM State/territory WA Qld Tas Qld ACT Date completed 2002 2005 2008 2011 2012 Capacity 59 GL 300 GL 43 GL‡ 103 GL 78 GL New dam or redevelopment of existing dam Replaces Harvey Weir (built 1916, extended 1931), capacity of ~10 GL New New New Replaces original Cotter Dam (built 1915, extended 1951), capacity of ~4 GL Primary use(s) proposed for water from dam Irrigated agriculture Irrigated agriculture, water supply Irrigated agriculture, environmental flows, hydro power Water supply to South East Queensland Water supply for Canberra Type of key project documents used for this review Proposed water allocation plans (no CBA available) CBA and economic impact assessment CBA Environmental Impact Statement (EIS) (no CBA available) EIS (which included CBA information but actual CBA report unavailable) †Constructed as part of wider Stirling-Harvey redevelopment scheme, the New Harvey Dam was to supply water to irrigated agriculture to free up water from the Stirling Dam to increase urban water supply ‡This Dam is listed on ANCOLD as having capacity of 24 GL. The Dam was constructed with a capacity of 43 GL but designed to make 24 GL of water available for irrigation. For each case study, publicly available documentation was obtained from government and other sources relating to: (i) initial plans and approval processes for the dams including environmental impact statements (EISs), economic justifications (including CBAs), sustainable water strategies etc.; and (ii) post-construction publications containing relevant information regarding the use of, and benefit flow from, the dams. Based on the information sourced, the forecast water demand in the project proposal was compared to the actual demand that emerged post construction, providing an ex-post evaluation of the accuracy of demand and benefit forecasts. Overall, the limited availability of data in the public domain (regarding specific quantity, timing and purpose) prevented a precise quantitative analysis of demand forecast (in)accuracies; instead, the information was qualitatively assessed to determine the likelihood of demand having been under or over estimated in the original dam proposals. This review does not seek to provide a systematic review of all relevant literature but focuses on those recent dams for which the best information is publicly available and most relevant to current water infrastructure planning in Australia. While the small sample size is a limitation, it is sufficient to highlight some of the most important CBA principles learned from recent past experience. A further key limitation relates to the limited availability of detailed reporting on dam developments, both ex-ante and ex-post, in the public domain. This is partially due to the commercialisation of the water authorities in Australia, and consequentially, the commercial-in- confidence nature of much of the data, which is compounded by difficulties in sourcing historical documents that may have been issued in limited hard copy rather than made widely available. This Assessment has focused purely on existing public documents and has not sought to collect independent primary data on actual water usage and benefits over time. 6.3 Proposed and realised outcomes for each case study development The context and summary of outcomes for each of the five dams selected are set out below. Further details on the expectations and outcomes arising from each development are presented in Table 6-2. Table 6-2 Summary of the expectations and reported outcomes for each dam reviewed DAM DEVELOPMENT PRIMARY OBJECTIVE FORECAST WATER DEMAND AND BENEFITS FROM DEVELOPMENT ACTUAL WATER DEMAND AND BENEFITS FROM DEVELOPMENT TO DATE New Harvey Dam as major component of the wider Stirling- Harvey redevelopment scheme (2002) To improve water security for the region while enabling continuation of irrigated agriculture in region Providing additional 34 GL of potable water to supply needs of estimated 350,000 people1 Supplying 68 GL of existing irrigation licence allocations, to continue to provide irrigation water for dairy and beef pasture and fodder, and some horticulture, whose use in preceding years was around 60 GL per annum2 Urban water demand within the region has grown, and has been supported by this development, alongside increasing reliance on the use of two seawater desalination plants (which started production in 2006 and 2011) and now produce 30% of Perth’s water supply11 Analysis comparing pre- and post-dam agriculture in region found small switch towards horticulture, and away from pasture-based agriculture, and that water usage by irrigators has not declined12 Water continues to be supplied to irrigators, and this water is now traded via an active online trading market13 Paradise Dam (2005) To stimulate regional economic growth and job creation via stimulating irrigated agriculture, plus provide additional water supply Provide 20 GL/y high-priority water for urban and industrial use3 Provide 124.2 GL/y medium-priority water for medium-priority use, being mainly agricultural irrigation, expected to be fully taken up by year 19 following commencement of construction of the dam (i.e. 2020)3 Forecast take up of additional irrigation water (i) in the short term by existing sugarcane farmers, alongside existing livestock industries, and (ii) in the longer term used to satisfy demand from growth of higher margin intensive horticultural crops (vegetables, citrus, other fruit and nuts), plus chicory, anticipated as a new crop for the region3 Anticipated substantial agricultural production increases, of 25% for sugarcane, and 5 to 6 fold increase for horticulture4 Flow-on benefits of construction and operation of a chicory plant, and a new sugarcane and bagasse pulp processing plant, anticipated as a consequence of development but not included in CBA3 Around 2.9 GL of high-priority water rights has been reported as being taken up by 201914 Around 15 GL of medium-priority water reported as taken up by 201415, and 24GL taken up by 201914. 24 GL represents ~19% of total anticipated yield of 124 GL anticipated by 20203 As of 2014 it was reported that the development and diversification of cropping had not matched expectations15 Rather than new sugar mill being opened as anticipated, an existing mill closed in 200516, and a further mill closure was announced in 202017 Predicted development of chicory plant in the region does not appear to have materialised Recent reports have argued that the anticipated increases in water demand for irrigated agriculture will materialise, but over a longer time frame than that anticipated in the original CBA18 Meander Dam (2008) To support environmental flows, enable expansion of irrigated agriculture, and produce hydro power Support increased environmental flows improving ecological health of river5 Recover agricultural water allocations that would be lost to improved environmental flows, and provide additional water allocations to irrigated agriculture, to be utilised by grazing (dairy), and cropping (poppies, potatoes, peas, beans, broccoli, carrots, onions and other crops), totalling around 24 GL/y by year 18 of project in most likely scenario5 Enable electricity generation from mini-hydro development of 10,000 MW hours per year5 Water supply for irrigation commenced during 2007–08 season, and increased substantially by 2008–0919 Almost immediately following construction was completed, further construction of an extension including four pipelines commenced, further increasing the water available from 24 GL to 28.8 GL19 The system operator now reports that 240 irrigators hold licences to access this water, and that 100% of licences have been sold20. While this supports the proposition that demand did exist to support the original scheme it is impossible to test if the demand would have arisen without the additional pipeline developments Other benefits including flood mitigation, improved water quality/reduced turbidity, increased recreation opportunities5 Mini-hydro scheme began generating and exporting electricity in 2008 in line with plan19 Wyaralong Dam (2011) To address water security concerns Water demand within SEQ was predicted to rise by almost 50% by 2026, continuing to grow, with the expectation that demand for water would have more than doubled by 20516 Dam expected to provide 18 GL/y, and to provide 8% of anticipated yield from all supply initiatives by 20156 When operating in conjunction with Cedar Grove weir and Bromelton Offstream Storage, development expected to provide up to 26 GL/y of additional water, an amount sufficient to meet the needs of more than 300,000 people7 Having been approved during drought which subsequently broke, and coupled with reduced per capita water demand that has endured post- drought, the dam’s water has not yet been required Currently is being used as a recreation facility 2016–2046 Water Security Plan identified need for further infrastructure including water treatment plant, pipelines and pumpstations, before dam could provide water to the local community or to the grid21 Construction of the required water treatment plan is in the early planning stage22 Enlarged Cotter Dam (2012) To address water security concerns Dam enlargement to increase capacity by 72 GL, and increase water storage capacity in ACT by 35%8 Plan would address water needs of anticipated population rises to 405,000 by 2017 and 500,000 by 20329 Plan would reduce the times when severe water restrictions would need to be applied, estimated range from $7m/y at stage 1 to $324.1m/y at stage 4 restrictions9 Further benefits anticipated during construction from employment opportunities, and from operation phase through improved workforce skills, enhanced infrastructure and amenities10 Population levels appear to be growing as anticipated23 Water consumption per head has reduced further than anticipated, moderated by voluntary and permanent water conservation measures24 (similar to stage 1 restrictions elsewhere)25 No temporary restrictions have been required since completion of this Dam, and the 2013 Murrumbidgee to Googong pipeline The developments, alongside efforts to manage consumption, are considered sufficient to supply ‘unrestricted demand for the ACT and Queanbeyan 95% of the time until at least 2030’26 CBA = cost–benefit analysis; SEQ = South East Queensland Sources for information: 1 Water and Rivers Commission (2000) 2 Water and Rivers Commission (1998) 3 NECG (2001) 4 National Competition Council (2003) 5 MJA (2003) 6 QWI (2007) 7 Queensland Government (2009) 8 https://www.iconwater.com.au/water-education/our-projects/water-security-projects/enlarged-cotter-dam.aspx 9 ACTEW (2009b) 10 ACTEW (2009a) 11 https://www.watercorporation.com.au/Our-water/Desalination 12 Resource Economics Unit (2007) 13 https://www.harveywater.com.au/ 14 Sunwater (2019) 15 Mainstream Economics and Policy (2014) 16 https://www.abc.net.au/news/2005-02-04/fairymead-sugar-mill-to-shut-doors/630858 17 https://www.abc.net.au/news/rural/2020-10-23/bingera-sugar-mill-closure-bundaberg-sugar-cane/12808948 18 Adept Economics (2020) 19 Davey and Maynard (2010) 20 https://www.tasmanianirrigation.com.au/schemes/greater-meander, information on irrigators and entitlements sold based on accessing webpage 12 July 2022 21 SEQWater (2017) 22 https://www.seqwater.com.au/news/wyaralong-water-treatment-plant 23 Based on population level at 2020 https://dbr.abs.gov.au/region.html?lyr=gccsa&rgn=8ACTE, and predictions for 2032 https://www.abs.gov.au/statistics/people/population/population-projections-australia/latest- release#australian-capital-territory 24 Icon Water (2018) 25 https://www.iconwater.com.au/my-home/saving-water/when-can-i-water/permanent-water-conservation-measures.aspx 26 ACT Government (2014, p. 20) New Harvey Dam The construction of the New Harvey Dam formed part of the wider Stirling-Harvey redevelopment scheme. It was designed to enable irrigated agriculture within the region to continue with business as usual while supplying significant additional water to the integrated water supply scheme for Perth and other towns in the region, and to meet the anticipated demand for high-priority water resulting from expected population growth in Perth and surrounding regions. Since completion of the development, the objectives appear to have been broadly met, with water use for irrigated agriculture being maintained while priority water uses have been met from a number of sources including the New Harvey Dam and from the construction of two desalination plants in the region. Overall, agricultural demand for irrigation water does appear to have met target. Paradise Dam This water infrastructure development was designed to facilitate regional development and to encourage wealth and job creation within the Burnett region, one of the least affluent and least developed locations across Queensland. The project comprised constructing Paradise Dam while also constructing some new weirs in the region and augmenting others. The development was predicted to stimulate substantial increases in agricultural production, to meet anticipated demand generated from both population growth across South East Queensland (SEQ) and export markets, and to contribute some high-priority water to the region. However, the development experienced difficulties following major flood events in 2011 and 2013 when structural problems with the construction of the dam wall emerged, requiring capacity to be restricted. Significant rectification works have been approved with early works expected to commence in 2023.4 Demand for water has emerged more slowly than anticipated in the CBA, revealing considerable shortfalls between actual and predicted water demand; further, anticipated knock-on developments (such as the construction of a chicory processing plant and a new cane and pulp mill) have also failed to materialise. A recent analysis (Adept Economics, 2020) has critiqued the assumptions in the original CBA as being overoptimistic regarding the trajectory of water demand, and to have failed to take account of possible climate variability. Overall, demand for water does not appear to have met target. 4 https://www.sunwater.com.au/projects/paradise-dam-improvement-project/ 5 https://www.tasmanianirrigation.com.au/schemes/greater-meander, information on irrigators and entitlements sold based on accessing webpage 12 July 2022 Meander Dam The proposal to dam the Meander River, prompted by a need to support environmental flows, described benefits including providing additional water for expansion of irrigated agriculture, to enable electricity generation from a mini-hydro development and other benefits including flood mitigation, savings from improved water quality/reduced turbidity, and improved recreation opportunities. Reviewing the actual experience, it appears these benefits have arisen, however, additional pipeline construction works (unforeseen in the original CBA) were required to enable farmers across the region to access the additional water. As of 2022, 100% of the irrigation licences available for the increased irrigation water have been sold.5 Thus, the predicted water demand in the CBA appears to have been reasonable but did require additional capital investment to enable the predictions to become reality. Overall, demand for water does appear to have met target, but additional enabling infrastructure spend was required to facilitate this. Wyaralong Dam The Wyaralong Dam was proposed as a means to improve the water security for the people of SEQ, stimulated by the millennium drought and the growing population in the region. A multi- faceted strategy was developed to address the predicted demand growth to provide water security for SEQ for the forecast period up to 2026. Key components of this strategy included traditional water infrastructure developments (dams and pipelines) and the development of climate-resilient water sources (desalination and recycled water projects). While a number of other components of the plan now contribute to the water supply of the region, the Wyaralong Dam has to date supplied no water and is currently used as a recreation facility. While the lack of demand for water from the dam can be partly attributed to the end of the severe drought and moderated by reductions in water consumption per head, post construction the dam was found to be unable to supply water of sufficient quality to the local community and to the grid without the construction of a water treatment plant, pipelines and pump stations. This capital investment was not within the initial project plans or CBA. Construction of the Wyaralong water treatment plant is reported to be in the early planning stages.6 It would appear that while the demand for water in the SEQ region has grown, and continues to grow, growth has been slower than expected and to date has been met from sources other than the Wyaralong Dam. While the dam may be used in the future as a water source for the region, this cannot occur without construction of additional infrastructure beyond that included in the original CBA. Overall, demand for water from this dam does not appear to have met target, and additional enabling infrastructure spend is required before this can occur. 6 https://www.seqwater.com.au/news/wyaralong-water-treatment-plant Enlarged Cotter Dam Against a background of population growth within Canberra and the ACT more widely, and increasing climate uncertainty, the ACT Government considered a range of initiatives to help secure Canberra and the region’s water supply into the future and unlock the potential to provide water through extended drought periods. The water security plan describes how water supply needed to be increased to meet the assumed population increase, and to reduce the times where severe water restrictions were required, estimating the economic cost of time spent on water restrictions to be $7 million per year for stage 1 temporary restrictions, rising to $324.1 million per year for stage 4 water restrictions (ACTEW, 2009b). Beyond this dam, the region has taken other significant steps to secure water, including constructing the Murrumbidgee to Googong pipeline (completed in 2013), taking steps towards water trading with other parts of the Murray–Darling Basin, and seeking to reduce consumption per capita (ACT Government, 2014). While the region has experienced population growth broadly in line with that forecast, the impact of this on the total demand for water has been moderated by reductions in water consumption per head (both voluntary and driven by permanent water conservation measures) over and above the reductions forecast. Since late 2010 the use of temporary water restrictions of differing levels has been replaced by permanent, year-round measures, similar to stage 1 temporary restrictions in other regions of Australia.7 No further restrictions, over and above these permanent measures have yet been required. The net impact of these factors suggest that the predicted increased demand for water has not been realised to the extent anticipated, due to the success of steps taken to moderate consumption. However, the dam has clearly made a contribution towards the objective of reducing the risk of having to implement severe water restriction measures, and thus has delivered this expected benefit. Overall, while demand for water has increased, the increase is less than anticipated due to the greater than anticipated success of encouraging voluntary water conservation measures. 7 https://www.iconwater.com.au/my-home/saving-water/when-can-i-water/permanent-water-conservation-measures.aspx 6.4 Key lessons Dams provide a complex mix of market and non-market benefits The contexts for proposing new dam developments vary significantly. The five case studies were not just geographically different but were also underpinned by different motivations and priorities. Some focused primarily on irrigated agriculture and regional economic development (including job creation), others focused primarily on providing water security, while others offered a mix of objectives. The dam developments were not always justified by purely financial (and hence easy to monetise) benefits. Non-market, non-financial, and social objectives (including water security, food security etc.) were frequently cited, but are harder to monetise and evaluate directly in CBAs. The prevailing circumstances at the time of the proposal also influenced the way that benefits were framed. For example, urban water security was prioritised more at times of drought. The term ‘monetise’ is defined here to mean assigning a dollar value to a (dis)benefit for purposes of quantitative analysis (without implying that it would necessarily be tradable in a financial transaction). The five case studies in this review were justified by a complex mix of market and non-market benefits. Some adverse impacts were also noted, hence named ‘disbenefits’, relating to reduced recreation opportunities necessary to protect water quality in the dam. While it is never simple to estimate future net benefit flows, quantifying market benefits for inclusion in proposal documents (which included CBAs in some but not all of the case studies selected) is less complex than quantifying non-market benefits. The market and non-market (dis)benefits and the approaches taken towards evaluating these are summarised in Table 6-3. Market benefits considered in the proposals included supporting and/or expanding irrigated agriculture (Stirling-Harvey redevelopment, Paradise, Meander) and for hydro-electric power (Meander). The monetary value of such expected benefits can be estimated (by predicting volume of demand that could be met from the dam development each year and the likely market prices) and included in the proposal and/or CBA. Table 6-3 Benefits (and disbenefits) included in proposals justifying the five dams reviewed BENEFITS (AND DISBENEFITS) INCLUDED IN PROPOSALS TO JUSTIFY DAM MARKET OR NON- MARKET CASE STUDY WHICH INCLUDED THIS (DIS)BENEFIT (DIS)BENEFIT INCLUDED IN NARRATIVE (DIS)BENEFIT QUANTIFIED Benefits Provide irrigation water for agriculture Market Stirling-Harvey1 Yes No2 Paradise Yes Yes4 Meander Yes Yes4 Generate hydro-electric power Market Meander Yes Yes4 Stimulate regional economic growth/job creation by encouraging new/existing industries to develop beyond direct impact on current farming activity Market Paradise Yes5 Yes6 Wyaralong Yes No12 Enlarged Cotter Yes10 No Provide additional urban water supply Market/ non-market Stirling-Harvey Yes No Paradise Yes Yes4 Wyaralong Yes No Enlarged Cotter Yes Yes11 Increase water security/improve reliability of water supply for the future Non-market Stirling-Harvey Yes No Paradise Yes7 No Meander Yes13 No Wyaralong Yes No Enlarged Cotter Yes Yes11 Support environmental flows improving ecological health of the river Non-market Meander Yes Yes8 Mitigate floods Non-market Paradise Yes No Reduce salinity Non-market Paradise Yes No Mitigate floods and related reduced water treatment costs Non-market Meander Yes Yes9 Increase recreation opportunities Market and non-market Stirling-Harvey Yes No Paradise Yes No Meander Yes No Wyaralong Yes No Disbenefits Reduce recreation opportunities to protect drinking water quality Market and non-market Stirling-Harvey Yes3 No Enlarged Cotter Yes14 No 1 As part of wider Stirling-Harvey redevelopment scheme, the New Harvey Dam was to supply water to irrigated agriculture to free up water from the Stirling Dam to increase urban water supply. As the two dams form an integrated scheme, the combined benefits are reflected in this table rather than a simple focus on either dam individually. 2 Project sought to maintain current water supply available for irrigated agriculture by replacing one source for another, rather than increasing quantity/value of agriculture in region. 3 Recreation disbenefits include applying additional restrictions to, or preventing, leisure activities both on water (marroning, fishing, swimming) and on land within the catchment (horse riding, motor rally, trail bikes, off-road driving, hunting). 4 Discounted cashflow estimated, based on quantified net benefit flow, and presented in CBA. 5 Includes anticipated new cane/bagasse pulp mill and new chicory processing plant. 6 Benefit quantified using I–O analysis but not included in CBA calculation of NPV. 7 Water security is not a key focus of this proposal, but discussion does note that demand for water will increase as the towns and communities in the region expand. 8 Proposal acknowledges that should the dam not be built, current temporary irrigated agriculture water licences would need to be revoked to protect the environmental health of the river. The value of this water to agriculture is incorporated in the CBA, recognising that as the development satisfies the environmental need without sacrificing this flow, then this value is a proxy for this benefit. 9 Estimated value of avoided damages due to reduced flooding, and reduced water treatment costs due to less need for treating turbidity and bacteriological problems. 10 Includes economic growth from improved workforce skills, improved capacity and capability of local firms, enhanced infrastructure and amenities. 11 Based on estimating the economic cost of imposing different levels of temporary water restrictions (from stage 1 to stage 4) and the expected reduction in time when such restrictions were expected to be required. 12 EIS quantified expected loss in regional Gross Domestic Product (GDP) and jobs if water supply were to fail due to failure to invest in water security project. 13 Increased water security and reliability of supply is described as an important benefit but framed via the lens of supporting agricultural and industrial uses rather than relating to urban drinking water. 14 Potential recreation disbenefits described but mitigation opportunities were considered such that only minor disbenefits were considered likely. Non-market benefits are more complex to quantify in biophysical and/or monetary terms, and could include motivations such as national security, water security, food security, (re)generation of a socio-economically disadvantaged/declining region, increased resilience etc. The particular non- market benefits anticipated in the five case studies varied significantly with regards to both the particular benefits considered and the estimation methods used. In some examples, attempts were made to quantify such benefits, while other examples discussed the anticipated benefit in the narrative text without attempting to estimate a monetary value for the benefit flow. Benefits that are particularly difficult to reflect in a CBA are those such as offering improved water security against changes in future rainfall patterns or periods of extreme drought. In these instances, the development is in effect like buying insurance – the benefits are intermittent and only apparent in times when a large adverse impact is avoided/mitigated. Estimating the timing and impact of events such as drought using stochastic analyses are particularly prone to error, and so too is estimating the ‘insurance’ benefit (in both ex-post and post-ante analyses) of having additional dam water for such periods. Decision support tools such as NPV and CBA are poorly suited to capturing the nuances of such vital, but intermittent, benefits. The case studies, all set in very different contexts, illustrate the challenges in quantifying different benefits (especially the intangible and non-market benefits where the dam acts as a form of insurance). Each dam proposal is trying to forecast the future, where the forecasts are hard to quantify, and harder for some objectives and contexts than others. This is particularly the case for projects where the primary development motivations are hard-to-value objectives (such as improved water security). It is likely that the values included in the analysis will in effect be a more easily monetised benefit that will serve as a proxy for the true underlying benefit. For example, it is easier to estimate the monetary impact of imposing specific water restrictions on businesses operating in a region than to estimate the monetary value of a lack of drinking water in a community at some unknown future date. These issues mean that a single financial metric from CBA is unlikely to be adequate for comparisons across projects in different contexts where different subsets of the full range of benefits may be captured in quantitative analyses. Additional information on the context and non- monetised costs and benefits would ideally be required. Systematic bias in overstating the anticipated net benefits The five cases studies used in this review all revealed varying degrees of discrepancies between forecast and realised future demand for water. This is not surprising; forecasting the future is difficult for the simplest of events, and more so for complex projects with a long useful life. Evaluations of water infrastructure projects need to consider the biophysical (e.g. rainfall, evaporation, river flow, extreme weather events including drought and floods), and socio- economic (e.g. population growth, changes to the mix of industries and agricultural products, economic growth and inflation) outcomes over many years. Furthermore, forecasts are complicated by needing to estimate both the timing and scale of benefits, including how quickly actual demand grows towards its potential. If the complexity of the task were the primary cause of forecasting errors, then an equal mix of under and overoptimistic estimates would be expected. However, the forecasts in the case studies tended to be consistently optimistic, favouring higher benefit–cost ratios (BCRs). This reflected optimism in both the forecast scale of demand once the developments reached their full potential and the rate at which that potential was achieved. Both biases contribute to over estimating the NPV of a project. International literature for ex-post evaluation of investments in public infrastructure provides several possible explanations for errors and biases in CBAs (for projects such as railways, bridges, tunnels and roads, in addition to dams) which are likely to be relevant here (Flyvbjerg et al., 2002, 2005; Nicolaisen and Driscoll, 2014; van Wee, 2007; World Commission on Dams, 2000b). First, there is a risk with all reviews such as this that success bias could influence findings. By definition, ex-post evaluation can only be done on project proposals that have been successful in attracting investment and where the developments actually go ahead. A project where net benefits in the CBA are overstated is far more likely to have been selected than a project that understated net benefits. Thus, a review of ‘successful’ projects is more likely to find over- rather than understated benefits. Secondly, systematic bias can be introduced by the views of and pressures on those preparing the CBA. For example, when an advocate/proponent of the project controls a CBA process, estimates of benefits/demand and of costs may be influenced (deliberately or sub-consciously) by a motivation to achieve a desired outcome. That is, the estimated NPV can be influenced by the decisions made regarding which costs and benefits to include in the analysis, and the scale and timing of those costs/benefits, resulting in inflated benefits and/or understated costs (where the desire is to facilitate the project, or the reverse bias if the desire is to obstruct a project). CBAs prepared by independent analysts (agnostic about whether the project proceeds) may appear pessimistic in comparison to those that are prepared by proponents to meet project selection criteria. When reading and comparing CBAs it is therefore important to consider the context within which they were prepared as this can have a substantial influence on their results. Summary of key issues This review has highlighted a number of issues with historical use of CBAs for recently built dams in Australia together with ways that they could be more rigorously addressed (Table 6-4). These issues arise both because of the complexity of the forecasts and estimates required to plan large infrastructure projects, and because of pressures on proponents that can introduce systematic biases. However, this report acknowledges that flaws with the use of CBAs in large public infrastructure investment decisions are not unique to regional Australia nor water infrastructure alone – they are systemic and occur in many different types of infrastructure globally. Under such circumstances it would be inequitable to apply more rigor to CBAs only for some select investments, geographic regions and infrastructure classes, before the same standards are routinely applied in all cases. And there is no incentive for individual proponents to apply more rigor to CBAs if those proposals would suffer from unfavourable comparisons to alternative/competing investments with exaggerated CBRs. In the short term the main value of the information provided here is to assist in more critically interpreting and evaluating CBAs, warts and all, so that more informed decisions can be made about the likely viability (and relative ranking) of projects in practice. In particular, it highlights several aspects of CBAs where the claims of proponents warrant critical scrutiny. In the longer term, this analysis supports many of the similar issues raised in past review cycles of Infrastructure Australia’s CBA best-practice guidelines and the recommendations that are being progressively added to those guidelines to improve how large public investments are evaluated (Infrastructure Australia, 2021a, 2021b). Table 6-4 Summary of key issues and potential improvements arising from a review of recent dam developments KEY ISSUE POTENTIAL IMPROVEMENTS 1 Lack of clear documentary evidence regarding the actual outcome of dam developments compared to assumptions made in ex-ante proposals, EISs and CBAs. Ex-post evaluations or post-completion reviews have either not been prepared, or not made publicly available. Conducting ex-post evaluations of developments and making these publicly available (as recommended by 2021 guidance from Infrastructure Australia, and in the 2022 National Water Grid Investment Framework) would enable lessons learned to be shared and to benefit future developments. 2 Predicted increases in water demand from specific developments generally do not appear to arise at the scale and/or within the time frame forecast. While the reasons for this are varied and context-dependent there does appear to be a systematic bias towards over estimating the magnitude and rate at which new benefit would flow. Recognising the tendency towards a systematic bias of over stating benefits and under stating costs, CBAs in project proposals could be improved by (i) further efforts to present unbiased financial analysis (e.g. independent review) and ensuring appropriate sensitivity analysis is included in all proposals, (ii) developing broadly applicable realistically achievable benchmarks for evaluating proponents’ assumptions and financial performance claims, (iii) using past experiences and lessons learned from previous projects with similar context to inform the analysis presented in the proposals (building on Issue 1 above), and (iv) presenting a like-for-like comparison of CBRs for the proposed case vs standard alternatives (such as water buybacks or a smaller dam, possibly better matched to realistic future demand). 3 The systematic bias towards optimism in proposals is exacerbated by mismatches of forecast demand and the full supporting infrastructure required to enable this demand to be realised, resulting in additional capital investment (pipelines, treatment plants etc.) being required that was not costed in the original proposal. The same improvements for Issue 2 in recognising and addressing inherent bias apply here. 4 Developments are justified based on a complex mix of multiple market and non-market benefits, many of which are hard to monetise and capture in a single NPV figure. CBAs could be improved by presenting clear information on the full portfolio of benefits (and costs and disbenefits) anticipated to arise from a project. While the quantitative part of the CBA would analyse the easily monetised costs and benefits (with metrics such as CBR and NPV), benefits that are hard to monetise could be formally presented alongside. This information would be presented in whatever form is most appropriate to the magnitude and nature of that particular benefit. This presentation would enable the relative importance of each element of the mix to be weighed and given appropriate consideration, rather than attention being focused on a single NPV figure, which may have omitted key elements of the project. KEY ISSUE POTENTIAL IMPROVEMENTS 5 Improved water security and reliability of supply is often the most important benefit offered by dam developments, while also being the hardest to monetise. Dams provide a form of insurance against the risk that water may not be available when needed in future. Assessing the value of this insurance requires consideration to be given to the cost of lack of water supply when needed, and the likelihood that this could occur. CBAs could be improved by providing clear information on exactly how the development will serve to improve water security, the likelihood that such insurance will be required (i.e. an estimate of the risk), and the estimated social and economic impacts if the insurance was not there when required. Such information could be presented alongside, and given equal precedence to, other information regarding the proposal including the estimated NPV, rather than attempts be made to ‘force’ the benefit into an NPV calculation which is ill equipped to deal with such a benefit. 7 New infrastructure demand and costs 7.1 Introduction This chapter is intended simply to serve as a reference of infrastructure costs for the range of components that would be required for new agricultural development in the Roper catchment, both for the component assets required for on-farm development and those for the supporting off-farm infrastructure. It serves three main purposes: (i) to provide a realistic benchmark of the rate of expansion of agriculture for forecasting demand for additional water (and other enabling infrastructure) in the NT, (ii) to provide benchmark indicators of the realistic costs of infrastructure for those wanting to independently assess the likely viability of development options, and (iii) to collate indicative costs for these different types of infrastructure as a reference for their use in financial analyses in other parts of this Assessment (including chapters 8 and 9). The information presented is particularly necessary given the systematic tendency of proponents of large infrastructure projects (including for new water supplies) to substantially under estimate development costs and over estimate trajectories of demand (see Chapter 6). This chapter also highlights the wide range of infrastructure assets (and associated private and public investors) that would be affected by new agricultural development. For a new scheme to function efficiently, the needs and responsibilities of investors in all keystone infrastructure assets would need to be considered, including the knock-on effects in creating demand for other types of enabling infrastructure. Large infrastructure projects, by their nature, are relatively rare and each has unique characteristics and challenges, making it difficult to extrapolate from one project to another. Even when case-specific details are taken into account, there are some challenges that cannot be known in advance and only become apparent once construction has begun. The costs provided here should therefore be taken as broadly indicative only. Actual costs incurred in any specific development project could differ substantially from those provided. A contingency would need to be factored in on top of the base costs presented to make allowance for these uncertainties. This chapter begins with an overview of growth trajectories in agricultural production and demand for irrigation water in the NT (Section 7.2) as context for why new infrastructure is required, and the rate at which it may need to be built. The chapter then presents costs for five types of new infrastructure that would be required to support an irrigation development and supply chains for new produce: • development costs of the water and land resources that investors in an irrigation scheme would have to cover (considering both large instream dams and on-farm sources of water) (Section 7.3) • costs of local processing facilities that may be required by new agricultural industries (built by private investors, who could be part of a vertically integrated project or separate investors) (Section 7.4) • costs of transport infrastructure (most likely publicly funded with a contribution from developers), and transport costs (Section 7.5) •costs of electricitytransmission and distribution infrastructure (built by energy providers withdeveloperspayingthe full or partial cost) (Section7.6) •costs ofcommunity infrastructure such as schools and hospitals (both publicly andprivatelyfunded) (Section7.7). 7.2Agriculturalgrowth and water demand trajectories If irrigated agricultural (particularlyhigh-value, water intensivehorticulture) is to continueto grow in the Roper catchmentand the rest of the NT, additional water will be required. Forecasting thatgrowth in demand is essential both forplanningnew waterinfrastructure and for evaluatingindividualwater infrastructure proposals to ensure that assumed demand trajectories for water(and theassociated (discounted) presentvalue that can be generatedfrom newhigh-valuehorticulture to justify thecosts of that infrastructure) are reasonable. ABS data series on historical agricultural production and water usewere analysedtoderive trendsandrelationshipsfor benchmarking realistic growthtrajectories for horticulturein the NT. Notethat GVAP includes bothirrigated and unirrigated agriculture,while thegross valueof irrigatedagricultural production(GVIAP) is restrictedto just irrigated agriculture; however,sincehorticulture is almost entirely irrigated,trendsfrom the longer GVAP data series areused belowfor estimating water demandtrajectories.Figure7-1shows growthtrends in various agricultural subsectors in Australia andtheNT. Growth trends for broadacre and horticultural crops in theNThave exceeded thoseforAustralia overall, while growth in cattle production forthe NT has been lowerthan the national trend. Thegrossvalue of NT horticulture morethan tripled in thedecadesof 1991–2000 (+233%)and 2001–10 (+210%) andincreased by35%in the decadeof2011–21. Thecurrent growthtrajectories for GVAP in Australia (with NT values followingin parentheses) are2.7billion (22million) per decade for horticulture, 8.9billion(37million)per decade for broadacrecrops, and6.8billion (288million) per decade forlivestock industries (step changes in GVAPfrom 2001–10 to2011–21 inFigure7-1).Horticultural produceis mainlysold domestically for consumption shortlyafter harvest (whilefresh). Growth in horticultural industries is thereforeconstrained by growth in demand from local consumers. Any new irrigated development would compete forsomeshare oftheabove growth values, providing abenchmark guide to what thepossible scale ofnew horticulture could be in anynew irrigation scheme, andthe trajectory for therate at which high-valuehorticulture (and associated waterdemand for high-value, high-prioritywater) could grow after a new irrigation schemewas completed. Chapter7 New infrastructure demand and costs|143 (a) Australia (b) Northern Territory For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au 010,00020,00030,0001981-901991-002001-102011-21GVAP ($M) DecadeCrops (horticulture)Crop (other)Livestock For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au 02004006001981-901991-002001-102011-21GVAP ($M) DecadeCrops (horticulture)Crop (other)Livestock Figure 7-1 Trends in gross value of agricultural production (GVAP) in (a) Australia and (b) the NT over 40 years (1981–2021) Data points are decade averages of annual values. The ‘Crop (other)’ category is predominantly broadacre farming. Source: ABS (2022) The scale of new horticultural farms is also limited by seasonal gaps in supply for each crop, so horticulture in any single location is typically a mix of products that fill the niche market gaps that that location can supply, rather than being a monoculture of the most valuable crop alone. This has implications for the value of new agricultural production that could pay for, and be used to justify the costs of, any new publicly funded irrigation scheme. Figure 7-2 shows the trends for how the GVIAP has increased in response to increasing supplies of irrigation water in Australia. The slopes of the trendlines give the increase in gross agricultural production per GL of new water used by different categories of Australian agriculture. Each extra new GL of water use could produce: • an extra $2.9 million of gross value in the fruit industries • an extra $7.9 million of gross value in the vegetable industries • an extra $3.8 million of gross value from mixed horticulture (from combined fruits and vegetables data) • an extra $1.2 million of gross value from a typical mix of agriculture overall. The horticultural component of proposed irrigation schemes deserves most scrutiny because financial viability of schemes is most sensitive to assumptions about the scale and rate of expansion of this more valuable form of agricultural production (and its requirement, typically, for higher security water than broadacre cropping). Currently about 30% of the total irrigation water applied in irrigated farming in Australia is used by horticultural production (ABS, 2021d). These values provide indicative benchmarks for the gross values that mixes of new agricultural activities could generate when planning new water supplies. (a) Fruits (d) Fruits and vegetables combined (b) Vegetables (d) Total agriculture For more information on this figure please contact CSIRO on enquiries@csiro.au y = 2.91x + 96.49R² = 0.8105001,0001,5000150300450GVAP ($ million) Water applied (GL) For more information on this figure please contact CSIRO on enquiries@csiro.au y = 3.76x + 145.49R² = 0.750750150022500125250375500GVAP ($ million) Water applied (GL) For more information on this figure please contact CSIRO on enquiries@csiro.au y = 7.94x -40.83R² = 0.7905001,0001,500020406080100GVAP ($ million) Water applied (GL) For more information on this figure please contact CSIRO on enquiries@csiro.au y = 1.23x + 560.73R² = 0.7102,0004,0006,00001000200030004000GVAP ($ million) Water applied (GL) Figure 7-2 Trends for increasing gross value of irrigated agricultural production (GVIAP) as available water supplies have increased for (a) fruits, (b) vegetables, (c) fruits and vegetables combined, and (d) total agriculture Source: ABS (2022) Irrigated agriculture represents about 25% of the Australian GVAP (ABS, 2021d). Agriculture used 7965 GL of water in 2018–19 and accounted for 59% of all water extractions. Most of the water used was applied to crops (70%) and pastures (30%). Out of the almost 8000 GL of water used in 2018–19, 28.6% came from groundwater sources (2280 GL), 1.5% from recycled water (115 GL) and the rest from other sources (rivers, creeks, lakes etc.) (ABS, 2021d). Figure 7-3 shows how much irrigation water is required by different types of horticultural farms, based on current national records of water use. In the NT, the horticultural farm types with the most intensive annual irrigation demand are ‘nurseries, cut flowers and cultivated turf’ (10.9 ML/ha) and ‘grapevines’ (9.8 ML/ha) (ABS, 2021d). For more information on this figure please contact CSIRO on enquiries@csiro.au 04812NSWVictoriaQueenslandSouth AustraliaWesternAustraliaNorthernTerritoryTasmaniaAustraliaWater application (ML/ha) JurisdictionFruit, nut, berriesVegetablesNurseries, flowers, turfGrapevines Figure 7-3 Mean annual water application rate by horticultural type across Australian states and territories Source: ABS (2021d) 7.3 Development costs for land and water resources Establishing new irrigated agriculture would involve the initial costs of developing water and land resources, and additional farm set-up costs for equipment and facilities on each new farm. There are many different options for where and how land and water resources are developed, each of which has implications for cost efficiencies and viability of a greenfield irrigation scheme. The analyses of scheme viability (Chapter 8) are not intended to prescribe particular scheme configurations or development pathways. Instead, the overall evaluation framework was designed to allow flexible comparisons across a wide range of different configurations (Figure 4-1), which required easy substitution of alternative land and water developments used in evaluations. To allow such arbitrary pairings of any land development option with any water development option, the individual options for developing each of these two agricultural resources had to be treated on a like-for-like basis. All water sources are therefore treated on a consistent basis where all capital and operating costs associated with delivering water to the farm at the farmland surface are treated as the costs of that water supply. This means that pumping costs for getting water from a weir to a farm, or pumping costs to lift groundwater to the farmland surface, are treated as costs of the water source (whereas pumping costs to then distribute and apply water on-farm, are treated as part of the costs of growing the crop, and were included in the costings of crop GMs in Chapter 5). This section covers the costs of developing new irrigated farms and the on- and off-farm water sources to supply them (following the distinction above in how they are costed). There may be additional costs, beyond those summarised below, to gain rights to land and water, particularly if an Indigenous land use agreement (ILUA) is required. For example, the Ord Final ILUA involved a compensation package worth $57 million to resolve several native title and heritage issues with the Miriuwung Gajerrong peoples over 1450 km2 of land in the Kimberley (Department of Regional Development and Lands, 2009). 7.3.1 Farm establishment costs The costs of developing new farms include capital expenditure on establishment and buildings (including approvals), farmland preparation (including clearing), irrigation systems (excluding the water source), and farm machinery and equipment. Capital costs of development are affected by the type of farm being developed, the siting of the farm (particularly soils and topography), the degree to which infrastructure is engineered, and choices about what activities are outsourced (particularly affecting the requirement for expensive packing and storage facilities on horticultural farms, and the requirement for owning specialised farm machinery). Indicative costs are provided for a range of farm development scenarios in Table 7-1. The base cases for broadacre farming are a typical furrow-irrigated farm (on clay soils, including water distribution and tailwater recycling) ($8,600/ha capital cost) and a farm on well-draining soils that would require a more expensive pressurised spray irrigation system (all other costs staying the same) ($11,600/ha). To bracket the range of establishment costs for broadacre crops, two other scenarios were used: a 5000-ha furrow-irrigated farm (capturing economies of scale in being able to use assets more efficiently) ($5,600/ha) and a higher cost spray irrigation development engineered to a higher standard and with complex approvals ($11,600/ha). Opportunities for very large (5000 ha) farms are limited in the Roper catchment, but the ‘Broadacre scale’ scenario indicates the potential efficiencies that scale can provide. These capital costs are also converted to an annualised equivalent (Table 7-1). Two scenarios are provided as indicators of the range of development costs for horticultural farms, both using high-pressure tape irrigation systems. The lower capital cost scenario (total capital costs $25,600/ha, Table 7-1) is based on direct packing of produce to bins in the field (e.g. for a row crop like melons) and assuming that nearby suitable off-farm accommodation is available for seasonal workers. If farm produce subsequently required grading, packing and cold storage by an off-farm service provider, the savings in upfront capital costs would be offset by additional ongoing costs of production from the outsourced service (that would reduce the farm’s GM). The higher capital cost scenario (total capital costs $71,000/ha, Table 7-1) includes the costs of modern packing and cold storage facilities, and on-site accommodation for seasonal workers (e.g. a remote fruit tree farm). Table 7-1 Indicative development costs for different types of irrigated farms All costs are standardised on a per hectare basis. Broadacre farms were based on a farm size of 500 ha, except for the ‘large scale’ scenario that was 5000 ha. Horticultural farms were based on farm size of 200 ha. The fixed component of maintenance costs was assumed to be 1% of the asset’s initial capital cost per year (and an additional variable cost of maintaining farm machinery and equipment was accounted for in crop gross margins in Chapter 5). A contingency would need to be factored in on top of these costs (e.g. an additional 10%). Equivalent annualised costs are based on a 10% discount rate. Costs of the irrigation water source are considered separately. ITEM UNITS BROADACRE LARGE SCALE BROADACRE FURROW BROADACRE PIVOTS BROADACRE HIGH STANDARD HORTICULTURE LOWER CAPITAL HORTICULTURE HIGHER CAPITAL Farm establishment and buildings $/ha 1,500 4,100 4,100 6,700 10,400 49,500 Farmland preparation $/ha 1,800 2,000 700 2,200 1,700 6,600 Irrigation system $/ha 800 1,000 5,300 5,300 4,500 6,600 Farm machinery and equipment $/ha 1,500 1,500 1,500 1,600 9,000 8,300 Total capital costs $/ha 5,600 8,600 11,600 15,800 25,600 71,000 Equivalent annualised cost $/ha/y 700 1,000 1,400 1,900 3,200 8,100 Maintenance costs $/ha/y 100 100 100 200 300 700 Total annualised costs $/ha/y 800 1,100 1,500 2,100 3,500 8,800 Source: Based on unit costs of component assets from Ash et al. (2018) and Stokes and Jarvis (2021), updated to $ June 2021 7.3.2 Costs for on-farm water sources Indicative costs for a range of scenarios for developing on-farm water sources are presented in Table 7-2. Costings were based on unit costs of component assets from Ash et al. (2018), including the delivery infrastructure to get water from the water source to the irrigation system (but not the costs of the irrigation system itself, which is already accounted for in the farm development costs above). The costs of developing on-farm water sources are highly dependent on characteristics of the location such as topography, soil texture and the success rate of bores. Each water source therefore included a more expensive and a less expensive scenario to illustrate some of this site- to-site variability. When compared on an equivalent basis (per unit area) indicative costs for developing on-farm water sources ranged from $3,800/ha to $14,200/ha (Table 7-2). Note that while the capital costs of developing bores is relatively low, pumping costs are typically high (depending on the TDH required to lift water to the soil surface). Likewise, high pumping costs would typically preclude water storages that are sited at a much lower elevation than the fields they are irrigating (noting, from like-for-like approach described before, that pumping costs to the farm surface are treated here as part of the costs of the water source). The companion technical report on water storage (Petheram et al., 2022) has much more detail on cost, siting and construction considerations for on-farm water storages, including maps of the locations in the Roper catchment most suited (topography and seepage) to building them. Table 7-2 Indicative capital costs for developing on-farm water sources (including distribution from source to cropped fields) Adapted from unit costings of farm development scenarios in Ash et al. (2018) and adjusted to June 2021 prices. Pumping costs for bores, or water storages that are below the height of the field they are irrigating, should allow about $2/ML/m TDH. WATER SOURCE FARM AREA (ha) CAPITAL COST ($/farm) COST PER HECTARE ($/ha) Gully dam, 6000 ML, well sited 500 1,900,000 3,800 Gully dam, 6000 ML, average siting 500 5,100,000 10,200 Stream diversion, gravity fed 5000 27,500,000 5,500 Stream diversion, requires pumping 5000 71,000,000 14,200 Flood flow pumping in floodplain, 6000 ML 500 6,450,000 12,900 Bores, good success and flow rate 500 1,900,000 3,800 Bores, moderate depth and success rate 500 3,850,000 7,700 7.3.3 Cost for large off-farm water infrastructure developments The companion technical report on water storage (Petheram et al., 2022) also evaluated the most cost-effective dam site locations in the Roper catchment, and the costs for building those dams and associated weir and reticulation infrastructure required to deliver that water to farms. Using information from Petheram et al. (2022), indicative costings are presented for two irrigation schemes based on the most cost-effective dam site locations (Table 7-3). This suggests that dams, together with supporting off-farm infrastructure, could supply water to new farms at a capital cost of about $50,000 to $80,000 per hectare of new irrigated farmland. Table 7-3 Indicative capital costs for developing two irrigation schemes based on the most cost-effective dam sites in the Roper catchment The dam costings already allow for a road and electricity grid connection to the dam: an indicative allowance is added for supporting off-farm roads and electricity distribution that farms can connect to (assumed 40 km of linear infrastructure for Waterhouse, and 15 km for Flying Fox, at a combined linear infrastructure cost of $2.3 million/km). ITEM WATERHOUSE COST ($) FLYING FOX COST ($) Capital costs Dam 253,000,000 318,000,000 Weir 0 89,000,000 Reticulation 126,400,000 12,000,000 Roads and electricity 90,000,000 35,000,000 Total 469,400,000 454,100,000 Summary metrics Irrigated area (ha) 10,100 5,485 Cost per hectare ($/ha) 46,500 82,800 Source: Dam, reticulation, and weir costings are from Petheram et al. (2022) and include contingencies, see that report for full details of cost breakdowns and assumptions The development of new agriculture would have flow-on consequences for local supply chains and demand for supporting infrastructure. These are considered in the following sections. 7.4 Processing costs 7.4.1 Dependence on new local processing facilities Due to the low value of some unprocessed farm commodities, particularly industrial crops like cotton and sugarcane, local processing is required for the total supply chain costs to be viable. This was demonstrated in the narrative risk analysis presented before that illustrated the influence of distance to gin on cotton GMs (Section 5.2.1). Sugarcane is even more reliant on local processing, because the unprocessed cane weighs about seven times as much as the processed sugar. For example, transporting sugarcane 100 km would cost about $26/t (see Table 7-6), more than half the gross cane revenue (about $40 to $45 per t). Investors in new local processing facilities would require economies of scale and security of supply (e.g. that farmers would not switch to other crops below the scale threshold) in order for their investments to be viable, and these would be essential considerations in the overall planning of a new irrigation scheme for these types of commodities. 7.4.2 Meatworks Meat processing capacity is concentrated in South East Queensland and on the eastern coast. Many cattle properties across northern Australia do not have access to local meatworks and have to transport cattle long distances (>1000 km) for processing (if they are not sold for live export). There have been several feasibility studies for the construction of abattoirs in western Queensland (e.g. Cloncurry, Hughenden, Roma) and other parts of northern Australia (e.g. Broome). A study by Meateng (2011) estimated the cost of constructing an abattoir at Broome would be around $33 million with an operational capacity of 100,000 head per year. Another study (Meateng, 2018) estimated the cost of constructing a 100,000 head per year abattoir in north-western Queensland to be about $100 million (not including the provision of land, power, water and road access) with operating costs of about $330/head. However, there has been a long history of meatworks being established in the NT but then struggling to remain viable. For example, AACo’s Livingstone Beef processing facility (situated about 50 km south of Darwin) has not been active since 2018. If the beef industry in the Roper catchment were to develop a boxed-meat market of sufficient scale, reviving a mothballed meatworks would probably be a more likely scenario than building a new one. 7.4.3 Cotton gin Indicative costs are provided for a cotton gin with maximum capacity of about 1500 bales/day (Table 7-4). Unprocessed seed cotton contains about 40% cotton lint, meaning that processed cotton bales are much lighter and cheaper to transport. Cotton seed is a by-product that can be used as a livestock feed supplement, with a ready market in the local Roper catchment cattle industry: trucks taking unprocessed cotton modules to the gin could return with cotton seed. The value of the cotton seed is generally about equal to the costs of processing charged to the grower. Harvested cotton can be stored, but susceptibility to spoilage in wet weather limits the length of the ginning season. An important consideration in remote locations would be how to power a new gin. A minimum area of about 15,000 ha irrigated cotton would be required to reach the scale of production necessary for a new gin to be viable. Higher cotton prices increase the distance that farms can profitably transport modules to the gin (Section 5.2.1), which increases the catchment area of a gin to attain threshold levels of supply, thereby increasing the chance of a cotton gin (and associated new cotton industry) navigating the challenging early years to become sustainable and profitable. A new gin, about 30 km north of Katherine, will become operational in 2023, providing opportunities for cotton growing in the western part of the Roper catchment. Table 7-4 Indicative capital and operating (fixed and variable) costs for a cotton gin from two sources ITEM SOURCE 1 SOURCE 2 COMMENTS Gin capacity ≈80,000 bales/y ≈95,000 bales/y Includes warehousing for 50,000 bales Capital cost $32,000,000 $30,000,000 Relocating an underutilised mill, if available, could be much cheaper Fixed costs $1,100,000/y $1,230,000/y Includes six full-time staff Variable costs $24 to $32/bale $35/bale Depends on scale and the source and cost of energy (on- or off-grid) Source 1: Rick Jones, Queensland Cotton (August 2017, pers. comm.); Stokes et al. (2017), adjusted to June 2021 prices Source 2: PwC (2019) (with input assumptions also from Queensland Cotton), adjusted to June 2021 prices 7.4.4 Sugar mill The amount of sugar that can be recovered by mills from harvested irrigated sugarcane is typically only about 15% by mass, a ratio known as the CCS (commercial cane sugar). Sugar mills are costly processing facilities that, depending on how they are configured, can produce different mixes of a range of products: sugar, molasses, renewable fuels (e.g. ethanol, biogas/methane or hydrogen), and/or baseload renewable power (from bagasse, the remaining fibre after crushing) (Jackson, 2013). Cane has to be crushed as it is harvested, so crushing operations are constrained by farming practices and trafficability of harvested fields (typically a 6-month crushing season between about mid-June and mid-December for irrigated cane). The standard practice in current sugarcane growing regions of Australia is for mills to pay for cane at the farm gate using a pricing formula that takes into account the quality (CCS) of the cane and the current sugar price (prices in $/t): cane price = raw sugar price * (CCS – 4)/100 * 90%. (i.e. millers get the first 4% of sugar extracted and 10% of the rest; growers get paid the value of 90% of sugar extracted above the first 4%). Processing of cane adds about 50% value in the sugar produced alone, and the bagasse (about 15% fibre) would be able to generate about 0.08 MW⋅h of exported power per tonne of cane (about another 15 to 30% value added to the value of the unprocessed cane). With appropriate management, including for pre-harvest water stress, irrigated cane reaches its peak quality around mid-November, and drops off rapidly either side of that date (with lower CCS and higher water content). Indicative costs are provided below for a basic sugar mill capable of processing about 1000 t cane per hour, or about 4 million t cane per year (for a 6-month crushing season and 90% mill reliability) (Table 7-5). Cane is first milled through crushers to separate the juice from the moist fibre (bagasse). Bagasse combustion produces steam to power the mill (and excess energy can be used for electricity generation). Juice is clarified to remove impurities before evaporating off water by boiling under partial vacuum. Crystallisation of sucrose occurs by further boiling, crystal seeding and centrifuging. Sugar and fibre can be further processed to produce ethanol. Throughput rates at different stages of processing depend on the quality of the cane, and hence affect the optimal configuration of mill components. Sugar mills are very large capital investments (about $400 million capital cost) and require a larger scale of farming than cotton to provide sufficient supply to justify such an investment. A minimum area of about 25,000 ha under irrigated sugarcane would be required to reach the scale of production necessary for a new mill to be viable. There are insufficient suitable land and water resources at locations within the Roper catchment to reliably support that scale of production. The information on costs of sugar mills and the scale of production required to support them is provided to show why sugarcane farming is not practicable in the Roper catchment despite the crop itself being agronomically suited to growing in these environments (and why sugarcane was therefore excluded from the set of crop options that were analysed in Part II of this report). Table 7-5 Indicative capital and operating costs for a basic sugar mill capable of processing 1000 t cane per hour Costs for cogeneration of electricity or ethanol production would be additional. Costs of each mill component depend on the quality of cane being processed (assumed 15% CCS, 15% fibre and 70% water content). See Jackson (2013) for a more detailed account of sugarcane processing. ITEM VALUE ($ million) Capital costs Crushers (extract and purify juice, separate fibre) 102 Evaporation (remove water from purified juice) 95 Pans and centrifugals (crystallise sucrose) 58 Utilities and balance of plant 155 Total capital costs 409 Operating costs (annual, recurrent) 34 Source: Stokes et al. (2017), adjusted for inflation to June 2021 pricing 7.5 Transport costs Indicative freight costs were estimated using the Transport Network Strategic Investment Tool (TraNSIT). TraNSIT (Higgins et al., 2015) is a modularised tool that uses detailed spatial information on the road (and rail) network in Australia (Figure 7-4) together with supply chain data on the movement of goods along this network for each agricultural industry. Freight estimates are based on detailed bottom-up modelling of the costs incurred by trains and trucks of different size classes moving different types of products along the transport network. It should be noted that in practice however, the actual prices charged to customers may not be split evenly in covering the trucking/rail costs of a round trip. Costs can be higher for the leg of the journey for which there is most demand and lower on the return leg (particularly if ‘backloading’ rates are charged on routes where some trucks would otherwise return empty or with loads below capacity). Costs for long distance trips (>1 day permitted driving time) do not scale completely linearly, as there are step changes each time the route crosses a threshold that requires drivers to take an overnight break. Figure 7-4 Road layer used in TraNSIT, showing road ranks and heavy vehicle restrictions Truck classes listed from shortest to longest in legend (left to right). Transport costs between Mataranka and key markets and ports are shown in Table 7-6 (with routes show in Figure 7-5). Transporting cattle from Mataranka to Darwin would cost about $47/t and a further $0.29 per t per km for the portion of the trip on the unsealed roads from within the Roper catchment to Mataranka. Estimated refrigerated freight costs to southern capital city markets (e.g. for most horticultural produce) range from $386/t (Adelaide) to $568/t (Perth). There would likely be opportunities for reduced backloading rates from Mataranka southwards for underutilised trucks on the return leg from supplying retail distribution centres in Darwin. Cost estimates do not include the disruptions from road closures that can cut off routes or require detours. The predominantly unsealed road network within the Roper catchment is susceptible to wet-season flooding. Table 7-6 Indicative road transport costs between the Roper catchment and key markets and ports The top section of the table gives trip costs from Mataranka to key destinations. The bottom section gives distance- based costs of getting goods from within the catchment to Mataranka (on unsealed roads) and approximate distance- based costs on sealed roads (to other destinations not specifically listed). DESTINATION TRANSPORT COST Unrefrigerated Refrigerated Cattle Transport costs from Mataranka ($/t) Adelaide 263.13 385.93 289.45 Brisbane 318.26 466.78 350.08 Broome Port 170.34 249.83 187.37 Cairns 245.84 360.57 270.42 Darwin 42.90 62.92 47.19 Karumba Port 177.30 260.04 195.03 Melbourne 371.20 544.43 408.32 Perth 391.38 574.02 430.51 Sydney 387.09 567.73 425.80 Townsville Port 220.23 321.43 241.92 Wyndham Port 73.53 107.84 80.88 Transport costs by distance ($/t/km) Properties to Mataranka 0.26 0.39 0.29 Mataranka to key markets/ports 0.17 0026 0.19 Se-R-512_TraNSIT_Aus_routes_v02 For more information on this figure please contact CSIRO on enquiries@csiro.au Figure 7-5 Freight paths from Mataranka to key ports and southern markets The freight path depends on the vehicle selection and heavy vehicle access (see Figure 7-4). Upgrading road networks can be an important enabler of regional development, improving the cost efficiencies and reliability of trucking routes. The cost of such upgrades, however, is substantial and highly variable depending on the route-specific works and bridges required. The Northern Australia Beef Roads Programme provided indicative costs of road upgrades across a range of scenarios (CSIRO, 2016; all prices quoted in this paragraph are adjusted to $ June 2021). For example, widening (9 m width) and sealing an existing unsealed road to state road standards was estimated to cost about $1.1 million per km (excluding bridges) in north-west Queensland. Construction costs of road upgrades could exceed $2.1 million per km in some cases, particularly when widening of floodways was required. Estimates of construction costs were as low as $270,000 per km for roads with lower volumes of traffic. In the NT, the cost of construction was about $860,000 per km for upgrading narrow sealed beef roads to two-lane sealed roads with flood immunity (e.g. Tableland Highway). Similar upgrades for beef roads in WA (e.g. Wyndham Spur) involving widening to 11 m, re-alignments and lengthening of culverts were estimated to cost about $1.6 million per km. The most expensive proposed upgrades were bridges and floodways, with a total cost of about $120 million for five bridges along the Great Northern Highway. Upgraded roads improve travel times (e.g. 80 km/h to 100 km/h), improve safety, reduce vehicle maintenance costs and reduce frequency of road closures. 7.6 Energy infrastructure costs Obtaining cost estimates for transmission infrastructure connections can be challenging, as costs are often borne by private companies and cost information is not shared publicly. Reliable cost data are also highly dependent on the location and requirements of the facility or load to be connected. A collaborative study by the CO2CRC and authored by the Electric Power Research Institute (EPRI, 2015) compiled energy infrastructure costs from a wide range of industry, government and research sources to develop estimates for its Least Cost of Energy (LCOE) methodology. This study provides credible technology cost and performance data and projections for Australian electricity over the period 2015 to 2030. It contains data ‘building blocks’ to use for policy and investment decisions and for further modelling of Australian electricity generation options. For a wide range of technologies, the study includes current and projected capital costs, operation and maintenance costs, and detailed performance data (EPRI, 2015). This reference has been heavily relied upon in the summary of electricity infrastructure costs below (with prices adjusted to $ June 2021). Transmission and distribution lines The delivery of electricity typically starts at a power generator from where a step-up transformer converts the electricity to higher voltages for more efficient long-distance transmission. Transmission lines provide for the bulk flow of electrical energy from generation sources to substations closer to end users, where step-down transformers convert the electricity to lower voltages for distribution. Distribution lines deliver electricity to consumers at voltages ready for use. The complex interconnected network of transmission lines, substations, distributions lines and control and conversion systems is often referred to collectively as a grid (such as the DKIS in the NT that extends as far south as Larrimah in the Roper catchment). High voltage (HV) transmission lines (132 to 330 kV lines with 50 to 3500 kVA power transfer capability) generally provide the backbone of Australian electricity transmission systems and deliver bulk energy directly from regional generation centres to load centres (EPRI, 2015). Lower voltage transmission lines (110 to 132 kV) are typically used to service mixed loads of residential, commercial and industrial demands and connect to the backbone 220 to 330 kV lines at bulk supply points that interface with the distribution network. Large industrial facilities such as mines, smelters and refineries can be directly connected to 220 to 330 kV transmission lines due to their high load requirements (100 to 900 MW). For HV transmission lines, there is also a wide range of nominal voltage levels and thermal capabilities between transmission lines from 132 to 330 kV, which can further vary final costs. For example, 132 to 330 kV transmission line costs can be $0.4 to $1.2 million per km depending on the voltage level and number of circuits, and the substation and switchgear can range from $10 million to $55 million depending on the arrangement of the substation (EPRI, 2015). An important consideration for the capital costs of network connection for both new generators and new loads is the influence of peak loads on capital costs. For generators, siting new power stations close to the existing grid can lower connection costs, but may constrain the technology options (EPRI, 2015). EPRI (2015) states that, ‘To use the full output of low-utilisation generators (such as intermittent renewables or peaking gas plants), network connections must be built to the peak capacity even though they might be used for only 20 to 40% of the time on average. Because connection costs have to be paid by the developer, this precludes all but short lines connecting to the existing grid without increasing an installed project’s LCOE. Traditional baseload generators may justify longer connections to the grid.’ This is true also for new load customers; their distribution lines must be sized to peak loads, even though there may be large portions of the day when the line is not delivering to capacity. Use of on-site storage may go some way to mitigate this, but the costs of on-site storage would need to be balanced with the avoided cost of capital for the larger distribution network capacity. Table 7-7 below provides some indicative transmission and distribution line costs from the EPRI study (EPRI, 2015). The 11 to 66 kV lines are most likely large enough and therefore most relevant for the kinds of developments likely to progress in the Roper catchment. Others have been included for the cases where projects may be economic for including larger cogeneration or renewables developments. Table 7-7 Indicative costs of transmission and distribution lines, for sizes relevant to this Assessment Acquisition of land and easement for the lines would be an additional cost. Costs are a rough guide only since they vary considerably depending on details of individual cases. ASSET DESCRIPTION TECHNICAL TRANSFER CAPABILITIES (MVA†) COST ($ million/km) Transmission line costs 220, 275, 330 kV single circuit 800 to 1300 0.75 132 kV double circuit 200 to 500 0.69 to 1.37 132 kV single circuit 45 to 234 0.30 to 0.76 Distribution line costs 11–33 kV single circuit 1 to 20 0.19 to 0.24 66 kV single circuit 10 to 100 0.21 to 0.43 †megavolt ampere (MVA) = 1 megawatt (MW) Source: EPRI (2015), adjusted for inflation to June 2021 pricing Transformers Substations connect two transmission or distribution lines of different voltage levels. Substations consist of transformers and associated switchgear and are a substantial part of the costs of connecting to the transmission system for a new-entrant generator (Table 7-8). Table 7-8 Indicative costs of transformer, for sizes likely to be relevant to developments in the Assessment area Transformers are categorised by the voltage pairs that they convert between. Excludes switchgear costs. na = not applicable. TRANSFORMER TECHNICAL TRANSFER CAPABILITIES (MVA†) COST ($ million) 275/132 kV 200 7.9 to 10.7 220/110 kV 150 5.4 132/22 kV na 7.0 to 7.6 110/33 kV 50 to 100 2.6 to 3.9 33/11 kV 5 to 20 1.1 to 2.1 †megavolt ampere (MVA) = 1 megawatt (MW) Source: EPRI (2015), adjusted for inflation to June 2021 pricing 7.7 Community infrastructure costs The availability of community services and facilities in remote areas can play an important role in attracting or deterring people from living in those areas. If local populations increase as a result of new irrigated developments, then there would be increased demand for public services in the region, and provision of those services would need to be anticipated and planned. Indicative costs for constructing a range of different facilities that may be required to support this growth are listed below (Table 7-9). Health care services in remote locations generally focus on primary and some secondary care, while the broadest range of tertiary services are concentrated in ‘principal referral hospitals’ that are mainly located in large cities but serve large surrounding areas by referral (AIHW, 2015). Each 1000 people in Australia require 2.3 (in ‘Major cities’) to 4.0 (in ‘Remote and Very remote areas’) hospital beds served by 16 full-time equivalent hospital staff and $3.5 million/year funding to maintain current mean national levels of hospital service (AIHW, 2023). Table 7-9 Indicative construction costs for different types of community facilities in Darwin Costs in remote areas like the Roper catchment are estimated to be about 30 to 60% higher than those quoted for Darwin. Cost ranges in columns two and three are per m2; costs in the last two columns are per hospital bed, house or apartment. na = not applicable BUILDING TYPE GFA† COST RANGE ($/m2) TOTAL COST RANGE ($) (low high) (low high) Private low-rise hospital, 45 to 60 m2/bed 3,900 4,650 225,000 350,000 Private low-rise hospital, 55 to 80 m2/bed + major operating theatre 4,700 5,700 350,000 500,000 House, single or double storey, 325 m2 1,800 2,800 585,000 850,000 Residential unit (townhouse), 90 to 120 m2 1,980 2,400 230,000 395,000 Offices, non-CBD, 1 to 3 stories 2,400 3,450 na na †GFA = gross floor area, the sum of covered and uncovered floor areas CBD = central business district Source: RLB (2021) Based on a small sample size, the indicative cost for building a new school is $9.9 million per school or about $27,000 per student (Table 7-10). For a larger sample size, the 2017 Queensland infrastructure plan (DILGP, 2017) (adjusted to $ June 2021) valued total public education assets for the state at $19 billion for 1239 state schools catering for 581,000 students. It is not clear on what basis the assets were valued, but these values equate to $15.5 million per school or $35,000 per student (which are slightly higher than the costs for the small sample of new schools). Demand for community services is growing both from population increases in Australia and rising community expectations (DILGP, 2017). New infrastructure that is built to service that demand would occur irrespective of what development occurs in particular parts of the country. However, if new irrigation projects shift some people to live in more remote parts of Australia, then this could shift the locations of where some services are delivered and associated infrastructure is built. The costs of delivering services and building infrastructure is generally higher in more remote locations. So, the net cost of any new infrastructure that is built to support regional developments is the difference in cost of shifting some infrastructure to more remote locations (not the full cost of facilities that would otherwise have been built elsewhere). Table 7-10 Indicative construction costs for new schools NAME STATE SUBURB ESTABLISHED COST ($ MILLION) STUDENTS TYPE SECTOR LOCATION Kingston Primary School WA Kingston 2009 12.7 768 Primary school Government Provincial South Halls Head Primary School WA Halls Head 2008 12.6 606 Primary school Government Inner regional Geographe Primary School WA Geographe 2002 12.0 542 Primary school Government Provincial Mackillop Catholic College Qld Mount Peter 2016 6.0 96 Combined Non- government Outer regional St Joseph’s Parish School Qld Weipa 2016 6.7 85 Primary school Non- government Very remote Holy Spirit College Cooktown Qld Cooktown 2015 9.2 89 Special Non- government Remote Mean (≈27,000 $/student) 9.9 364 Source: Stokes et al. (2017) based on all schools built between 2002 and 2017 in WA, NT and Queensland (Qld) for which construction costs could be found; updated to June 2021 prices 8 Financial viability of new irrigated development 8.1 Introduction There is a growing emphasis in Australia on greater accountability and transparency for large new infrastructure projects. This includes planning and building of new water infrastructure, and the way water resources are managed and priced (e.g. Infrastructure Australia, 2021a, 2021b; NWGA, 2022, 2023). Part of this shift has involved greater scrutiny of the costs and benefits of potential large new public dams. Large infrastructure projects, such as new irrigation developments in the Roper catchment, would be complex and costly investments. The difficulty in accurately estimating costs and the chance of incurring unanticipated expenses during construction, or not achieving projected water demand and revenue trajectories when completed, means that there are risks to the viability of developments if they are not thoroughly planned and assessed (as discussed in Chapter 6). This chapter therefore provides financial tools to assist in planning and evaluating irrigated development options (and easily comparing alternative configurations). New irrigation schemes in the Roper catchment would be costly to develop, such that even when technically feasible options are found, many of these are unlikely to be profitable at the returns and over the time periods expected by many investors. The amount of area in the Roper catchment that it would be technically feasible to farm (in terms of the scale of suitable land and water resources) is vastly greater than the area where it would be commercially sensible to do so. For example, the current area of irrigated agriculture in tropical Australia west of the Great Dividing Range uses less land area than mining (both <0.1%) (Watson et al., 2021a). Ultimately, financial factors will determine the types and scale of development. This chapter continues the overarching multi-scale agricultural viability framework introduced in Figure 4-1. Part II provided a bottom-up evaluation of farm performance for different crop options and this chapter provides a top-down analysis to determine the farm performance that would be required to pay for different ways of developing farms and water resources. The costs of developing water and land resources can vary widely, depending on a range of case- specific factors that are dealt with in other parts of this Assessment. These factors include the nature of the water source, the type of water storage, geology, topography, soil characteristics, the water distribution system, the type of irrigation system, the type of crop to be grown, local climate, land preparation requirements and the level to which infrastructure is engineered. The scale and pathways of development are therefore uncertain, so the analyses in this chapter were designed to be flexible and able to accommodate very different scales and configurations of development options. Rather than analysing the cost–benefit of specific irrigation scheme proposals, this chapter presents generic tables for evaluating multiple alternative development configurations, providing threshold farm gross margins and water costs/pricing that would be required to cover infrastructure costs. These provide a powerful (if slightly abstract) set of tools that allows users to answer their own questions about whether various aspects of agricultural land and water developments could be financially viable in the Roper catchment. Some examples of the questions that can be asked, and which tools to use to answer them, are summarised below (Table 8-1). Table 8-1 Types of questions that users can answer using the tools in this chapter For each question the relevant table number is given together with an example answer for a specific development scenario. More questions can be answered with each tool by swapping around the factors that are known and the factor being estimated. (All initial estimates assume farm performance is 100% in all years, i.e. before accounting for risks. See Table 8-2 for supporting generalised assumptions.) QUESTION (WITH EXAMPLE ANSWER) RELEVANT TABLE 1) How much can different types of farms afford to pay per ML of water they use? Table 8-3 A broadacre farm with a gross margin (GM) of $4,000/ha and water consumption of 8 ML/ha could afford to pay $135/ML while achieving a 10% internal rate of return (IRR). 2) How much would the operator of a large off-farm dam have to charge for water? Table 8-5 If off-farm water infrastructure had a capital cost of $5,000 for each ML/y supply capacity (yield) at the dam wall, the (public) water supplier would have to charge $537 for each ML to cover its costs (at a 7% target IRR). 3) For an on-farm dam with known development cost, what is the equivalent $/ML price of water? Table 8-7 A farm dam that had a capital cost of $1,500 for each ML/y supply capacity (yield) to develop would be equivalent to purchasing water at cost of $190 for each ML (at a 10% target IRR). 4) What farm gross margin (GM) would be required to fully cover the costs of an off-farm dam? What proportion of the costs of off-farm water infrastructure could farms cover? Table 8-4 If off-farm infrastructure had a capital cost of $50,000/ha to build, broadacre farms would need to generate a GM of $5,701/ha in order to fully cover the water supplier costs (while meeting a target 7% IRR for the water supplier (public investor) and a 10% IRR for the irrigator (private investor)). A broadacre farm with a GM of $4,000/ha could contribute the equivalent of $20,000 to $30,000 per ha towards the capital costs of building the same $50,000/ha dam (about 50% of the full costs of building and operating that infrastructure). 5) What GM would be required to cover the costs of developing a new farm, including a dam or bores? Table 8-6 A horticultural farm with low overheads ($1,500/ha) that cost $40,000/ha to develop (e.g. $30,000/ha to establish the farm and $10,000/ha to build the on-farm water supply to irrigate it) would require a GM of $6,702/ha to attain a 10% IRR. 6) How would risks associated with water reliability affect the farm GMs above? Table 8-8 If an on-farm dam could fully irrigate the farm in 70% of years and could irrigate 50% of the farm in the remaining years, all farm GMs in the answers above would need to multiplied by 1.18 (18% higher), and the price irrigators could afford to pay for water would need to be divided by 1.18. For example, in Q4, the GM required to cover the costs of the farm development would increase from $5,701/ha to $6,727 after accounting for risks of water reliability. 7) How would risks associated with ‘learning’ (initial farm underperformance) affect estimates? Table 8-10 If a farm achieved a GM that was 50% of its full potential in the first year, and gradually improved to achieve its full potential over 10 years, then GMs above would need to be higher by a factor of 1.26 (26% higher). For example, in Q6, the required farm GM would increase to $8,476/ha after accounting for risks of both water reliability and learning (a combined 49% higher than the value before accounting for risks). The next section describes the discounted cashflow (DCF) analysis approach used in financial analyses (Section 8.2). As set out in the rationale above, rather than using the DCF for a traditional cost–benefit analysis of specific development proposals/scenarios (as in Chapter 6) the analyses are used in a less prescriptive way to provide flexible tools that allow users to evaluate their own development scenarios. The analyses are first used to calculate the water price that irrigators can afford, as a useful common point of reference in subsequent analyses for identifying water sources that farms could pay for (Section 8.3). Analyses then consider the case of irrigation schemes built around a large dam and associated supporting off-farm infrastructure (Section 8.4). Then the case of self-contained, modular farm developments, with their own on-farm source of water, is considered (Section 8.5). The next section considers how different types of risks would affect the viability of irrigation schemes and provides adjustment factors that can be applied to previous analyses to account for the effects of these risks (Section 8.6). The chapter concludes by summarising the opportunities and principles for achieving sustainable and viable new irrigation developments in the Roper catchment (Section 8.7). 8.2 Balancing scheme-scale costs and benefits Designing a new irrigation development in the Roper catchment would require balancing three key determinants of irrigation-scheme financial performance to find combinations that might collectively constitute a viable investment: 1. Farm financial performance (relative to development costs and water use) (Chapter 5) 2. Capital cost of development, for both water resources and farms (Chapters 7) 3. Risks (and associated required level of investment return) (Section 8.6). Other assumptions were limited as much as possible, restricting these to factors with greater certainty and/or lower sensitivity, so that the results can be applied to a wide range of potential development scenarios. 8.2.1 Terminology Scheme financial evaluations use a DCF approach to evaluate the commercial viability of irrigation developments. The approach, following that of Stokes and Jarvis (2021), is intended to provide a purely financial evaluation of the conditions required to produce an acceptable return from an investor’s perspective. It is not a full economic evaluation of the costs and benefits to other industries, nor does it consider ‘unpriced’ impacts that are not the subject of normal market transactions, or the equity of how costs and benefits are distributed. (Non-market impacts are covered in the companion technical reports on ecological assets and ecological modelling (Stratford et al., 2022; 2023)). For the discussion that follows, an irrigation scheme was taken to be all the costs and benefits from the development of the land and water resources to the point of sale for farm produce. The DCF was applied in a non-standard generic manner to back-calculate threshold criteria for different development configurations to break even (rather than the traditional CBA approach of estimating financial performance of a few specific, detailed options). The section below explains the terminology and standard assumptions used. A discounted cashflow (DCF) analysis considers the lifetime of costs and benefits following capital investment in a new project. Costs and benefits that occur at different times are expressed in constant real dollars (June 2021 dollars), with a discount rate applied to streams of costs and benefits. The discount rate is the percentage by which future cost and benefits are discounted each year (compounded) to convert them to their equivalent present value. For an entire project, the net present value (NPV) can be calculated by subtracting the present value of the stream of all costs from the present value of the stream of all benefits. The benefit– cost ratio (BCR) of a project is the present value of all the benefits of a project divided by the present value of all the costs involved in achieving those benefits. To be commercially viable (at the nominated discount rate), a project would require an NPV that is greater than zero (in which case the BCR would be greater than one). The internal rate of return (IRR) is the discount rate at which the NPV is zero (and the BCR is one). For a project to be considered commercially viable it needs to meet its target IRR, where the NPV is greater than zero at a discount rate appropriate to the risk profile of the development and alternate investment opportunities available to investors. A target IRR of 7% is typically used when evaluating large public investments (with sensitivity analysis at 3% and 10%) (Infrastructure Australia, 2021b), while private agricultural developers usually target an IRR of 10% or more (to compensate for the investment risks involved). A back-calculation approach is used in the tables below to present threshold GMs and water prices that are required for investors to achieve specified target IRRs (and therefore, equivalently, NPV is zero at these discount rates). Project evaluation periods used in this chapter matched the lifespans of the main infrastructure assets: 100 years for large off-farm dams, and 40 years for on-farm developments. To simplify the tracking of asset replacements, four categories of life spans were used: 15 and 40 years for farms, and 25 and 100 years for off-farm infrastructure. It was assumed the shorter life span assets would be replaced at the end of their life, and costs were accounted for in full in the actual year of their replacement. At the end of the evaluation period, a residual value was calculated to account for any shorter life span assets that had not reached the end of their working life. Residual values were calculated as the proportional asset life remaining multiplied by the original asset price. Discounted residual values were trivially small (because the evaluation period matched the lifespans of the principal, dominant, longer lifespan assets) and hence analyses were not sensitive to the choice of method for how they were calculated. Capital costs of infrastructure were assumed to be the costs at completion (accounted for in full in the year of delivery), such that the assets commenced operations the following year. In some cases, the costs of developing the farmland and setting up the buildings and equipment were considered separately from the costs of the water source, so that different water sources could be compared on a like-for-like basis. Where an off-farm water source was used, this was treated as a separate investor receiving payments for water at a price that the irrigator could afford to pay. The main costs for operating a large dam and associated water distribution infrastructure are fixed costs for administering and maintaining the infrastructure, expressed here as percentage of the original capital cost, and variable costs associated with pumping water into distribution channels. At the farm scale, fixed overhead costs are incurred each year whether or not a crop is planted in a particular field that year. Fixed costs are dominated by the fixed component of labour costs, but also include maintenance, insurance, professional services and registrations. An additional allowance is made for annual operation and maintenance (O&M) budgeted at 1% of the original capital value of all assets (with an additional variable component to maintenance costs when machinery was used for cropping operations). A farm annual gross margin (GM) is the difference between the total revenue from crop sales and variable costs of growing a crop each year. Net farm revenue is calculated by subtracting fixed overhead costs from the GM. Variable costs vary in proportion to the area of land planted, the amount of crop harvested and/or the amount of water and other inputs applied. Farm GMs can vary substantially within and between locations, and as socioeconomic conditions change over time, as indicated in Chapter 5. GMs presented here are the values before subtracting the variable costs of supplying water to farms, with these costs instead accounted for in the capital costs of developing water resources. (Equivalent unit costs of supplying each ML of water are presented separately below.) 8.2.2 Threshold gross margins and water pricing to achieve target IRRs Financial analyses in this chapter used a generic approach to explore the consequences of variation in development costs, and other key factors that determine whether or not an irrigation scheme would be viable, such as farm performance and the level of returns sought by investors. The analyses used the DCF framework described above to back-calculate and fit the water prices and farm GMs that would be required for respective public (off-farm) and private (irrigators) investors to achieve their target IRRs. The results are then summarised as tables showing threshold criteria that would be required for a pair of water development and farm development options to combine together and meet investors’ target returns. The tables allow viable pairings to be identified in either of two ways: based on the threshold costs of water or farm GMs required. Financial viability for these threshold values was defined and calculated as investors achieving their target IRR (or, equivalently, that the investment would have an NPV of zero and a BCR of one at the specified discount rate). 8.2.3 Accounting structure Analyses first considered the case of irrigation schemes built around public investment in a large off-farm dam in the Roper catchment, and then considered the case of developments using on- farm dams and bores. Cost and benefit streams across the scheme were tracked for the separate components described in Figure 8-1. For farms, the streams were (i) the capital costs of land development, farm buildings and equipment (including replacement and maintenance costs, and residual values); (ii) the fixed overhead costs, applied to the full area of developed farmland; and (iii) the total farm GM (across all farms in the scheme), applied to the mean proportion of land in production each year (Figure 8-1). If a development scenario used an on-farm water source, then the costs of building and operating that water source were added to the overall farm costs (in the three categories above). Farm developers were treated as private investors who would seek a commercial return. In cases where an off-farm water source (large dam >25 GL/year) was evaluated, this was treated as a separate public investor whom farmers paid for water supplied (which served as an additional stream of costs for farmers and a stream of benefits for the water supplier at their respective target IRRs). For the public off-farm developer, the streams of costs were (i) the capital costs of developing the water and associated enabling infrastructure (including replacement and O&M costs and residual values), and (ii) the costs of maintaining and operating those assets (Figure 8-1). Accounting within a water infrastructure CBA needs to rigorously associate each benefit with all the costs and land and water resources used to attain it, and conversely, ensure that each cost and use of resources flows through to the benefit that is generated. To assist with such accounting, it is useful to have a framework that clearly defines the bounds of the overall irrigation development and of the component investments with it (Figure 8-1). For the purposes of this analysis, the irrigation scheme is defined as all the costs, benefits, use of land and transfers of water from when water is extracted by the scheme until agricultural produce is transported to, and revenue received at, the point of sale. The water source could either be part of a single on-farm investment (the green highlighted section of Figure 8-1, where water would be supplied from on-farm dams or bores), or there could be additional separate investors in the off-farm water infrastructure development (mainly in the blue highlighted section of Figure 8-1, where water would be supplied from a large off-farm dam and farms would pay the operator of the dam and water reticulation infrastructure). SCHEME All infrastructure and costs for capturing and diverting water, establishing a new irrigation area, and growing produce … to farmers receiving payment for produce Scheme accounting (quantities accounted for in each structural component below): Costs: Initial capital costs of developed assets Renewal/replacement costs of assets (based on lifespans) O&M costs of assets (recurrent – annual) Other recurrent costs for each asset (pumping to farm gate/surface for water source) Annual production costs (for each source of revenue) Revenue (benefits): Gross revenue paid to farmers for all agricultural produce Resource use: Water use (and transfers, with losses, between components) each year Area of farmland in production (using water and generating revenue) each year Scheme structure / Investment components: Off-farm (public) Everything for water storage and reticulation down to point of discharge Dams Dams and associated infrastructure (other than diversion and reticulation) Diversion Channels etc. used to divert water to irrigation area Irrigation area Roads, transmission lines, water reticulation to connect with farms in irrigation area Unaccounted Additional enabling infrastructure excluded from initial project, but required afterwards Revenue Payments received for water Farms (private, three types) Water costs, all other farming cost … to sale of produce Water source Costs of on-farm water source (or water payments for off-farm water source) Farm development (excluding water source) Capital costs of greenfield farm establishment (land development, irrigation system, buildings and structures, farm machinery and equipment) Crop production Crop growing and marketing costs (other than costs of water supply) to point of sale Revenue Sale of farm produce Figure 8-1 Financial structure of irrigation scheme used in accounting for costs, revenue and use of land and water resources Standardised accounting rules allow analyses to interchangeably pair any on-farm or off-farm water source with any farm development option. For on-farm water sources, no off-farm water infrastructure would be required, only supporting infrastructure such as roads and electricity supplies (blue highlighted section). O&M = operation and maintenance. 8.2.4 Assumptions To keep the results as relevant as possible to a wide range of different development options and configurations, the analyses here do not assume what scale a water development would be. Instead, all costs are expressed (i) per hectare of irrigated farmland and (ii) per megalitre per year of water supply capacity, facilitating comparisons between scenarios (that can differ substantially in size). Section 7.3 provided illustrations of how this approach was used for indicative costing of a range of farmland development options, on-farm water sources, and for the off-farm infrastructure costs for developments configured around the two most cost-effective dam sites in the Roper catchment. Those capital costs of development are referred to extensively in the analyses below. To further assist in making like-for-like comparisons across different development scenarios, a set of standard assumptions are made about the breakdown of development costs (by lifespan) and associated ongoing operating costs (Table 8-2). Three indicative types of farming enterprise are used to represent different levels of capital investment associated with the intensity of production and the extent to which farming operations are performed on-farm or outsourced (Table 8-2). Capital costs and fixed costs are higher for horticulture than broadacre farming, but the more expensive irrigation systems used (such as drippers) apply water more precisely and efficiently to crops. The indicative ‘Broadacre’ farm could, for example, represent hay or cotton farming using furrow irrigation on heavier clay soils. The indicative capital-intensive ‘Horticulture-H’ farm could, for example, represent high-value fruit-tree orchards with a high standard of on-farm packing and cold room facilities, and include accommodation for seasonal workers travelling to remote Roper catchment farms. The indicative less capital-intensive ‘Horticulture-L’ farm option could, for example, represent a row crop like melons, with packing directly to bins and using off-farm accommodation for seasonal workers (which reduces the upfront capital cost of establishing the farm, but increases ongoing costs for outsourced services that reduces farm GMs). For consistency, all costs required to deliver water to the farm at the level of the soil surface, are treated as the costs of the water source (so that different water sources can be substituted for each other on a like-for-like basis: see Section 7.3). Subsequent farm pumping costs to distribute and apply the supplied water to crops are treated as part of the variable costs of growing crops, and are already accounted for in the crop gross margins presented in Section 5.2. Pumping costs for the water source are highly situation-specific for different water sources: in particular, these pumping costs are affected by the elevation of the water source relative to the point of distributing to the farm, for example, the height water needs to be pumped from a weir to a distribution channel, from a farm dam to a field, or the dynamic head required to lift bore water to the field surface. For this reason, water source pumping costs are not included in summary tables of water pricing but should be added separately as required at a cost of about $2 per ML per m dynamic head (which is mainly a consideration for groundwater bores, but also applies where water needs to lifted from rivers or irrigation channels). For more information on water infrastructure costs see Chapter 6 (and companion technical reports referenced there) and for crop GMs see Chapter 5. Table 8-2 Assumed indicative capital and operating costs for new off- and on-farm irrigation infrastructure Three types of farming enterprise were represented to cover a range of increasing intensity, value and cost of production. Indicative base capital costs for establishing new farms (excluding water costs) allow on- and off-farm water sources to be added and compared on an equal basis. Annual operation and maintenance (O&M) costs are expressed as a percentage of the capital costs of assets. The ‘Horticulture-H’ farm with higher development costs includes on-farm packing facilities, cold storage and accommodation for seasonal workers. The ‘Horticulture-L’ farm with lower development costs does not include these assets and would have to outsource these services if required (reducing the farm gross margin). IRR = internal rate of return. SCHEME COMPONENT ITEM VALUE UNIT O&M COST (% capital cost/y) Off-farm infrastructure development capital and operating costs (large dam and enabling infrastructure) Capital costs Total capital costs (split by life span below) indicative >50,000 (analysed range: 20,000 to 150,000) $/ha Longer lifespan infrastructure (100 year) 85 % 0.4 Shorter lifespan infrastructure (40 year) 15 % 1.6 Operating costs O&M (by lifespan categories) % capital cost $/ha/y Off-farm water source pumping costs additional, ~2 $/ML/m Target IRR Base (with sensitivity range) 7 % Farm development capital and operating costs Broadacre Horticulture-L (low capital) Horticulture-H (high capital) Capital costs Base (excluding water source) 9,000 25,000 70,000 $/ha Water source (on- or off-farm) indicative >4,000 (analysed range: 3,000 to 15,000) $/ha Longer lifespan infrastructure (40 year) 50 50 50 % 1.0 Shorter lifespan infrastructure (15 year) 50 50 50 % 1.0 Operating costs O&M (by lifespan categories) % capital cost $/ha/y Farm water source pumping costs ~2 (additional) $/ML/m Fixed costs 600 1,500 6,500 $/ha/y Water use Crop water use (before losses) 6 6 6 ML/ha/y On-farm water use efficiency 70 90 90 % Gross margin Indicative gross margin 4,000 7,000 11,000 $/ha/y Target IRR Base (with sensitivity range) 10 10 10 % Analyses presented below first consider the case of irrigation schemes built around a large dam and associated supporting off-farm infrastructure (Section 8.4). Then the case of self-contained, modular farm developments, with their own on-farm source of water, is considered (Section 8.5). For both cases, the water price that irrigators can afford provides a useful common point of reference for identifying suitable water sources that different farm developments would be able to pay for (Section 8.3). Initial analyses assumed all farmland was in full production and performed at 100% of its potential (including 100% reliable water supplies) from the start of the development. Section 8.6 then provides a set of adjustment factors that quantify risks of several sources of anticipatable underperformance. 8.3 Price irrigators can afford to pay for a new water source Table 8-3 shows the price that the three different types of farms would be able to afford to pay for water, while meeting a target 10% IRR, for different levels of farm water use and productivity. For the prices to be sustained at this level throughout the life of the water source, the associated farm GM (in the row headings of Table 8-3) would also need to be maintained over this period. The table is therefore most useful when assessing the long-term price that can be sustained to pay off long-lived water infrastructure (rather than temporary spikes in farm GMs during runs of favourable years). Table 8-3 Price irrigators can afford to pay for water based on the type of farm, the farm water use, and annual gross margin (GM) of the farm Analyses assume water volumes are measured on delivery to the farm gate/surface: pumping costs involved in getting water to the farmland surface would be an additional cost of supplying the water (indicatively $2 per ML per m dynamic head) while pumping costs in distributing and applying the water to the crop are considered part of the variable costs included in the GM. Indicative GMs that the three types of farms could attain in the Roper catchment are $4,000, $7,000 and $11,000 per ha per year, respectively (highlighted rows): note however that the third type of farm cannot pay anything for water until it achieves a GM above $17,000 per ha per year. GROSS MARGIN PRICE IRRIGATORS CAN AFFORD TO PAY ($/ha/y) ($/ML at farm gate/surface) Farm water use (ML/ha including on-farm distribution and application losses) 4 5 6 7 8 9 10 12 Broadacre ($9,000/ha development costs, $600/ha/y fixed costs, 70% on-farm efficiency) 2,000 25 20 17 14 12 11 10 8 2,500 86 69 57 49 43 38 34 29 3,000 147 118 98 84 74 65 59 49 3,500 209 167 139 119 104 93 83 70 4,000 270 216 180 154 135 120 108 90 5,000 392 314 262 224 196 174 157 131 Horticulture-L ($25,000/ha development costs, $1,500/ha/y fixed costs, 90% on-farm efficiency) 5,000 39 31 26 22 19 17 16 13 6,000 241 193 161 138 121 107 97 80 7,000 444 355 296 254 222 197 178 148 8,000 646 517 431 369 323 287 259 215 10,000 1051 841 701 601 526 467 421 350 12,000 1456 1165 971 832 728 647 583 485 Horticulture-H ($70,000/ha development costs, $6,500/ha/y fixed costs, 90% on-farm efficiency) Below 16,000 Farms cannot afford to pay for water (or their other costs) at GMs lower than this 17,000 203 162 135 116 101 90 81 68 20,000 810 648 540 463 405 360 324 270 GROSS MARGIN PRICE IRRIGATORS CAN AFFORD TO PAY ($/ha/y) ($/ML at farm gate/surface) 25,000 1823 1458 1215 1042 911 810 729 608 30,000 2835 2268 1890 1620 1418 1260 1134 945 40,000 4860 3888 3240 2777 2430 2160 1944 1620 50,000 6885 5508 4590 3934 3443 3060 2754 2295 The lowest GM in the first column of Table 8-3 for each farm is the value below which the farm would not be viable even if water was free. This does not necessarily mean that such GMs could readily be achieved in practice: for the capital-intensive ‘Horticulture-H’ farm in particular, it would be challenging in the Roper catchment to reach the $17,000 per ha per year GM to cover the farm’s other costs, even before considering the costs of water. These water prices are likely most useful for public investors in large dams, because the sequencing of development creates asymmetric risks between the water supplier and irrigators. Irrespective of the water pricing that was planned for a dam project, once the dam is built irrigators have the choice of whether to develop new farms or not, and are unlikely to act to their own detriment in making that investment if they cannot do so at a water price that will allow them to attain a commercial rate of return. These water prices, together with estimates of likely attainable farm GMs in other parts of the Assessment, provide a useful benchmark for checking assumptions on any potential public dam developments in the Roper catchment. For on-farm water sources, these water prices can be used to assist in planning water development options that cropping operations could reasonably be expected to afford. Tables in the next sections allow these comparisons by converting capital costs of developing on- and off- farm water sources to volumetric costs ($/ML supplied). All water prices are based on volumes supplied to the farm gate/surface (after losses getting to that point) per metered ML supplied. 8.4 Financial targets required to cover costs of large, off-farm dams The first generic assessment considered the case of public investment in a large dam in the Roper catchment, and whether the costs of that development could be covered by water payments from irrigators (priced at their capacity to pay). The public costs of development include the cost of the dam and water distribution, and any other supporting infrastructure required. Costs are standardised per unit of farmland developed, noting that a smaller area could be developed for a crop with a higher water use (so the water development costs per hectare would be higher). 8.4.1 Farm gross margins to cover full costs of off-farm public water infrastructure Table 8-4 shows what farm annual GM would be required for different costs of water infrastructure development at the public investors’ target IRR. As expected, higher farm GMs are required to cover higher capital costs and attain a higher target IRR. These tables can be used to assess whether water development opportunities and farming opportunities in the Roper catchment are likely to pair together in financially viable ways. Indicative farm GMs that could be achieved in the Roper catchment are about $4,000, $7,000 and $11,000 per ha per year for ‘Broadacre’, less capital-intensive 'Horticulture-L’ (including penalty to GM for outsourcing), and capital-intensive ‘Horticulture-H’, respectively (see Section 5.2). A dam and supporting infrastructure would likely require at least $50,000/ha of capital investment (see Table 7-3). None of the three farming types are likely to be viable at these farm GMs and water development costs (at a 7% target IRR for the public investor). However, broadacre and less capital-intensive ‘Horticulture-L’ farming might be marginally viable at a 3% target IRR for the public investor. Alternatively, broadacre and lower cost ‘Horticulture-L’ could both achieve a target 10% IRR for the farm investments while contributing $20,000 to $30,000 per ha (40% to 60%) towards the cost of a dam (including enabling infrastructure and ongoing O&M costs) that cost $50,000/ha to build. That is a higher proportion of costs than irrigators have historically contributed towards irrigation schemes in some other parts of Australia (about a quarter of capital costs; Vanderbyl, 2021), but would be a decision for the Commonwealth and Northern Territory governments based on their expectations, priorities and investment criteria. Table 8-4 Farm gross margins (GMs) required to cover the costs of off-farm water infrastructure (at the suppliers’ target internal rate of return (IRR)) Assumes 100% farm performance on all farmland in all years once construction is complete. Costs of supplying water to farms are consistently treated as costs of water source development (and not part of the farm GM). Risk adjustment multipliers are provided in Section 8.6. Blue shading of rows indicates the capital costs that could be afforded by farms with GMs of 4,000, $7,000 and $11,000 per ha per year, respectively, for the farm types in the three sections of the table below. Blue shading of columns indicates the range of the most cost-effective dam development options in the Roper catchment (Table 7-3). TARGET IRR FARM GROSS MARGIN REQUIRED TO PAY FOR OFF-FARM WATER INFRASTRUCTURE (%) ($/ha/y) Total capital costs of off-farm water infrastructure ($/ha) 20,000 30,000 40,000 50,000 70,000 100,000 125,000 150,000 Broadacre ($9,000/ha development costs, $600/ha/y fixed costs, 70% on-farm efficiency) 3 2,604 3,016 3,428 3,840 4,664 5,900 6,930 7,960 5 2,977 3,569 4,160 4,751 5,933 7,707 9,185 10,663 7 3,359 4,139 4,920 5,701 7,263 9,605 11,558 13,510 10 3,941 5,013 6,085 7,157 9,301 12,516 15,196 17,876 12 4,333 5,601 6,869 8,137 10,673 14,478 17,648 20,818 Horticulture-L ($25,000/ha development costs, $1,500/ha/y fixed costs, 90% on-farm efficiency) 3 5,584 5,996 6,408 6,820 7,645 8,881 9,911 10,941 5 5,985 6,576 7,167 7,759 8,941 10,715 12,193 13,671 7 6,370 7,150 7,931 8,712 10,274 12,616 14,569 16,521 10 6,952 8,024 9,096 10,168 12,312 15,528 18,208 20,887 12 7,345 8,613 9,881 11,149 13,685 17,489 20,659 23,829 Horticulture-H ($70,000/ha development costs, $6,500/ha/y fixed costs, 90% on-farm efficiency) 3 16,618 17,068 17,518 17,967 18,867 20,217 21,342 22,467 5 17,164 17,789 18,413 19,038 20,288 22,162 23,724 25,286 7 17,610 18,416 19,222 20,027 21,638 24,055 26,070 28,084 10 18,215 19,301 20,387 21,472 23,644 26,901 29,615 32,330 12 18,607 19,884 21,161 22,438 24,992 28,823 32,015 35,207 8.4.2 Target water pricing for off-farm public water infrastructure Table 8-5 shows the price that a public investor in off-farm water infrastructure would have to charge to fully cover the costs of development of off-farm water infrastructure, expressed per unit of supply capacity at the dam wall. Pricing assumes that the full supply of water (i.e. reservoir yield) would be used and paid for every year over the entire lifetime of the dam, after accounting for water losses between the dam and the farm. It can be challenging for farms to sustain the high levels of revenue over such long periods (100 years) to justify the costs of building expensive dams. For these base analyses, the water supply is assumed to be 100% reliable; risk adjustment multipliers to account for reliability of supply are provided in Section 8.6. Table 8-5 Water pricing required to cover costs of off-farm irrigation scheme development (dam, water distribution, and supporting infrastructure) at the investors target internal rate of return (IRR) Assumes the conveyance efficiency from dam to farm is 70% and that supply is 100% reliable. Risk adjustment multipliers for water supply reliability are provided in Section 8.6. Pumping costs between the dam and the farm would need to be added (e.g. about $30/ML extra to lift water about 15 m from weir pool to distribution channels). ‘$ CapEx per ML/y at dam’ is the capital expenditure on developing the dam and supporting off-farm infrastructure expressed per ML per year of the dam’s supply capacity measured at the dam wall. Highlighted values are indicative of the most cost-effective large dam options available in the Roper catchment (Table 7-3). TARGET IRR WATER PRICE THAT WOULD NEED TO BE CHARGED TO COVER OFF-FARM INFRASTRUCTURE COSTS (%) ($/ML charged at farm gate) Capital costs of off-farm infrastructure ($ CapEx per ML/y at dam) 3,000 4,000 5,000 6,000 8,000 10,000 12,000 14,000 16,000 3 162 215 269 323 431 538 646 754 861 5 239 319 399 479 638 798 958 1117 1277 7 322 429 537 644 859 1073 1288 1502 1717 10 448 598 747 897 1196 1495 1794 2093 2392 For example, in the Roper catchment some of the most cost-effective dam opportunities would cost about $5000 for each ML/year of supply capacity at the dam wall after including the required supporting off-farm water infrastructure (see Table 7-3). This would require farms to pay $537 for each ML extracted to fully cover the costs of the public investment (at the base 7% target IRR for public investments, Table 8-2). Comparisons against what irrigators can afford to pay (Table 8-3), show that it is unlikely any farming options would be able to cover the costs of a dam in the Roper catchment at the GMs farms are likely to be able to achieve (see Section 5.2). In cases where a scheme is not viable (BCR <1), the water cost and pricing tables can be used as a quick way of estimating the BCR and likely proportion of public development costs that farms would be able to cover. For example, a broadacre farm that uses 8 ML/ha (measured at delivery to the farm) with a GM of $4000 per ha per year could afford to pay $135/ML extracted, which would cover 25% ($135/$537) of the $537/ML price required to cover the full costs of the public development: the BCR would therefore be 0.25 (the ratio of the full costs of the scheme to the proportion the net farm benefits can cover). As for the example discussed for Table 8-4, it would be a decision for the public investor as to what proportion of the capital costs of infrastructure projects they would realistically expect to recover from users. 8.5 Financial targets required to cover costs of on-farm dams and bores The second generic assessments considered the case of on-farm sources of water. Indicative costs for on-farm water sources, including supporting on-farm distribution infrastructure, vary between $4,000 and $15,000 per hectare of farmland, depending on the type of water source, how favourable the local conditions are for its development, and the irrigation requirement of the farming system. Since the farm and water source would be developed by a single investor, the first analyses considered the combined cost of all farm development together (without separating out the water component). 8.5.1 Gross margins to cover full costs of farm development with water source Table 8-6 shows the farm GMs that would be required to cover different costs of farm development at the investors target IRR. Note that private on-farm water sources are typically engineered to a lower standard than public water infrastructure, and have lower upfront capital costs, higher recurrent costs (higher O&M and asset replacement rates) and lower reliability. Based on the same indicative farm GMs as before (Table 8-2) and 10% target IRR, a broadacre farm with $4,000 per ha per year GM could cover total on-farm development capital costs of about $20,000/ha, a lower capital cost ‘Horticulture-L’ farm with GM of $7,000 per ha per year could afford about $40,000/ha of initial capital costs, and a capital-intensive ‘Horticulture-H’ farm with GM of $11,000 per ha per year could pay about $30,000/ha for farm development (Table 8-6). This indicates that on-farm water sources may have more prospects of being viable than large public dams in the Roper catchment, particularly for broadacre farms and horticulture with lower development costs, if good sites can be identified for developing sufficient on-farm water resources at low-enough cost. Table 8-6 Farm gross margins (GMs) required to achieve target internal rate of return (IRR) given different capital costs of farm development (including an on-farm water source) Assumes 100% farm performance on all farmland in all years once construction is complete. Risk adjustment multipliers are provided in Section 8.6. Blue shading of rows indicates the capital costs that could be afforded by farms with GMs of $4,000, $7,000 and $11,000 per ha per year, respectively, for the farm types in the three sections of the table below. TARGET IRR FARM GROSS MARGIN REQUIRED TO ACHIEVE FA’MER'S TARGET IRR (%) ($/ha/y) Total capital costs of farm development, including water source ($ CapEx/ha) 10,000 15,000 20,000 30,000 40,000 50,000 70,000 100,000 Broadacre ($600/ha/y fixed costs, 70% on-farm efficiency) 5 1,516 1,957 2,398 3,279 4,160 5,042 6,804 9,449 7 1,669 2,181 2,694 3,718 4,742 5,767 7,815 10,888 10 1,923 2,554 3,185 4,447 5,709 6,972 9,496 13,282 12 2,105 2,821 3,537 4,968 6,400 7,832 10,696 14,991 15 2,389 3,238 4,087 5,785 7,483 9,181 12,578 17,672 20 2,882 3,963 5,044 7,206 9,368 11,530 15,854 22,340 TARGET IRR FARM GROSS MARGIN REQUIRED TO ACHIEVE FA’MER'S TARGET IRR (%) ($/ha/y) Total capital costs of farm development, including water source ($ CapEx/ha) 10,000 15,000 20,000 30,000 40,000 50,000 70,000 100,000 Horticulture-L ($1500/ha/y fixed costs, 90% on-farm efficiency) 5 2,469 2,909 3,350 4,231 5,113 5,994 7,757 10,401 7 2,637 3,149 3,661 4,685 5,710 6,734 8,783 11,856 10 2,915 3,546 4,177 5,439 6,702 7,964 10,488 14,274 12 3,114 3,830 4,546 5,978 7,409 8,841 11,705 16,001 15 3,424 4,273 5,122 6,820 8,519 10,217 13,613 18,708 20 3,962 5,043 6,124 8,286 10,448 12,610 16,934 23,420 Horticulture-H ($6500/ha/y fixed costs, 90% on-farm efficiency) 5 7,760 8,201 8,642 9,523 10,404 11,286 13,048 15,692 7 8,012 8,524 9,036 10,060 11,085 12,109 14,158 17,231 10 8,427 9,058 9,689 10,951 12,213 13,475 15,999 19,785 12 8,720 9,436 10,152 11,584 13,016 14,448 17,312 21,607 15 9,177 10,026 10,875 12,573 14,271 15,970 19,366 24,461 20 9,963 11,044 12,125 14,287 16,449 18,611 22,935 29,421 8.5.2 Volumetric water cost equivalent for on-farm water source Table 8-7 converts the capital cost of developing an on-farm water source (per ML of annual supply capacity) into an equivalent cost for each individual ML of water supplied by the water source. The table can be used to estimate how much a farm could spend on developing required water resources by comparing the $/ML costs against what farms can afford to pay for water (Table 8-3). For example, a broadacre farm with a GM of $4000 per ha per year and annual farm water use of 8 ML/ha could afford to pay $135/ML for its water supply (Table 8-3), which would allow capital costs of $700 to $1000 for each ML/year supply capacity for developing an on-farm supply. Indicative costs for developing on-farm water sources range from about $500/ML to $2000/ML (based on the range of per hectare costs above) which confirms, by this alternative approach, that there are likely to be viable farming opportunities using on-farm water development in the Roper catchment. Table 8-7 Equivalent costs of water per megalitre for on-farm water sources with different capital costs of development, at the internal rate of return (IRR) targeted by the investor Assumes the water supply is 100% reliable. Risk adjustment multipliers for water supply reliability are provided in Section 8.6. Pumping costs to the field surface would be extra (e.g. about $2 per ML per m dynamic head for bore pumping). TARGET IRR WATER VOLUMETRIC COST EQUIVALENTUNIT FOR DIFFERENT CAPITAL COSTS OF WATER SOURCE (%) ($/ML) Capital costs for on-farm water infrastructure ($ CapEx per ML/y at farmland surface) 300 400 500 700 1000 1250 1500 1750 2000 3 22 29 37 51 74 92 110 129 147 TARGET IRR WATER VOLUMETRIC COST EQUIVALENTUNIT FOR DIFFERENT CAPITAL COSTS OF WATER SOURCE (%) ($/ML) Capital costs for on-farm water infrastructure ($ CapEx per ML/y at farmland surface) 300 400 500 700 1000 1250 1500 1750 2000 5 26 35 44 61 87 109 131 153 175 7 31 41 51 72 102 128 154 179 205 10 38 51 63 89 127 159 190 222 254 12 43 58 72 101 144 180 216 252 288 15 51 68 85 120 171 213 256 299 342 20 65 87 109 152 217 271 326 380 434 8.6 Risks associated with variability in farm performance This section assessed the impacts of two types of risks on scheme financial performance: those that reduce farm performance through the early establishment and learning years, and those occurring periodically throughout the life of the development. The effect of these negative risks is to reduce the expected revenue and expected GM. Setbacks that occur early on after a scheme is established were found to have the largest effect on scheme viability, particularly at higher target IRRs. There is a strong incentive to start any new irrigation development with well-established crops and technologies, and to be thoroughly prepared for the anticipatable agronomic risks of establishing new farmland. Analyses showed that delaying full development for longer periods than the learning time had only a slight negative effect on IRRs, whereas proceeding to full development before learning was complete had a much larger impact. This implies that it would be prudent to err on the side of delaying full development (particularly given that in practice, it would only be possible to know when full performance was achieved in retrospect, not in advance). An added benefit of staging would be limiting losses where small-scale testing proves initial assumptions of benefits to be overoptimistic and that full- scale development could never be profitable, even after trying to overcome unanticipated challenges. For an investment to be viable, farm GMs need to be sustained at high levels over long periods. Thus, variability in farm performance poses risks that need to be considered and managed. GMs can vary between years either because of short-term initial underperformance or because of periodic shocks. Initial underperformance is likely to be associated with learning as farming practices are adapted to local conditions, overcoming initial challenges to reach their long-term potential. There would be further unavoidable periodic risks associated with water reliability, climate variability, flooding, outbreaks of pests and diseases, periodic technical/equipment failures, and fluctuations in commodity prices and market access. Periodic risks, such as reliability of water supply, are less easy to avoid. Results for analyses of both periodic and learning risks are shown below. Throughout this section, farm performance in a given year is quantified as the proportion of the long-term mean GM a farm attains, where 100% performance is when this level is reached and zero % equates to a performance where revenues only balance variable costs (GM = zero). 8.6.1 Risks from periodic underperformance Analyses considered periodic risks generically, without assuming any of the particular causes listed above. Periodic risks were characterised in terms of three components to quantify their effects on scheme financial performance: • reliability: the proportion of ‘good’ years where the ‘full’ 100% farm performance was achieved, with the remainder of years being ‘failures’ where some negative impact was experienced • severity: the farm performance in a ‘failed’ year where some type of setback occurred • timing: for ‘early’ timing a 10-year cycle was used where, for example, with 80% reliability failures would occur in the first 2 years of the scheme and the first 2 years of each 10 years in a cycle after that. For ‘late’ timing, the ‘failures’ came at the end of each 10-year cycle. Where ‘random’ timing was used, each year was represented as having the long-term mean farm performance of ‘good’ and ‘failed’ years (frequency weighted). Table 8-8 summarises the effects of a range of different reliabilities and severities for periodic risks on scheme viability. Periodic risks had a consistent proportional effect on target GMs, irrespective of development options or costs, so results were simplified as a set of risk adjustment multipliers. The multipliers can therefore be applied to the target farm GMs in the previous section (required to cover capital costs of development at investors’ target IRRs at 100% farm performance) to account for the effects of various risks. These same adjustment factors can be applied to the water prices that irrigators can afford to pay (Table 8-3) but would be used as divisors to reduce the price that irrigators could pay for water. As would be expected, the greater the frequency and severity of ‘failed’ years, the greater the impact on scheme viability and the greater the increase in farm GMs that would be required to offset these impacts. As an example, the reliability of water supply is one of the more important sources of unavoidable variability in productivity of irrigated farms. In such cases, water reliability (proportion of years where the full supply of water is available) is the same as the ‘reliability’ in Table 8-8, and the mean percentage of water available in a ‘failed’ year (where less than the full supply is available) is equivalent to the ‘failed year performance’ in Table 8-8 (assuming the area of farmland planted is reduced in proportion to the amount of water available). For example, if a water supply was 85% reliable and provided on average 75% of its full supply in ‘failed’ years, a risk adjustment factor of 1.04 (Table 8-8) would have to be applied to baseline target GMs (Table 8-4 and Table 8-6) and the prices irrigators can afford to pay for water (Table 8-3). This means that a 4% higher GM would be required to achieve a target IRR (and irrigators’ capacity to pay for water would be ~4% lower) than if water could be supplied at 100% reliability. For crops where the quality of produce is more important than the quantity, such as annual horticulture, the approach of reducing planted land area in proportion to available water in ‘failed’ years would be reasonable. However, for perennial horticulture or tree crops it may be difficult to reduce (or increase) areas on an annual basis. Farmers of these crops would therefore tend to opt for systems with a high degree of reliability of water supply (e.g. 95%). For many broadacre crops, deficit irrigation could partially mitigate impacts on farm performance in years with reduced water availability, as could carryover effects from inputs (such as fertiliser) in a failed year that reduce input costs the following year. Table 8-8 Risk adjustment factors for target farm gross margins (GMs), accounting for the effects of reliability and severity (level of farm performance in ‘failed’ years) of periodic risks Results are not affected by discount rates. ‘Good’ years = 100% farm performance; ‘Failed’ = <100% performance. ‘Failed year performance’ is the mean farm GM in years where some type of setback is experienced relative to the mean GM when the farm is running at ‘full’ performance. FAILED YEAR PERFORMANCE (%) RISK ADJUSTMENT MULTIPLIER FOR TARGET FARM GROSS MARGINS (VS BASE 100% RELIABILITY TABLES) (unitless ratio) Reliability (Proportion of ‘good’ years) 1.00 0.90 0.85 0.80 0.70 0.60 0.50 0.40 0.30 0.20 85 1.00 1.02 1.02 1.03 1.05 1.06 1.08 1.10 1.12 1.14 75 1.00 1.03 1.04 1.05 1.08 1.11 1.14 1.18 1.21 1.25 50 1.00 1.05 1.08 1.11 1.18 1.25 1.33 1.43 1.54 1.67 25 1.00 1.08 1.13 1.18 1.29 1.43 1.60 1.82 2.11 2.50 0 1.00 1.11 1.18 1.25 1.43 1.67 2.00 2.50 3.33 5.00 Table 8-9 summarises how timing of periodic impacts affects scheme viability, providing risk adjustment factors for a range of reliabilities for an impact that had 50% severity with late timing, early timing, and no (long-term frequency, weighted mean performance) timing. These results show that any negative disturbances that reduce farm performance will have a larger effect if they occur early on after the scheme is established, and that this effect is greater at higher target IRRs. For example, at a 7% target IRR and 70% reliability with ‘late’ timing (where setbacks occur in the in the last three of every 10 years) the GM multiplier is 1.13, meaning the annual farm GM would need to be 13% higher than if farm performance were 100% reliable. In contrast, for the same settings with ‘early’ timing, the GM multiplier is 1.23, so impacts of early setbacks are more severe and the farm GM would have to be 23% higher than if farm performance were 100% reliable. Table 8-9 Risk adjustment factors for target farm gross margins (GMs), accounting for the effects of reliability and timing of periodic risks Assumes 50% farm performance during ‘failed’ years, where 50% farm performance means 50% of the GM at ‘full’ potential production. IRR = internal rate of return. TARGET IRR (%) TIMING OF FAILED YEARS RISK ADJUSTMENT MULTIPLIER FOR TARGET FARM GROSS MARGINS (VS BASE 100% RELIABILITY TABLES) (unitless ratio) Reliability (proportion of ‘good’ years) 1.00 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0.20 3 Late 1.00 1.05 1.10 1.16 1.22 1.30 1.39 1.50 1.63 Random – no bias 1.00 1.05 1.11 1.18 1.25 1.33 1.43 1.54 1.67 Early 1.00 1.06 1.13 1.20 1.28 1.37 1.47 1.58 1.70 7 Late 1.00 1.04 1.08 1.13 1.19 1.26 1.35 1.46 1.59 Random – no bias 1.00 1.05 1.11 1.18 1.25 1.33 1.43 1.54 1.67 Early 1.00 1.07 1.15 1.23 1.32 1.41 1.51 1.62 1.74 10 Late 1.00 1.03 1.07 1.12 1.17 1.24 1.32 1.42 1.56 Random – no bias 1.00 1.05 1.11 1.18 1.25 1.33 1.43 1.54 1.67 TARGET IRR (%) TIMING OF FAILED YEARS RISK ADJUSTMENT MULTIPLIER FOR TARGET FARM GROSS MARGINS (VS BASE 100% RELIABILITY TABLES) (unitless ratio) Early 1.00 1.08 1.16 1.25 1.35 1.45 1.55 1.66 1.77 8.6.2 Risks from initial ‘learning’ period Another form of risk arises from the initial challenges in establishing new agricultural industries in the Roper catchment, and includes setbacks from delays, such as gaining regulatory approvals and adapting farming practices to Roper catchment conditions. Some of these risks are avoidable if investors and farmers learn from past experiences of development in northern Australia (e.g. Ash et al., 2014), avoid previous mistakes, and select farming options that are already well proven in analogous northern Australian locations. However, even if developers are well prepared, there are likely to be initial challenges in adapting to the unique circumstances of a new location. Newly developed farmland can take some time to reach its productive potential as soil nutrient pools are established, soil limitations are ameliorated, suckers and weeds are controlled, and pest and weed management systems are established. • ‘Learning’ (used here to broadly represent all aspects of overcoming initial sources of farm underperformance) was assessed in terms of two simplified generic characteristics: • initial level of performance: represented as described before, as the proportion of the long-term mean GM that the farm achieves in its first year • time to learn: the number of years taken to reach the long-term mean farm performance. Performance was represented as increasing linearly over the learning period from the starting level to the long-term mean performance level (100%). The effect of learning on scheme financial viability was considered for a range of initial levels of farm performance and learning times. As before, learning had consistent proportional effects on target GMs, so results were simplified as a set of risk adjustment factors (Table 8-10). As would be expected, the impacts on scheme viability are greater the lower the starting level of farm performance, and the longer it takes to reach the long-term performance level. Since these impacts, by their nature, are weighted to the early years of a new development, they have more impact at higher target IRRs. To minimise risks of learning impacts, there is a strong incentive to start any new irrigation development with well-established crops and technologies, and to be thoroughly prepared for the anticipatable agronomic risks of establishing new farmland. Higher- risk options (e.g. novel crops, equipment or practices that are not currently in profitable commercial use in analogous environments) could be tested and refined on a small scale until locally proven. As indicated in the examples above, the influence of each risk individually can be quite modest. However, it is the combined influence of all foreseeable risks that need to be accounted for in planning and the cumulative effect of these risks can be substantial. For example, see the last question in Table 8-1 for the combined effect of just two risks, and see Stokes and Jarvis (2021) for the effects of a common suite of risks on the financial performance of a Bradfield-style irrigation scheme. Table 8-10 Risk adjustment factors for target farm gross margins (GMs), accounting for the effects of learning risks Learning risks were expressed as the level of initial farm underperformance and time taken to reach full performance levels. Initial farm performance is the initial GM as a percentage of the GM at ‘full’ performance. IRR = internal rate of return. TARGET IRR (%) INITIAL FARM PERFORMANCE (%) RISK ADJUSTMENT MULTIPLIER FOR TARGET FARM GROSS MARGINS (VS BASE 100% RELIABILITY TABLES) (unitless ratio) Learning time (years to 100% performance) 2 4 6 8 10 15 3 85 1.01 1.02 1.03 1.03 1.04 1.05 75 1.02 1.03 1.04 1.05 1.07 1.10 50 1.04 1.06 1.09 1.12 1.14 1.21 25 1.06 1.10 1.14 1.19 1.23 1.35 0 1.08 1.14 1.20 1.26 1.33 1.53 7 85 1.02 1.03 1.04 1.05 1.05 1.07 75 1.03 1.05 1.06 1.08 1.09 1.13 50 1.06 1.10 1.13 1.17 1.21 1.29 25 1.09 1.15 1.22 1.28 1.35 1.51 0 1.12 1.21 1.31 1.41 1.52 1.83 10 85 1.02 1.03 1.05 1.06 1.07 1.09 75 1.04 1.06 1.08 1.10 1.11 1.15 50 1.08 1.12 1.17 1.21 1.26 1.35 25 1.12 1.20 1.28 1.36 1.44 1.65 0 1.16 1.28 1.41 1.55 1.69 2.10 8.7 Achieving financial viability in a new irrigation development Four key factors determine the financial performance and viability of irrigation schemes: capital costs of development, farm performance (that determines trajectories of future water demand and associated benefits from increased GVAP), risk (and associated required level of investment return), and value adding beyond the farm gate (Stokes et al., 2017). Designing a new irrigation project would require balancing these four factors to find combinations that might collectively constitute a viable investment. As demonstrated by lessons from recent dam developments in Australia (Chapter 6), it can be difficult to fully balance the costs of new water infrastructure with the direct new benefits they generate. In concluding this chapter, the broad principles for balancing each of the factors analysed is discussed below. Lowest capital costs of development – cheapest water As highlighted in the companion reports in this Assessment, developing water resources suitable for irrigation in northern Australia is technologically challenging and opportunities are limited. The costs of developing these resources vary widely (Petheram et al., 2022; Chapter 6), such that even when technically feasible options are found, many of these are likely to be too expensive for irrigation schemes. Capital costs of developing new water sources are high and a key determinant of scheme financial viability. Results suggest broadacre farms with GMs of $4000 per ha per year would generate sufficient revenue (while providing a 10% IRR to farmers) to cover the costs of about $20,000 to $30,000/ha of off-farm water infrastructure, before accounting for the negative effects of risks. This would cover about 50% of the costs of the most cost-effective large dam development option in the Roper catchment (about $50,000/ha off-farm water infrastructure cost; at a 7% target IRR for the public investor). Although irrigators are therefore unlikely to be able to pay the full costs of a publicly developed large dam, they may be able to cover a greater proportion of costs (>25%) than in many existing irrigation developments (Vanderbyl, 2021). On- farm water sources appear to provide good opportunities for affordable water that could support broadacre and cost-efficient horticulture but developing these resources would need to concentrate on the most cost-effective sites. Highest farm gross margins – best crops, soils, and niche opportunities The companion report on land suitability (Thomas et al., 2022) highlights where the best soils for various farming options are likely to be found (summarised in Section 3.2), and Chapter 5 assessed a range of farming options, including opportunities and constraints on maximising farm performance (including farm GMs). There are likely to be niche opportunities for farmers to improve GMs by taking advantage of cost savings and price premiums, but these are unlikely to be scalable. Periods of high prices for agricultural commodities (such as the recent high prices for cotton) provide opportunities for new industries to establish by creating a buffer for learning during the crucial startup years when farms and associated supply chains will not yet be performing at their full sustainable potential. Reducing investor risk – making lower investor returns acceptable There are numerous risks that confront large infrastructure projects, such as new irrigation schemes. The higher these risks, the higher the return an investor would likely require, raising the performance thresholds a project would have to attain to be commercially viable. Conversely, lowering those risks lowers the target revenues that scheme investors would need to generate, which could contribute to making a potential investment viable. One of those risks is the paucity of background information required to develop new irrigation schemes in northern Australia. The information provided in the companion reports in this Assessment is targeted at addressing this information gap and reducing the uncertainty about the physical resources in the Roper catchment, and how they might be developed. Some risks can be avoided through careful planning, learning from past cropping experiences in northern Australia, and starting with well-established crops, technologies and management practices. Risks that cannot be avoided need to be managed, mitigated where possible and accounted for in determining the realistic returns that can be expected from a scheme. This would include having adequate capital buffers to survive through challenging periods that may exacerbate negative cashflows in the initial years of establishment. Another perceived risk for investors is that of uncertainty around future policy, regulation changes, and tenure rights for land and water. Reducing this, or any other sources of risk, would contribute to making marginal investment opportunities more attractive. Value adding and synergies Value adding and synergies could contribute to the viability of a new irrigation scheme. The establishment of a new cotton gin near Katherine provides opportunities for local processing and provides natural synergies for the local use of cotton seed as a cattle feed supplement within the Roper catchment. Other synergies that could also be considered to improve scheme revenues or reduce costs would include: sequential cropping systems (increasing net farm revenue by growing two or more compatible crops from the same field each year; Section 5.3); integration of irrigated forages into existing beef enterprises (Section 5.4); including small-scale, high-value crops in the mix of farms in a scheme; expanding the scale of a scheme with extra dryland/opportunistic cropping around the irrigated core; and improving transport infrastructure and supply chains (reducing the disadvantages of remote locations). Location-appropriate production systems would need to be developed and proven for some of these options. Conclusion Ultimately no single one of the above factors is likely to provide a silver bullet to meet the substantial challenge of designing a commercially viable new irrigation scheme. Achieving financial viability will likely require contributions from each of the above factors, with careful selection to piece together a workable combination. 9 Regional economics 9.1 Multiplier and input–output (I–O) approach When new economic activity begins in a region, such as with the development of a water infrastructure project, there will be knock-on effects to the wider regional economy, over and above the impacts directly related to the development scheme itself, through the way the new activities affect the flows of local goods and services. These effects can be both positive and negative. This section uses regional multiplier and input–output (I–O) analysis to estimate the regional economic benefits that could arise if new irrigated development were to occur in the Roper catchment. When evaluating the regional economic impact of new irrigated agricultural development within the Roper catchment, there are two separate analyses required for each of the two distinct phases of the scheme. Firstly, the initial temporary impact from the construction activity at the start of the project. This is followed by the ongoing impacts arising from the increased agricultural production within the region once the development becomes operational and the new farming operations are up and running. The approach closely follows the regional economic analyses used in previous similar water resource assessments (Stokes et al., 2017; Stokes and Jarvis, 2021), and further details of the approach, including discussion of the relative strengths and weakness of I–O analyses, are covered in those reports. To briefly summarise, I–O multipliers are widely used to quantify economic impacts of projects (at regional or national level), offering clear advantages of transparency and ease of use compared to other methods. Simplistically, the method enables an estimate to be made of the total regional or national impact of the development project including the direct spend of the project itself, plus all the production and consumption-induced (knock-on) impacts on other businesses and households within the region. The I–O multiplier approach recognises that the full impact of the economic stimulus provided by an irrigated agricultural development project extends far beyond the impact on those businesses and workers directly involved in either the short term (the construction phase) or longer term (the ongoing agricultural production phase). Those businesses directly benefitting from the increased construction (short term) and agricultural activity (longer term) would need to increase their purchases of goods and services, which would stimulate economic growth in the regions where those products were purchased. These impacts are known as production-induced effects. Furthermore, household incomes increase when local residents are employed as a consequence of the direct and/or production-induced business stimuli. A proportion of this additional income is spent within the region, creating additional demand, which serves to further stimulate regional economic activity. This additional economic activity is known as a consumption-induced effect. The size of the production-induced and consumption-induced benefits can be quantified by the economic multiplier. Regional (or national) I–O multipliers are summary measures used to estimate the total economic impact on all industries within a region (or nation), from a change in demand for the output of any particular industry (McLennan, 1996). The key output from the I–O models is the estimated value of the increased economic activity (including, when focusing on an irrigated agricultural development, the original increase to construction or agriculture), where larger multipliers generate larger regional benefits. The models also estimate the increase in household incomes in the region. From this estimate the approximate number of jobs represented by this increased economic activity can be calculated (including those directly related to the increase in construction or agriculture and those generated by the indirect production and consumption effects). Thus, I–O models can be used to estimate the impact of new irrigation development on employment, income and regional economic activity during each phase (development and operational), encompassing all of the direct and indirect impacts expected as a result of the development. I–O tables and associated multipliers can be constructed at a national or regional scale. With national models, inputs and outputs by industry sector reflect national production and spending patterns, while additional data reflect international imports and exports. For Australia, the ABS releases national I–O tables at regular intervals, with the latest release being for the financial year 2018–19 (ABS, 2021a). However, despite publishing the national I–O tables, the ABS has not compiled and published national I–O multipliers based on these tables since 1998–99 (leaving such a step to data users who can use the published national I–O tables to calculate multipliers at national level) due to concerns that provided multipliers could be used for purposes where they are unsuitable, or where lack of consideration is given to their inherent shortcomings and limitations; these limitations are discussed further below (Section 9.1.2). The ABS does not prepare or publish I–O tables at sub-national scale. Regional models focus on a specific region and thus contain a spatially delimited subset of the expenditure patterns used in national models. They also require additional data to identify inter- regional imports and exports and to quantify other regional-specific spending patterns (Jarvis et al., 2018). This is necessary as relationships between industries within a region are not identical to those at the national scale. Typically, smaller and more remote geographic areas have smaller multipliers as inter-industry linkages tend to be shallow and the region’s capacity to produce a wide range of goods is low, meaning that inputs and final household consumption are less likely to be locally sourced than in regions with larger urban centres (Jarvis et al., 2018; Stoeckl and Stanley, 2009). Furthermore, firms in rural and remote areas may have disparate access to production technologies, are often less able to take advantage of economies of scale, and face different relative input prices than their counterparts in developed urban areas (Stoeckl, 2012). In addition to differences in economic scales between regions, different industries are also more or less prominent in different regions; these differences can have an impact on the relative size of multipliers comparing region to region and industry to industry (Jarvis et al., 2018). Accordingly, where available, regional-specific models should be selected for use in the analysis. The regional context is vital, particularly in rural remote areas that likely have different characteristics compared with the rest of the country. Unfortunately, regional I–O tables are rare, and infrequently prepared; the lack of a recent and regionally specific model is accepted as a limitation of this work. When considering the regional economic impact of such a development it is important to be aware that not all of the expenditure generated by the scheme will occur within the local region. The greater the leakage (that is the amount of direct and indirect spend made outside the region), the smaller the resulting economic benefit that will be enjoyed by the region. Conversely, the more of the initial spend, and subsequent indirect spend, that is retained within the region the greater the economic benefit, and the number of jobs created, within the local region. Accordingly, where there is leakage to other regions the local knock-on benefits would be reduced, but there would be benefits in the other regions where the expenditure occurred instead; thus, the irrigated agricultural development would provide benefits to those other regions across Australia who were the recipients of the additional demand for goods and services stimulated by the irrigated agricultural development. However, in instances where the leakage is to other countries, such as when capital items are imported, the benefits would flow outside of Australia. Thus, the economic impact of the project that remains within the local region, or within the country, is dependent on the skills and resources available locally and nationally, and leakage issues can be mitigated by careful design of the project, in both the construction and operational phases. Leakage from the local region can be minimised if local resources, including local workers and local businesses, can be employed; leakage from the region to elsewhere in the nation represents lost benefit locally but is offset by benefit gains elsewhere which may be viewed as ‘no net loss’ overall if the project is funded at the national level. However, leakage from the local region to international suppliers is a true loss, which may be minimised by careful project design but may be unavoidable if particular resources and skills can only be sourced overseas. Another important consideration for model selection, beyond the specific geographic location, was the demographic characteristics of the region. The Roper catchment population includes a much larger proportion of people identifying as Indigenous compared to Australia as a whole, and this characteristic can have a significant effect of any development in the region. Research based on small, remote communities has found that the expenditure patterns of Indigenous communities differ from the typical patterns elsewhere in Australia (Stoeckl et al., 2013). Additionally, Indigenous Australians are less likely to be in formal employment (government- sponsored employment schemes often involve a transfer of public funds from outside the region) and are proportionally more likely to be employed in the public and health sectors than non- Indigenous residents (Stoeckl et al., 2011). Accordingly, the greater proportion of Indigenous people within the region compared to the national average further underpins the necessity of using I–O tables derived from local data, or from data as close to local as possible, rather than basing analyses on models drawn from dissimilar regions. As noted above, recent regional I–O models are rare, and unfortunately no model exists that is specific to the Roper catchment. Hence, there was a need to source model(s) that could provide an approximation of the likely impact of this development for this region. The analyses used two different I–O models to reflect the nature of the region (rural, remote and with a significant proportion of the population identifying as Indigenous). These models were prepared independently at different times and using different approaches. Each model is used to provide insights into the likely economic impact of this development, and to reinforce the robustness of the findings by triangulating results. Details of each of these models, and the appropriateness of each for providing insights into the likely regional economic impacts from a development within the Roper catchment is discussed in more detail in the following section (Section 9.1.1). This report focuses on the total output multipliers (referred to as Type II multipliers). Type II multipliers consider initial (direct) expenditure and intra-industrial ‘knock-on’ benefits along the business supply chain (as measured by simple output Type I multipliers) as well as ‘knock-on’ effects linked to the local expenditure of (household) wages and income (McLennan, 1996; Gretton, 2013). 9.1.1 Description of the two regional models used for I–O analysis For the analysis presented here, regional multipliers were derived and modified from two separate publicly available sources: (i) a regional I–O table developed by the Office of Resource Development of the NT Government providing coverage for the entire NT region (Murti and Northern Territory Office of Resource Development, 2001); and (ii) a regional I–O table developed specifically for the Daly River catchment, adjacent to the Roper River catchment in the NT (Stoeckl et al., 2011). Figure 9-1 shows the relative geographic locations of these I–O regions and Table 9-1 summarises their socio-economic characteristics for comparison. For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au Figure 9-1 Regions used in the input–output (I–O) analyses relative to the Roper catchment Assessment area The first model covers a much wider geographic scale (1,348,094.3 km2 for NT compared to 77,352.2 km2 for Roper) and also includes a large city, the territory capital Darwin (population 118,456 as at 2016 census for the urban centre). These differences can have an impact on the relative complexity of the economic structures in each region. Agriculture is more important to the Roper catchment than to the whole of the NT, providing 14.0% and 2.0%, respectively, of employment within each region. Due to the larger scale (on geographic, population and level of economic activity measures), the multipliers estimated using the whole of NT model will likely be larger than the multipliers for a small region within the NT. This is because of a number of factors (including less opportunities to take advantage of economies of scale, increased input prices and reduced access to production technologies, as described above), compared to firms in developed urban centres such as Darwin (Stoeckl, 2012). Accordingly, the estimated multipliers are likely to provide upper bounds estimates of the multipliers for the Roper catchment region. However, it is likely to be a more appropriate estimate of the magnitude of the impact of a water development on the economic activity within the wider region, including Darwin; it is likely that the potentially lower impact within the Roper catchment itself will ‘leak’ into the more urbanised locations within the NT. Table 9-1 Key 2016 data comparing the Roper catchment with the related I–O analysis regions Population statistics for Roper and Daly regions have been estimated based on the weighted average of 2016 census data (ABS, 2016) obtained by SA2 statistical region, with weighting based on the proportion of relevant ABS SA2 statistical regions falling within each of the catchment regions. ROPER CATCHMENT1 DALY CATCHMENT I–O REGION† NT I–O REGION‡ Land area (km2) 77,352.2 53,088.5 1,348,094.3 Population 2,512 11,312 228,833 % male 51.10% 52.07% 51.82% % Indigenous 73.35% 28.66% 25.45% Median age 28 32 32 Median household income $61,852 $84,328 $99,580 Contribution of agriculture, forestry and fishing to employment in the region 14.0% 6.2% 2.0% Major industries of employment – top three industries in region as % of employment 2016 • Largest employer in region Public administration and safety Public administration and safety Public administration and safety • 2nd largest employer in region Education and training Health care and social assistance Health care and social assistance • 3rd largest employer in region Agriculture, forestry and fishing Education and training Construction Gross value of total agricultural in region $60 million $49 million $697 million † Statistics for Roper (ABS, 2016a) and Daly (ABS, 2016b) regions have been estimated using the weighted average of ABS 2016 census data obtained by SA2 statistical region, with weighting based on the proportion of relevant ABS SA2 statistical regions falling within each of the catchment region. ‡ ABS 2016 census data (ABS, 2016c). § ABS Value of agricultural commodities produced 2015-16 by region, report 75030DO005_201516 (ABS, 2017). The NT I–O table was originally prepared by the Centre for International Economics (1997), and subsequently updated by Murti and the Northern Territory Office of Resource Development (2001). This I–O table utilises data from 1997–98 and incorporates inputs and outputs relating to 50 industry sectors operating within the region. See Murti and Northern Territory Office of Resource Development (2001) for additional detail on the methods and data used to prepare this table. While more recently compiled I–O table data would have been desirable, in general industry relationships within regions change slowly and the multipliers generally remain fairly stable over time (McLennan, 1996). The analyses presented here aggregated the 50 industries in the model to a smaller subset of 22 industry classes, both to reflect the nature of economic activity in the Roper catchment and for consistency with similar previous analyses done for northern Australian catchments that used the same industry aggregations (Stokes et al., 2017; Stokes and Jarvis, 2018). The agricultural sectors included in this aggregated model were ‘beef cattle’; ‘agriculture excluding beef cattle’; and ‘aquaculture, forestry and fishing’. For this study the NT I–O model used by Stokes and Jarvis (2018) was further amended to disaggregate of the impacts on household incomes and job creation between Indigenous and non-Indigenous households, using disaggregation methods established by Jarvis et al. (2018). The second I–O model focuses on the smaller geographic region of the Daly catchment, which is directly adjacent to the Roper catchment and of similar geographic size (Daly catchment ~53,100 km2, Roper ~77,400 km2). This I–O model was developed using a different methodology to the NT model, being based on survey data; prior research has established the appropriateness of such an approach (Stoeckl, 2007, 2012). Further, this model was specifically developed to explore the economic impacts of different types of development in remote, sparsely populated regions of northern Australia; such regions generally include a far greater proportion of Indigenous people compared to Australia as a whole. The model was developed based on highly aggregated industry sectors, containing one agriculture sector, rather than any subdivision of the economic impacts of different types of agriculture within the region. As with the NT model, the Daly catchment model separately estimates the impacts on incomes for Indigenous and non-Indigenous households. Although the Daly and Roper catchments are comparable in terms of their geographic location and demography, there are also some notable differences. The Daly catchment is physically located closer to the major city of Darwin, (see Figure 9-1), suggesting the Daly region is more likely to realise spill-over benefit from Darwin’s regional economic activity than is the Roper catchment; however, the better road links to the Roper than the Daly are likely to negate this issue. The Daly includes the important regional centre of Katherine (population 6303), while the Roper’s largest settlement is Ngukurr with a population of 1149 people, based on the 2016 census data; however, Katherine is located close to the Daly/Roper border and is connected to both regions via road and electricity infrastructure. Agriculture is more important to the Roper catchment than to the Daly catchment, providing, respectively, 14.0% and 6.2% of employment within each region, although this is likely due to the impact of those employed and residing in Katherine. Annual agricultural production in the Roper catchment is similar in value to that of the Daly catchment ($60 million and $49 million GVAP, respectively). Thus, overall, while the Daly catchment may not be a perfect comparator to the Roper catchment, the regions are sufficiently similar that it is likely that the use of this model will provide a good estimate of the impact of the development to the Roper catchment and the surrounding regions overall; that is spill-overs from the Roper catchment to the adjacent catchments are likely to be reflected in the results. Using each of these models in turn, the total regional benefits from the operations of an irrigation development including all multiplier effects (indirect production effects, and the consumption effects linked to the local expenditure of (household) wages and income, in addition to initial direct effects) is estimated using I–O analysis. The regional economic impact from the construction phase and from the ongoing agricultural production phase of the development are estimated separately. The I–O analysis incorporates the value of the anticipated additional agricultural output directly driven by new development as an exogenous shock to the appropriate industry then estimates how much additional activity is generated within each I–O region as a result of the exogenous shock. The I–O analysis also estimates the likely increases to household incomes in the region. This increase in income was used to estimate the increase in jobs created in the I–O region (directly, and indirectly through production and consumption effects), by dividing the total increase in household incomes by the average income in the I–O region. Specifically, the estimated number of jobs was calculated as follows: 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑗𝑗𝑗𝑗𝑗𝑗𝑗𝑗 = 𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖 ℎ𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜ℎ𝑜𝑜𝑜𝑜𝑜𝑜 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑖𝑖𝑖𝑖 𝑁𝑁𝑁𝑁 Where ‘Estimated mean employee income in NT’ has been calculated based on latest available mean employee income data for the NT (as at June 2021) from the ABS (ABS, 2021b) updated using wage price indices to more current wage rates based on the ABS wage price index data series (ABS, 2021c). Specifically, the estimated mean employee income in NT was calculated to be $71,329 based on the following calculation: 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖=𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐽𝐽𝐽𝐽𝐽𝐽 ‘18× 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 ‘21 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 𝐽𝐽𝐽𝐽𝐽𝐽 ‘18 Because the purpose of the analysis was to estimate the number of new jobs created, incomes were specifically estimated only for employees (because including income from pensions or other non-employment sources would distort job estimates). It should be noted that this method results in an imperfect estimate of the number of jobs created and is affected both by the limitations of I–O analysis and by the assumption that the mean income from additional direct and indirect jobs created will be the same as the current mean income level in the NT. However, it provides some guidance to the likely employment opportunities that could result from development within the region. As with all estimates resulting from I–O analysis, this should be considered as an upper bound estimate. 9.1.2 Strengths, limitations and inherent weaknesses of using I–O multipliers to assess regional economic impacts When using I–O analysis and I–O-derived multipliers for analysis of regional economic impact of developments, such as an irrigated agriculture development within the Roper catchment, it is important to be aware of the inherent weaknesses and limitations of the approach, in addition to the factors that make the approach appropriate for use in such a location. Firstly, the approach fails to recognise or incorporate any supply-side constraints or budget constraints, the omission of which results in estimates overstating the likely economic impacts. Secondly, the approach assumes the structure of the regional economy, including interlinkages between different sectors of the economy and ratios of leakages from the region, does not change either over time or as a result of policy or technological advancements. That is, the approach assumes fixed prices, fixed ratios for intermediate inputs and production, and makes no allowance for purchasers’ marginal responses to change. Again, such omissions are likely to result in biased estimates of economic impacts. More detailed discussion of the limitations of I–O multiplier analysis can be found in Stokes et al. (2017) and ABS (2021a). These inherent limitations should be borne in mind when considering the results set out below. Specifically, the limitations and weaknesses inherent in the assumptions underpinning the approach results in multipliers providing biased estimates of the benefits or costs of a project. However, despite these acknowledged limitations, there are aspects of the I–O multiplier approach that make it well-suited for scoping assessments of the potential regional benefits of greenfield developments in remote parts of Australia, which is how multipliers are used here. Firstly, alternate approaches to estimating regional economic impacts, such as Computable General Equilibrium (CGE) models, are also imperfect and frequently suffer from similar limitations. Furthermore, models using alternate approaches are generally unavailable for small rural and remote regions such as the Roper catchment, and CGE modelling has proven unsuitable, or to offer little benefit compared to I–O models, in previous similar applications for estimating the regional impacts of small agricultural developments that represent such a minor perturbation to regional economies. To provide two examples: • The use of the TERM dynamic multi-regional CGE model was trialled for assessing the impact of water resource developments within the Flinders and Gilbert catchments (Brennan McKellar et al., 2013). This model has been described in great deal (see Wittwer, 2012), and when developed was theoretically an improvement on prior models due to providing finer regional divisions and disaggregating the economy across a wider number of industrial sectors, and indeed, appears to have been successfully used in a number of examples such as within the southern Murray–Darling Basin (Wittwer, 2012). However, in practice, for small remote data- poor regions such as the Roper catchment, the model suffers limitations similar to the simpler (and less costly) I–O approach. For example, the model was developed based on historical data (a national I–O table supplemented with some regional data, with a base year of 1997), and, despite the model disaggregating Australia into 57 different regions, these regions are insufficiently fine scale to match northern river catchments (the NT as a whole, for example, forms one region in the model) (Horridge, 2012). • The use of the ABARES’ AusRegion dynamic regional CGE has been trialled for assessing the regional economic impact of water resource developments across a number of different water resource developments (Ash et al., 2014), seeking to demonstrate the impact of development compared to a reference scenario, on output, employment and wages at the regional level, and on output at state and national levels. This approach offered the advantage of providing bottom-up estimates of impact at different scales (regional, state, national), but suffered limitations similar to the I–O approach, in that the model was based on historical data (2005–06) regarding the structure of the economy and interrelationships between industries and regions, and also that the model failed to disaggregate regional data to the fine scale required to match northern river catchments. Further, the usefulness of the approach was limited by the model outputs, which represented, for one future time period only (2029–30) the percentage difference between the economic indicators for that year under the development scenario compared to the reference scenario. These results appeared to emerge from the complex model as if from a ‘black box’ approach, with no quantification, or indication of relative importance, of the different drivers underpinning the cumulative effect of construction and operational phases over time, or of the changing impacts over time; thus the results provide little to no guidance on how these component regional economic impacts could change in response to variations in the assumptions underpinning the scenario (Ash et al., 2014). Thus, taking into account the practical considerations of working with a small, remote region, I–O analysis is the most suitable approach for assessing regional economic impacts in the Roper catchment (and similar river catchments in northern Australia). Secondly, a region such as the Roper catchment provides a data-sparse environment with few local precedents or alternate developments that can act as proxies to guide the likely outcomes from irrigated agricultural developments within the region. Alternate approaches to I–O, such as the CGE models, are data intensive, requiring detailed information on the structure of the economic system in the region under study to enable the regional economic model to be specified and parameterised. This includes structural information on production, consumption and trade, for example, and additionally, further data on behaviours, describing how this system will respond to changes; this usually requires information in the form of elasticities of demand, production and trade. The problems of gathering reliable data for remote regional areas are well known. Beyond the physical challenges of collecting data from remote, hard-to-access regions (due to poor transport and communication infrastructure, difficult terrain and extreme/variable weather conditions), the reporting of socio-economic data is also frequently suppressed or distorted for confidentiality purposes.8 These data problems are exacerbated for regions with smaller and highly Indigenous populations. Taylor (2013) describes some of the flaws in data relating to Indigenous people and their communities, which raise questions of data accuracy and also comparability of data over time. Some of the reasons noted as underpinning data issues include a changing propensity of individuals to identify as Indigenous over time and more frequent mobility of Indigenous people (Taylor, 2013); these problems are exacerbated by the acknowledged significant undercounting of Indigenous people in official statistics such as census data, thought to be 17.5% for the 2016 Census (ABS, 2018b). In such scenarios, the less data hungry I–O approach can be reasonably robust, compared to theoretically preferred but more data hungry approaches such as CGE, thus making the approach attractive in such a region as the Roper catchment (data- sparse, remote and with a large proportion of the population identifying as Indigenous). 8 An explanation from the ABS can be found at https://www.abs.gov.au/statistics/detailed-methodology-information/concepts-sources- methods/survey-income-and-housing-user-guide-australia/2019-20/using-survey#confidentiality. Finally, I–O approaches can usefully provide an estimate of the upper bound of regional knock-on effects at relatively low cost. By indicating the likely upper bound of these added benefits, they enable the easy identification of schemes that are likely to produce benefits too small to be able to justify substantial public subsidies for schemes that are not close to being financially viable in the first place. For such schemes, a more precise measure of regional benefits would be unlikely to change substantive decisions about whether a scheme would go ahead or not (e.g. even if the knock-on benefit was 50% lower or 50% higher this would not be enough to tilt a decision on a financially non-viable scheme becoming viable or vice versa). For situations where CBA indicated that a scheme was very close to being viable and regional benefits were to be a critical deciding factor, then it would be appropriate to go beyond indicative I–O analysis and invest in additional bespoke regional economic analysis in those specific cases. This is in line with the activities in the other parts of this Assessment which, as a combined scoping study, is primarily aimed at assisting investors and government planners in identifying where potential opportunities lie (distinguishing development options that might be viable from those that can easily be ruled out), with the expectation that specific project proposals would need to conduct additional feasibility analysis. In summary, the lack of better alternate approaches for many rural and remote Australian regions, combined with their adaptability and ease of use, makes I–O multipliers a popular and suitable tool for scoping-level economic impact analysis. When used appropriately, I–O multipliers can provide valid and useful information, provided results are carefully interpreted with due consideration for the key assumptions and limitations that underpin the models (Gretton, 2013; Office of the Government Statistician Queensland Government, 2004). Accordingly, when using the results presented in the section below (Section 9.2) it is important to recognise the effect of the limitations is that regional benefit values are likely to represent an upper bound estimate of potential outcomes (ABS, 2019). This becomes a more significant issue with larger and more complex developments, as smaller and fairly simple developments are less likely to distort current markets, place price pressures on supply chains and labour markets, and require imports (materials, equipment, skilled labour) from overseas. 9.2 Regional economic benefits I–O analysis was applied to two distinctly separate streams of benefits: the ‘one off’ benefits that arise during the construction phase of a new irrigation scheme and the ongoing benefits that arise during the operational phase when new farming production begins. Accordingly, the multipliers and estimated regional impacts of the two distinct phases are considered separately below. Estimates of regional benefits included all Type II multiplier effects (indirect production-induced effects, and the consumption-induced effects linked to the local expenditure of household wages and income, in addition to initial direct effects). The analysis is not based on any specific construction project and subsequent irrigated agricultural operations. Rather, the analysis provides information illustrating potential regional outcomes from a range of alternate types/scales of construction projects and the agricultural operations that could flow from these. The initial source of funding for the capital cost of the water infrastructure development is not addressed in this analysis, that is, the analysis does not explicitly address the upfront source of the capital funding (be it private sector, public sector or public-private partnership). Similarly, the analysis during the operational phase doesn’t explicitly address the size/proportion of contribution to construction cost required from farmers compared to possible public subsidies provided. However, the implications of funders and funding mechanisms are discussed in each of the sections below. 9.2.1 Regional economic benefits arising during construction phase (one off) While the initial cost of building irrigation infrastructure (and other related infrastructure such as new roads) is a cost, by creating additional expenditure within a region (thus putting additional cash into people’s/firm’s pockets) this increases regional economic activity. Thus, this creates a fairly short-term (non-recurring) economic benefit to the region during the construction phase. Construction industry multipliers should be applied to the annual expenditure on construction over the duration of the construction phase of the project, that is, they estimate the impact on the regional economy of the construction activity for the year in which that construction activity takes place. This regional economic impact will be of a ‘one off’ nature; that is, the benefit will not be repeated in subsequent years. A scenario approach was adopted for the scales of development considered in estimating the regional impact of the construction phase of potential developments. The analyses modelled regional impacts for five different indicative sizes of developments in the Roper catchment, with capital costs from $250 million to $4 billion. These total capital costs include costs of labour and materials required by the project. The smallest scale of development in Table 9-2, with a capital cost of $250 million, would broadly represent about new 20 new farm developments with their own on-farm water sources enabling around 10,000 ha of irrigation for horticulture and broadacre farming (based on the costing information previously presented for on-farm establishment costs in Table 7-1 and for on-farm water source developments in Table 7-2). The second-smallest scale scenario, $500 million capital cost, could represent a similar development to the first but with 20,000 ha of new irrigated farmland; this level of investment could also include a new processing facility (such as a cotton gin) that could be required and supported from this scale of agricultural development. Alternatively, the $500 million scale of development could represent a large off- farm water infrastructure development (based on indicative costings for the most cost-effective dam locations in the Roper catchment, Table 7-3) along with related farm establishment costs (Table 7-1). The larger scales of development, at $1 or $2 billion shown in Table 9-2, indicate outcomes from combining potential developments in different ways (such as one large off-farm dam and multiple on-farm water sources), and also including investment in indirect supporting infrastructure across the region, such as investment in roads, electricity and community infrastructure (as described in sections 7.5, 7.6 and 7.7). Careful consideration was given to estimating the appropriate proportions of initial spend during the construction phase that would actually be spent within the region. The costs incurred during this phase would include labour, materials and equipment costs. For labour costs, it is likely that the wages would be paid to workers sourced from within the region and from elsewhere, with the likely proportion of labour costs relating to each source of workers being dependent on the availability of appropriately skilled labour within the region. For example, a highly populated region with a high unemployment rate is likely to be able to supply a large proportion of the workers required from within the region; however, a sparsely populated region like the Roper catchment with fewer trained construction workers would likely to need to attract many workers from outside the region, on a fly-in fly-out (FIFO) and/or drive-in drive-out (DIDO) basis. Similarly, for materials and equipment, some regions may be better able to supply a large proportion of these items from within the region whereas construction projects in other locations may find they are unable to source what they need locally and instead import a significant proportion into the region from elsewhere. A scenario approach was again adopted, indicating the impact that would result from three different proportions of local construction spend (labour, equipment and materials) that could be sourced within the region (as opposed to being imported which has no impact on the regional economy): 65% (i.e. low leakage scenario), 50%, and 35% (i.e. high leakage scenario) spent locally. For a very remote region such as the Roper catchment, the potential exists for leakage to be higher than this high leakage scenario (i.e. <35% spent locally), resulting in a lower benefit to the local Roper catchment regional economy; however, when considering the wider region encompassing the Roper and adjacent regions within the northern NT, this range of likely leakage scenarios is likely reasonable. In cases of high leakage, the knock-on benefits would instead occur in the regions supplying the goods and services (like the wider NT I–O region). Utilising four possible scales for development capital construction costs together with three possible levels of spend to be made locally resulted in 12 different construction scenarios. Each of these scenarios was processed through the two separate I–O models to estimate the potential regional benefit from the construction phase (including the Type II multiplier effects). The results of this analysis are set out in Table 9-2. The values are the total benefits over the duration of construction, so annual benefits would be split according to the expenditure in each year and would cease once the construction phase was complete. Table 9-2 Regional economic impact estimated by I–O analysis for the total construction phase of an irrigated agricultural development based on estimated Type ll multipliers determined from two independent I–O models Estimates represent an upper bound because some assumptions of I–O analysis are violated in the case of such a large public investment in a region where existing agricultural activity is so low. Leakage to other regions and other countries is accounted for by reducing the proportion of expenditure (and benefits) within the region. DEVELOPMENT CAPITAL COST ($ billion) TOTAL REGIONAL ECONOMIC ACTIVITY WITHIN I–O REGION AS A RESULT OF THE CAPITAL COST OF THE DEVELOPMENT ($ billion) Roper catchment based on NT I–O model Roper catchment based on Daly catchment I–O model Proportion of total scheme-scale capital cost made locally within the I–O region 65% 50% 35% 65% 50% 35% 0.250 0.33 0.26 0.18 0.35 0.27 0.19 0.500 0.67 0.52 0.36 0.71 0.55 0.38 1.000 1.34 1.03 0.72 1.42 1.09 0.76 2.000 2.68 2.06 1.44 2.83 2.18 1.53 As can be seen from these results, the proportion of scheme construction costs spent within the region, (indicating how much of the initial exogenous shock is retained within the region rather than being lost in leakage to elsewhere) has a significant impact on the size of the regional economic benefit experienced. If a large proportion of the initial spend leaks from the region, then the benefit of the initial construction investment will be less concentrated in the local Roper catchment economy and would spread to other locations from where goods and services are sourced. Comparing the results of the two separate I–O models reveals the estimates to be fairly similar: the construction multiplier for the Daly catchment model estimates 5.8% higher regional benefits than the NT model (noting that rounding slightly affects comparisons of presented values). It is clear that the more significant differences relate to the proportion of spend that is spent locally, and the absolute capital cost of construction; differences resulting from use of the different regional multiplier models are relatively minor in comparison. The combined direct and indirect impacts on household incomes, resulting from each of the scenarios were also estimated using the two models. Based on the NT I–O model, only 6% of the increased household incomes flows to Indigenous households, despite Indigenous people comprising around 25% of the population of the NT. Based on the Daly catchment model, 8% of the increased household incomes flow to Indigenous households, despite Indigenous people comprising around 29% of the population of that region. Thus, both models clearly indicate that the benefits flow disproportionately towards non-Indigenous households. This reflects the lower level of Indigenous engagement (compared to that of non-Indigenous peoples) with the economic activity of these regions; this is not unexpected based on findings of previous research. This indicates that if the irrigated agricultural development is to contribute towards the government’s Indigenous Advancement Strategy (NIAA, 2021), and contribute towards achieving the ‘closing the gap’ targets (Australian Government, 2021) then specific interventions are likely to be required to increase Indigenous involvement in the construction phase of the project by specifically seeking to provide employment opportunities to the Indigenous people of the region where possible. Based on the estimated increase in household incomes, the number of direct and indirect jobs created during the construction phase were estimated (Table 9-3). The estimated number of jobs has been presented in total, rather than presenting the additional jobs likely to be created for Indigenous and non-Indigenous workers separately due to the many additional assumptions that would be required for such an analysis (over and above the assumptions on which the I–O analysis are based). Note, however, that Indigenous workers are likely to only fill a small proportion of new jobs created, because only 6% to 8% of additional household incomes is estimated by the I–O models to flow to Indigenous households. Table 9-3 Estimated full time equivalent (FTE) number of jobs created for the construction phase of an irrigated agricultural development Based on estimates of impact on household incomes calculated from I–O analysis (using Type ll multipliers determined from two independent I–O models) and average incomes per person, for the construction phase of an irrigated agricultural development. Analyses assume the construction phase and duration of jobs would be within one year: for longer construction periods the annual FTE would be lower but spread over more years. Estimates represent an upper bound because some assumptions of I–O analysis are violated in the case of such a large public investment in a region where existing agricultural activity is so low. Leakage to other regions and other countries is accounted for by reducing the proportion of expenditure (and benefits) within the region. SCHEME COST CAPITAL COST ($ billion) ADDITIONAL JOBS ESTIMATED TO BE CREATED AS A RESULT OF THE CAPITAL COST OF THE SCHEME (FTE) Roper catchment based on NT I–O model Roper catchment based on Daly catchment I–O model Proportion of total scheme-scale capital cost made locally within the I–O region 65% 50% 35% 65% 50% 35% 0.250 643 495 346 592 456 319 0.500 1,287 990 693 1,185 911 638 1.000 2,574 1,980 1,386 2,369 1,823 1,276 2.000 5,147 3,959 2,772 4,739 3,645 2,552 As expected, the estimated employment outcomes are closely related to those for impacts on regional economic activity, with a larger number of jobs created when there is assumed to be lower leakage rates and when the initial capital spend is larger. The analysis based on each of the two separate I–O models present different but similar results. Employment predictions based on the Daly catchment model are 8% smaller than the predictions based on the NT model; such differences are less significant than the differences due to assuming different leakage rates. The similarity between the estimated levels of economic activity that could result from the Roper catchment development from the two independently developed I–O models provides some reassurance of the robustness of the findings presented here. The regional benefits from the construction phase are estimated based on the assumption that the capital funding is an exogenous injection of spending into the region rather than some (or all) of the funding representing a diversion of current spend away from other construction projects within the region towards the new development; that is, the capital funding is treated as new expenditure over and above current activity. The benefits to the region under this assumption would be unaffected by whether the funding was derived from the public sector (federal or state) or private sources, or a combination. Should any of the construction represent a diversion from current regional spend, then the analysis should be considered to be based on the net injection (total spend on new project less current regional spend diverted) rather than gross expenditure to avoid overstatement of regional benefits. Further, the opportunity cost of the capital spend is not incorporated in this analysis, but the best alternate use(s) of the funding should also be evaluated before scarce available funding is committed to the development. 9.2.2 Regional economic benefits during the operational phase (recurrent) For assessing the regional economic benefits arising during the operational (farming) phase of an irrigated agricultural development, analyses used four scenarios as indicators of the possible scales of investment and types of development. These scenarios evaluated the impacts of increases in gross value of agricultural production from new agricultural development of $25, $50, $100 and $200 million per year. At the low end ($25 million per year), this could represent 10,000 ha of new plantation timber, while the high end ($200 million per year) could represent 10,000 ha of mixed broadacre cropping and horticulture (based on farm financial estimates for these crops presented in Section 5.2, with other crop options falling in between the two ends of this range). In each scenario, the additional agricultural output is considered once developments have reached their full potential. The different scales of increased economic output from agriculture, resulting from new water development, are stated net of any contribution the farmers are required to make towards the costs of building off-farm infrastructure. In practice farmers may be charged for the infrastructure development as part of their cost of acquiring a water entitlement and/or ongoing payments for water extraction, and these contributions may be subsidised to some extent by government grants towards the construction. Impacts were quantified in terms of the total increased economic activity (Table 9-4) in the region, followed by analysis based on the associated impact on household incomes and employment (using the approach described above). The multipliers estimated from the I–O analysis that were used to estimate the increased economic activity are summarised in Table 9-5. The estimated impact on household incomes, disaggregated between Indigenous and non-Indigenous households, is shown alongside the estimated increased number of jobs represented by that increased income (Table 9-6). Note that all results scale linearly as the economic output of each type of agricultural activity increases; likewise, a linear decrease in economic activity would result from a decrease in agricultural activity. As before, estimates were made based two independent I–O models: (i) the wider NT model estimated economic impacts from increased activity from within each of three categories of agricultural activity (‘beef cattle industry’, ‘agriculture excluding beef cattle’, and ‘aquaculture, forestry and fishing’), and (ii) the Daly catchment model estimated economic impacts from increased activity from within a general category of ‘agriculture’, encompassing all different possible agricultural activities. Table 9-4 Estimated regional economic impact per year resulting from four scales of direct increase in agricultural output (rows) in the Roper catchment, for the different categories of agricultural activity for two I–O models (columns) Increases in agricultural output are assumed to be net of the annualised value of contributions towards the construction costs. Estimates are based on Type ll multipliers determined from two independent I–O models for each year of agricultural production. Estimates represent an upper bound because some assumptions of I–O analysis are violated in the case of such a large public investment in a region where existing agricultural activity is so low. Leakage to other regions and other countries is accounted for by reducing the proportion of expenditure (and benefits) within the region. DIRECT INCREASE IN AGRICULTURAL OUTPUT PER YEAR NET OF CONTRIBUTION TO CONSTRUCTION COSTS ($ million) TOTAL VALUE OF INCREASED ECONOMIC ACTIVITY IN I–O REGION – DIRECT, PRODUCTION-INDUCED AND CONSUMPTION-INDUCED ($ million) Roper catchment based on NT I–O model Roper catchment based on Daly catchment I–O model Type of agricultural development Beef cattle Agriculture excluding beef cattle Aquaculture, forestry and fishing Agriculture of all types 25 51 37 70 51 50 103 73 141 102 100 205 146 282 203 200 411 292 563 406 Table 9-5 Type II regional economic multipliers applicable to the ongoing agricultural production phase of the Roper catchment development Estimates represent an upper bound because some assumptions of I–O analysis are violated in the case of such a large public investment in a region where existing agricultural activity is so low. Leakage to other regions and other countries is accounted for by reducing the proportion of expenditure (and benefits) within the region. ECONOMIC MULTIPLIER ($/$) Estimated using the NT I–O model Beef cattle 2.05 Agriculture excluding beef cattle 1.46 Aquaculture, forestry and fishing 2.82 Estimated using the Daly catchment I–O model Agriculture of all types 2.03 When applying the results of this analysis to a new irrigation scheme, based on the estimated increased value of agricultural output, it is important to be aware of all underlying assumptions, as explained previously. For example, the actual outcome may be quite different to that predicted by the analysis if the mix of agricultural activities within the I–O region is changed significantly from that in existence when the original I–O table was derived. Furthermore, the I–O method is generally considered to over estimate economic impacts, so the results are best used for relative comparisons among development options, or for providing an indication of the upper bound of the absolute magnitude of the regional benefit. Table 9-6 Estimated impact on annual household incomes and full time equivalent (FTE) jobs within the Roper catchment resulting from four scales of direct increase in agricultural output (rows) for the different categories of agricultural activity (columns) Increases in agricultural output are assumed to be net of the annualised value of contributions towards the construction costs. Estimates are based on Type ll multipliers determined from two independent I–O models for each year of agricultural production. Estimates represent an upper bound because some assumptions of I–O analysis are violated in the case of such a large public investment in a region where existing agricultural activity is so low. Leakage to other regions and other countries is accounted for by reducing the proportion of expenditure (and benefits) within the region. DIRECT INCREASE IN AGRICULTURAL OUTPUT PER YEAR NET OF ANY CONTRIBUTION TO CONSTRUCTION COSTS ($ million) TOTAL VALUE OF INCREASED ECONOMIC ACTIVITY IN I–O REGION – DIRECT, PRODUCTION-INDUCED AND CONSUMPTION-INDUCED ($ million or FTE) Roper catchment based on NT I–O model Roper catchment based on Daly catchment I–O model Type of agricultural development Beef cattle Agriculture excluding beef cattle Aquaculture, forestry and fishing Agriculture of all types Additional incomes expected to flow to Indigenous households from development ($ million) 25 0.8 0.1 0.9 0.5 50 1.6 0.2 1.7 1.0 100 3.3 0.4 3.4 2.0 200 6.5 0.8 6.8 4.0 Additional incomes expected to flow to non-Indigenous households from development ($ million) 25 7.1 1.7 14.3 6.75 50 14.2 3.3 28.7 13.5 100 28.4 6.7 57.4 27.0 200 56.8 13.4 114.7 54.0 Additional jobs estimated to be created (FTE) 25 111 25 213 102 50 222 50 426 203 100 444 100 852 407 200 888 199 1,704 813 As can be seen from the estimated regional economic impacts (Table 9-4), based on the model for the NT as a whole, an irrigation scheme that increases the output of the ‘beef cattle’ industry could have a larger impact on regional economic activity than a scheme that promotes ‘agriculture excluding beef cattle’, while the largest regional economic benefit would derive from an aquaculture, forestry and fishing focused development. These differences result from the different multipliers estimated for the different types of activities, as set out in Table 9-5. Using the alternate model for the Daly catchment, the estimated benefits from an agricultural development are similar, but slightly below, the estimates for a beef cattle focused development from the NT model. This finding is as expected, as the agricultural activity within the Daly catchment was heavily skewed towards beef cattle rather than other types of agriculture, as is the Roper catchment currently. Further, the estimate using the Daly catchment I–O model is smaller than that from the NT model which is also as expected based on economic theory (described above), as it is based on a much smaller (in geographic, population and existing level of economic activity scales) and more rural and remote region (including no major town or city). Given the most common type of agricultural activity across the Daly catchment (and the Roper catchment) is beef cattle, it is not surprising that the estimates from the Daly model are closest to the estimates arising from the NT model for beef cattle; the estimates from the Daly model are only 1% below the beef cattle estimates from the NT model. The similarity between the estimates of the level of economic activity that could result from Roper catchment development from the two independently developed I–O models provides some reassurance of the robustness of the presented findings. The analysis also estimated increases to household incomes within the region that would result from the exogenous boost in demand (assumed to equal the increased production facilitated by the development) to the agricultural industries using each of the I–O models. This increase in income was used to estimate the increase in jobs created in the region (directly, and indirectly through production and consumption effects), by dividing the total increase in household incomes by the estimated mean annual incomes in the region (calculated using ABS wage price index data for the NT and the mean income within the NT region for the year ended June 2018, being the most recent regional data available, as described in Section 9.1.1 above). The disaggregated impact on Indigenous and non-Indigenous household incomes was estimated using each model; these disaggregated income effects are set out in Table 9-6; the same table also reflects the estimated number of jobs (Indigenous and non-Indigenous combined) that could be created directly and indirectly by such developments. Based on the Daly catchment model, 7% of the increase in household incomes is estimated to flow to Indigenous households, despite Indigenous people representing around 29% of the population within the region. Using the NT model, where around 25% of the population identify as Indigenous, the proportion of household income increase flowing to Indigenous households varies according to agricultural type. A beef cattle based agricultural development would result in 10% of the increase in household incomes flowing to Indigenous households, compared to 6% for the other two categories (agriculture excluding beef cattle, and aquaculture, forestry and fishing). This is not unexpected, as Indigenous people are known to have been involved in working with cattle on cattle stations across the country since colonisation, and thus have higher levels of involvement within this sector compared to others. Accordingly, it should be noted that a proportion of these jobs are unlikely to be filled by people currently residing within the Roper catchment as it is unlikely that a sufficient pool of suitable workers is currently available in the catchment. A proportion of the jobs could be filled by people from the wider region, that is, people could migrate from other parts of northern Australia to take up some of these opportunities, however, foreign and domestic migratory workers from outside the region may also take up some of these opportunities. Accordingly, some of the benefits would accrue to people currently outside the catchment, and potentially outside of Australia, which could increase the leakage of benefits from the scheme outside of the region, and potentially, outside of the country. 9.2.3 Summary of estimated regional economic impact of a Roper catchment irrigated agriculture development While I–O based methods result in imperfect estimates, the approach provides some useful guidance to the likely upper bounds of the regional economic activity and employment opportunities that could result from development within the Roper catchment region, in both the construction and operational phases. Based on the Daly catchment model (as providing slightly lower estimates and representing a region with greater similarity to the Roper catchment than the NT as a whole), a large agricultural development providing $200 million of ongoing net additional output each year (after subtracting any payments farmers are required to contribute to the capital costs of the development to enable the scheme to be fully self-funded) could provide up to $406 million of regional benefit (with almost half of this representing the direct benefit of the new farming) (Table 9-4) and create about 800 jobs (Table 9-6) (considering just irrigated cropping; higher regional benefits could be possible from aquaculture). This represents $1.03 of additional benefit to the direct benefit of each dollar of new agricultural production generated by such a scheme (Table 9-5). Should the Commonwealth and/or NT governments choose to cover part of the costs of the development then the value of additional output will increase by the amount of the subsidy and the total regional economic benefit (direct and indirect) would increase by just over double the amount of that publicly funded contribution. Policy makers would need to consider the benefit generated by such a public investment compared to alternate uses of public funds (the opportunity cost) when determining the amounts and value of public contributions to new developments. In the construction phase, based on the consideration of the results of both I–O models, a medium-scale agricultural development requiring a capital cost of $2 billion could provide a one- off (temporary) regional benefit of $2.7 billion, based on an optimistic estimate of the likely leakage outside the region (with the majority of this representing the direct benefit of the construction work) (Table 9-2), and create about 5000 jobs (Table 9-3). This represents around $0.35 of additional benefit to the direct benefit of each dollar invested in the construction of the scheme. It should be noted that the above approach of summarising regional benefits of both the construction and agricultural phases of the project essentially represent upper bound estimates for the likely outcomes, and particularly, that the magnitude of regional benefits arising during the construction phase is likely to be small relative to the actual capital cost of a development. Regional benefits, in terms of sustained increases in economic activity, incomes and jobs from new farming, are expected to flow during the operational phase. 9.3 Water multipliers and environmental input–output (EI–O) analysis Environmental input–output analysis (EI–O) is an extension of the I–O approach (described in Section 9.1) that quantifies how natural resource use and pollutant emissions are embodied in the production, supply chains and final consumption of goods and services within a region or country (Daniels et al., 2011). It can be used to estimate the response of emissions and/or resource use to changes in consumption and production across the economy that result from the direct and knock-on effects of a development scheme. In many countries data are frequently published setting out the volume of water used by different sectors in the economy. For example, the ABS publishes annual water accounts which present information on the physical and monetary supply and use of water within the Australian economy, including details by industry sector (ABS, 2021d). However, while the direct water use by different industries can provide useful information, this information alone doesn’t describe the full impact that growth in a particular industry can have on total water use. Just as regional I–O analysis can be used to calculate the total (direct and indirect) change in economic activity that is likely to occur within the region in response to a change in final demand for the output of a particular sector, water EI–O analysis can be used to estimate the total change in demand for water (including direct and indirect impacts) that will result from the change in economic activity. Simplistically, the total change in water demand resulting from a specified change in demand to one or more sectors of the economy is calculated by multiplying the total change in regional output derived from I–O analysis by a vector that describes the sectoral (direct) water use. The water EI–O analysis used an extension of the Daly catchment regional I–O model (Stoeckl et al., 2011) described in Section 9.1.1. The water EI–O model provides a useful way of estimating the impacts on water use from the construction and operational phases of an agricultural development within the Roper catchment. Rather than present an exact prediction of the likely water needs, the analysis estimates likely lower and upper bounds of the impact on water demand in the region following an irrigated agricultural development. The analysis indicates the additional water required to not only meet the anticipated direct increase in consumption from new agricultural production, but also the additional indirect demand associated with the other industries and households in the region that benefit from the new farming. The following sections, describe water EI–O analysis in more detail, together with some contextual statistics on historical patterns of water use, and an analysis of a range of development scenarios using the EI–O approach. 9.3.1 Introduction to water EI–O analysis EI–O analysis has been widely used to assess different types of environmental impacts, including the impact on emissions of a wide range of pollutants (such as carbon dioxide, nitrous oxide, methane, waste etc.) and the increased use of natural resources (such as water). For example, O’Doherty and Tol (2007) analysed the impact of growth in different industry sectors on 17 different emissions plus water use in Ireland, while Sheng et al. (2019) analysed the increased water requirements resulting from increased demand in different sectors of the economy in the south-eastern USA, and Lenzen and Foran (2001) focused on estimating water multipliers for different industry sectors within Australia. When developing an EI–O model to estimate the impact in water use, it is first necessary to estimate the direct volumes of water used by each sector of the economy to meet current demand (described here as ‘water use coefficients’). These coefficients are added as a set of row vectors to an I–O model to estimate the amount of additional direct water usage that results from increased activity in a particular sector of the economy. This approach has been widely adopted for investigating the increased resource consumption and/or pollutant emissions that result from economic development (Guan and Hubacek, 2008; Stoeckl et al., 2011). Put simply, the I–O approach is used to estimate the total regional change in output and household incomes that occurs where there is a change to final demand within one or more sectors of the economy, such as agriculture; the resulting direct and indirect changes in water demand is then estimated by multiplying the vector of direct water use per sector by the total change in regional output from the I–O model (Stoeckl et al., 2011). The approach estimates water multipliers that quantify the additional volume of water required to meet a $1 increase in final demand for the production by each industry (Sheng et al., 2019). It has been noted that characteristics of water use make it a particularly good candidate for analysis via regional EI–O models as, unlike many pollutants (such as greenhouse gases), the impacts of water extraction and use are primarily restricted to those regions where the water is extracted and used, rather than having wider international impacts or global impacts (Daniels et al., 2011). Furthermore, water is a renewable resource drawn from local natural systems (which vary from region to region) and used by many different sectors with different water needs and its presence varies significantly from region to region; accordingly, water demands from different regions cannot be simply summed but require analysis that reflects the water scarcity and sector usage within each source region (Daniels et al., 2011). Benefits of EI–O approaches, compared to other methods of evaluating resource use, include ensuring that any particular sector development is considered in the context of the broader economic system; that linkage between sectors are incorporated in the analysis, that regionally specific relationships are recognised; and that it can contribute towards triple-bottom-line analysis, with the I–O and EI–O analysis together combining economic (economic activity), social (proxied by job creation) and environmental (water use) perspectives of likely outcomes from development in a region (Daniels et al., 2011). Prior water EI–O research has found that the water use per dollar of income varies greatly by industry sector and by geographic location. Agriculture is frequently found to be the sector with the highest water use by a large margin. For example, a study in Ireland found the water use coefficient per euro of consumption was 27.1 litres for agriculture, followed by 11.42 litres for food, beverage and tobacco, while many other sectors (including construction) were below 1 litre per euro consumed (O’Doherty and Tol, 2007). Within Australia, water multipliers were found to range from around 1,000 litres for $1 additional demand for sugarcane and cotton, and around 200 litres for $1 additional demand for agricultural products in general (including unirrigated agriculture), down to around 6 litres per additional $1 demand for construction and less than 3 litres for $1 demand increase in the insurance sector, all in 1994–95 Australian dollars (Lenzen and Foran, 2001). This study also noted the variation in embodied water within particular sectors across the country, for example agricultural water multipliers can vary due to differing levels of irrigation use across the country, and the far higher evapotranspiration rates in tropical compared to temperate regions (Lenzen and Foran, 2001). These findings were echoed in work by Sheng et al. (2019), who used a multi-regional EI–O model to illustrate that regional differences also have an impact on multiplier values, demonstrating that due to locally specific factors different US states had different water multiplier values, but generally the water multipliers for non- agricultural sectors were much lower in relative magnitude than those for primary agriculture and for utility sectors. Clearly, for the sectors for which water use is measured (noting that water use data by the mining sector is generally poorly collected), agricultural water use exceeds other industry sectors substantially, indeed by an order of magnitude, while the exact water coefficients will also depend on local climate conditions and agricultural practices. Thus, water EI–O models need to be tailored as closely as possible to the geographic region of study, just as region-specific models are required for using I–O analysis to estimate regional economic impacts flowing from stimuli applied to different industries. 9.3.2 Water use in the Northern Territory by sector The ABS present annual data on water supply and usage within Australia as a whole, and by state and territory. Data are provided regarding use and supply of self-extracted and mains water as well as water reuse, use of wastewater and water discharged to the environment. This Assessment was primarily interested in consumptive water use rather than total use, as the latter measure includes water that an industry or household uses temporarily prior to returning to the environment or elsewhere, where it is available for use by others. Focusing on consumptive use for analysis reflects the economic opportunity cost concept, in that the volume of water consumed by one sector of the economy is not available for consumption by other sectors (thus a cost of using in one sector is the lost benefit that would result from use elsewhere). The ABS defines the consumptive water use measure as follows: Water consumption = Distributed water use + Self-extracted water use + Reuse water use – Water supplied to other users – Instream (non-consumptive) use The EI–O analyses used ABS data on annual water supply and use in the NT from 2001–02 to 2017–18. Over this period the data series used a consistent methodology, but data were provided intermittently rather than every year for years prior to 2008–2009, and the methodology changed for years after 2017–18. These dates also approximately matched those used for the economic data in the underpinning I–O model (2005–06). The agricultural sector has been the dominant consumer of water in the NT for most years (Figure 9-2). Exceptions relate to the mining boom, when water consumption by the mining industry peaked in 2012–2013, and also in 2008–2009 when agricultural consumption dipped to the lowest point across the years where data are available. For more information on this figure please contact CSIRO on enquiries@csiro.au 0204060801002000–012004–052008–092009–102010–112011–122012–132013–142014–152015–162016–17Annual water consumption (GL) YearHousehold consumptionAgricultural consumptionMining and manufacturing consumptionOther industries consumption Figure 9-2 Water consumption in the NT analysed by ‘household’, ‘agricultural’, ‘mining and manufacturing’ and ‘other industries’ consumption from 2000–01 to 2016–17 Data were not provided on an annual basis until 2008–09. Original data provided for 2000–01 were corrected and represented alongside data for 2004–05; the corrected data provided in 2001–05 are shown here. From 2017–18 onwards water consumption has not been included as the methodology, terminology and concepts used to present data were revised and as a result data are not comparable to that presented above for earlier years. Source: ABS water accounts for the NT (ABS, 2021d) To illustrate the regional variations in the intensity of water use (which has an impact on water multipliers) across space, Figure 9-3 compares water consumption for Queensland against that for the NT. The data for the NT and Queensland show clear differences in scale and the relative importance of different water users, between locations and over time. For more information on this figure please contact CSIRO on enquiries@csiro.au 01000200030002000–012004–052008–092009–102010–112011–122012–132013–142014–152015–162016–17Annual water consumption (GL) YearHousehold consumptionAgricultural consumptionMining and manufacturing consumptionOther industries consumption Figure 9-3 Water consumption in Queensland analysed by ‘household’, ‘agriculture’, ‘mining and manufacturing’ and ‘other industries’ consumption from 2001–01 to 2016–17 Data were not provided on an annual basis until 2008–09. Original data provided for 2000–01 were corrected and represented alongside data for 2004–05; the corrected data provided in 2001–05 are shown here. From 2017–18 onwards water consumption has not been included as the methodology, terminology and concepts used to present data were revised and as a result data are not comparable to that presented above for earlier years. Source: ABS water accounts for Queensland (ABS, 2021d) Focusing on agriculture specifically, the total value of agricultural commodities produced in Queensland in 2019–20 was a gross value of $13.6 billion, compared to $0.9 billion for the NT (ABS, 2021e). There are significant differences in the agricultural types and practices in the two regions: Queensland has more irrigated cropping and more intensive animal and plant production, for example (Stoeckl et al., 2011), in addition to Queensland having a much larger population and covering a larger geographic area. Water consumption is far higher in Queensland in total, and for agricultural consumption alone, and also agricultural water consumption is a much larger percentage of the total, than that in the NT, as shown in Figure 9-4. This emphasises the importance of considering regional-specific water usage models to ensure regional variations do not distort the analysis. For more information on this figure please contact CSIRO on enquiries@csiro.au 0% 20% 40% 60% 80% 100% 01,0002,0003,0004,0005,0002000–012004–052008–092009–102010–112011–122012–132013–142014–152015–162016–17Proportion consumed by agriculture (%) Water consumption (GL) YearTotal water consumption NTTotal water consumption QldAgricultural water consumption NTAgricultural water consumption QldAgricultural water consumption % of total NTAgricultural water consumption % of total Q.d Figure 9-4 Water consumption for the NT and for Queensland (Qld), in total and agricultural consumption alone, illustrating percentage of total consumption arising from agriculture from 2000–01 to 2016–17 Data were not provided on an annual basis until 2008–09. Original data provided for 2000–01 were corrected and represented alongside data for 2004–05; the corrected data provided in 2001–05 are shown here. From 2017–18 onwards water consumption has not been included as the methodology, terminology and concepts used to present data were revised and as a result data are not comparable to that presented above for earlier years. Source: ABS water accounts for the NT and Qld (ABS, 2021d) 9.3.3 Applying water EI–O analysis to a Roper catchment irrigated agriculture development: developing the water EI–O model The base EI–O mode is described in detail by Stoeckl et al. (2011). To update and adapt the original model for the analysis here of the Roper catchment, a number of amendments were made. First, the new model incorporates water accounts that have been improved from the 2000–01 data used in the original model. As part of a later 2004–05 water accounts data release, the ABS revised their water use estimates for 2000–01, explaining that, … increased ABS survey activity, better business reporting and greater access to State, Territory and Australian government data have led to improvements in data quality for the 2004–05 Water Account. Improved data for 2004-05 has also enabled a greater understanding of the data used in 2000–01 and these data have been updated to reflect this as well as some changes in accounting treatments. During this process some errors in data and estimation procedures used in 2000–01 were identified and these too have been corrected. The main revisions are outlined below. For the AGRICULTURE industry estimates of water were revised downwards for all jurisdictions for 2000– 01. Based on data from ABS surveys of water use by irrigators for 2002–03 to 2004–05, the majority of crop application rates used to produce estimates for the 2000–01 Water Account were found to be high. In 2000–01, the Agricultural Census only collected data on irrigated area, whereas latter surveys collected additional data on the volume of water applied to irrigated crops and pastures as well as sources of water… (ABS, 2005, p. 6) Corrections were also made to the 2000–01 data for other industries beyond agriculture, including mining, manufacturing, electricity and gas supply, and water supply, sewerage and drainage services (ABS, 2005). Accordingly, the restated data for 2000–01 presented in the 2004–05 Water Account (ABS, 2005) form the basis for the analysis presented here. A further complexity introduced by the ABS corrections to previously reported data for 2000–01 related to the industry sectors used in the water supply and use tables. The number of industry sectors listed in the I–O table were reduced for water use reporting from 2004–05 onwards (including the restated 2000–01 data), with a number of industrial sectors combined into a single group named ‘Other industries’. Water use is required for the analysis to be segregated into different industries to enable the water use to be mapped to the same industry sectors used in the I–O model. Unfortunately, the sectors combined by the ABS to form the ‘Other industries’ category were: ‘Construction; Wholesale & retail trade’; ‘Accommodation cafes & restaurants’; ‘Transport & storage’; ‘Finance property & business services’; ‘Government administration’; ‘Education’; ‘Health & community services’; and ‘Cultural, recreational & personal services’. Separate information on a number of these sectors was required to populate the water EI–O model, and hence assumptions regarding the analysis of the revised water use data needed to be made. In the absence of any better alternative, the assumption was made to calculate the proportion of water use relating to each of these industries based on the relative proportion of each of these industries in the originally reported water use data for 2000–01. The re-analysis of water use data for NT for 2000–01, used in the water EI–O model, can be seen in Table 9-7. Note that the original water use data in the I–O table is that used by the model in Stoeckl et al. (2011, Table 10, p. 58). Water accounts data are provided by the ABS at national, state and territory level, but data are not available from ABS at a finer regional scale. To address this limitation, water data for the NT and Queensland have been used as proxies in this EI–O analysis to represent lower and upper bound estimates, respectively, for the impact of development on the water demanded. The NT data provide a reasonable lower bound, being based on a region with relatively little irrigated agriculture, while the Queensland data represent a region with far more irrigated agriculture and more intensive agricultural production systems, providing an approximation of the upper bounds of the likely impact to water demand in the region. Table 9-7 Re-analysis of corrected water use data by industry in the NT for 2000–01 The table sets out assumptions underlying the estimated use for those industries where ABS has changed the data presentation (now combined by ABS within one category whereas previously presented separately) INDUSTRY 2000–01 ORIGINAL WATER USE BY INDUSTRY FOR NT (ML) METHOD ADOPTED FOR ESTIMATING CORRECTED 2000– 01 WATER USE SECTOR’S WATER USE AS % OF COMBINED TOTAL FOR OTHER INDUSTRIES GROUPING (%) REVISED ESTIMATED 2000–01 WATER USE BY SECTOR USED IN I–O ANALYSIS (ML) Agriculture 70,377 ABS provided na 38,975 Mining and manufacturing 13,687 ABS provided na 22,385 Electricity & gas, and Water supply, sewerage & drainage services 9,607 ABS provided na 9,786 Sectors combined within ‘Other industries’ • Construction 26 % combined total 0.1 37 • Wholesale & retail trade 1,742 % combined total 8.0 2,242 • Accommodation, cafes & restaurants 433 % combined total 2.0 609 • Transport & storage 1,474 % combined total 6.7 2,074 • Finance, property & business services 599 % combined total 2.7 843 • Government administration, Education, and Health & community services 16,001 % combined total 73.2 22,519 • Cultural, recreational & personal services 1,593 % combined total 7.3 2,242 Total ‘Other industries’ 21,868 % combined total 100.0 30,776 Total industry water consumption 115,539 ABS provided na 101,922 na = not applicable Sources: Original 2001–01 data for ABS Water Account, Australia 2001–01 (ABS, 2004), corrected data from ABS Water Account, Australia 2004–05 (ABS, 2005) Corrected 2000–01 water use data were compared with the gross value of each industry sector for the same period to calculate water use per dollar of gross value per industry. Water use dollar values were adjusted for inflation between June 2001 and June 2021 using the CPI (ABS, 2021f). The vector of water use by industry was then supplemented with data for water use by Indigenous and by non-Indigenous households. These data were sourced directly from Stoeckl et al. (2011); the original data had been sourced from detailed surveys with householders within the Daly region of the NT and from the Mitchell catchment in Queensland. 9.3.4 Estimated total additional water usage arising from a Roper catchment irrigated agriculture development based on the water EI–O model Figure 9-5 clearly illustrates that aggregate water multipliers (including all direct and knock-on effects on water usage) for agriculture are many times larger than for other industries. It is also clear that the type of agricultural development (intensity and dependence on irrigation) is important, as illustrated by the large variation between the upper and lower bound estimates. The results of the EO-I analyses using these multipliers are presented below, first for the construction phase of the project and then for the operational (farming) phase. As described in the methods, all results are provided for lower and upper bounds (based on the NT and Queensland water use data, respectively). For more information on this figure please contact CSIRO on enquiries@csiro.au 050100150200250300350400AccommodationAgricultureConstructionCultureElectricityFinanceGovernmentMiningTradeTransportIndigenousHouseholdsNon-IndigenousHouseholdsLower bound multiplier: change in water demand (ML per $1 million expansion of sector) Upper bound multiplier: change in water demand (ML per $1 million expansion of sector) Figure 9-5 Water multipliers estimated for Roper catchment, demonstrating direct and indirect (production and consumption induced) demand for additional ML of water resulting from additional $1 million stimulus applied to expanding a particular sector of the economy Lower bound estimates are based on water use coefficients derived from water use in NT; upper bound estimates use coefficients derived from water use in Queensland. Source: Author calculations based on adapting model developed by Stoeckl et al. (2011) Construction phase The increased demand for water during the construction phase of any irrigated agricultural developments in the Roper catchment (Table 9-8) was estimated for the same set of scenarios used in the I–O analyses (Section 9.2.1). The construction industry has a relatively low direct need for water, and also does not stimulate a large ‘knock-on’ impact on agriculture (the industry with the largest direct water use). Accordingly, low water multipliers are estimated for construction developments. As a result, the total estimated increase in water usage triggered by the development is fairly modest, especially in the smaller investment scenarios. However, scenarios based on the larger capital investment scenarios and using the upper bound estimates do produce some fairly significant water needs, which would need to be considered in planning any construction project. Table 9-8 Estimated increase in demand for water (ML) based on increased demand for construction activity, for the construction phase of an irrigated agricultural development Lower bound estimates developed from NT data, upper bound estimates from Queensland data. SCHEME COST CAPITAL COST ($ billion) TOTAL VOLUME OF ADDITIONAL WATER DEMAND IN I–O REGION AS A RESULT OF THE CAPITAL COST OF THE DEVELOPMENT SCHEME – DIRECT, PRODUCTION-INDUCED AND CONSUMPTION-INDUCED (ML) Lower bound estimates Upper bound estimates Proportion of total scheme-scale capital cost made locally within the I–O region 65% 50% 35% 65% 50% 35% 0.250 679 522 366 2,196 1,689 1,183 0.500 1,358 1,045 731 4,392 3,379 2,365 1.000 2,716 2,089 1,463 8,785 6,758 4,730 2.000 5,433 4,179 2,925 17,570 13,515 9,461 4.000 10,865 8,358 5,851 35,139 27,030 18,921 Operational phase of an irrigated agricultural development in the Roper catchment The estimates of increased annual demand for water during the operational phase of potential irrigated agricultural developments are shown in Table 9-9 for the same scenarios used in the I–O analyses (Section 9.2.2). As would be expected, the direct and indirect increases in water usage stimulated by an agricultural development are significant (as quantified in the agronomic analyses in Chapter 5 across a range of different farming options). Accordingly, the scale of the additional agricultural stimulus will have a significant impact on the water use, but also the type of agricultural development. If the additional agriculture was predominantly fairly low intensity grazing then water usage would be towards the lower bound estimate. However, if, as is likely, the development focuses on higher intensity irrigated cropping and horticulture, then the water demand could approach, or even exceed, the upper bounds estimate (as the upper bound is based on data for Queensland, which includes a mix of irrigated and non-irrigated types of agriculture). For water developments that seek to promote agriculture (as opposed to other less water intensive industries) the upper bound estimate should thus be given greater weight. As a reference for the scale of potential development, total mean annual water usage in the entire NT was 161 GL across the period 2000–01 to 2016–17. In this context an increase in agricultural output of $200 million per year could represent an approximate increase in the NT water use of up to 50% (79 GL per year for the upper bounds water usage estimates), so developments in the Roper catchment may be more likely to be towards the lower end of the scenarios in Table 9-9, unless the development focuses on particularly high water usage crops. The output of the EI–O analysis clearly identifies the amount of (direct and indirect) additional water required for the regional economic development that can arise from a new irrigated agriculture within the region. It is important to note that this analysis requires ground-truthing against estimates of the physical amount of water that could be made available, taking into account the climate and quantity/quality of available land across the region. Thus, the analysis should be considered alongside agronomic factors that affect agriculture in the region (see Chapter 3). Table 9-9 Estimated increase in total demand for water (ML) based on increased demand for agricultural output during the operational phase of an irrigated agricultural development Lower bound estimates developed from NT data, upper bound estimates from Queensland data. Queensland has a larger proportion of land with irrigated cropping and irrigated agriculture compared to NT, and thus should better indicate water use coefficients after a development that increases irrigated agriculture within the region. DIRECT INCREASE IN AGRICULTURAL OUTPUT PER YEAR ($ million) TOTAL VOLUME OF ADDITIONAL WATER DEMAND IN I–O REGION AS RESULT OF OPERATIONAL PHASE OF THE DEVELOPMENT SCHEME – DIRECT, PRODUCTION-INDUCED AND CONSUMPTION- INDUCED (ML) Lower bound estimate Upper bound estimate Total additional water demand 25 1,877 9,893 50 3,755 19,785 100 7,509 39,571 200 15,018 79,141 Direct water demand from agricultural development included in total additional water demand 25 1, 608 8,610 50 3,216 17,220 100 6,432 34,440 200 12,864 68,879 Indirect water demand (beyond direct water demand from agriculture) arising from agricultural development and included in total additional water demand 25 269 1,283 50 539 2,566 100 1,077 5,131 200 2,155 10,262 The direct water needs of agriculture far exceed those of other industries for which reliable water usage is available (noting that data on the usage of the mining industry can be sparse), and thus the direct effects on water usage far exceed the indirect effects. However, the indirect impacts on water demand (from knock-on impacts to other industries and induced effects) also have to be considered within the local context of water supply and planning in the region. States and territories within Australia each have their own legislative water planning framework, which establish processes for allocating water between competing uses, and provide for how water is to be shared between high and medium-priority allocations. Thus, the regional water planning context is an important factor to be borne in mind when considering the feasibility of new water- dependent developments, and the additional knock-on water multiplier effects for other water users analysed here. Part IV Concluding comment 10 The ‘sweet spot’ for northern development The purpose of this report was to provide information on the costs, risks and benefits of new irrigated development in the Roper catchment, at farm to scheme and regional scales, and supply chains beyond. The overall conclusion is that large public dams would be marginal, but on-farm water sources, suitably sited, could provide good prospects for viable new farms. There is a range of cropping options that could be suitable, of which the most likely to be profitable (if development costs can be kept low enough) are annual horticulture, cotton, forages and peanut. Sequential cropping systems present opportunities for combining crops that might not be profitably grown alone and/or or to generate additional net revenue from the same capital investment. There are many potential cropping sequences that show agronomic potential for matching back-to-back crop requirements with Roper catchment growing conditions, particularly on well-draining loamy soils (like the Kandosols on the Sturt Plateau that dominate the arable areas in the western Roper catchment), but these would need to be developed and proven locally. Trafficability constraints on finer-textured clay soils (mainly dissected and distributed patches of alluvial Vertosols) would make scheduling back-to-back crops in the same season more difficult, so would restrict the choice of crops to those with shorter growing seasons and would likely be opportunistic. The farm-scale performance of cropping systems will be determined by: • finding markets and supply chains that can provide a sufficient price and reliability of demand, while being able to supply those markets at adequate scale and an affordable cost (see Section 2.2 and Chapter 7 of this report) • the skill of farmers and investors in managing the operational and financial complexity of adapting crop mixes and production systems to Roper catchment environments (including soils, water resources and climates), particularly in managing cash flows and ‘learning’ through the challenging establishment years (see parts II and III of this report) • the nature of water resources in terms of their costs to develop, the volume and reliability of supply, and the timing of when water is available relative to optimal planting windows (particularly for sequential cropping) (see companion technical reports on river modelling (Hughes et al., 2023), surface water storage (Petheram et al., 2022), and groundwater modelling (Knapton et al., 2023)) • the nature of the soil resources in terms of their scale and distribution, their proximity to water sources and supply chains, their farming constraints, crops they can support with viable yields, and their costs to develop; where the best opportunities are supported by Kandosols (which provide good wet-season trafficability, but require higher fertiliser inputs and more expensive pressurised irrigations systems) and smaller, scattered pockets of Vertosols (which are less expensive to develop but have poor trafficability in the wet season and are limited in the scale of localised production they could support) (see companion technical reports on digital soils mapping and land suitability (Thomas et al., 2022) and flood modelling (Kim et al., 2023), and Part III of this report). Long supply chains and distant processing facilities have typically put northern Australia agriculture at a competitive disadvantage (relative to southern farming regions). However, some of those constraints have recently been alleviated creating new opportunities from: (i) the construction of a new cotton gin near Katherine, providing closer processing for possible new farms in the western parts of the Roper catchment; (ii) recent high cotton prices, that provide a financial buffer while farmers learn to grow the crop to its full potential, and develop supply chains to process and market their product; and (iii) recently built phytosanitary facilities at Darwin Airport, which makes export of horticultural produce easier. As market, regulatory, infrastructure and other conditions in the Roper catchment change from those prevailing at the time this report was written, growers would be expected to adapt and respond to opportunities and challenges accordingly. Ultimately the crops (if any) that can be successfully and sustainably grown will have to find sweet spots where investors can simultaneously address all three of the following questions (Stokes et al., 2019) (Figure 10-1; Table 10-1): • Markets: Where is the investor going to sell their produce and how are they going to set up the supply chains to get their products, at low-enough cost, from the Roper catchment to those who want to buy them? • Production systems: What is the investor going to grow and do they understand how this needs to be grown differently in tropical Australia (and the soils, water resources and climates of Roper catchment environments specifically) to where they have gained their previous experience? • Competition: Why is it better to grow the chosen product(s) in tropical Australia, relative to alternative options of growing the same product elsewhere, or growing different products in the chosen location? For more information on this figure please contact CSIRO on enquiries@csiro.au Thriving local agricultureSpatial diversificationDry season plantingSequential croppingOveroptimismUnproven production systemsPreservationist attitudesApprovals processesProductionSWEETSPOT Figure 10-1 Viable irrigated agriculture investments in the Roper catchment require a combination of capturing opportunities and mitigating risks in three critical areas: markets, production systems and competition Adapted from Stokes et al. (2019). Details for each risk and opportunity are expanded in Table 10-1. There is a wide variety of potential investors in northern Australia agriculture, each of whom will come with different strengths and blind spots (Stokes et al., 2017). Each may initially be drawn by an opportunity in a particularly strong area of competence for one of the three criteria above (be it a new market where they can fill an unmet demand, a crop product with particular promise, or identifying a prospect for gaining a competitive advantage within an industry) but will likely not initially be completely aware of the full scale of the challenge in one of the other areas. Successful investments have typically been able to address all three of the above criteria, while failures have not. Table 10-1 Opportunities and risks across three key criteria for the success of irrigated development in the Roper catchment Adapted from Stokes et al. (2019), which provides details of the methods and supporting literature. These points are further supported by analyses and literature presented in this Assessment. MARKETS PRODUCTION SYSTEMS COMPETITION Where is the investor going to sell their produce and how are they going to set up the supply chains, at low- enough cost, to get their products to those who want to buy them? What is the investor going to grow and do they understand how this needs to be grown differently in tropical Australia to where they have gained their previous experience? Why is it better to grow the chosen product(s) in tropical Australia, relative to alternative options of growing the same product elsewhere, or different products in the chosen location? Opportunities/Strengths Opportunities/Strengths Opportunities/Strengths Capacity to expand Thriving local agricultural industries Safe, clean and green produce Northern Australia is relatively un- developed with capacity and natural resources to expand Intensive agricultural businesses are growing and maturing in the NT Gives access to markets with high health and environmental standards that some competitors are unable to meet; also meets consumer preferences in some markets Growing demand from Asia and Middle East Spatial diversification Timing of seasonal production Market analyses have identified a range of products with unmet demand that northern Australia could produce from horticulture and broadacre crops, including cotton, grain sorghum and sesame seeds More uniform supply of agricultural products by spreading exposure to weather events, such as floods, destructive winds, drought, temperatures, climate change; (e.g. offset risks for melon production concentrated in Queensland) Out of season production (relative to the rest of Australia), broadens the national seasonal supply and can provide price premiums for local produce (e.g. early season mangoes from Katherine/Mataranka) Production system/supply chain integration Dry-season planting allows better seasonal planning Biosecurity advantages of isolation Opportunities to integrate agricultural production systems and supply chains with other regions/countries (e.g. live export of cattle to South-East Asia for fattening and supply chains with little refrigeration) Planting at end of wet season in the north (vs start of wet in south) allows better seasonal planning (available soil and stored water are known at time of planting) Remoteness from other areas growing the same crop reduces the risks of spreading diseases between them (e.g. Panama TR4 fungus in Cavendish bananas) Freer trade agreements Sequential cropping High-value horticulture and aquaculture Opportunities in 17 markets from free trade agreements, including recent agreements with Indonesia and India (with ongoing initiatives for additional potential agreements) Advantage of tropics is length, not quality, of growing season; sequential broadacre cropping systems can make use of the longer seasons, but require tuning to local conditions (e.g. Cerrado in Brazil) Proportionally less affected by higher costs of remoteness, and better suited to niche, small-scale, localised opportunities Risks/Weaknesses Risks/Weaknesses Risks/Weaknesses Processing facilities Greenfield risks (overoptimism) Length and quality of supply chains Processors require assured scale and reliability of primary produce for investment in new processing infrastructure to be viable An entrepreneurial spirit is required, but enthusiasm can exceed capacity and planning (e.g. under estimating development costs and time required to learn and adapt to local greenfield conditions, and over estimating farm production and profitability) Higher transport costs and spoilage overall resulting from large distances to market and poorer quality of many regional roads and some storage/processing facilities Biosecurity facilities for export Novel/adapted production systems required for greenfield development Labour availability and capacity MARKETS PRODUCTION SYSTEMS COMPETITION To meet quarantine requirements and certification for some target markets (e.g. irradiation of mangoes); as with processing facilities, this requires assured scale and reliability of primary produce and market demand Novel elements are required to enable and adapt production systems for the particular challenges of northern agriculture (e.g. tropical vs subtropical agronomy, sequential cropping, variability in climates and prices, and biosecurity) Intensive production has high, seasonal demands for labour relative to local population (e.g. demand in peak week of NT mango fruiting requires equivalent of ~2% of resident working population in Greater Darwin) Scale of production Preservationist attitudes High input costs Chicken and egg: need to achieve scale of production to cover required infrastructure costs and establish new markets and supply chains, but hard to scale production efficiently until that infrastructure is built Attitudes from intensively developed parts of the south inappropriately exported and applied to sparsely developed north: irrigated agriculture in tropical Australia west of the Great Dividing Range only occupies about the same area as mining (both <0.1%) Input costs are high relative to competitors (generally Australia vs international, and remote Australia vs regions closer to dominant southern markets and labour sources) Trade policy risk (market access) Approvals process Full cost recovery for new public water infrastructure development Access to foreign markets can become more restricted (e.g. live cattle export restrictions and policy changes by major trading partners) Approvals process can be protracted, costly and inefficient: a definitive decision in a reasonable time frame, either way, provides investors with certainty It is challenging for new irrigated development to compete on the basis of full cost recovery against existing developments where water costs are subsidised This Assessment (including companion reports) has focused primarily on ‘production system’ challenges by filling knowledge gaps on the land and water resources in the Roper catchment. This report has evaluated the farming options that could be sustainably and profitably developed on that resource base, and has provided additional supporting information for overcoming the competitive disadvantages and market constraints for northern Australia. Widespread expansion of agriculture in the Roper catchment is unlikely to occur in the near term. However, smaller scale opportunities will continue to emerge (as they have done throughout northern Australia before) for those able to find niches for cost savings and suitable markets, and who have the capital and capacity to persist through the challenging establishment years. References ABARES (2022) Agricultural commodities: September quarter 2022. Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra. September CC BY 4.0. DOI: 10.25814/zs85-g927. ABARES (2023) Australian horticulture prices. Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra. Viewed 10 March 2023, https://www.agriculture.gov.au/abares/data/weekly-commodity-price-update/australian- horticulture-prices#daff-page-main. ABARES Trade dashboard (beta) (2022) Australia’s agricultural export data by commodity, year and destination. Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra. Viewed 4 March 23, https://www.agriculture.gov.au/abares/research- topics/trade/dashboard. Abrecht DG and Bristow KL (1996) Coping with climatic hazards during crop establishment in the semi-arid tropics. Australian Journal of Experimental Agriculture 36, 971–983. ABS (2004) Water Account, Australia 2000–01. Australian Bureau of Statistics, Canberra. Viewed 15 December 2022, https://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/4610.02000- 01?OpenDocument. ABS (2005) Water account, Australia 2004–05. Australian Bureau of Statistics, Canberra. Viewed 15 December 2022, https://www.ausstats.abs.gov.au/ausstats/subscriber.nsf/0/3494F63DFEE158BFCA257233001CE732/$File/46100_2004-05.pdf. ABS (2006) Census of population and housing time series profile. Catalogue number 2003.0 for various SA regions falling partly within Roper catchment. Australian Bureau of Statistics, Canberra. Viewed 19 October 2021, https://www.abs.gov.au/census. ABS (2011) Census of population and housing time series profile. Catalogue number 2003.0 for various SA regions falling partly within Roper catchment. Australian Bureau of Statistics, Canberra. Viewed 19 October 2021, https://www.abs.gov.au/census. ABS (2016) Census of population and housing time series profile. Catalogue number 2003.0 for various SA regions falling partly within Roper catchment. Australian Bureau of Statistics, Canberra. Viewed 19 October 2021, https://www.abs.gov.au/census. ABS (2017) 75030DO005_201516 Value of agricultural commodities produced, Australia – 2015– 16, datacube released 31 October 2017. Australian Bureau of Statistics, Canberra. Viewed 11 November 2021, https://www.abs.gov.au/statistics/industry/agriculture/value-agricultural- commodities-produced-australia. ABS (2018a) 2033.0.55.001 Socio-economic Indexes for Areas (SEIFA) (released 27 March 2018). Findings based on use of Data Cube. Australian Bureau of Statistics, Canberra. Viewed 30 November 2021, https://www.abs.gov.au/AUSSTATS/abs@.nsf/DetailsPage/2033.0.55.0012016?OpenDocument. ABS (2018b) Estimates of Aboriginal and Torres Strait Islander Australians methodology, released 31 August 2018. Australian Bureau of Statistics, Canberra. Viewed 11 November 2021, https://www.abs.gov.au/methodologies/estimates-aboriginal-and-torres-strait-islander- australians-methodology/jun- 2016#:~:text=6%20In%20the%202016%20Census,the%202011%20Census%20(17.2%25). ABS (2019) 5209.0.55.001 – Australian national accounts: input–output tables, 2015–16. Australian Bureau of Statistics, Canberra. Viewed 15 December 2022, https://www.abs.gov.au/ausstats/abs@.nsf/7d12b0f6763c78caca257061001cc588/5607443df596b7e3ca25842500147bda!OpenDocument. ABS (2021a) Australian national accounts: input-output tables, 2018–19. Australian Bureau of Statistics, Canberra. Viewed 15 December 2022, https://www.abs.gov.au/statistics/economy/national-accounts/australian-national- accounts-input-output-tables/latest-release#data-download. ABS (2021b) Region summary: Northern Territory. Australian Bureau of Statistics, Canberra. Viewed 15 December 2022, https://dbr.abs.gov.au/region.html?lyr=ste&rgn=7. ABS (2021c) 6345.0 Wage price index, Australia. Australian Bureau of Statistics, Canberra. Viewed 15 December 2022, https://www.abs.gov.au/statistics/economy/price-indexes-and- inflation/wage-price-index-australia/latest-release#data-download. ABS (2021d) Water account, Australia 2019–20. Australian Bureau of Statistics, Canberra. Viewed 15 December 2022, https://www.abs.gov.au/statistics/environment/environmental- management/water-account-australia/latest-release. ABS (2021e) 75030DO001_201920 Value of agricultural commodities produced, Australia, 2019– 20. Australian Bureau of Statistics, Canberra. Viewed 15 December 2022, https://www.abs.gov.au/statistics/industry/agriculture/value-agricultural-commodities- produced-australia/latest-release. ABS (2021f) Consumer price index, Australia, September 2021. Australian Bureau of Statistics, Canberra. Viewed 15 December 2022, https://www.abs.gov.au/statistics/economy/price- indexes-and-inflation/consumer-price-index-australia/sep-2021. ABS (2022) Agricultural commodities, Australia. In: Historical selected agricultural commodities by state. Australian Bureau of Statistics, Canberra. Viewed 15 June 2022, https://www.abs.gov.au/statistics/industry/agriculture/value-agricultural-commodities- produced-australia/latest-release#data-download. ACARA (2022) MySchool web site. Australian Curriculum, Assessment and Reporting Authority, Sydney. Viewed 13 July 2022, https://www.myschool.edu.au. ACT Government (2014) ACT water strategy 2014–44 – striking the balance. Viewed 20 December 2022, https://www.environment.act.gov.au/__data/assets/pdf_file/0019/621424/ACT- Water-Strategy-ACCESS.pdf. ACTEW (2009a) Enlargement of the Cotter Reservoir and associated works. Environmental impact statement. ACTEW Corporation Limited, Australia. Viewed 20 December 2022, https://www.iconwater.com.au/water-education/our-projects/water-security- projects/enlarged-cotter-dam/project-approval-documents/enlarged-cotter-dam- environmental-impact-statement.aspx. ACTEW (2009b) Enlargement of the Cotter Reservoir and associated works Final public environment report 2008/4524. ACTWE Corporation Limited, Australia. Viewed 20 December 2022, https://www.iconwater.com.au/water-education/our-projects/water- security-projects/enlarged-cotter-dam/project-approval-documents/enlarged-cotter-dam- public-environment-report.aspx. Adept Economics (2020) Economic costs of inaction on Paradise Dam. Report prepared for Bundaberg Regional Council, Wide Bay Burnett Regional Organisation of Councils, Regional Development Australia Wide Bay Burnett, Bundaberg CANEGROWERS, CANEGROWERS Isis, and Bundaberg Fruit and Vegetable Growers, Australia. Viewed 20 December 2022, https://www.bdbcanegrowers.com.au/wp-content/uploads/2020/02/Economic-Costs-of- Inaction-on-Paradise-Dam-report-24-February-20-FINAL.pdf. AFR (Australian Financial Review) (2023). Beef cattle prices slump 20pc in a boon for exports. Viewed 24 June 2023, https://www.afr.com/markets/commodities/beef-cattle-prices-slump- 20pc-in-a-boon-for-exports-20230522-p5da7d. AIHW (2015) Australian hospital peer groups. Health services series no. 66. Cat. no. HSE 170. Australian Institute of Health and Welfare, Canberra, Australia. AIHW (2023) Hospitals at a glance: web report. Australian Institute of Health and Welfare, Canberra. Viewed 1 March 2023, https://www.aihw.gov.au/reports/hospitals/australias- hospitals-at-a-glance. Akbar S, Abbas A, Hanjra MA and Khan S (2013) Structured analysis of seepage losses in irrigation supply channels for cost-effective investments: case studies from the southern Murray- Darling Basin of Australia. Irrigation Science 31(1), 11–25. Allen RG, Pereira LS, Raes D and Smith M (1998) Crop evapotranspiration: guidelines for computing crop water requirements. FAO irrigation and drainage paper 56. Food and Agriculture Organization of the United Nations, Rome. Ansar A, Flyvbjerg B, Budzier A and Lunn D (2014) Should we build more large dams? The actual costs of hydropower megaproject development. Energy Policy 69, 43–56. DOI: 10.1016/j.enpol.2013.10.069. Arndt W, Phillips LJ and Norman MJT (1963) Comparative performance of crops on three soils of the Tipperary region NT. Division of Land Research and Regional Survey Technical Paper No. 23. CSIRO, Canberra. Ash A and Watson I (2018) Developing the north: learning from the past to guide future plans and policies. The Rangeland Journal 40, 301–314. Ash A, Cossart R, Ham, C, Laing A, MacLeod N, Paini D, Palmer J, Poulton P, Prestwidge D, Stokes C, Watson I, Webster T and Yeates S (2018) Agricultural viability: Fitzroy catchment. A technical report from the CSIRO Northern Australia Water Resource Assessment, part of the National Water Infrastructure Development Fund: Water Resource Assessments, Australia. Ash A, Gleeson T, Cui H, Hall M, Heyhoe E, Higgins A, Hopwood G, MacLeod N, Paini D, Pant H, Poulton P, Prestwidge D, Webster T and Wilson P (2014) Northern Australia: food and fibre supply chains study project report. CSIRO and ABARES, Australia. Ash AJ, Hunt L, McDonald C, Scanlan J, Bell L, Cowley R, Watson I, McIvor J and Macleod N (2015) Boosting the productivity and profitability of northern Australian beef enterprises: exploring innovation options using simulation modelling and systems analysis. Agricultural Systems 139, 50–65. Auditor General Western Australian (2016) Ord-East Kimberley Development. Report 20: September 2016. Office of the Auditor General Western Australia, Perth. Viewed 15 December 2022, https://audit.wa.gov.au/wp-content/uploads/2016/09/report2016_20- OrdEastKimberley.pdf. Australasia Aquaculture (2005) Low intensity prawn farming. Australasia Aquaculture (March 2005): 30-36. Australian Industry and Skills Committee (2022) Production horticulture. National Industry Insights Report. Viewed 9 August 2022, https://nationalindustryinsights.aisc.net.au/industries/agriculture/production-horticulture. Australian Government (2021) Closing the gap in partnership – national agreement on closing the gap. Viewed 15 December 2022, https://www.closingthegap.gov.au/national-agreement. Beach DF (1995) Australian sesame industry – an overview. In: Bennett MR and Wood IM (eds) Proceedings of first Australia sesame workshop, Darwin-Katherine NT, March 1995. Northern Territory Department of Primary Industry and Fisheries, Darwin. Bega Peanut Butter (2022) Bega peanut butter. Viewed 11 October 2022, https://begapeanutbutter.com.au/. BITRE (2018) Ex-post economic evaluation of national road investment projects – volume 1 synthesis report. Report 145. Bureau of Infrastructure, Transport and Regional Economics, Canberra. Viewed 20 December 2022, https://www.bitre.gov.au/publications/2018/rr_145. Brann M (2021) Mangoes picked in July as Northern Territory keeps title of mango king. ABC News. Viewed 12 November 2022, www.abc.net.au/news/rural/2021-07-10/mangoes-in-july- picked-in-%0Anorthern-territory/100279176. Brennan McKellar L, Monjardino M, Bark R , Wittwer G, Banerjee O, Higgins A, MacLeod N, Crossman N, Prestwidge D and Laredo L (2013) Irrigation costs and benefits. A technical report to the Australian Government from the CSIRO Flinders and Gilbert Agricultural Resource Assessment, part of the North Queensland Irrigated Agriculture Strategy. CSIRO Water for a Healthy Country and Sustainable Agriculture flagships, Australia. Buckworth RC, Venables WN, Lawrence E, Kompas T, Pascoe S, Chu L, Hill F, Hutton T and Rothlisberg PC (2014) Incorporation of predictive models of banana prawn catch for MEY- based harvest strategy development for the Northern Prawn Fishery. Final report to the Fisheries Research and Development Corporation, Project 2011/239. CSIRO Marine and Atmospheric Research, Brisbane. Carberry PS, Cogle AL and McCown RL (1991) Future prospects for cropping in the semi-arid tropics of north Queensland. Land Management Research, CSIRO Division of Tropical Crops and Pastures, St Lucia, Queensland. Carberry PS, Hochman Z, Hunt JR, Dalgleish NP, McCown RL, Whish JPM, Robertson MJ, Foale MA, Poulton PL and Van Rees H (2009) Re-inventing model-based decision support with Australian dryland farmers. 3. Relevance of APSIM to commercial crops. Crop and Pasture Science 60, 1044–1056. CIE (1997) An economy-wide modelling framework for the Northern Territory: theory, data and application. Centre for International Economics, Canberra. Chapman AL, Sturtz JD, Cogle AL, Mollah WS and Bateman RJ (1996) Farming systems in the Australian semi-arid tropics-a recent history. Australian Journal of Experimental Agriculture 36(8), 915–928. Chauhan BS, Mahajan G, Sardana V, Timsina J and Jat ML (2012) Productivity and sustainability of the rice-wheat cropping system in the Indo-Gangetic plains of the Indian subcontinent: problems, opportunities, and strategies. Advances in Agronomy 117, 315–369. Chudleigh F, Oxley T and Bowen M (2019) Improving the performance of beef production systems in northern Australia. Queensland Government and Northern Territory Government. Clonan M, Hernaman V, Pearce K, Hopkins M, Moise A and McConchie C (2020) Impact of climate change on flowering induction in mangoes in the Northern Territory. Earth Systems and Climate Change Hub Report No. 16. NESP Earth Systems and Climate Change Hub, Australia. Cowley T (2014) The pastoral industry survey – Katherine region. Northern Territory Government, Australia. CRCNA (2020) Northern Australian broadacre cropping situational analysis. ST Strategic Services and Pivotal Point Strategic Directions (Issue July). Cooperative Research Centre for Developing Northern Australia, Townsville. CSIRO (2016) Transport Network Strategic Investment Tool (TraNSIT) – application to northern Australia beef roads submissions. CSIRO, Australia. CSIRO and BoM (2015) Climate change in Australia information for Australia’s natural resource management regions: technical report. CSIRO and Bureau of Meteorology, Australia. Culpitt R (2011) Healthy HeadWaters. Farm level economic and financial modelling of alternative irrigation practices. Synergies Economic Consulting, Brisbane. DAFF (2022) Horticulture fact sheet – Department of Agriculture, Fisheries and Forestry. Viewed 15 November 2022, https://www.agriculture.gov.au/agriculture-land/farm-food- drought/hort-policy. Dambacher JM, Rothlisberg PC and Loneragan NR (2015) Qualitative mathematical models to support ecosystem-based management of Australia’s Northern Prawn Fishery. Ecological Applications 25(1), 278–298. Daniels PL, Lenzen M and Kenway SJ (2011) The ins and outs of water use – a review of multi- region input-output analysis and water footprints for regional sustainability analysis and policy. Economic Systems Research 23(4), 353–370. DOI: 10.1080/09535314.2011.633500. Davey and Maynard (2010) Meander Valley agriculture study – report for Meander Valley Council (MVC) and the Department of Economic Development Tourism & Arts (DEDTA). Davey & Maynard Agricultural & Resource Management Consulting, Devonport, Tasmania. Viewed 20 December 2022, https://stors.tas.gov.au/au-7-0095-04081. DCCEEW (2022) National water grid investment framework. Australian Government Department of Climate Change, Energy, the Environment and Water, Canberra. Department of Environment Parks and Water Security (2018) Water Licensing - Groundwater and Surface Water Extraction points. https://ftp- dlrm.nt.gov.au/main.html?download&weblink=f75f2af704805fb44b9484017fc19bfd&realfilename=WaterManagement_GW_Licences.zip. Department of Environment Parks and Water Security (2019) Water Control Districts in the Northern Territory. https://ftp- dlrm.nt.gov.au/main.html?download&weblink=139913360c21f560ae66df0b5c113205&realfilename=WaterManagement_ControlDistricts.zip. Department of Regional Development and Lands (2009) Ord-East Kimberley development plan. Government of Western Australia, Perth. DeSantis D, James BD, Houchins C, Saur G and Lyubovsky M (2021) Cost of long-distance energy transmission by different carriers. iScience 24(12). DOI: https://doi.org/10.1016/j.isci.2021.103495. Dichmont CM, Deng A, Punt AE, Ellis N, Venables WN, Kompas T, Ye Y, Zhou S and Bishop J (2008) Beyond biological performance measures in management strategy evaluation: bringing in economics and the effects of trawling on the benthos. Fisheries Research 94, 238–250. DILGP (2017) State infrastructure plan, part B: program – 2017 update (July 2017). Queensland Department of Infrastructure, Local Government and Planning, Brisbane. Entegra Signature Structures (2022) How big is the horticulture industry in Australia? Entegra Signature Structures. Viewed 10 September 2022, https://www.entegra.com.au/how-big-is- the-horticulture-industry-in-australia/. EPRI (Electric Power Research Institute) (2015) Australian energy generation technology report. CO2CRC Ltd, Melbourne, Australia. Flyvbjerg B, Holm MS and Buhl S (2002) Underestimating costs in public works projects: error or lie? Journal of the American Planning Association 68, 279–295. Flyvbjerg B, Skamris Holm MK, and Buhl SL (2005) How (in)accurate are demand forecasts in public works projects? The case of transportation. Journal of the American Planning Association 71, 131–146. Giovi Agriculture (2018) Mowanjum irrigation trial industry report. Western Australia Department of Primary Industries and Regional Development, Perth. Gleeson T, Martin P and Mifsud C (2012) Northern Australian beef industry: assessment of risks and opportunities. ABARES report to client prepared for the Northern Australia Ministerial Forum, Canberra. Gleeson, J and Richards, M (1985) Mataranka and the Daly, two studies in history of settlement in Northern Territory. Northern Research Unit Monograph, Australian National University, Darwin. GPA (2021) Pipelines vs powerlines – a technoeconomic analysis in the Australian context. Report commissioned by the Australian Pipelines and Gas Association. GPA document no: 210739- REP-001. GPA Engineering, Sydney. GRDC (2017) Peanut profitability dependent on reducing yield gaps. Grains Research and Development Corporation, Canberra. Viewed October 2017, https://grdc.com.au/resources- and-publications/groundcover/ground-cover-issue-34-wa/huge-peanut-profits-await- closing-of-yield-gap. Gretton P (2013) On input-output tables: uses and abuses. Staff research note. Productivity Commission, Canberra. Viewed 2 December 2022, https://www.pc.gov.au/research/supporting/input-output-tables/input-output-tables.pdf. Grundy P, Yeates S and Grundy T (2012) NORpak cotton production and management guidelines for the Burdekin and north Queensland coastal dry tropics region. Cotton Catchment Communities Cooperative Research Centre, Canberra. Guan D and Hubacek K (2008) A new and integrated hydro-economic accounting and analytical framework for water resources: a case study for north China. Journal of Environmental Management 88(4), 1300–1313. DOI: 10.1016/j.jenvman.2007.07.010. Gunn, J (1908) We of the never-never, 15th edition (2003). Macmillan, New York. Guy JA, McIlgorm A and Waterman P (2014) Aquaculture in regional Australia: responding to trade externalities. A northern NSW case study. Journal of Economic & Social Policy 16(1), 115. Hausler P, Gara FO and Price T (2002) Agnote Growing grain sorghum in the Northern Territory. Trials, March 1–4. Hendrey GR, Lewin KF and Nagy J (1993) Free air carbon dioxide enrichment: development, progress, results. Vegetation 104, 17–31. Higgins A, McFallan S, Laredo L, Prestwidge D and Stone P (2015) TRANSIT– a model for simulating infrastructure and policy interventions in agriculture logistics: application to the northern Australia beef industry. Computers and Electronics and Agriculture 114, 32–42. Hoffman GJ, Evans RG, Jensen ME, Martin DL and Elliott RL (2007) Design and operation of farm irrigation systems, 2nd edition. American Society of Agricultural and Biological Engineers, St. Joseph, Michigan. Holzworth DP, Huth NI, deVoil PG, Zurcher EJ, Herrmann NI, McLean G, Chenu K, van Oosterom E, Snow V, Murphy C, Moore AD, Brown H, Whish JPM, Verrall S, Fainges J, Bell LW, Peake AS, Poulton PL, Hochman Z, Thorburn PJ, Gaydon DS, Dalgliesh NP, Rodriguez D, Cox H, Chapman S, Doherty A, Teixeira E, Sharp J, Cichota R, Vogeler I, Li FY, Wang E, Hammer GL, Robertson MJ, Dimes J, Carberry PS, Hargreaves JNG, MacLeod N, McDonald C, Harsdorf J, Wedgwood S and Keating BA (2014) APSIM − evolution towards a new generation of agricultural systems simulation. Environmental Modelling and Software 62, 327–350. Horridge M (2012) The TERM model and its database. In: Wittwer G (eds) Economic modeling of water: the Australian CGE experience. Global issues in water policy, volume 3. Springer, Dordrecht, 13–36. https://link.springer.com/chapter/10.1007/978-94-007-2876-9_2. Hort Innovation (2021) NTMs of commercial significance to mangoes. March, 1100–1102. Hughes J, Yang A, Marvanek S, Wang B, Petheram C and Philip S (2023) River model calibration and scenario analysis for the Roper catchment. A technical report from the CSIRO Roper River Water Resource Assessment for the National Water Grid. CSIRO, Australia. Icon Water (2018) Source water strategy 2018–2030. Icon Water Limited, Australia. Viewed 20 December 2022, https://www.iconwater.com.au/~/media/files/icon-water/water-and- sewerage-files/source-water/source-water-strategy- 20182030.pdf?la=en&hash=992CA2433177B8861B549289123A2AED1D32DE01. Infrastructure Australia (2021a) Post completion review Stage 4 of the Assessment Framework. Infrastructure Australia, Canberra. Viewed 1 March 2023, https://www.infrastructureaustralia.gov.au/sites/default/files/2021- 07/Assessment%20Framework%202021%20Stage%204.pdf. Infrastructure Australia (2021b) Guide to economic appraisal. Technical guide of the Assessment Framework. Infrastructure Australia, Canberra. Viewed 1 March 2023, https://www.infrastructureaustralia.gov.au/guide-economic-appraisal. Irvin S, Coman G, Musson D and Doshi A (2018) Aquaculture viability. A technical report to the Australian Government from the CSIRO Northern Australia Water Resource Assessment, part of the National Water Infrastructure Development Fund: Water Resource Assessments. CSIRO, Australia. Jackson D, Rolfe JW, English BH, Holmes B, Matthews R, Dixon RM, Smith P and MacDonald N (2012, reprinted 2015) Phosphorus management of beef cattle in northern Australia. Meat and Livestock Australia. Jackson P (2013) Energy cane. In: Saha MC, Bhandari HS and Bouton JH (eds) Bioenergy feedstocks: breeding and genetics. John Wiley and Sons. Jakku E, Thorburn PJ, Marshall NA, Dowd AM, Howden SM, Mendham E, Moon K and Brandon C (2016) Learning the hard way: a case study of an attempt at agricultural transformation in response to climate change. Climatic Change 137, 557–574. Jarvis D, Stoeckl N, Hill R and Pert P (2018) Indigenous land and sea management programs: can they promote regional development and help ‘close the (income) gap’? Australian Journal of Social Issues 53(3), 283–303. Jeffrey SJ, Carter JO, Moodie KB and Beswick AR (2001) Using spatial interpolation to construct a comprehensive archive of Australian climate data. Environmental Modelling and Software 16, 309–330. Jones (1995) Production of juvenile redclaw crayfish, Cherax quadricarinatus (von Martens) (Decapoda, Parastacidae) III. Managed pond production trials. Aquaculture 138(1), 247–255. DOI: https://doi.org/10.1016/0044-8486(95)00067-4. Jones (2007) Redclaw Package 2007. Introduction to Redclaw Aquaculture. Department of Primary Industries and Fisheries, Brisbane. Jones C, Grady J-A and Queensland Department of Primary Industries (2000) Redclaw from harvest to market: a manual of handling procedures. Department of Primary Industries, Brisbane. Jones C, Mcphee C and Ruscoe I (1998) Breeding redclaw: management and selection of broodstock. QI98016. Department of Primary Industries, Brisbane. Keating BA, Carberry PS, Hammer GL, Probert ME, Robertson MJ, Holzworth D, Huth NI, Hargreaves JNG, Meinke H, Hochman Z, McLean G, Verburg K, Snow V, Dimes JP, Silburn M, Wang E, Brown S, Bristow KL, Asseng S, Chapman S, McCown RL, Freebairn DM and Smith CJ (2003) An overview of APSIM, a model designed for farming systems simulation. European Journal of Agronomy 18, 267–288. Kim S, Ticehurst C, Hughes J, Marvanek S, Petheram C, Wang B and Vaze J (2023) Floodplain inundation mapping and modelling for the Roper catchment. A technical report from the CSIRO Roper River Water Resource Assessment for the National Water Grid. CSIRO, Australia. Knapton A, Taylor AR, Petheram C and Crosbie RS (2023) An investigation into the effects of climate change and groundwater development scenarios on the water resources of the Roper catchment using two finite element groundwater flow models. A technical report from the CSIRO Roper River Water Resource Assessment for the National Water Grid. CSIRO, Australia. KPMG and The University of Sydney China Studies Centre (2013) Demystifying Chinese investment in Australia. Australia still a priority destination, but the world is catching up. Viewed 2 March 2023, https://assets.kpmg.com/content/dam/kpmg/pdf/2013/10/Demystifying- Chinese-Investment-in-Australian-O-201303.pdf. Kumar V and Telfer D (2022) Developing a northern Australia industrial hemp industry value chain AgriFutures Australia publication no. 22-109. AgriFutures Australia, Wagga Wagga. L.E.K. Consulting (2021) Horticulture brief – international supply chain benchmarking sectoral assessment (Issue August). Lenzen M and Foran B (2001) An input–output analysis of Australian water usage. Water Policy 3(4), 321–340. DOI: 10.1016/S1366-7017(01)00072-1. Lockie S (2015) Australia’s agricultural future: the social and political context. Australian Council of Learned Academies, 71. Lopes AS and Guilherme LRG (2016) A career perspective on soil management in the Cerrado region of Brazil. Advances in Agronomy 137, 1–72. Lopes AS, Guilherme LRG and Ramos SJ (2012) The saga of the agricultural development of the Brazilian Cerrado. e-ifc, no. 32, 29–57, November, IPI 60 anniversary issue. Viewed 10 September 2017, https://www.ipipotash.org/publications/eifc-261. Mainstream Economics and Policy (2014) The economics of new water supply infrastructure for irrigated agriculture. Technical drafting notes for WWF prepared by MainStream Economics and Policy, Australia. Viewed 20 December 2022, https://www.etheridge.qld.gov.au/downloads/file/232/wwf-irrigation-infrastructure- whitepaperpdf. Mann D (2012) Impact of aerator biofouling on farm management, production costs and aerator performance. Mid project report to farmers. Seafood CRC, Adelaide. Marsden Jacob Associates (2003) Improving water-use efficiency in irrigation conveyance systems: a study of investment strategies. Marsden Jacob Associates, Land and Water Australia, Australia. Masser M and Rouse B (1997) Australian red claw crayfish. The Alabama Cooperative Extension Service, USA. Mathew S, Zeng B, Zander KK and Singh RK (2018) Exploring agricultural development and climate adaptation in northern Australia under climatic risks. Rangeland Journal 40(4), 353-364. DOI: https://doi.org/10.1071/RJ18011. McJannet D, Yang A and Seo L (2023) Climate data characterisation for hydrological and agricultural scenario modelling across the Victoria, Roper and Southern Gulf catchments. A technical report from the CSIRO Victoria, Roper and Southern Gulf Water Resource Assessments for the National Water Grid. CSIRO, Australia. McKellar L, Bark RH and Watson I (2015) Agricultural transition and land-use change: considerations in the development of irrigated enterprises in the rangelands of northern Australia. The Rangeland Journal 37, 445–457. McKeon G, Ash A, Hall W and Stafford Smith M (2000) Simulation of grazing strategies for beef production in north-east Queensland. In: Hammer GL, Nicholls N and Mitchell C (eds) Applications of seasonal climate forecasting in agricultural and natural ecosystems. Kluwer Academic Publishers, The Netherlands, 227–252. McLean I and Holmes P (2015) Improving the performance of northern beef enterprises, 2nd edition. Meat and Livestock Australia, Sydney. McLennan W (1996) Information paper: Australian national accounts: introduction to input-output multipliers. Catalogue no. 5246.0. Australian Bureau of Statistics, Canberra. Meat and Livestock Australia (2006) Beef cattle nutrition. An introduction to the essentials. Meat and Livestock Australia, Sydney. Meateng (2011) Pre-feasibility study into a northern Western Australian beef abattoir. Hyper Consulting Pty Ltd. Meateng (2018) Analysis of beef cattle supply and evaluation of commercial viability of locations for processing facilities in Queensland. Meateng, Australia. Meyer WS (2005) The irrigation industry in the Murray and Murrumbidgee basins. Cooperative Research Centre for Irrigation Futures. MJA (2003) Meander dam project economic analysis Report for Department Of Primary Industries Water & Environment. Marsden Jacob Associates, Financial & Economic Consultants, Victoria. Viewed 20 December 2022, https://nla.gov.au/nla.obj-1837979351/view. Moore G, Revell C, Schelfhout C, Ham C and Crouch S (2021) Mosaic agriculture. A guide to irrigated crop and forage production in northern WA. Bulletin 4915. Department of Regional Industries and Regional Development, Perth. Moulden JH, Yeates SJ, Strickland GR and Plunkett GM, (2006) Developing an environmentally responsible irrigation system for cotton in the Ord River Irrigation Area. In: Proceedings of ANCID 2006, 16–19 October, Darwin, NT. The Australian National Committee on Irrigation and Drainage. Murti S and Northern Territory Office of Resource Development (2001) Input-output multipliers for the Northern Territory 1997–1998. Office of Resource Development, Darwin. National Competition Council (2003) Assessment of governments’ progress in implementing the national competition policy and related reforms: volume one – overview of the national competition policy and related reforms. AusInfo, Canberra. Viewed 20 December 2022, http://ncp.ncc.gov.au/docs/2003%20assessment.pdf. NECG (2001) Indicative economic impacts of additional water storage infrastructure in the Burnett Region. Network Economics Consulting Group Pty Ltd, Australia. Viewed 20 December 2022, https://wbbec.files.wordpress.com/2013/03/paradise_eis_appl_ecoreport.pdf. NESP Earth Systems and Climate Change Hub (2019) Impact of climate change on mango production in the Northern Territory. Earth Systems and Climate Change Hub Report No. 9. NESP Earth Systems and Climate Change Hub, Australia. NIAA (2021) Indigenous advancement strategy. National Indigenous Australians Agency, Canberra. Viewed 15 December 2022, https://www.niaa.gov.au/indigenous-affairs/grants-and- funding/indigenous-advancement-strategy. Nicolaisen MS and Driscoll PA (2014) Ex-post evaluations of demand forecast accuracy: a literature review. Transport Reviews 34, 540–557. Northern Territory Economy (2022) Budget 2022–23. Northern Territory Government, Darwin. Northern Territory Government (2022) 2022–23 budget papers. Viewed 3 June 2022, https://budget.nt.gov.au/budget-papers. NT Department of Environment Parks and Water Security (2018) Water Licensing - Groundwater and Surface Water Extraction points. Viewed 13 March 2023, https://ftp- dlrm.nt.gov.au/main.html?download&weblink=f75f2af704805fb44b9484017fc19bfd&realfilename=WaterManagement_GW_Licences.zip. NT Department of Environment Parks and Water Security (2019) Water Control Districts in the Northern Territory. Viewed 13 March 2023, https://ftp- dlrm.nt.gov.au/main.html?download&weblink=139913360c21f560ae66df0b5c113205&realfilename=WaterManagement_ControlDistricts.zip. NT Department of Treasury and Finance (2022) Agriculture, forestry and fishing – Northern Territory economy. Viewed 6 June 2022, https://nteconomy.nt.gov.au/industry- analysis/agriculture,-foresty-and-fishing. NT Farmers (2019) Northern Territory plant industries economic impact analysis. NT Farmers Association, Coolalinga. NT PHN (2020) Katherine Region data report: overview of selected demographic and health data for the Katherine region of the Northern Territory. Australian Government, Northern Territory Primary Health Network, Department of Health and Aged Care, Darwin. NWGA (2022) National Water Grid investment framework. Australian Government, National Water Grid Authority, Department of Climate Change, Energy, the Environment and Water, Canberra. Viewed 1 March 2023, https://www.nationalwatergrid.gov.au/framework. NWGA (2023) Project administration manual: National Water Grid Fund. Australian Government, National Water Grid Authority, Department of Climate Change, Energy, the Environment and Water, Canberra. Viewed 1 March 2023, https://www.nationalwatergrid.gov.au/framework. O’Gara FP (2010) Striking the balance – conservation farming and grazing systems for the semi- arid tropics of the Northern Territory. Second Edition. Northern Territory Government, Darwin. Odeck J and Skjeseth T (1995) Assessing Norwegian toll roads. Transportation Quarterly 49(2), 89– 98. O'Doherty J and Tol RSJ (2007) An environmental input-output model for Ireland. The Economic and Social Review 38(2), 157–190. Office of the Government Statistician Queensland Government (2004) Queensland regional input- output tables 1996–97 34 industries. Queensland Government, Brisbane. Parry ML, Rosenzweig C, Iglesias A, Livermore M and Fischer G (2004) Effects of climate change on global food production under {SRES} emissions and socio-economic scenarios. Global Environmental Change 14(1), 53–67. Paterson B and Miller S (2013) Energy use in shrimp farming, study in Australia keys on aeration and pumping demands. Global Aquaculture Advocate. Peake AS, Carberry PS, Raine SR, Gett V and Smith RJ (2016) An alternative approach to whole- farm deficit irrigation analysis: evaluating the risk-efficiency of wheat irrigation strategies in sub-tropical Australia. Agricultural Water Management 169, 61–76. Pearson L and Langridge J (2008) Climate change vulnerability assessment: review of agricultural productivity. CSIRO Climate Adaptation Flagship Working paper no.1. CSIRO, Canberra. Petheram C and McMahon TA (2019) Dams, dam costs and damnable cost overruns. Journal of Hydrology X 3, 100026. DOI: 10.1016/j.hydroa.2019.100026. Petheram C, Watson I and Stone P (eds) (2013a) Agricultural resource assessment for the Flinders catchment. A report to the Australian Government from the CSIRO Flinders and Gilbert Agricultural Resource Assessment, part of the North Queensland Irrigated Agriculture Strategy. CSIRO Water for a Healthy Country and Sustainable Agriculture Flagships, Australia. Petheram C, Watson I and Stone P (eds) (2013b) Agricultural resource assessment for the Gilbert catchment. A report to the Australian Government from the CSIRO Flinders and Gilbert Agricultural Resource Assessment, part of the North Queensland Irrigated Agriculture. Petheram C, Yang A, Seo L, Rogers L, Baynes F, Marvanek S, Hughes J, Ponce Reyes R, Wilson P, Stratford D, Philip S (2022) Assessment of surface water storage options and reticulation infrastructure in the Roper catchment. A technical report from the CSIRO Roper River Water Resource Assessment for the National Water Grid. CSIRO, Australia. Pettit C (undated) Sturt Plateau District. Land condition guide. Understanding the productivity of grazing lands. Northern Territory Government. Piaui C (2010) Brazilian agriculture the miracle of the Cerrado. The Economist. Viewed 17 November 2017, http://www.economist.com/node/16886442?story_id=16886442. Plant Health Australia (2021) Mangoes. Viewed 7 September 2022, www.planthealthaustralia.com.au/%0Aindustries/mangoes. Pratt Water (2004) The business of saving water: the report of the Murrumbidgee Valley Water Efficiency Feasibility Project. Pratt Water, Campbellfield, Victoria. PwC (PricewaterhouseCoopers) (2019) Business case for the construction of a cotton gin in the Northern Territory. NT Farmers Association, Darwin. QDAF (2017) Queensland-relevant information on field crops and pastures. Queensland Department of Agriculture and Fisheries, Brisbane. QDPIF (2006) Australian prawn farming manual: health management for profit. Dept. of Primary Industries and Fisheries, Brisbane, Qld Queensland Government (2009) South East Queensland infrastructure plan and program 2009 2026. Department of Infrastructure and Planning, Queensland Government, Brisbane. Viewed 20 December 2022, https://cabinet.qld.gov.au/documents/2009/jun/seqipp%202009- 26/Attachments/seqipp.pdf. QWI (2007) Wyaralong Dam environmental impact statement. Queensland Water Infrastructure Pty Ltd, Australia. Viewed 20 December 2022, https://www.statedevelopment.qld.gov.au/coordinator-general/assessments-and- approvals/coordinated-projects/completed-projects/wyaralong-dam. Raine SR and Bakker DM (1996) Increased furrow irrigation efficiency through better design and management of cane fields. Proceedings Australian Society of Sugar Cane Technologists 1996, 119–124. Rashed MRU (2016) Substrate effects on plant transpiration rate under several vapour pressure deficit (VPD) levels. Journal of Plant Pathology and Microbiology 7, 369. DOI:10.4172/2157- 7471.1000369. Resource Economics Unit (2007) Inquiry into competition in the water and wastewater services sector: Water trading issues. Prepared for the Economic Regulation Authority of Western Australia. Viewed 20 December 2022, https://www.erawa.com.au/cproot/6231/2/Resource%20Economics%20Unit%20- %20Water%20Trading%20Issues.pdf. Riahi K, Rao S, Krey V, Cho C, Chirkov V, Fischer G, Kindermann G, Nakicenovic N and Rafaj P (2011) RCP 8.5—a scenario of comparatively high greenhouse gas emissions. Climatic Change 109, 33. Rickert KG, Stuth JW and McKeon GM (2000) Modelling pasture and animal production. In: L't Mannetje and Jones RM (eds) Field and laboratory methods for grassland and animal production research. CABI Publishing, New York, 29–66. RIRDC (2013) Fodder crops. AgriFutures Australia, RIRDC Publication No 03/072. Rural Industries Research and Development Corporation, Canberra. RLB (2021) Riders Digest 2021: Darwin, Australia edition. Rider Levett Bucknall, Darwin. Robertson M, Kirkegaard J, Peake A, Creelman Z, Bell L, Lilley J, Midwood J, Zhang H, Kleven S, Duff C and Lawes R (2016) Trends in grain production and yield gaps in the high-rainfall zone of southern Australia. Crop and Pasture Science 67(9), 921–937. Rogers CDW and Beringer J (2017) Describing rainfall in northern Australia using multiple climate indices. Biogeosciences 14(3), 597–615. Viewed 13 November 2017, http://doi.org/10.5194/bg-14-597-2017. Sangha KK, Ahammad R, Mazahar MS, Hall M, Owens G, Kruss L, Verrall G, Moro J and Dickinson G (2022) An integrated assessment of the horticulture sector in northern Australia to inform future development. Sustainability (Switzerland) 14(18), 1–18. DOI: https://doi.org/10.3390/su141811647. Schipp G, Humphrey JD, Bosmans J and Northern Territory Department of Primary Industry Fisheries and Mines (2007) Northern Territory barramundi farming handbook. Department of Primary Industry, Fisheries and Mines, Darwin. Sefton and Associates (2013) Blueprint for Australian agriculture, 2013–2020. National Farmers’ Federation and Sefton and Associates. SEQWater (2017) Water for life South East Queensland’s water security program 2016–2046. Queensland Bulk Water Supply Authority, trading as Seqwater, Australia. Viewed 20 December 2022, https://www.seqwater.com.au/sites/default/files/2019- 08/Water%20for%20life_Water%20Security%20Program.pdf. Sheng D, Owen S, Lambert DM, English BC, Menard RJ, Hughes DW and Clark CD (2019). A multiregional input-output analysis of water withdrawals in the southeastern United States. The Review of Regional Studies 49(2), 323. DOI: 10.52324/001c.8259. Shuttleworth WJ (1993) Evaporation. In: Maidment DR (ed) Handbook of hydrology. McGraw-Hill, Inc., New York, 4.1–4.53. Steven, AH, Dylewski, M and Curtotti, R (2021) Australian fisheries and aquaculture statistics 2020. Fisheries Research and Development Corporation, ABARES, Canberra. August CC BY 4.0. DOI: https://doi.org/10.25814/0wzy-re76. Stoeckl N (2007) Using surveys of business expenditure to draw inferences about the size of regional multipliers: a case-study of tourism in Northern Australia. Regional studies 41(7), 917–931. DOI: 10.1080/00343400601142803. Stoeckl N (2012) Comparing multipliers from survey and non-survey based IO models: an empirical investigation from northern Australia. International Regional Science Review 35(4), 367–388. DOI: 10.1177/0160017610385452. Stoeckl N and Stanley O (2009) Maximising the benefits of development in Australia’s far north. Australasian Journal of Regional Studies 15(3), 255–280. Stoeckl N, Esparon M, Farr M, Delisle A and Stanley O (2013) The great asymmetric divide: an empirical investigation of the link between Indigenous and non-Indigenous economic systems in northern Australia. Papers in Regional Science, n/a-n/a. DOI: 10.1111/pirs.12028. Stoeckl N, Esparon M, Stanley O, Farr M, Delisle A and Altai Z (2011) Socio-economic activity and water use in Australia’s tropical rivers: a case study in the Mitchell and Daly river catchments. Charles Darwin University, Darwin. Viewed 15 December 2022, https://www.nespnorthern.edu.au/wp-content/uploads/2016/02/TRaCK-Project-3.1-Final- Report-March-2011.pdf. Stokes C, Addison J, Macintosh A, Jarvis D, Higgins A, Doshi A, Waschka M, Jones J, Wood A, Horner N, Barber M, Bruce C, Austin J and Lau J (2017) Costs, benefits, institutional and social considerations for irrigation development. A technical report to the Australian Government from the CSIRO Northern Australia Water Resource Assessment, part of the National Water Infrastructure Development Fund: Water Resource Assessments. CSIRO, Australia. Stokes C and Jarvis D (2018) Chapter 6: Overview of economic opportunities and constraints. In: Petheram C, Chilcott C, Watson I and Bruce C (eds) (2018) Water resource assessment for the Darwin catchments. A report to the Australian Government from the CSIRO Northern Australia Water Resource Assessment, part of the National Water Infrastructure Development Fund: Water Resource Assessments. CSIRO, Australia. Stokes C and Jarvis D (2021) Economic assessment. In: Petheram C, Read A, Hughes J, Marvanek S, Stokes C, Kim S, Philip S, Peake A, Podger G, Devlin K and Hayward J. An assessment of contemporary variations of the Bradfield Scheme. A technical report to the National Water Grid from the Bradfield Scheme Assessment. CSIRO, Canberra. Stokes C, Phillip S, McConnell A, Pert P, Zhoa Y, Robinson C, Barber M and Watson I (2019) Integrated analysis for sustainable regional development: northern Australia: a methods report supporting the key project insights. CSIRO, Canberra. https://publications.csiro.au/rpr/pub?pid=csiro:EP196609. Stratford D, Kenyon R, Pritchard J, Merrin L, Linke S, Ponce Reyes R, Blamey L, Buckworth R, Castellazzi P, Costin B, Deng R, Gannon R, Gilbey S, King D, Kopf K, Kopf S, McGinness H, McInerney P, Perna C, Plaganyi E and Waltham N (2022) Ecological assets of northern Australia to inform water resource assessments. A technical report from the CSIRO Roper River Water Resource Assessment for the National Water Grid. CSIRO, Australia. Stratford D, Merrin L, Linke S, Kenyon R, Ponce Reyes R, Buckworth R, Deng R, McGinness H, Pritchard J, Seo L and Waltham N (2023) Assessment of the potential ecological outcomes of water resource development in the Roper catchment. A technical report from the CSIRO Roper River Water Resource Assessment for the National Water Grid. CSIRO, Australia. Sunwater (2019) Paradise Dam improvement project overview report. Viewed 20 December 2022, https://www.sunwater.com.au/wp-content/uploads/Home/Projects/Paradise-Dam- Essential-works/Sunwater_Overview_Report_-_November_2019.pdf. Taylor AR, Crosbie RS, Turnadge C, Lamontagne S, Deslandes A, Davies PJ, Barry K, Suckow A, Knapton A, Marshall S, Hodgson G, Tickell S, Duvert C, Hutley L and Dooley K (2023) Hydrogeological assessment of the Cambrian Limestone Aquifer and the Dook Creek Aquifer in the Roper catchment, Northern Territory. A technical report from the CSIRO Roper River Water Resource Assessment for the National Water Grid. CSIRO, Australia. Taylor J (2013) Data for better Indigenous policy evaluation: achievements, constraints and opportunities. Productivity Commission, Canberra. Viewed 15 December 2022, https://www.pc.gov.au/research/supporting/better-indigenous-policies/better-indigenous- policies.pdf#page=125. Thomas M, Philip S, Stockman U, Wilson PR, Searle, R, Hill J, Bui E, Gregory, L, Watson, I, Wilson PL and Gallant G (2022) Soils and land suitability for the Roper catchment, Northern Territory. A technical report from the CSIRO Roper River Water Resource Assessment for the National Water Grid. CSIRO, Australia. Timbal B, Abbs D, Bhend J, Chiew FHS, Church J, Ekstrom M, Kirono D, Lenton A, Lucas C, McInnes K, Moise AF, Monselesan D, Mpelasoka F, Webb L and Whetton PH (2015) Murray Basin cluster report. In: Climate change in Australia – projections for Australia’s natural resource management regions: technical report. CSIRO and Bureau of Meteorology, Australia. Tubiello FN, Soussana JF and Howden SM (2007) Crop and pasture response to climate change. Proceedings of the National Academy of Sciences 104(50), 19686–19690. Tyler R, English B, Sullivan M, Jackson D, Matthews R, Holmes B, MacDonald N, Oxley T, Leigo S and Smith P (2012) Weaner management in northern beef herds. Meat and Livestock Australia. van Wee B (2007) Large infrastructure projects: a review of the quality of demand forecasts and cost estimations. Environment and Planning B: Planning and Design, 34, 611–625. Vance DJ, Bishop J, Dichmont CM, Hall N, McInnes K and Taylor BR (2003) Management of common banana prawn stocks of the Gulf of Carpentaria: separating the effects of fishing from the environment. CSIRO Report 1998/0716, Brisbane. Vance DJ, Haywood MDE, Heales DS, Kenyon RA and Loneragan RN (1998) Seasonal and annual variation in abundance of postlarval and juvenile banana prawns Penaeus merguiensis and environmental variation in two estuaries in tropical northeastern Australia: a six year study. Marine Ecology Progress Series 163, 21–36. Vanderbyl T (2021) Southern Gulf: Queensland water plans and settings. A report from the CSIRO Southern Gulf Water Resource Assessment to the Government of Australia. CSIRO, Australia. Verburg K, Bond WJ and Smith CJ (2003) Use of APSIM to simulate water balances of dryland farming systems in south eastern Australia. CSIRO Land and Water, Canberra. Wachs M (1990) Ethics and advocacy in forecasting for public policy. Business and Professional Ethics Journal 9, 1–2. Water and Rivers Commission (1998) Proposed Harvey Basin surface water allocation plan. Water and Rivers Commission, Water Resource Allocation and Planning Series No WRAP 14. Viewed 20 December 2022, https://www.water.wa.gov.au/__data/assets/pdf_file/0016/4561/10026.pdf. Water and Rivers Commission (2000) Stirling Dam catchment area water source protection plan: Perth metropolitan and Harvey town water supply. Water and Rivers Commission, Water Resource Protection Series No WRP 34. Viewed 20 December 2022, https://www.water.wa.gov.au/__data/assets/pdf_file/0014/4262/11495.pdf. Watson I, Ash A, Petheram C, Barber M and Stokes C (2021a) Development in the northern rivers of Australia. In: Ferrier B and Jenkins A (eds) Handbook of catchment management, 2nd edition. John Wiley & Sons, 465–497. DOI: 10.1002/9781119531241. Watson, I, Austin J and Ibrahimi T (2021b) Other potential users of water. Chapter 8 in Petheram C, Read A, Hughes J, Marvanek S, Stokes C, Kim S, Philip S, Peake A, Podger G, Devlin K, Hayward J, Bartley R, Vanderbyl T, Wilson P, Pena Arancibia J, Stratford D, Watson I, Austin J, Yang A, Barber M, Ibrahimi T, Rogers L, Kuhnert P, Wang B, Potter N, Baynes F, Ng S, Cousins A, Jarvis D and Chilcott C (2021) An assessment of contemporary variations of the Bradfield Scheme. A technical report to the National Water Grid Authority from the Bradfield Scheme Assessment. CSIRO, Australia. Webster T, Poulton P, Yeates SJ, Cocks B, Hornbuckle J, Gentle J, Brennan McKellar L, Mayberry D, Jones D and Wixon A (2013) Agricultural productivity in the Flinders and Gilbert catchments. A technical report to the Australian Government from the CSIRO Flinders and Gilbert Agricultural Resource Assessment, part of the North Queensland Irrigated Agriculture Strategy. CSIRO Water for a Healthy Country and Sustainable Agriculture flagships, Australia. Williams J (2009) Final report. The water and salt balances of the Burdekin River Irrigation Area. Prepared for Queensland Canegrowers Pty Ltd, Brisbane, QLD. Williams J, Day KJ, Isbell RF and Reddy SJ (1985) Soils and climate. In: Muchow RC (ed.) Agro- research for the semi-arid tropics: north-west Australia. University of Queensland Press, St Lucia. Winter WH (1988) Supplementation of steers grazing Stylosanthes hamata pastures at Katherine, Northern Territory. Australian Journal of Experimental Agriculture 28, 669-682 Wittwer G (ed) (2012) Economic modeling of water: the Australian CGE experience (vol. 3). Springer, Dordrecht, Netherlands. Wood M, Wang Q and Bethune M (2007) An economic analysis of conversion from border-check to centre-pivot irrigation on dairy farms in the Murray dairy region. Irrigation Science 26(1), 9–20. Woodhams J, Larcombe J and George D (2011) Northern Prawn Fishery. In: Woodhams J, Stobutzki I, Vieira S, Curtotti R and Begg GA (eds) Fishery status reports 2010: status of fish stocks and fisheries managed by the Australian Government. ABARES, Canberra. Woodward E, Jarvis D, Maclean K (2019) The traditional owner-led bush products sector: an overview. CSIRO, Australia. World Commission on Dams (2000a) Cross-check survey: final report. World Commission on Dams, Cape Town, South Africa. Viewed 20 December 2022, https://open.uct.ac.za/handle/11427/4835. World Commission on Dams (2000b) Dams and development A new framework for decision making. Earthscan Publications Ltd, UK. Viewed 20 December 2022, https://www.ern.org/wp- content/uploads/sites/52/2016/12/2000_world_commission_on_dams_final_report.pdf Yeates SJ (2001) Cotton research and development issues in northern Australia: a review and scoping study. Australian Cotton Cooperative Research Centre, Darwin. Yeates SJ and Martin CC (2006) Progress in predicting the hydrology of cotton farming systems and the impact on the hydrology of the largely uncleared Katherine – Daly catchments in the NT. In: Proceedings of ANCID 2006, 16–19 October, Darwin. The Australian National Committee on Irrigation and Drainage. Yeates S, Nolen S and Goldsmith P (2022) Cotton irrigation water requirement and response to water stress 2019 & 2020. In: Cotton production system report for CRCP-P586604 ‘Developing sustainable cropping systems for cotton, grains and fodder’ Yeates S.J. (ed), CRC Business. Yeates SJ and Poulton PL (2019) Determining dryland cotton yield potential in the NT: Preliminary climate assessment and yield simulation. Report to NT Farmers, Queensland Cotton and the Cotton Research and Development Corporation. CSIRO, Canberra. Yeates SJ, Strickland GR and Grundy PR (2013) Can sustainable cotton production systems be developed for tropical northern Australia? Crop and Pasture Science 64, 1127–1140. Zaar, U (2009) Gulf water study: Roper River region. Report 16/2009D. Northern Territory Government, Darwin. Part V Appendices Aquaculture opportunities and viability A.1 Introduction There are considerable opportunities for aquaculture development in northern Australia given its natural advantages of a climate suited to farming valuable tropical species, the large areas identified as suitable for aquaculture, political stability and proximity to large global markets. The main challenges to developing and operating modern and sustainable aquaculture enterprises are regulatory barriers, global cost competitiveness and the remoteness of much of the suitable land area. This appendix draws on a recent assessment of the opportunities for aquaculture in northern Australia (Irvin et al., 2018) summarising: the three most likely candidate species (Section A.2); an overview of production systems (Section A.3); and the financial viability of different types of aquaculture developments (Section A.4). A.2 Candidate species The three species with the most aquaculture potential in the Roper catchment are black tiger prawns (Penaeus monodon), barramundi (Lates calcarifer), and red claw (Cherax quadricarinatus). The first two species are suited to many marine and brackish water environments of northern Australia and have established land-based culture practices and well-established markets for harvested products. Prawns could potentially be cultured in either extensive (low density, low input) or intensive (higher density, higher inputs) pond-based systems in northern Australia, whereas land-based culture of barramundi would likely be intensive. Red claw is a freshwater crayfish that is currently cultured by a much smaller industry than the previous two species. Black tiger prawns Black tiger prawns are found naturally at low abundances across the waters of the western Indo- Pacific region, with wild Australian populations making up the southernmost extent of the species. Within Australia, the species is most common in the tropical north, but does occur at lower latitudes. Barramundi Barramundi is the most highly produced and valuable tropical fish species in Australian aquaculture. Barramundi inhabit the tropical north of Australia from the Exmouth Gulf in WA through to the Noosa River on Queensland’s east coast. It is also commonly known as ‘Asian sea bass’ or ‘giant sea perch’ throughout its natural areas of distribution in the Persian Gulf, the western Indo-Pacific region and southern China (Schipp et al., 2007). The attributes that make barramundi an excellent aquaculture candidate are: fast growth (1 kg or more in 12 months); year- round fingerling availability; well-established production methods; and hardiness (i.e. they have a tolerance to low oxygen levels, high stocking densities and handling, as well as a wide range of temperatures) (Schipp et al., 2007). In addition, barramundi are euryhaline (able to thrive and be cultured in fresh and marine water) but freshwater barramundi can have an earthy flavour. Red claw Red claw is a warm-water crayfish species that inhabits still or slow-moving water bodies. The natural distribution of red claw ranges from the tropical catchments of Queensland and the NT to southern New Guinea. The name ‘red claw’ is derived from the distinctive red markings present on the claws of the male crayfish. The traits of red claw that make them attractive for aquaculture production are: a simple life cycle, which is beneficial in that complex hatchery technology is not required (Jones et al., 1998); they can tolerate low oxygen levels (<2 mg/L), which is beneficial in terms of handling, grading and transport (Masser and Rouse, 1997); they have a broad thermal tolerance, with optimal growth achievable between 23 and 31 °C; and they can remain out of the water for extended periods. A.3 Production systems Overview Aquaculture production systems can be broadly classified into extensive, semi-intensive and intensive systems. Intensive systems require high inputs, with expected high outputs: they require high capital outlay; high running costs; specially formulated feed; specialised breeding, water quality and biosecurity processes; and have high production per hectare (in the order of 5,000 to 20,000 kg/ha per crop). Semi-intensive systems involve stocking seed from a hatchery, routine provision of a feed, and monitoring and management of water quality. Production is typically 1000 to 5000 kg/ha per crop. Extensive systems are characterised by low inputs and low outputs: they require less sophisticated management and often require no supplementary feed because the farmed species live on naturally produced feed in open-air ponds. Extensive systems produce about half the volume of global aquaculture production (but there are few commercial operations in Australia). Water salinity and temperature are the key parameters that determine species selection and production potential for any given location. Suboptimal water temperature (even within tolerable limits) will prolong the production season (slow growth) and increase the risk of disease, reducing profitability. The primary culture units for land-based farming are purpose-built ponds. Pond structures typically include an intake channel, production pond, discharge channel and a bioremediation pond (Apx Figure A-1). The function of the pond is to be a containment structure, an impermeable layer between the pond water and the local surface water and groundwater. Optimal sites for farms are flat and have sufficient elevation to enable ponds to be completely drained between seasons. It is critical that all ponds and channels can be fully drained during the off (dry-out) season to enable machinery access to sterilise and undertake pond maintenance. For more information on this figure please contact CSIRO on enquiries@csiro.au Apx Figure A-1 Schematic of marine aquaculture farm Most production ponds in Australia are earthen. Soils for earthen ponds should have low permeability and high structural stability. Ponds should be lined if the soils are permeable. Synthetic liners have a higher capital cost but are often used in high-intensity operations, which require high levels of aeration (conditions that would lead to significant erosion in earthen ponds). Farms use aerators (typically electric paddlewheels and aspirators) to help maintain optimal water quality in the pond, provide oxygen, and create a current that consolidates waste into a central sludge pile (while keeping the rest of the pond floor clear). A medium-sized 50-ha prawn farm in Australia uses around 4 GW⋅h annually, with pond aeration accounting for most of an enterprise’s energy use (Paterson and Miller, 2013). Back-up power capacity sufficient to run all the aerators on the farm, usually via a diesel generator, is essential to be able to cope with power failures. Black tiger prawns For black tiger prawns, a typical pond in the Australian industry would be rectangular in shape, about 1 ha in area and about 1.5 m in depth. The ponds are either wholly earthen, lined on the banks with black plastic and earthen bottoms, or (rarely in Australia) fully lined. Pond grow-out of black tiger prawns typically operates at stocking densities of 25–50 individuals per m2 (termed ‘intensive’ in this report). These pond systems are fitted with multiple aeration units (that could double from 8 to 16 units as the biomass of the prawn crop increases) (Mann, 2012). At the start of each prawn crop, pond bottoms are dried and unwanted sludge from the previous crops is removed, and if needed, additional substrate is added. Prior to filling the ponds, lime is often added to buffer pH, particularly in areas with acid sulfate soils. The ponds are then filled with filtered seawater and left for about 1 week prior to postlarval stocking. Algal blooms in the water are encouraged through addition of organic fertiliser to provide shading for prawns, discourage benthic algal growth, and stimulate growth of plankton as a source of nutrition (QDPIF, 2006). Postlarvae are purchased from hatcheries and grow rapidly into small prawns in the first month after stocking, relying mainly on the natural productivity (zooplankton, copepods and algae) supported by the algal bloom for their nutrition. Approximately 1 month after the prawns are stocked, pellet feed becomes the primary nutrition source. Feed is a major cost of prawn production; around 1.5 kg of feed is required to produce 1 kg of prawns. Prawns typically reach optimal marketable size (30 g) within 6 months. After harvest, prawns are typically processed immediately, with larger farms having their own production facilities that enable grading, cooking, packaging and freezing activities. Effective prawn farm management involves maintaining optimal water quality conditions, which becomes progressively more complex as prawn biomass and the quantity of feed added to the system increases. As prawn biomass increases, so too does the biological oxygen demand required by the microbial population within the pond in breaking down organic materials. This requires increases in mechanical aeration and water exchanges (either fresh or recycled from a bioremediation pond). In most cases water salinity is not managed, except through seawater exchange, and will increase naturally with evaporation and decrease with rainfall and flooding. Strict regulation of the quality and volume of water that can be discharged means efficient use of water is standard industry practice. Most Australian prawn farms allocate up to 30% of their productive land for water treatment by pre-release containment in settlement systems. Barramundi The main factors that determine productivity of barramundi farms are the provision of optimal water temperature, dissolved oxygen, effective waste removal, expertise of farm staff, and the overall health of the stock. Barramundi are susceptible to a variety of bacterial, fungal and parasitic organisms, and are at highest risk of disease when exposed to suboptimal water quality conditions (e.g. low oxygen or temperature extremes). Due to the cost and infrastructure required, many producers elect to purchase barramundi fingerlings from independent hatcheries, moving fish straight into their nursery cycle. Regular size grading is essential during the nursery stage due to aggressive and cannibalistic behaviour. Size grading helps to prevent mortalities and damage from predation on smaller fish and assists with consistent growth. Ponds are typically stocked to a biomass of about 3 kg per 1000 L. Under optimal conditions barramundi can grow to over 1 kg in 12 months and to 3 kg within 2 years (Schipp et al., 2007). A pellet feed is produced by the two largest Australian aquafeed manufacturers (located in Brisbane and Hobart), providing a specific diet promoting efficient growth and feed conversion. The industry is heavily reliant on these mills to provide a regular supply of high-quality feed. The cost of feed transport would be a major cost to barramundi production in the Roper catchment. As a carnivorous species, high dietary protein levels, with fishmeal as a primary ingredient, is required for optimal growth. Barramundi typically require between 1.2 and 1.5 kg of pelleted feed for each 1 kg of body weight produced. Warm water temperatures in northern Australia enable fish to be stocked in ponds year round. Depending on the intended market, harvested product is processed whole or as fillets and delivered fresh (refrigerated, ice slurry) or frozen. Smaller niche markets for live barramundi are available for Asian restaurants in some capital cities. Red claw Water temperature and feed availability are the variables that most affect crayfish growth. Red claw are a robust species but are most susceptible to disease (including viruses, fungi, protozoa, bacteria) when conditions in the production pond are suboptimal (Jones, 1995). In tropical regions, mature females can be egg bearing year round. Red claw breed freely in production ponds, so complex hatchery technology (or buying juvenile stock) is not required. However, low fecundity, and the associated inability to source high numbers of quality selected broodstock, is an impediment to intensive expansion of the industry. Production ponds are earthen lined, rectangular in design and average 1 ha, and are sloping in depth from 1.2 m to 1.8 m. Sheeting is used on the pond edge to keep the red claw in the pond (migration tendency) and netting surrounds the pond to protect stock from predators (Jones et al., 2000). At the start of each crop, ponds are prepared (as for black tiger prawns above) then filled with fresh water and left for about 2 weeks prior to stocking. During this period, algal blooms in the water are encouraged through addition of organic fertiliser. Ponds are then stocked with about 250 females and 100 males that have reached sexual maturity. Natural mating results in the production of around 20,000 advanced juveniles. Red claw are omnivorous, foraging on natural productivity such as microbial biomass associated with decaying plants and animals. Early-stage crayfish rely almost solely on natural pond productivity (phytoplankton and zooplankton) for nutrition. As the crayfish progress through the juvenile stages, the greater part of the diet changes to organic particulates (detritus) on the bottom of the pond. Very small quantities of a commercial feed are also added on a daily basis to assist with the weaning process and provide an energy source for the pond bloom. The provision of adequate shelters (net bundles) is essential at this stage to improve survival (Jones, 2007). Approximately 4 months after stocking, the juveniles are harvested and graded by size and sex for stocking in production ponds. Juveniles are stocked in production ponds at 5–10 per m2. Shelters are important during the grow- out stage, with 250/ha recommended. During the grow-out phase, pellet feed becomes an important nutrition source, along with the natural productivity provided by the pond. Current commercial feeds are low-cost and provide a nutrition source for natural pond productivity as much as for the crayfish. Most Australian farmers use diets consisting of 25 to 30% protein. Effective farm management involves maintaining water quality conditions within ranges optimal for crayfish growth and survival as pond biomass increases. As with barramundi, management involves increasing aeration and water exchanges, while strictly managing effluent discharges. Red claw are harvested within 6 months of stocking to avoid reproduction in the production pond. At this stage the crayfish will range between 30 and 80 g. Stock are graded by size and sex into groups for market, breeding or further grow-out (Jones, 2007). Estimated water use An average crop of prawns farmed in intensive pond systems (8 t/ha over 150 days) is estimated to require 127 ML of marine water, which equates to 15.9 ML of marine water for each tonne of harvested product (Irvin et al., 2018). For pond culture of barramundi (30 t/ha over 2 years), 562 ML of marine water, or fresh water, is required per crop, equating to 18.7 ML of water for each tonne of harvested fish. For extensive red claw culture (3 t/ha over 300 days) 240 ML of fresh water is required per pond crop, equating to 16 ML of water for each harvested tonne of crayfish (Irvin et al., 2018). A.4 Aquaculture viability This section provides a brief, generic analysis of what would be required for new aquaculture developments in the Roper catchment to be financially viable. The analyses follow the same approach as those conducted in Irvin et al. (2018) but have been updated. First, indicative costs are provided for a range of four possible aquaculture enterprises that differ in species farmed, scale and intensity of production. The cost structure of the enterprises is based on established tools available from the Queensland Government for assessing the performance of existing or proposed aquaculture businesses (https://publications.qld.gov.au/dataset/agbiz-tools-fisheries- aquaculture). Based on the ranges of indicative capital and operating costs for the four types of enterprises, gross revenue targets are calculated that a business would need to attain to be commercially viable. Enterprise-level costs for aquaculture development Costs of establishing and running a new aquaculture business are divided here into the initial capital costs of development and ongoing operating costs. The four enterprise types analysed were chosen to portray some of the variation in cost structures between potential development options, not as a like-for-like comparison between different types of aquaculture (Apx Table A-1). Capital costs include all land development costs, construction, and plant and equipment, accounted for in the year production commences. The types of capital development costs are largely similar across the aquaculture options with costs of constructing ponds and buildings dominating the total initial capital investment. Indicative costs were derived from Guy et al. (2014), and consultation with experts familiar with the different types of aquaculture, including updating to 2021 dollar values (Apx Table A-1). Operating costs cover both overheads, which do not change with output, and variable costs that increase as the yield of produce increases. Fixed overhead costs in aquaculture are a relatively small component of the total costs of production. Overheads consist of costs relating to licensing, approvals and other administration (Apx Table A-1). The remaining operating costs are variable (Apx Table A-1). Feed, labour and electricity typically dominate the variable costs. Aquaculture requires large volumes of feed inputs, and the efficiency with which this feed is converted to marketed produce is a key metric of business performance. Labour costs consist of salaries of permanent staff and casual staff who are employed to cover intensive harvesting and processing activities. Aerators require large amounts of energy, increasing as the biomass of produce in the ponds increase, which accounts for the large costs of electricity. Transport, although a smaller proportional cost, is important because this puts remote locations at a relative disadvantage to aquaculture businesses that are closer to feed suppliers and markets. In addition, transport costs may be higher at times if roads are cut (requiring much more expensive air freight or alternate, longer road routes) or if the closest markets become oversupplied. Packing is the smallest component of variable costs in the breakdown categories used here. Revenue for aquaculture produce typically ranges between $10 and $20 per kg (on a harvested mass basis), but prices vary depending on the quality and size classes of harvested animals and how they are processed (e.g. live, fresh, frozen or filleted) and farms are likely to deliver a mix of products targeted to the specifications of the markets they supply. Note that the mass of sold product may be substantially lower than the harvested product (e.g. fish fillets are about half the mass of harvested fish), so prices of sold product may not be directly comparable to the costs of production below (which are on a harvest mass basis) (Apx Table A-1). Apx Table A-1 Indicative capital and operating costs for a range of generic aquaculture development options Costs are provided both per ha of grow-out pond and per kg of harvested produce, although capital costs scale mostly with the area developed and operating costs scale mainly with yield at harvest. Capital costs have been converted to an equivalent annualised cost assuming a 10% discount rate and that a quarter of the developed infrastructure was assets with a 15-year lifespan and the remainder had a 40-year lifespan. Indicative breakdowns of cost components are provided on a proportional basis. PARAMETER UNITS PRAWN (EXTENSIVE) PRAWN (INTENSIVE) BARRAMUNDI RED CLAW (SMALL SCALE) Scale of development Grow-out pond area ha 20 100 30 4 Total farm area ha 25 150 100 10 Yield at harvest t/y 30 800 600 32 Yield at harvest per pond area t/ha/y 1.5 8.0 20.0 3.0 Capital costs of development (scale with area of grow-out ponds developed) Land and buildings % 56% 26% 23% 30% Vehicles % 5% 2% 2% 11% Pond-related assets % 27% 67% 70% 41% Other infrastructure and equipment % 11% 6% 5% 17% Total capital cost (year 0) $/ha 65,000 125,000 129,000 143,000 Equivalent annualised cost $/kg 4.75 1.71 0.71 5.22 $/ha/y 7,122 13,695 14,134 15,668 Operating costs (vary with yield at harvest, except overheads) Nursery/juvenile costs % 12% 9% 7% 1% Feed costs % 0% 26% 30% 8% Labour costs % 47% 13% 12% 57% Electricity costs % 16% 24% 30% 9% Packing costs % 2% 4% 3% 2% Transport costs % 6% 16% 16% 11% Overhead costs (fixed) % 17% 8% 1% 12% Total annual operating costs $/kg 16.88 10.90 10.89 15.56 $/ha/y 25,321 87,227 217,854 46,683 Total costs of production Total annual cost $/kg 21.63 12.62 11.60 20.78 $/ha/y 32,400 100,900 232,000 62,400 Commercial viability of new aquaculture developments Capital and operating costs differ between different types of aquaculture enterprises (Apx Table A-2), but these costs may differ even more between location (depending on case- specific factors such as remoteness, soil properties, distance to water source and type of power supply). Furthermore, there can be considerable uncertainty in some costs, and prices paid for produce can fluctuate substantially over time. Apx Table A-2 Gross revenue targets required to achieve target internal rates of return (IRRs) for aquaculture developments with different combinations of capital costs and operating costs All values are expressed per ha of grow-out ponds in the development. Gross revenue is the yield per ha of pond multiplied by the price received for produce (averaged across products and on a harvest mass basis). Capital costs were converted to an equivalent annualised cost assuming a quarter of the developed infrastructure was assets with a 15-year lifespan and the remainder had a 40-year lifespan. Targets would be higher after taking into account risks such as initial learning and market fluctuations. OPERATING ($/ha/y) GROSS REVENUE REQUIRED TO ACHIEVE TARGET IRR ($/ha/y) Capital costs of development ($/ha) 60,000 70,000 80,000 90,000 100,000 110,000 125,000 150,000 175,000 7% target IRR 20,000 25,022 25,859 26,696 27,533 28,371 29,208 30,463 32,556 34,648 50,000 55,022 55,859 56,696 57,533 58,371 59,208 60,463 62,556 64,648 100,000 105,022 105,859 106,696 107,533 108,371 109,208 110,463 112,556 114,648 150,000 155,022 155,859 156,696 157,533 158,371 159,208 160,463 162,556 164,648 100,000 105,022 105,859 106,696 107,533 108,371 109,208 110,463 112,556 114,648 200,000 205,022 205,859 206,696 207,533 208,371 209,208 210,463 212,556 214,648 250,000 255,022 255,859 256,696 257,533 258,371 259,208 260,463 262,556 264,648 10% target IRR 20,000 26,574 27,669 28,765 29,861 30,956 32,052 33,695 36,434 39,174 50,000 56,574 57,669 58,765 59,861 60,956 62,052 63,695 66,434 69,174 100,000 106,574 107,669 108,765 109,861 110,956 112,052 113,695 116,434 119,174 150,000 156,574 157,669 158,765 159,861 160,956 162,052 163,695 166,434 169,174 100,000 106,574 107,669 108,765 109,861 110,956 112,052 113,695 116,434 119,174 200,000 206,574 207,669 208,765 209,861 210,956 212,052 213,695 216,434 219,174 250,000 256,574 257,669 258,765 259,861 260,956 262,052 263,695 266,434 269,174 14% target IRR 20,000 28,776 30,238 31,701 33,163 34,626 36,089 38,283 41,939 45,596 50,000 58,776 60,238 61,701 63,163 64,626 66,089 68,283 71,939 75,596 100,000 108,776 110,238 111,701 113,163 114,626 116,089 118,283 121,939 125,596 150,000 158,776 160,238 161,701 163,163 164,626 166,089 168,283 171,939 175,596 100,000 108,776 110,238 111,701 113,163 114,626 116,089 118,283 121,939 125,596 200,000 208,776 210,238 211,701 213,163 214,626 216,089 218,283 221,939 225,596 250,000 258,776 260,238 261,701 263,163 264,626 266,089 268,283 271,939 275,596 Given the variation among possible aquaculture developments in the Roper catchment, a generic approach was taken to determine what would be required for new aquaculture enterprises to become commercially viable. The approach used here was to calculate the gross revenue that an enterprise would have to generate each year to achieve a target internal rate of return (IRR) for given operating costs and development costs (both expressed per hectare of grow-out ponds). Capital costs were converted to annualised equivalents on the assumption that developed assets equated to a mix of 25% 15-year assets and 75% assets with a 40-year life span (using a discount rate matching the target IRR). The target gross revenue is the sum of the annual operating costs and the equivalent annualised cost of the infrastructure development (Apx Table A-2). In order for an enterprise to be commercially viable, the volume of produce grown each year multiplied by the sales price of that produce would need to match or exceed the target values provided above. For example, a proposed development with capital costs of $90,000/ha and operating costs of $200,000 per ha per year would need to generate gross revenue of $213,695 per ha per year to achieve a target IRR of 10% (Apx Table A-2). If the enterprise received $12/kg for produce (averaged across product types, on a harvest mass basis), then it would need to sustain average long-term yields of 18 t/ha (= $213,695/ha/y ÷ $12/kg × 1t/1000kg) from the first harvest. However, if prices were $20/kg, average long-term yields would require 11 t/ha (= 213,695/ha/y ÷ $20/kg × 1t/1000kg) for the same $125,000 capital costs per hectare, or only 8 t/ha harvests if the capital costs could be reduced to $100,000/ha. Target revenue would be higher after taking into account risks, such as learning and adapting to the particular challenges of a new location and periodic setbacks that could arise from disease, climate variability, changes in market conditions or new legislation. Key messages From this analysis, a number of key points are apparent about achieving commercial viability in new aquaculture enterprises: • Operating costs are very high and the amount spent each year on inputs can exceed the upfront (year 0) capital cost of development (and the value of the farm assets). This means that the cost of development is a much smaller consideration for achieving profitability than ongoing operations and costs of inputs. • High operating costs also mean that substantial capital reserves are required, beyond the capital costs of development, as there will be large cash outflows for inputs in the start-up years before revenue from harvested product starts to be generated. This is particularly the case for larger size classes of product that require multi-year grow-out periods before harvest. Managing cash flows would therefore be an important consideration at establishment and as yields are subsequently scaled up. • Variable costs dominate the total costs of aquaculture production so most costs will increase as yield increases. This means that increases in production, by itself, would contribute little to achieving profitability in a new enterprise. What is much more important is increasing production efficiency, such as feed conversion rate or labour-efficient operations, so that inputs per unit of produce are reduced (and profit margins per kg are increased). • Small changes in quantities and prices of inputs and produce would have a relatively large impact on net profit margins. These values could differ substantially between different locations (e.g. remoteness, available markets, soils and climate), and can depend on the experience of managers. Even small differences from the indicative values provided above could render an enterprise unprofitable. • Enterprise viability would therefore be very dependent on the specifics of each particular case and how the learning, scaling up and cash flow were managed during the initial establishment years of the enterprise. It would be essential for any new aquaculture development in the Roper catchment to refine the production system and achieve the required levels of operational efficiency (input costs per kg of produce) using just a few ponds before scaling any enterprise. As Australia’s national science agency and innovation catalyst, CSIRO is solving the greatest challenges through innovative science and technology. CSIRO. Unlocking a better future for everyone. Contact us 1300 363 400 +61 3 9545 2176 csiroenquiries@csiro.au www.csiro.au For further information Environment Dr Chris Chilcott +61 8 8944 8422 chris.chilcott@csiro.au Environment Dr Cuan Petheram +61 467 816 558 cuan.petheram@csiro.au Agriculture and Food Dr Ian Watson +61 7 4753 8606 Ian.watson@csiro.au