Australia’s NationalScience Agency
Floodplain inundation mappingand modelling for the Southern
Gulf catchments
A technical report from the CSIROSouthern GulfWater ResourceAssessmentfor theNational Water Grid
Fazlul Karim,ShaunKim, CatherineTicehurst,Matt Gibbs, Justin Hughes,Steve Marvanek,
AngYang, Bill Wang,Cuan Petheram
’
ISBN 978-1-4863-2057-8 (print)
ISBN 978-1-4863-2058-5 (online)
Citation
Karim F, Kim S, Ticehurst C, Gibbs M, Hughes J, Marvanek S, Yang A, Wang B and Petheram C (2024) Floodplain inundation mapping and modelling
for the Southern Gulf catchments. A technical report from the CSIRO Southern Gulf Water Resource Assessment for the National Water Grid. CSIRO,
Australia.
Copyright
© Commonwealth Scientific and Industrial Research Organisation 2024. 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 Southern Gulf 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 and Queensland governments.
The Assessment was guided by two committees:
i. The Governance Committee: CRC for Northern Australia/James Cook University; CSIRO; National Water Grid (Department of Climate
Change, Energy, the Environment and Water); Northern Land Council; NT Department of Environment, Parks and Water Security; NT
Department of Industry, Tourism and Trade; Office of Northern Australia; Queensland Department of Agriculture and Fisheries;
Queensland Department of Regional Development, Manufacturing and Water
ii. The Southern Gulf catchments Steering Committee: Amateur Fishermen’s Association of the NT; Austral Fisheries; Burketown Shire;
Carpentaria Land Council Aboriginal Corporation; Health and Wellbeing Queensland; National Water Grid (Department of Climate
Change, Energy, the Environment and Water); Northern Prawn Fisheries; Queensland Department of Agriculture and Fisheries; NT
Department of Environment, Parks and Water Security; NT Department of Industry, Tourism and Trade; Office of Northern Australia;
Queensland Department of Regional Development, Manufacturing and Water; Southern Gulf NRM
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 their release.
This report was reviewed by Dr Zaved Khan and Mr Mahdi Montazeri of CSIRO.
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
Nicholson River. Source: CSIRO
Director’s foreword
Sustainable development and regional economic prosperity are priorities for the Australian,
Queensland and Northern Territory (NT) governments. However, more comprehensive
information on land and water resources across northern Australia is required to complement
local information held by Indigenous Peoples and other landholders.
Knowledge of the scale, nature, location and distribution of likely environmental, social, cultural
and economic opportunities and the risks of any proposed developments is critical to sustainable
development. Especially where resource use is contested, this knowledge informs the consultation
and planning that underpin the resource security required to unlock investment, while at the same
time protecting the environment and cultural values.
In 2021, the Australian Government commissioned CSIRO to complete the Southern Gulf Water
Resource Assessment. In response, CSIRO accessed expertise and collaborations from across
Australia to generate data and provide insight to support consideration of the use of land and
water resources in the Southern Gulf catchments. The Assessment focuses mainly on the potential
for agricultural development, and the opportunities and constraints that development could
experience. It also considers climate change impacts and a range of future development pathways
without being prescriptive of what they might be. The detailed information provided on land and
water resources, their potential uses and the consequences of those uses are carefully designed to
be relevant to a wide range of regional-scale planning considerations by Indigenous Peoples,
landholders, citizens, investors, local government, and the Australian, Queensland and NT
governments. By fostering shared understanding of the opportunities and the risks among this
wide array of stakeholders and decision makers, better informed conversations about future
options will be possible.
Importantly, the Assessment does not recommend one development over another, nor assume
any particular development pathway, nor even assume that water resource development will
occur. 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 Southern Gulf Water Resource Assessment Team
Project Director
Chris Chilcott
Project Leaders
Cuan Petheram, Ian Watson
Project Support
Caroline Bruce, Seonaid Philip
Communications
Emily Brown, Chanel Koeleman, Jo Ashley, Nathan Dyer
Activities
Agriculture and socio-
economics
Tony Webster, Caroline Bruce, Kaylene Camuti1, Matt Curnock,
Jenny Hayward, Simon Irvin, Shokhrukh Jalilov, Diane Jarvis1, Adam Liedloff,
Stephen McFallan, Yvette Oliver, Di Prestwidge2, Tiemen Rhebergen,
Robert Speed3, Chris Stokes, Thomas Vanderbyl3, John Virtue4
Climate
David McJannet, Lynn Seo
Ecology
Danial Stratford, Rik Buckworth, Pascal Castellazzi, Bayley Costin,
Roy Aijun Deng, Ruan Gannon, Steve Gao, Sophie Gilbey, Rob Kenyon,
Shelly Lachish, Simon Linke, Heather McGinness, Linda Merrin,
Katie Motson5, Rocio Ponce Reyes, Jodie Pritchard, Nathan Waltham5
Groundwater hydrology
Andrew R. Taylor, Karen Barry, Russell Crosbie, Margaux Dupuy,
Geoff Hodgson, Anthony Knapton6, Stacey Priestley, Matthias Raiber
Indigenous water
values, rights, interests
and development goals
Pethie Lyons, Marcus Barber, Peta Braedon, Petina Pert
Land suitability
Ian Watson, Jenet Austin, Bart Edmeades7, Linda Gregory, Ben Harms10,
Jason Hill7, Jeremy Manders10, Gordon McLachlan, Seonaid Philip,
Ross Searle, Uta Stockmann, Evan Thomas10, Mark Thomas, Francis Wait7,
Peter Zund
Surface water hydrology
Justin Hughes, Matt Gibbs, Fazlul Karim, Julien Lerat, Steve Marvanek,
Cherry Mateo, Catherine Ticehurst, Biao Wang
Surface water storage
Cuan Petheram, Giulio Altamura8, Fred Baynes9, Jamie Campbell11,
Lachlan Cherry11, Kev Devlin4, Nick Hombsch8, Peter Hyde8, Lee Rogers,
Ang Yang
Note: Assessment team as at September, 2024. All contributors are affiliated with CSIRO unless indicated otherwise. Activity Leaders are
underlined.
1James Cook University; 2DBP Consulting; 3Badu Advisory Pty Ltd; 4Independent contractor; 5 Centre for Tropical Water and Aquatic Ecosystem
Research. James Cook University; 6CloudGMS; 7NT Department of Environment, Parks and Water Security; 8Rider Levett Bucknall; 9Baynes Geologic;
10QG Department of Environment, Science and Innovation; 11Entura
Shortened forms
SHORT FORM
FULL FORM
AEP
annual exceedance probability
AGDC
Australian Geoscience Data Cube
AHD
Australian Height Datum
AMTD
Adopted Middle Thread Distance
AR6
Sixth Assessment Report
AWRA-L
Australian Water Resource Assessment – Landscape model
AWRA-R
Australian Water Resource Assessment – River model
AWRC
Australian Water Resources Council
BRDF
Bidirectional Reflectance Distribution Function
CMIP
Coupled Model Intercomparison Project
CAWCR
Collaboration for Australian Weather and Climate Research
DCFR
diversion commencement flow requirement
DEA
Digital Earth Australia
DEM
digital elevation model
DOI
digital object identifier
ERP
equivalent Riemann problem
ESA
European Space Agency
ETM
Enhanced Thematic Mapper
ETS
Equitable Threat Score
FAR
False Alarm Ratio
FB
Frequency Bias
GCM
global climate model
GCM-PS
global climate model – pattern scaling
GIS
Geographic Information System
GPU
graphics processing unit
HAND
Height Above Nearest Drainage
HDF
hierarchical data format
IDL
Interactive Data Language
IPCC
Intergovernmental Panel on Climate Change
LiDAR
Light Detection and Ranging
LP DAAC
Land Processes Distributed Active Archive Center
MGA
Map Grid of Australia
MODIS
Moderateâ€Resolution Imaging Spectroradiometer
NBAR
Nadir BRDF-Adjusted Reflectance
SHORT FORM
FULL FORM
NCI
National Computing Infrastructure
NDWI
Normalized Difference Water Index
OLI
Operational Land Imager
OWL
Open Water Likelihood
PE
potential evaporation
POD
Probability Of Detection
PS
pattern scaling
RCP
Representative Concentration Pathway
SAR
Synthetic Aperture Radar
SILO
scientific information for land owners
SRTM
Shuttle Radar Topography Mission
SSP
Shared Socioeconomic Pathway
TM
Thematic Mapper
URBS
Unified River Basin Simulator
USGS
United States Geological Survey
Units
UNIT
DESCRIPTION
cm
centimetre
GL
gigalitre
km
kilometre
m
metre
ML/d
megalitres per day
GL/d
gigalitres per day
mm
millimetre
s
second
Preface
Sustainable development and regional economic prosperity are priorities for the Australian, NT
and Queensland governments. In the Queensland Water Strategy, for example, the Queensland
Government (2023) looks to enable regional economic prosperity through a vision that states
‘Sustainable and secure water resources are central to Queensland’s economic transformation and
the legacy we pass on to future generations.’ Acknowledging the need for continued research, the
NT Government (2023) announced a Territory Water Plan priority action to accelerate the existing
water science program ‘to support best practice water resource management and sustainable
development.’
Governments are actively seeking to diversify regional economies, considering a range of factors,
including Australia’s energy transformation. The Queensland Government’s economic
diversification strategy for North West Queensland (Department of State Development,
Manufacturing, Infrastructure and Planning, 2019) includes mining and mineral processing; beef
cattle production, cropping and commercial fishing; tourism with an outback focus; and small
business, supply chains and emerging industry sectors. In its 2024–25 Budget, the Australian
Government announced large investment in renewable hydrogen, low-carbon liquid fuels, critical
minerals processing and clean energy processing (Budget Strategy and Outlook, 2024). This
includes investing in regions that have ‘traditionally powered Australia’ – as the North West
Minerals Province, situated mostly within the Southern Gulf catchments, has done.
For very remote areas like the Southern Gulf catchments (Preface Figure 1-1), the land, water and
other environmental resources or assets will be key in determining how sustainable regional
development might occur. Primary questions in any consideration of sustainable regional
development relate to the nature and the scale of opportunities, and their risks.
How people perceive those risks is critical, especially in the context of areas such as the Southern
Gulf catchments, where approximately 27% of the population is Indigenous (compared to 3.2% for
Australia as a whole) and where many Indigenous Peoples still live on the same lands they have
inhabited for tens of thousands of years. About 12% of the Southern Gulf catchments are owned
by Indigenous Peoples as inalienable freehold.
Access to reliable information about resources enables informed discussion and good decision
making. Such information includes the amount and type of a resource or asset, where it is found
(including in relation to complementary resources), what commercial uses it might have, how the
resource changes within a year and across years, the underlying socio-economic context and the
possible impacts of development.
Most of northern Australia’s land and water resources have not been mapped in sufficient detail
to provide the level of information required for reliable resource allocation, to mitigate
investment or environmental risks, or to build policy settings that can support good judgments.
The Southern Gulf Water Resource Assessment aims to partly address this gap by providing data
to better 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.
Preface Figure 1-1 Map of Australia showing Assessment area (Southern Gulf catchments) and other recent CSIRO
Assessments
FGARA = Flinders and Gilbert Agricultural Resource Assessment; NAWRA = Northern Australia Water Resource
Assessment.
The Assessment differs somewhat from many resource assessments in that it considers a wide
range of resources or assets, rather than being a single mapping exercises of, say, soils. It provides
a lot of contextual information about the socio-economic profile of the catchments, and the
economic possibilities and environmental impacts of development. Further, it considers many of
the different resource and asset types in an integrated way, rather than separately.
The Assessment has agricultural developments as its primary focus, but 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, urban and aquaculture development, in relevant locations. The outcome of
no change in land use or water resource development is also valid.
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. Policy and regulations can change, so this flexibility
enables the results to be applied to the widest range of uses for the longest possible time frame.
It was not the intention of – 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
For more information on this figure please contact CSIRO on enquiries@csiro.au
not directly examined in the Assessment are discussed with reference to and in the context of the
existing literature.
CSIRO has strong organisational commitments to Indigenous reconciliation and to conducting
ethical research with the free, prior and informed consent of human participants. The Assessment
allocated significant time to consulting with Indigenous representative organisations and
Traditional Owner groups from the catchments to aid their understanding and potential
engagement with its requirements. The Assessment did not conduct significant fieldwork without
the consent of Traditional Owners. CSIRO met the requirement to create new scientific knowledge
about the catchments (e.g. on land suitability) by synthesising new material from existing
information, complemented by remotely sensed data and numerical modelling.
Functionally, the Assessment adopted an activities-based approach (reflected in the content and
structure of the outputs and products), comprising activity groups, each contributing its part to
create a cohesive picture of regional development opportunities, costs and benefits, but also risks.
Preface Figure 1-2 illustrates the high-level links between the activities and the general flow of
information in the Assessment.
Preface Figure 1-2 Schematic of the high-level linkages between the eight activity groups 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 for each Assessment most reliably informs
discussion and decisions concerning regional development when read as a whole.
For more information on this figure please contact CSIRO on enquiries@csiro.au
The Assessment has produced a series of cascading reports and information products:
• Technical reports present scientific work with sufficient detail for technical and scientific experts
to reproduce the work. Each of the activities (Preface Figure 1-2) has one or more corresponding
technical reports.
• A catchment report, which synthesises key material from the technical reports, providing well-
informed (but not necessarily scientifically trained) users with the information required to
inform decisions about the opportunities, costs and benefits, but also risks, associated with
irrigated agriculture and other development options.
• A summary report provides a shorter summary and narrative for a general public audience in
plain English.
• A summary fact sheet provides key findings for a general public audience in the shortest possible
format.
The Assessment has also developed online information products to enable users to better access
information that is not readily available in print format. All of these reports, information tools and
data products are available online at https://www.csiro.au/southerngulf. The webpages give users
access to a communications suite including fact sheets, multimedia content, FAQs, reports and
links to related sites, particularly about other research in northern Australia.
Executive summary
This report focuses on flooding characteristics of the Southern Gulf catchments in Queensland.
The impact of flooding on agricultural production can be significant, potentially leading to the loss
of livestock, fodder and topsoil, and damage to crops and infrastructure. However, flooding is
generally favourable to floodplain wetlands and coastal ecosystems. For example, flood pulses
create opportunities for offstream wetlands to connect with the main river channels, allowing the
exchange of water, sediments, organic matter and biota.
This report provides an overview of inundation duration and depth across the floodplains of the
major rivers in the Southern Gulf catchments. It also details the hydrodynamic modelling tools
utilised to assess flood inundation. This includes information on data acquisition, model
configuration, and the evaluation process, and a comparison of the model results with satellite-
based flood inundation maps and water levels at gauging sites.
Hydrodynamic models offer several advantages over satellite-based approaches and conceptual
node-link river system models when it comes to evaluating flood inundation. Hydrodynamic
models enable the assessment of not only the extent of inundation but also water depth and
velocity, with the ability to analyse these factors at very fine time intervals, often in the order of
seconds. Furthermore, satellite-based approaches primarily focus on analysing historical flood
events, whereas hydrodynamic models can be used to assess how flood characteristics might
change under future climate and development scenarios.
The outputs derived from the hydrodynamic modelling play a crucial role, including:
• identifying areas prone to flooding under historical climates and the current level of
development, commonly known as the baseline scenario
• estimating changes in inundation area under projected future climate and hypothetical
development scenarios
• estimating changes in inundation depth and duration across the floodplain under future climate
and development scenarios.
Hydrodynamic model configuration and calibration
In the Assessment, a two-dimensional flexible-mesh hydrodynamic model, MIKE 21 Flow Model
FM, was used to simulate floodplain hydraulics (e.g. depth, velocity) and inundation dynamics
across the floodplains of three large river systems: Nicholson, Gregory and Leichhardt.
The boundary conditions were derived from the daily discharge from a calibrated river system
model called the Australian Water Resources Assessment – River model (AWRA-R), the hourly tide
gauge information, and Sacramento rainfall-runoff model simulations. A two-dimensional
hydrodynamic model (MIKE 21 FM) was configured for the middle and downstream reaches of the
Nicholson, Gregory and Leichhardt rivers and their tributaries. The model domain includes areas
downstream of Doomadgee on the Nicholson River, Gregory on the Gregory River and Lorraine on
the Leichhardt River, and encompasses an area of 17,130 km2. Flood inundation maps for
individual flood events from 2000 to 2023 were created using Terra (EOS AM-1), Aqua (EOS PM-1),
Landsat, Sentinel-1 and Sentinel-2 satellite imagery. These maps were used to calibrate the
hydrodynamic model.
High-resolution Light Detection and Ranging (LiDAR) data (5 m) was acquired over the floodplains
of the Albert, Gregory, Leichhardt and Nicholson rivers and only a small proportion of the
Alexandra River, as part of the Assessment. For the remainder of the model domain, where floods
are less frequent, a 30 m Forest And Buildings removed Copernicus digital elevation model (DEM)
was used for land topography. The highest resolution publicly available topographic data covering
the entire Southern Gulf catchments includes 1-second (i.e. ~30 m) Shuttle Radar Topography
Mission (SRTM) digital elevation model (DEM) and 1-second FABDEM. These two global DEMs
were compared, with the FABDEM being chosen due to its superior vertical accuracy in the
Assessment area. The final combined DEM was created by resampling the FABDEM to 5 m, to
match the original LiDAR resolution. The area covered by LiDAR is 7490 km2, which is
approximately 43.7% of the hydrodynamic model domain.
The hydrodynamic model was calibrated for the 2005 (annual exceedance probability (AEP) of 1 in
2), 2016 (AEP of 1 in 3), 2018 (AEP of 1 in 5), 2019 (AEP of 1 in 10) and 2023 (AEP of 1 in 38) flood
events. The model was calibrated primarily by adjusting the roughness coefficient and the
infiltration rate. While a good match was attained for the flood peaks, there were differences in
the rising and falling limbs of the flood hydrograph. The model demonstrated reasonable
simulation of spatial inundation patterns when compared with the Landsat, MODIS and Sentinel
water maps. Overall, the model performed better for large floods, followed by medium-sized
events, and then small events. There are some limitations to the model, and a lack of good-quality
satellite imagery restricts rigorous calibration of the model results. Moreover, there are
uncertainties in river model simulations that were used to specify the inflow boundaries of the
hydrodynamic model.
The calibrated hydrodynamic models were utilised to investigate flood characteristics under future
climate and development scenarios. Due to the extensive computation time, a limited number of
simulation runs were conducted to explore the impact of future climate and hypothetical
developments.
Flood characteristics
Intense seasonal rains from monsoonal bursts and tropical cyclones from November to March
create flooding in parts of the Southern Gulf catchment and inundate large areas of floodplains,
mostly in the downstream reaches between the Nicholson, Gregory and Leichhardt rivers.
Floodplains along the Alexandra and Albert rivers are also heavily flooded. Rivers in Southern Gulf
regions are unregulated, and its overbank flow is generally governed by the topography of the
floodplain.
Flooding is widespread in parts of Lawn Hill Creek and the Gregory, Leichhardt and Albert rivers
near Burketown. In the last 40 years (1984 to 2023), there have been 41 floods ranging from small
to large in the catchments. This is based on an overbank threshold of 709 m3/second, which was
estimated by obtaining the daily streamflow at Floraville on the Leichhardt River (913007) that
corresponded to floodplain inundation in available satellite imagery. While floods can occur in any
month between October and May, the majority of the historical floods have occurred between
December and March.
Additional observations of flooding under the historical climate are as follows:
• Flood peaks typically take about 2 days to travel from Gregory to Burketown, with a mean speed
of 3.2 km/hour.
• For flood events of AEP of 1 in 2, 1 in 5 and 1 in 10, the peak discharge at Floraville on the
Leichhardt River is 1220, 3350 and 7140 m3/second, respectively (i.e. 105.4, 289.4 and 616.9
GL/day, respectively).
• Between 1984 and 2023 (40 years), events with a discharge greater than or equal to AEP of 1 in
1 occurred during all months from October to May, with approximately 90% of historical floods
occurring between December and March and the maximum in January (33.3%).
• Of the ten largest flood peak discharges at Floraville, four events occurred during January, four
in March, one in February and one in December.
Scenario analysis
A limited number of simulations were conducted to investigate the effects of future climate and
development on inundation duration and depth. Three dam sites with the total capacity of
2560 GL at full supply level and five water harvesting sites with a total annual maximum diversion
of 150 GL were implemented in the model for impact assessment.
The model results revealed that the impacts of future projected wet (Scenario Cwet) and dry
(Scenario Cdry) climates on floodplain inundation were more pronounced than the modelled
impacts of water resource development. The decrease in floodplain inundation under Cdry was
larger than the increase under Cwet, which is consistent with the changes in modelled streamflow
under Cdry and Cwet scenarios.
The inclusion of three hypothetical dams resulted in a 33.8% decrease in the inundated area
downstream for an event with an AEP of 1 in 3, and a 5.1% decrease for an event with an AEP of 1
in 38. In general, impacts are higher for smaller flood events, because dams store a substantial
proportion of floodwater. The smaller relative impact found for the 1 in 38 AEP event during 2023
was due to a large flood volume (~21,000 GL) compared with the combined dam capacity
(2560 GL). The impacts are also influenced by antecedent conditions at the beginning of the flood
event and the flow hydrograph (e.g. impacts are less for multipeak floods).
Water harvesting (150 GL) resulted in a 5.4% decrease in inundation area and very minor changes
to inundation duration for an event with an AEP of 1 in 3 (2016 flood). For the event with an AEP
of 1 in 38, the decrease in inundation area was only 0.8%, and the changes in inundation duration
were negligible. As expected, impacts were relatively large for the smaller flood event, given the
same amount of water was extracted for both flood events.
Hydrodynamic models are computationally demanding and therefore only a limited number of
events can be analysed. This in turn means that the characteristics and timing (particularly
antecedent effects) of chosen events can influence the apparent response to various scenarios. To
counter this, a flood area emulator was derived using river model daily flows that could predict
flooded area across the entire 133-year time series. Using the emulator, the annual maximum
flooded area was calculated for various scenarios. The mean annual maximum flooded area across
133 years of simulation was 1132 km2 under Scenario A. This was reduced to 785 km2 under the
three dams scenario, while the water harvesting scenario with an irrigation target of 150 GL was
associated with only a very small reduction in the annual maximum flooded area, to 1086 km2.
Contents
Director’s foreword .......................................................................................................................... i
The Southern Gulf Water Resource Assessment Team .................................................................. ii
Shortened forms .............................................................................................................................iii
Units ............................................................................................................................... v
Preface ............................................................................................................................... vi
Executive summary .......................................................................................................................... x
1 Introduction ........................................................................................................................ 1
1.1 Objectives .............................................................................................................. 2
1.2 Previous flood studies in the Southern Gulf catchments ...................................... 2
1.3 Overview of flood modelling frameworks used in the Assessment ...................... 3
1.4 Report overview and structure ............................................................................. 4
1.5 Key terminology and concepts .............................................................................. 5
2 Floodplain inundation mapping .......................................................................................... 8
2.1 Satellite imagery acquisition and pre-processing ................................................. 8
2.2 Inundation mapping using MODIS ........................................................................ 9
2.3 Inundation mapping using Landsat ..................................................................... 11
2.4 Inundation mapping using Sentinel-2 ................................................................. 11
2.5 Inundation mapping using Sentinel-1 ................................................................. 12
2.6 Combined summary maps ................................................................................... 12
2.7 Summary .............................................................................................................. 12
3 Floodplain inundation modelling ...................................................................................... 15
3.1 Hydrodynamic models ......................................................................................... 15
3.2 Data requirement for model configuration......................................................... 16
4 Southern Gulf catchments hydrodynamic model calibration .......................................... 17
4.1 Physical and hydro-meteorological properties ................................................... 17
4.2 Model configuration ............................................................................................ 25
4.3 Model input ......................................................................................................... 26
4.4 Flood frequency and selected events for model calibration .............................. 29
4.5 Hydrodynamic model simulation and outputs .................................................... 31
4.6 Hydrodynamic model calibration ........................................................................ 31
4.7 Results and discussion ......................................................................................... 33
4.8 Summary .............................................................................................................. 38
5 Flood modelling under future climate and development scenarios ................................ 40
5.1 Introduction ......................................................................................................... 40
5.2 Future climate scenarios ..................................................................................... 41
5.3 Potential development scenarios ........................................................................ 44
5.4 Floodplain inundation scenario analysis ............................................................. 47
5.5 Floodplain inundation emulator .......................................................................... 62
6 Summary ........................................................................................................................... 65
References ............................................................................................................................. 67
Figures
Preface Figure 1-1 Map of Australia showing Assessment area (Southern Gulf catchments) and
other recent CSIRO Assessments ................................................................................................... vii
Preface Figure 1-2 Schematic of the high-level linkages between the eight activity groups and
the general flow of information in the Assessment ..................................................................... viii
Figure 1-1 Flowchart illustrating the method used to calibrate a hydrodynamic model
(MIKE 21 FM) and scenario modelling for future climate and dam impact assessment ............... 4
Figure 2-1 MODIS satellite–based flood inundation map of the Southern Gulf catchments....... 10
Figure 2-2 Combined Landsat, Sentinel-2 and Sentinel-1 satellite–based flood inundation map
of the Southern Gulf catchments .................................................................................................. 13
Figure 4-1 Southern Gulf catchments map showing physiography and river network. From Zund
et al. (2024) ................................................................................................................................... 18
Figure 4-2 Historical monthly rainfall (showing the range in values between the 20% and 80%
monthly exceedance rainfall) and annual rainfall in the Southern Gulf catchments at
Doomadgee, Gregory and Burketown (from McJannet et al., 2023) ........................................... 20
Figure 4-3 Location of streamflow gauges in the Southern Gulf catchments .............................. 22
Figure 4-4 Monthly flow distribution at Floraville on the Leichhardt River, based on the
observed flow data for 1984 to 2023 ........................................................................................... 23
Figure 4-5 Annual maximum daily discharge at Floraville on the Leichhardt River, based on the
observed flow data for 1984 to 2023 ........................................................................................... 24
Figure 4-6 Monthly flood frequency in the Southern Gulf catchments (floods defined as an AEP
of ≥1 in 1, based on flow data for 1984 to 2023) at Floraville on the Leichhardt River ............... 24
Figure 4-7 Hydrodynamic model configuration of the Southern Gulf catchments, showing the
river network, model boundaries, and local runoff points in the model domain ........................ 25
Figure 4-8 LiDAR data coverage in the hydrodynamic model domain of the Southern Gulf
catchments .................................................................................................................................... 27
Figure 4-9 Peak flood discharge and annual exceedance probability at gauge: (a) 912107
(Nicholson River at Connolly’s Hole, (b) 912105 (Gregory River at Riversleigh) and (c) 913007
(Leichhardt River on Floraville) ..................................................................................................... 30
Figure 4-10 Classification at the grid cell level using a contingency table ................................... 32
Figure 4-11 Comparison of the model-simulated stage height and the observed stage height at
Floraville (913007B) on the Leichhardt River in the Southern Gulf catchments .......................... 34
Figure 4-12 Comparison of (MODIS and Sentinel) satellite–based inundation maps with
hydrodynamic model results for the Southern Gulf catchments ................................................. 37
Figure 5-1 Percentage change in mean annual rainfall and potential evaporation under Scenario
C relative to Scenario A ................................................................................................................. 42
Figure 5-2 River flow under Cdry and Cwet scenarios relative to Scenario A (Baseline) (for 2000
to 2023) at the boundary of Southern Gulf catchments hydrodynamic model, gauge 91210104
on the Gregory River at Gregory, 91210703 on the Nicholson River at Connolly’s Hole and
91390002 on the Leichhardt River at Lorraine ............................................................................. 43
Figure 5-3 Simulated aggregated streamflow used as inflows in the hydrodynamic model
(91210104 on the Gregory River, 91210703 on the Nicholson River and 91390002 on the
Leichhardt River) for two different flood events – 2016 (AEP of 1 in 3) and 2023 (AEP of 1 in 38)
– under scenarios A (Baseline), Cdry (future dry climate) and Cwet (future wet climate) .......... 44
Figure 5-4 Locations of existing water users under scenarios A and C, and additional
hypothetical options considered under scenarios B and D .......................................................... 45
Figure 5-5 Mean monthly dam storage in different months at three dam sites in the Southern
Gulf catchments ............................................................................................................................ 46
Figure 5-6 Percentage inundation frequency in the Southern Gulf hydrodynamic model domain
under scenarios A (Baseline) and B (3-dams) ............................................................................... 48
Figure 5-7 Depth at maximum inundation extent in the Southern Gulf hydrodynamic model
domain under scenarios A (Baseline) and B (Dam) ...................................................................... 49
Figure 5-8 Comparison of inundated area (in square kilometres) under scenarios A (Baseline)
and B (Dam) in the Southern Gulf hydrodynamic model ............................................................. 50
Figure 5-9 Percentage inundation frequency in the Southern Gulf catchments under scenarios A
(Baseline) and B (Water Harvesting of 150 GL) ............................................................................ 51
Figure 5-10 Depth at maximum inundation extent in the Southern Gulf hydrodynamic model
domain under scenarios A (Baseline) and B (Water Harvesting of 150 GL) ................................. 52
Figure 5-11 Comparison of inundated area (in square kilometres) in the Southern Gulf
hydrodynamic model domain under scenarios A (Baseline) and B (Water Harvesting of 150 GL)
....................................................................................................................................................... 53
Figure 5-12 Percentage inundated frequency in the Southern Gulf hydrodynamic model domain
under scenarios A (Baseline) and C (Future Climate) ................................................................... 54
Figure 5-13 Depth at maximum inundation extent in the Southern Gulf hydrodynamic model
domain under scenarios A (Baseline) and C (Future climate) ...................................................... 55
Figure 5-14 Comparison of inundated area (in square kilometres) (left) in the Southern Gulf
hydrodynamic model domain under scenarios A (Baseline) and C (Future Climate) ................... 56
Figure 5-15 Percentage inundation frequency in the Southern Gulf hydrodynamic model
domain under scenarios A (Baseline) and D (Dry Climate and Dam) ........................................... 57
Figure 5-16 Depth at maximum inundation extent in the Southern Gulf hydrodynamic model
domain under scenarios A (Baseline) and D (Dry Climate and Dam) ........................................... 58
Figure 5-17 Comparison of inundated area (in square kilometres) in the Southern Gulf
catchments under scenarios A (Baseline) and D (Dry Climate and Dam) ..................................... 59
Figure 5-18 Percentage inundation frequency in the Southern Gulf hydrodynamic model
domain under scenarios A (Baseline) and D (Dry Climate and Water Harvesting) ...................... 60
Figure 5-19 Depth at maximum inundation extent in the Southern Gulf hydrodynamic model
domain under scenarios A (Baseline) and D (Dry Climate and Water Harvesting) ...................... 61
Figure 5-20 Comparison of inundated area (in square kilometres) in the Southern Gulf
catchments under scenarios A (Baseline) and D (Dry Climate and Water Harvesting) ................ 62
Figure 5-21 Relationship between flood discharge and inundation area for the Southern Gulf
catchments .................................................................................................................................... 63
Figure 5-22 Estimated annual maximum flooded area for the various climate and development
scenarios for the Southern Gulf catchments ................................................................................ 64
Tables
Table 4-1 Manning’s roughness coefficient (n) for various types of land cover occurring in the
Southern Gulf catchments ............................................................................................................ 28
Table 4-2 List of stream gauges that were used for the Southern Gulf hydrodynamic model
configuration and calibration ........................................................................................................ 28
Table 4-3 Flood events used for calibration ................................................................................. 30
Table 4-4 Flood event dates and number of satellite images (Landsat, MODIS and Sentinel)
processed for the Southern Gulf catchments hydrodynamic model calibration ......................... 35
Table 4-5 Detection statistics for the Landsat, MODIS and Sentinel images considered in the
analysis for the Southern Gulf hydrodynamic model calibration ................................................. 38
Table 5-1 Summary of selected future climate and development scenarios ............................... 40
Table 5-2 Surface area and reservoir capacity at full supply level (FSL) of the short-listed
hypothetical dams in the Southern Gulf catchments ................................................................... 46
Table 5-3 Comparison of the inundated area and associated changes under Scenario C (Future
Climate) relative to Scenario A (Baseline) .................................................................................... 56
Table 5-4 Emulator estimates of the flooded area for 133 years of simulation .......................... 64
1 Introduction
The most frequent, and often most damaging type of natural disaster has for many years been
floods (Kron, 2015; Wang and Gao, 2022; Yu et al., 2022). The changes in climate and land use
(including rapid urbanisation) that have occurred in recent times have caused flood events to
become even more frequent and disastrous (Arnell and Gosling, 2016; Dottori et al., 2018; Tabari,
2020). In Australia, floods are one of costliest types of natural disasters (Rice et al., 2022;
Ulubasoglu et al., 2019).
While floods are generally perceived as natural disasters, they can provide many environmental
and ecological benefits (Opperman et al., 2009; Tockner et al., 2008). Floodplain inundation
contributes to species diversity and relative abundance, aquatic biota growth (Phelps et al., 2015),
groundwater recharge (Doble et al., 2012) and soil fertility (Ogden and Thoms, 2002). During
floods, there is an exchange of water, sediments, chemicals, organic matter, and biota between
the main river channels and their floodplains (Bunn et al., 2006; Thoms, 2003; Tockner et al.,
2010). Since the Flood Pulse Concept first appeared in the scientific literature (Junk et al., 1989),
the importance of floodplain inundation for these exchanges and for the productivity of diverse
aquatic biota in river–floodplain systems has been emphasised in many studies (Bayley, 1991;
Gallardo et al., 2009; Heiler et al., 1995; Middleton, 2002). However, our knowledge of the
frequency and duration of floodplain inundation, of the associated connectivity between water
bodies, and of its impact on the ecological functioning of many of the world’s largest floodplain
systems is very limited. To date, the published knowledge is insufficient to adequately inform
water management for biodiversity protection or adaptation to future climates (Arthington et al.,
2015; Beighley et al., 2009).
Despite centuries of human activities that altered river floodplains worldwide, remnant
permanent water bodies still exist on the floodplains, but they are diminishing at increasing rates
(Bayley, 1995; Tockner et al., 2008). An important requirement for the management of floodplain
water bodies, including the management of wetlands of historical, cultural, economic and other
biodiversity values, is knowledge of the extent, frequency and duration of floodplain inundation
and of the hydrological connectivity between them. This is essential in deriving strategies for
maintaining, or even enhancing to an optimal level, the biophysical exchanges between rivers and
floodplains. The catchments of the Southern Gulf, the Assessment area, has large floodplains in
their middle and lower reaches, and they support a larger number of offstream wetlands with high
ecological, cultural and biodiversity values. Therefore, it is important to quantify the inundation
dynamics (in terms of extent, frequency and duration) and the hydrological connectivity between
the offstream wetlands and the main channel (or several channels) under the historical climate,
and to assess how the inundation and connectivity could be impacted under future climate and
development.
However, the quantification of floodplain inundation dynamics and hydrological connectivity
between water bodies remains a great challenge. A number of studies have used a combination of
remotely sensed inundated area and concurrent river flow to predict the impact of river flow on
flooded area (e.g. Frazier and Page, 2009; Overton, 2005; Peake et al., 2011; Townsend and Walsh,
1998). The same approach has also been used to quantify how river flow affects the number of
inundated wetlands (e.g. Frazier et al., 2003; Shaikh et al., 2001). However, this approach is not
dynamic. It cannot produce a continuous time series of predicted inundation extent, and it is not
possible to cannot predict the duration of wetland connectivity. Inundation extent and the
duration of wetland connectivity can have an important influence on wetland ecology. With the
development of computational methods and computer technology, hydrodynamic modelling has
become popular for the study of floodplain hydraulics and for quantifying the time course of flood
inundation with high spatial and temporal resolution (Nicholas and Mitchell, 2003; Schumann et
al., 2009). By combining these modelling techniques with high-resolution topography data, the
duration, frequency and timing of wetland connectivity can be quantified (Karim et al., 2012,
2015). Previous studies have used a combination of hydrological and hydrodynamic models using
simplified one-dimensional (e.g. Beighley et al., 2009; Chormanski et al., 2009) to more complex
two-dimensional (Tuteja and Shaikh, 2009) modelling. In this Assessment, a two-dimensional
flexible-mesh hydrodynamic module (MIKE 21 Flow Model FM, hereafter referred to as
MIKE 21 FM) has been used with advanced model configuring with flexible-mesh modelling tool to
simulate floodplain inundation.
1.1 Objectives
The Southern Gulf Water Resource Assessment flood modelling activity seeks to answer the
following questions:
• What areas on the floodplains are susceptible to flooding under the historical climate scenario?
• What are the extent, duration and frequency of floodplain inundation under the historical
climate scenario?
• What changes could be expected in inundation dynamics under future climate and development
scenarios?
• What changes could be expected in hydrological connectivity between floodplain water bodies
due to flow regime change under future climate and development scenarios?
This report describes the configuration and calibration of the MIKE 21 FM and scenario modelling
for the future climate and water infrastructure development scenarios.
1.2 Previous flood studies in the Southern Gulf catchments
Large-scale flood modelling for the Southern Gulf catchments has been very limited. However,
some local-scale flood studies have been conducted by the Queensland Reconstruction Authority
as a part of the Queensland Flood Mapping Program 2013. The Queensland Reconstruction
Authority (2012) investigated flood extent and depth for the town of Gregory using Geographic
Information Systems (GIS)-based methods, by combining topography and flood magnitude. They
produced inundation maps for floods of various magnitudes up to an annual exceedance
probability (AEP) of 1 in 100. Under the same flood mapping program, the Queensland
Reconstruction Authority (2013) investigated flood level, velocity and hazard in the vicinity of
Doomadgee township. The Assessment was conducted using a combination of hydrological input
and a two-dimensional hydrodynamic model (TUFLOW). Engeny (2020) undertook a flood study
for Burketown, which is located on a remnant of the main channel of the Albert River and
represents the most eastward extent of a very flat ridgeline that forms the highest ground (~5 m
above sea level) on the western bank of the river. The study was conducted using TUFLOW and
produced inundation maps for floods of various magnitudes.
1.3 Overview of flood modelling frameworks used in the Assessment
Hydrodynamic models are considered to be very useful tools for detailed flood inundation
modelling, and they have been utilised for several decades (Bulti and Abebe, 2020; Liu et al., 2015;
Teng et al., 2017). Based on the complexity of the river–floodplain network and the availability of
input data for model configuration, one can select one-dimensional, two-dimensional or coupled
one- and two-dimensional models (Horritt and Bates, 2002; Teng et al., 2017). However, it is
extremely difficult to represent complex floodplain features using one-dimensional models
because of the one-directional representation of river–floodplain system. Two-dimensional
models avoid much of the conceptualisation required for building an accurate one-dimensional
model by using gridded topography data (Pinos and Timbe, 2019). Nonetheless, the application of
traditional fixed-grid two-dimensional models is not always sufficient for reproducing river
conveyance. This is because model grids are not aligned with the riverbanks, and in many cases
the lowest points in the river are not adequately represented in the model (Bomers et al., 2019;
Teng et al., 2017). More recently, flexible-mesh (also called irregular-grid) models have been found
to be superior to regular-grid models in terms of accuracy and computational time (Kim et al.,
2014; Mackay et al., 2015; Pinos and Timbe, 2019). The use of a flexible-mesh model can
overcome many of the limitations of regular-grid models, as they allow complex floodway
geometries to be modelled with precision. They do not require the remainder of the floodplain to
be modelled at the same spatial resolution, since they allow the computational mesh to be aligned
and refined to suit the geometry of the floodplain (Mackay et al., 2015; Symonds et al., 2016).
The hydrodynamic models need to be calibrated against historical streamflow and inundation data
before they can be applied with a degree of confidence. Traditionally, flood models are calibrated
by comparing instream water heights (commonly gauge records) with floodplain inundation
(commonly water marks on trees, buildings and electric poles). However, for relatively remote and
sparsely populated catchments, it is often not possible to collect the field data that are necessary
to robustly calibrate the model. This serves as a major constraint in the use of hydrodynamic
models in remote and data-sparse areas. In recent years, there have been major advances in flood
inundation mapping using satellite and airborne remote sensing. While the satellite imagery–
based approaches have some limitations, including spatial and temporal resolutions, these
techniques provide very useful data for hydrodynamic model calibration. In the Assessment, a
combination of field-based observed stage heights and satellite-based inundation maps were used
to calibrate the hydrodynamic model. Figure 1-1 shows the general steps in configuring and
calibrating the two-dimensional hydrodynamic model (MIKE 21 FM) and scenario modelling for the
future climate and dam impact assessment.
Figure 1-1 Flowchart illustrating the method used to calibrate a hydrodynamic model (MIKE 21 FM) and scenario
modelling for future climate and dam impact assessment
DEM = digital elevation model.
1.4 Report overview and structure
This report has been prepared to:
• document the methods that were used to calibrate the MIKE FLOOD hydrodynamic models for
the Assessment area
• report on the assessment of hydrodynamic model performance relative to satellite-based flood
inundation mapping
• report on potential flood inundation extent under the historical climate and current level of
development scenario
• report on potential changes to flood inundation and wetland connectivity under future climate
and development scenarios.
This report is structured as follows. Chapter 2 describes the inundation mapping approach using
satellite data and provides a summary of long-term inundation extent for the three study areas.
Chapter 3 describes the hydrodynamic modelling approach, including the rationale for the
selection of the MIKE 21 FM model, its input/output data requirements and the model calibration
algorithm. Chapters 4 describes the hydrodynamic model configuration and the calibration of
hydrodynamic model parameters for the Southern Gulf catchments. Chapter 5 provides the results
For more information on this figure, please contact CSIRO on enquiries@csiro.au
Inflow at
the
boundaryRainfallLand use
mapWetland
dataModel gridHydraulic
roughnessHydrodynamic
model
configurationFlood frequency
analysisConnectivity
assessmentHydrodynamic
model
calibrationSimulated
streamflowWetland
selection for
connectivityInundation mapFuture
climate/
damStream
gauge dataDEMSatellite
imageryPublished
literature
(roughness)
Impact assessment
Surface
topographyInundation
simulation for
selected flood
eventsStep 1Step 2Step 3Step 4Local runoff
generationOpen water
likelihood
algorithm
Step 5
and discussion on impacts of future climate and infrastructure scenarios on floodplain inundation.
Chapter 6 summarises the key findings of the Assessment.
1.5 Key terminology and concepts
1.5.1 WATER YEAR AND WET AND DRY SEASONS
Northern Australia has a highly seasonal climate, with most rain falling from December to March.
Unless otherwise specified, the Assessment defines the wet season as the 6-month period from
1 November to 30 April, and the dry season as the 6-month period from 1 May to 31 October.
All results in the Assessment are reported over the water year, defined as the period 1 September
to 31 August, unless otherwise specified. This allows each individual wet season to be counted in a
single 12-month period, rather than being split over two calendar years (i.e. counted as two
separate seasons). This is more realistic for reporting climate statistics from a hydrological and
agricultural assessment viewpoint.
1.5.2 SCENARIO DEFINITIONS
The Assessment considered four scenarios, reflecting a combination of different levels of
development and historical and future climates, much like those used in the Northern Australia
Water Resource Assessment projects (Charles et al., 2016) and the Victoria, Roper and Southern
Gulf Water Resource Assessment (McJannet et al., 2023):
• Scenario A – historical climate and current development
• Scenario B – historical climate and hypothetical future water resource development
• Scenario C – future climate and current development
• Scenario D – future climate and hypothetical future water resource development.
SCENARIO A
Scenario A is a historical climate and current development scenario. The historical climate series is
defined as the observed climate (rainfall, temperature and potential evaporation (PE) for the
water years from 1 January 1889 to 1 July 2023). All baseline data presented in this report has
been calculated from data for this period unless otherwise specified. Justification for use of this
period is provided in the companion technical report on climate (McJannet et al., 2023). The
current level of surface water, groundwater and economic development were assumed (as of
1 July 2023). This scenario was referred to as Scenario AE in the river model scenario analysis
(Gibbs et al., 2024b), as distinct from Scenario A, which assumed full use of the existing
entitlements in that report. Scenario A was used as the baseline against which assessments of
relative change were made. Historical tidal data were used to specify downstream boundary
conditions for the hydrodynamic modelling.
SCENARIO B
Scenario B is a historical climate and future development scenario, as generated in the
Assessment. Scenario B used the same historical climate series as Scenario A and assumed full use
of existing entitlements. River inflow, groundwater recharge and flow, and agricultural
productivity were modified to reflect future development. Two types of hypothetical future
development were considered. The first was an increase in water harvest extraction directly from
watercourses, typically assumed to supply water to nearby farm-scale developments. The second
hypothetical future development considered was the construction of large instream dams,
typically assumed to supply water to large contiguous irrigation districts. The impacts of changes
in flow due to this future development were assessed, including impacts on:
• instream, riparian and near-shore ecology
• Indigenous water values
• economic costs and benefits
• opportunity costs of expanding irrigation
• institutional, economic and social considerations that may impede or enable the adoption of
irrigated agriculture.
SCENARIO C
Scenario C is a future climate with current levels of surface water and ground development
(assuming full use of existing entitlements as under Scenario A) assessed at approximately the
year 2060. It is based on the 133-year climate series (as in Scenario A) derived from global climate
model (GCM) projections for an approximately 1.6 °C global temperature rise (by ~2060) relative
to the 1990 scenario. This climate projection represents Shared Socioeconomic Pathway (SSP) 2-
4.5, as defined in the United Nations Intergovernmental Panel on Climate Change (IPCC) Sixth
Assessment Report (IPCC, 2022). McJannet et al. (2023) provides further on the definition and
selection of SSPs. As in Scenario B, full use of existing surface water entitlements was assumed,
along with current level of groundwater and economic development.
SCENARIO D
Scenario D is a future climate and future development scenario. It used the same future climate
series as Scenario C, but river inflow was modified to reflect future development, as in Scenario B.
Therefore, in this report, the climate data for scenarios A and B were the same (based on historical
observations from 1 January 1889 to 1 July 2023), and the climate data for scenarios C and D were
the same (the above historical data scaled to reflect a plausible range of future climates).
1.5.3 HYPOTHETICAL DEVELOPMENT TERMINOLOGY
Water harvesting – an operation in which water is pumped or diverted from a river into an
offstream storage, assuming no instream structures.
Offstream storages – usually fully enclosed circular or rectangular earthfill embankment structures
situated close to major watercourses or rivers to minimise the cost of pumping.
Large engineered instream dam – a barrier across a river for storing water in the created reservoir,
usually constructed from earth, rock or concrete materials. In the Southern Gulf catchments, most
hypothetical dams were assumed to be concrete gravity dams with a central spillway (see
companion technical report on water storage (Yang et al., 2024)).
Annual diversion commencement flow requirement (DCFR) – the sum of the cumulative flow that
must pass the most downstream nodes sum in the Nicholson (node number 9121090), Albert
(node number 9129040) and Leichhardt rivers (node number 9130071) during a water year
(1 September to 31 August) before pumping can commence. It is usually implemented as a
strategy to mitigate the ecological impact of water harvesting.
Pump start threshold – a daily flow rate threshold above which pumping or diversion of water can
commence. It is usually implemented as a strategy to mitigate the ecological impact of water
harvesting.
Pump capacity – the capacity of the pumps expressed as the number of days it would take to
pump the entire node irrigation target.
Reach irrigation volumetric target – the maximum volume of water extracted in a river reach over
a water year. Note, the end use is not necessarily limited to irrigation. Users could also be involved
in aquaculture, mining, urban or industrial activities.
System irrigation volumetric target – the maximum volume of water extracted across the entire
study area over a water year. Note, the end use is not necessarily limited to irrigation. Users could
also be involved in aquaculture, mining, urban or industrial activities.
Transparent flow – a strategy to mitigate the ecological impacts of large instream dams by
allowing all reservoir inflows below a flow threshold to pass ‘through’ the dam.
2 Floodplain inundation mapping
Spatial maps of water in the landscape were derived from satellite imagery for dates coinciding
with flood events. They were useful in calibrating and post-auditing the hydrodynamic models
used to simulate floodplain inundation in the Southern Gulf catchments. These maps were
produced using satellite imagery from the optical sensors of Moderate Resolution Imaging
Spectroradiometer (MODIS), Landsat and Sentinel-2, and from the radar sensor Sentinel-1.
However, frequent cloud occurrence across northern Australia during the wet season limits the
optical remote-sensing opportunities for capturing inundation extents over the various rivers and
floodplains in the hydrodynamic model domain, particularly during flood peaks. Refer to the
technical report on Earth observation methods, Sims et al. (2016), for further details on the
satellite data, processing methods, and calibration of the inundation maps.
2.1 Satellite imagery acquisition and pre-processing
MODIS satellite data were used for producing daily maps of surface water. The MODIS sensor is an
optical/infrared sensor from the National Aeronautics and Space Administration. There are two
MODIS sensors currently orbiting the Earth (TERRA since 2000 and AQUA since 2002), although
they are approaching their end of life. They acquire daytime images of Australia at around 10 am
(TERRA) and 2 pm (AQUA). MODIS surface reflectance data are available from early 2000 until the
present from the United States Geological Survey (USGS) Land Processes Distributed Active
Archive Center (LP DAAC) (https://lpdaac.usgs.gov/) as gridded tiles, and they are also stored at
CSIRO for the whole of Australia. These data are available in hierarchical data format (HDF), in a
sinusoidal projection, with a pixel size of 0.004697 degrees (~500 m). Daily images of surface
reflectance from the TERRA MODIS sensor (MOD09GA) and an 8-day composite product (based on
cloud-free, good-quality images; from TERRA – MOD09A1) were used. (Data from the AQUA
sensor were not used, due to a detector failure in Band 6 (Gladkova et al., 2012) that resulted in a
striped pattern in the data.)
Landsat data, where available, are also useful for mapping surface water. These data are at a much
finer spatial resolution (30-m pixels) than MODIS, which is better suited to identifying narrow or
small water features. However, Landsat images are only available every 8 to 16 days at best
(depending on the number of operating sensors). This is usually less frequent due to cloud cover
and missing data. Landsat 5 data (Thematic Mapper or TM), Landsat 7 data (Enhanced Thematic
Mapper or ETM), and Landsat 8 and Landsat 9 data (Operational Land Imager or OLI) are available
from Digital Earth Australia (DEA) from 1987 until the present. The DEA provides consistent pre-
processing, organisation and analytics of Landsat data for the Australian continent (Dhu et al.,
2017). This processing involves corrections for illumination and observation angles, the
Bidirectional Reflectance Distribution Function (BRDF, which influences relative pixel brightness
across large scene areas) and atmospheric conditions. Refer to the companion technical report on
Earth observation methods, Sims et al. (2016), for further details on Landsat data processing.
The European Space Agency (ESA) operates the Sentinel-2 satellites. Sentinel-2 has two operating
sensors (2A since 2015 and 2B since 2017), and its data has a spatial resolution of 10 to 20 m and a
temporal frequency of every 5 days. Sentinel-2A and -2B are optical remote-sensing instruments,
so cloud cover reduces the amount of useful data for identifying inundation. These data are
available from DEA (https://www.ga.gov.au/scientific-topics/dea) in a similar analysis-ready
format to the Landsat data.
To help overcome the negative impact of frequent cloud cover during flood events, the ESA’s
Sentinel-1 Synthetic Aperture Radar (SAR) sensors were also used. Sentinel-1 SAR operates in the
microwave wavelength range, so is not affected by cloud cover (although heavy rain can influence
its return signal). Sentinel-1A (launched in 2014) and Sentinel-1B (operating from 2016 until 2022)
have a pixel size of 10 m and a temporal frequency of (generally) every 12 days within Australia.
These data are currently available through the Sentinel Australasia Regional Access (SARA) hub
(Sentinel Australasia Regional Access (SARA) website
) as a level 1 product in their native radar
coordinates.
2.2 Inundation mapping using MODIS
The Open Water Likelihood (OWL) algorithm (Guerschman et al., 2011) was used for mapping
open surface water with MODIS imagery at a 500 m pixel resolution. The OWL was developed
using empirical statistical modelling and calculates the fraction of water within a MODIS pixel. A
cloud mask was applied using the MODIS state band associated with each product, which contains
information on cloud and cloud-shadow locations. Refer to the technical report on Earth
observation methods, Sims et al. (2016), for further details on the MODIS OWL algorithm. Using
Python code, the daily MODIS OWL water maps (from TERRA – MOD) and the 8-day MODIS OWL
water maps (also from TERRA – MOD) for the Assessment area were extracted from the Australia-
wide products.
A limitation of MODIS mapping of surface water is that it is not of sufficient detail for mapping
narrow water features of less than 1 pixel in width (~500 m). This problem is even more
exaggerated when the narrow river channel is covered by vegetation along the banks or floating
vegetation, which effectively obscures the water from the sensor.
2.2.1 EVENT MAPS
The daily MODIS maps were extracted and subset for the Southern Gulf hydrodynamic model
domain for flood events used for inundation modelling. The years 2005, 2016, 2018, 2019 and
2023 were considered. The MODIS OWL water maps were converted into a map of water and non-
water pixels, and a threshold was used to stratify each MODIS OWL water fraction into water/non-
water. Ticehurst et al. (2015) showed that, in the Flinders catchment, a 10% threshold resulted in
the best match when comparing MODIS and Landsat inundation maps. Thus, all pixels above the
OWL threshold of 10% were mapped as water, and the images reprojected onto the geographic
latitude/longitude coordinate system (coordinate system code EPSG:4326), before conversion to a
GeoTIFF format for use with the hydrodynamic models. To help reduce the commission errors
throughout the Southern Gulf catchments, the Height Above Nearest Drainage (HAND) algorithm
(Nobre et al., 2011) was used to identify areas that were unlikely to flood. Pixels for which the
HAND value was above 40 were masked as nulls.
2.2.2 SUMMARY MAPS
Summary maps were also produced for the flood events processed for the hydrodynamic
modelling (Figure 2-1). These summary maps used composites of the MOD09A1 MODIS OWL
water maps to show maximum inundation extent, and the percentage of clear observations (i.e.
observations without clouds and/or nulls).
Figure 2-1 MODIS satellite–based flood inundation map of the Southern Gulf catchments
Data captured using MODIS satellite imagery. This figure illustrates the maximum percentage of MODIS pixel
inundation between 2000 and 2020.
For more information on this figure please contact CSIRO on enquiries@csiro.au
2.3 Inundation mapping using Landsat
2.3.1 ALGORITHM USED
Landsat 5 TM, 7 ETM, 8 OLI and 9 OLI data were extracted from DEA. For mapping flood
inundation, the Normalized Difference Water Index (NDWI)Xu (Xu, 2006) – an index based on the
green and mid infrared wavelengths – was used. Water was separated from other features using a
threshold of NDWIXu ≥ −0.3, which is consistent with Xu (2006) and balances errors of omission
and commission in the Southern Gulf catchments. Masking of cloud cover was undertaken by
extracting the pixel quality band (‘fmask’) available with each Nadir BRDF-Adjusted Reflectance
(NBAR) product. As for the MODIS water maps, to reduce the commission errors that were
detected throughout the catchment, the HAND algorithm (with a threshold of 15) was applied to
mask pixels in steep terrain. This threshold worked well for the Landsat spatial resolution along
the narrow, steep valleys.
2.3.2 EVENT MAPS
Landsat water maps were selected for the same hydrodynamic model domains and flood events as
the MODIS data. These were examined to find those images least affected by cloud cover, which
greatly limited the number of available images, given that deep clouds and sustained rain prevail
during the wet season across northern Australia, and the infrequent satellite overpass. These
images were then converted to the geographic latitude/longitude (EPSG:4326) coordinate system
and GeoTIFF format for use with the hydrodynamic models.
2.4 Inundation mapping using Sentinel-2
2.4.1 ALGORITHM USED
All Sentinel-2 imagery available during the flood event dates were extracted from DEA. The
NDWIXu was used to separate water from non-water, and the same threshold as used for Landsat
was used for the Sentinel-2 imagery. The ‘fmask’ band was used to mask pixels affected by cloud
or cloud shadow. As for the Landsat data, the HAND algorithm (with a threshold of 15) was applied
to mask pixels in steep terrain.
2.4.2 EVENT MAPS
Sentinel-2 water maps were selected for the same hydrodynamic model domains and flood events
as the MODIS data. These were examined to find those images least affected by cloud cover,
which greatly limited the number of images available, given that deep clouds and sustained rain
prevail during the wet season across northern Australia. The Fmask quality band for Sentinel-2
does not always detect clouds, so images affected by significant remnant clouds were removed.
The remaining images were then converted to the geographic latitude/longitude (EPSG:4326)
coordinate system and GeoTIFF format for use with the hydrodynamic models.
2.5 Inundation mapping using Sentinel-1
2.5.1 ALGORITHM USED
Sentinel-1 backscatter data for the flood events were processed to normalised radar backscatter,
an analysis-ready format, and filtered to reduce speckle effects. The radar backscatter was
converted from intensity to decibels for better contrast between water and land. A low
backscatter threshold was used to identify surface water. After testing various methods, this
threshold was calculated based on the likelihood of flooding generated from the HAND algorithm.
Areas with a low HAND value (i.e. along the valley floor) were more likely to be flooded, so the
backscatter threshold was increased there compared with those pixels further from the valley
bottom. To reduce commission errors, small ‘clumps’ of pixels (up to 10 pixels in size) that were
misclassified as water bodies were removed using a sieve filter. The HAND algorithm was applied
to mask pixels in steep terrain.
2.5.2 EVENT MAPS
Sentinel-1 images for 14 dates were processed to water maps. Due to the large catchment size, no
single date had images available for the entire Southern Gulf catchments. The images were
provided in the geographic latitude/longitude (EPSG:4326) coordinate system and GeoTIFF format
for use with the hydrodynamic model.
2.6 Combined summary maps
There were a limited number of images identifying flooding among the Landsat, Sentinel-2 and
Sentinel-1 images. However, because they were of a similar spatial resolution and quality, these
images were able to be combined to produce a summary map. Those water maps produced for
the flood events used in the hydrodynamic modelling were combined to delineate the maximum
flooding extent, and to calculate the percentage of clear observations in which a pixel was
inundated (Figure 2-2).
2.7 Summary
Flood inundation maps were produced using MODIS, Landsat, Sentinel-2 and Sentinel-1 imagery
for the Southern Gulf catchments. MODIS surface reflectance data are available since 2000 for the
whole of Australia. Daily images of surface reflectance from the TERRA sensor (MOD09GA), and 8-
day composite images of surface reflectance from the TERRA MODIS sensor (MOD09A1) were
used in the analysis. Landsat data are available from DEA from 1987 until the present in a
consistent analysis-ready format. Sentinel-2 data are available from DEA from 2015 until the
present.
The OWL algorithm was used for mapping open surface water with MODIS imagery at a 500 m
pixel resolution. The OWL was developed using empirical statistical modelling and calculates the
fraction of water within a MODIS pixel. The daily MODIS OWL water maps (from TERRA – MOD)
were extracted and subset for the Southern Gulf catchments hydrodynamic model domain, and a
Figure 2-2 Combined Landsat, Sentinel-2 and Sentinel-1 satellite–based flood inundation map of the Southern Gulf
catchments
Data captured using Landsat, Sentinel-2 and Sentinel-1 satellite imagery. This figure illustrates the percentage of clear
observations in which a pixel was inundated during the flood events used in the hydrodynamic modelling.
For more information on this figure please contact CSIRO on enquiries@csiro.au
series of flood maps were produced for selected flood days for the hydrodynamic model
calibration.
Sentinel-1 data, which are not affected by cloud cover, were also used to identify flooding, based
on a low radar backscatter. These images were able to capture some flooding in parts of the
Southern Gulf catchments.
The NDWI algorithm was used for mapping flood inundation based on Landsat and Sentinel-2
imagery. Water was separated from other features using a threshold of NDWI ≥ −0.3, which
balances errors of omission and commission, in the Southern Gulf catchments. Similarly, for
MODIS, a set of event-based and summary maps were produced using the combined Landsat,
Sentinel-2 and Sentinel-1 data.
Water maps generated from satellite imagery are particularly useful for encompassing large areas
at a reasonable temporal frequency. While MODIS can provide daily water maps, it is of a poorer
spatial resolution (~500 m) compared with Landsat (30 m pixels), Sentinel-2 (10–20 m pixels) and
Sentinel-1 (10 m pixels), and surface water of less than 1 pixel in width (~500 m) cannot be
mapped, particularly if there is vegetation along the riverbanks. In general, it was found that in
northern Australia’s wet–dry tropical/monsoonal climate, MODIS produces better flood maps for
large floodplains and/or big flood events. Care must be taken when interpreting the MODIS water
maps, due to artefacts in the imagery and confusion with dark soils, dark rocks, residual cloud and
topographic shadow. Unusual water features appearing in only one image need to be treated with
caution, and a flood-likelihood mask would be of great benefit in interpretation of the data.
The Landsat and Sentinel-2 water maps are very useful for detecting fine water features. However,
the temporal frequency of the imagery made it difficult to analyse flood events, as the flood peak
was often missed due to the timing of satellite overpass or cloud cover. As for MODIS, the Landsat
and Sentinel-2 water maps will be affected by the difficulty of detecting water under flooded
vegetation, and by confusion with dark features (e.g. residual cloud and topographic shadow).
The maximum percentage water cover based on MODIS imagery is higher than that determined
from combined Landsat/Sentinel-2/Sentinel-1 imagery. The reason for this is that MODIS OWL
water maps can identify wet soil as water (Sims et al., 2016). Another reason for the difference
could be that MODIS has a much higher temporal frequency (daily) than Landsat (every 16 days),
Sentinel-2 (every 5 days) and Sentinel-1 (every 12 days), which means that these sensors will fail
to coincide with as many flood events as MODIS, especially as there can be high cloud cover in the
Southern Gulf catchments.
3 Floodplain inundation modelling
3.1 Hydrodynamic models
Two-dimensional hydrodynamic models (e.g. MIKE 21 FM, TUFLOW, LISFLOOD-FP) are commonly
used to simulate flood levels and inundation extent in a river–floodplain system (Kumar et al.,
2023; Kvocka et al., 2015; Neal et al., 2012). The main strengths of the hydrodynamic models are
that they produce floodplain hydraulics that can be used to estimate inundation extent and
duration, and depth and frequency of wetting and drying at desired spatial (e.g. 5 to 10 m–grid)
and temporal (e.g. hourly) scales (Horritt and Bates, 2002). Based on the modelling objectives and
the availability of input data, either a two-dimensional regular-grid model (DHI, 2012) or a two-
dimensional flexible-grid model (DHI, 2016) can be selected. Technical considerations include the
size of the hydrodynamic model domain, irregularity in land topography, the availability of
topography data, and the complexity of the hydraulic regime. These two-dimensional models can
be coupled with one-dimensional river models, which allows finer-scale representation of the
highly dynamic river processes, including river cross-sections.
For the Assessment, a two-dimensional flexible-mesh model (MIKE 21 FM) was selected for the
Southern Gulf catchments, primarily based on the availability of fine-scale laser altimetry (LiDAR)
data. A brief description of the MIKE 21 FM model is presented in the following sections.
3.1.1 TWO-DIMENSIONAL HYDRODYNAMIC MODEL – MIKE 21 FM
The hydrodynamic module of the MIKE 21 FM is based on the two-dimensional incompressible
Reynolds-averaged Navier–Stokes equations (DHI, 2016). The model simulates the water level and
velocity flux in response to a variety of forcing functions in floodplains, lakes, estuaries, bays and
coastal areas. The boundary conditions in MIKE 21 FM can vary in both time and space. Point
sources and sinks can also be incorporated into the model. The model has been widely used across
the world, including in Australia, for flood inundation modelling (Teng et al., 2017). The main
strength of the MIKE 21 FM model is its ability to simulate wetting and drying of a floodplain
during a flood event, and the large number of computational cells that it can handle (in the range
of millions). Model input includes river bathymetry, floodplain topography, land surface roughness
(constant or spatially variable), inflow and outflow conditions, eddy viscosity and radiation
stresses. Rainfall, evaporation and surface infiltration can also be incorporated in MIKE 21 FM. The
model output includes time series of water depth, velocity and discharge for the entire
computational domain and user-specified time intervals. In the Assessment, a triangular mesh of
varying size was used, and the modelling domain was divided into three subzones based on mesh
size. The flexible-mesh model is preferable over the classic MIKE 21 FM regular-grid model,
because it allows selection of a very small grid at the area of interest and also the alignment of
model grids to the riverbanks.
3.1.2 SOLVING GOVERNING EQUATIONS AND HARDWARE REQUIREMENTS
The primitive variables of the governing equations are discretised using an element-centred finite-
volume method. The spatial domain is discretised into non-overlapping elements, which can be
either triangular or quadrilateral (DHI, 2016). The finite-volume method sets up an equivalent
Riemann problem (ERP) across each element interface and solves it to determine the variable
fluxes between the elements. The technique used in MIKE 21 FM determines an exact solution to
an approximate Riemann problem. The approach treats the problem as one-dimensional in the
direction perpendicular to each element interface. MIKE 21 FM has two options for time
integration accuracy: a first-order explicit Euler method (referred to as the lower temporal order
scheme) and a second-order Runge–Kutta method (referred to as the higher temporal order
scheme). There are also two options for spatial integration, with the second-order (higher-order)
accuracy being achieved through a variable gradient reconstruction technique prior to the ERP
formulation (DHI, 2016). The model can be simulated on either a central processing unit (CPU)
machine or a graphics processing unit (GPU) machines. In the present Assessment, the models
were run using GPU machines containing 16 CPU cores and 3 GPU cards (each with 4 P100 Nvidia
GPUs). Each run took about 3 days of computer time to simulate a 30-day flood event.
3.2 Data requirement for model configuration
Hydrodynamic models are data intensive. A large amount of temporal and spatial data is required
for setting up a hydrodynamic model for flood inundation modelling. While some input data (e.g.
stream network, water level, discharge) can be extracted from the secondary sources, most input
data are case-specific and need to be prepared before running the model. To configure the
MIKE 21 FM model, data needed across each of the hydrodynamic model domains included:
• topography
• surface roughness
• stream network
• Inflow/outflow
• water level
• local runoff.
4 Southern Gulf catchments hydrodynamic
model calibration
4.1 Physical and hydro-meteorological properties
4.1.1 PHYSIOGRAPHIC CHARACTERISTICS
The Southern Gulf catchments comprise four northerly draining catchments defined by the
Australian Water Resources Council (AWRC) river basin boundaries – Settlement Creek
(17,600 km2), Nicholson River (52,200 km2), Leichhardt River (33,400 km2) and Morning Inlet
(3700 km2) – and the islands within the AWRC Mornington Island Basin (total 1200 km2) (Figure
4-1). The catchments are characterised by several uplands and plains. The upland area in the south
and west reaches 620 m above sea level. The uplands have been eroded into a complex pattern of
easterly flowing streams and valleys separated by ranges and outcrops of sedimentary formations.
The Nicholson and South Nicholson rivers are the primary systems draining this area.
Musselbrook, Lagoon, Settlement, Gold and Running creeks also drain this area The Cloncurry
Plain consists of gently sloping colluvial and fluvial sedimentary plains. Streams are few and incised
into the pediments, with narrow alluvial plains. The Cloncurry Plain extends from the middle reach
of the Leichhardt River to Lawn Hill Creek. In the north, the Doomadgee Plain lies below and
adjacent to the Cloncurry Plain. Widely spaced creeks drain the plains towards the coast. In the
south, the Armraynald Plain lies below and adjacent to the Cloncurry Plain. Stream channels are
few, widely spaced, and deeply incised due to sea-level changes. The plains extend up the Lawn
Hill Creek, and Gregory and Leichhardt valleys. Lawn Hill Creek and Gregory River are spring-fed
permanent running streams. The Gregory River splits into a giant braid (20 km at its widest) of
permanent streams (consisting of the Gregory River, Beames Brook, Barkly River and Running
Creek) downstream of the Gregory Crossing. Downslope of both the Doomadgee Plain and the
Armraynald Plain lies the coastal Karumba Plain. This coastal unit extends 10 to 35 km inland from
the Gulf of Carpentaria coast, and the plain is most extensive near the Albert River mouth. Some
of the inland plains only flood when the rivers are in spate or when the north-westerly winds
cause exceptionally high tides during the monsoon. Because the plain is wide and the tidal range is
moderate (~3.5 m), and because the plain is generally flat, tidal waters can rapidly inundate the
land. Mangroves and tidal flats dominate the coastline (Gibbs et al., 2024a). For the hydrodynamic
modelling study, the focus was the floodplains of the Nicholson, Gregory and Leichhardt rivers.
The main land use in the Southern Gulf catchments is grazing of beef cattle (84%), including on
productive black soil plains, with nature conservation through national parks (including
Boodjamulla (formerly Lawn Hill) National Park and Indigenous Protected Areas) comprising 13%
of the catchments’ area. Century Zinc Mine – formerly one of the world’s largest zinc mines – is
located near Lawn Hill, and there are large mining operations in and near Mount Isa. The
catchments incorporate the edge of the Barkly Tableland and are characterised by extensive
alluvial grassy plains, distinctive marine plains, dissected hilly regions, and erosional plains.
Figure 4-1 Southern Gulf catchments map showing physiography and river network. From Thomas et al. (2024)
For more information on this figure, please contact CSIRO on enquiries@csiro.au
4.1.2 CLIMATE
The Southern Gulf catchments are characterised by a distinctive wet and dry season due to their
location in the northern Australia tropics. The mean annual rainfall, averaged over the Southern
Gulf catchments for the 133-year historical period (1 September 1890 to 31 August 2022) is
602 mm. Rainfall totals are highest near the coast and decline in a southerly direction. In the
Southern Gulf catchments, approximately 94% of rain falls during the wet-season months
(1 November to 30 April). The mean and median annual rainfall is highest near the coast, primarily
due to the monsoonal activity, which generates significant rainfall during the wet season
(McJannet et al., 2023). The highest rainfall totals typically occur during January and February
(Figure 4-2).
Mean areal potential evapotranspiration in the Southern Gulf catchments is approximately
1900 mm. Evaporation is high all year round, but exhibits a strong seasonal pattern, ranging from
about 200 mm per month during November and December to about 100 mm per month during
the middle of the dry season (June). The high mean annual PE and moderate mean annual rainfall
result in a mean annual rainfall deficit across all of the Southern Gulf catchments (Figure 4-2).
Consequently, most of the study area is semi-arid.
Tropical cyclones and lows contribute large quantities of rainfall over the Southern Gulf
catchments in some years and result in high daily rainfall values. Increased rainfall, storm surge,
and wind speeds are associated with tropical cyclones. The cyclone season in the Southern Gulf
catchments falls between November and April. From 1969 to 2022, the Southern Gulf catchments
experienced 53 tropical cyclones. Sixty percent of seasons experienced no tropical cyclones, 36%
one, and 4% two (McJannet et al., 2023).
Approximately a third of GCMs project an increase in mean annual rainfall by more than 5%, a fifth
of GCMs project a decrease in mean annual rainfall by more than 5%, and about half project ‘little
change’ (McJannet et al., 2023).
Figure 4-2 Historical monthly rainfall (showing the range in values between the 20% and 80% monthly exceedance
rainfall) and annual rainfall in the Southern Gulf catchments at Doomadgee, Gregory and Burketown (from
McJannet et al., 2023)
The left-hand column shows the monthly rainfall, and the right-hand column shows time series of the annual rainfall.
(The range in the left-hand column is the 10th to 90th percentile for the monthly rainfall, and the blue line in the right-
hand column represents the 10 years moving mean).
"\\fs1-cbr\{lw-rowra}\work\1_Climate\4_S_Gulf\2_Reporting\plots\climate_report\annual_monthly_rainfall_range_4_station.png"
For more information on this figure, please contact CSIRO on enquiries@csiro.au
4.1.3 STREAMFLOW
Streamflow across the Southern Gulf catchments varies significantly based on river network and
location. There were 23 gauging stations, but only 7 stations are currently operating across the
Leichhardt, Gregory and Nicholson catchments. The only site with streamflow gauging over the
past 35 years and maximum gauging covering at least 75% of the total volume is the gauge on
Gunpowder Creek at Gunpowder (913006A). The streamflow gauging station on the Gregory River
at Riversleigh (912105A) also has relatively extensive gauging data. The Nicholson River at
Connolly’s Hole (912007A) has gauging data that cover most of the flow regime (86% of the total
estimated volume); however, this station was closed in 1988. Most gauges are located towards the
southern portion of the Assessment area, in the upper reaches of the catchments (Figure 4-3). The
only gauge available in the lower reaches of the Assessment area is on the Leichhardt River at
Floraville Homestead (913007B). There are a number of anabranches on the Gregory River near
the gauge at Gregory (912101A). The proportion of flow occurring in each of the flow paths will
have an influence on the flow remaining in the Gregory River, and ultimately flowing towards the
Nicholson River, and the proportion that splits off into Beames Brook and ultimately the Albert
River at Burketown.
Driven by the highly seasonal climate, streamflow in the Southern Gulf catchments displays very
strong seasonal patterns (Figure 4-4). The Southern Gulf catchments comprise four northerly
draining catchments: the catchments of Settlement Creek, the Nicholson River, the Leichhardt
River and Morning Inlet. The rivers and creeks in the area are seasonal (January to March) due to
the vast majority of rainfall occurring during the wet season, with cease-to-flow periods for 30%–
60% of the time. The watercourses in the Gregory River catchment are a notable exception, with
perennial flow that is associated with limestone and dolostone aquifers.
For the period of 1890 to 2022, the mean (median) annual end-of-system volume for the
Leichhardt River catchment is 1721 GL (804 GL). For the Gregory–Nicholson catchment, the mean
(median) annual end-of-system volume was 2448 GL (1159 GL). Across all catchments in the
Assessment area, the mean (median) annual end-of-system volume was estimated to be 7360 GL
(3816 GL) (Gibbs et al., 2024a).
Figure 4-3 Location of streamflow gauges in the Southern Gulf catchments
Colours indicate the proportion of the total estimated volume that occurred below the maximum gauging at the site,
with the size of the symbol indicating the number of years with satisfactory data (defined as having a good- or fair-
quality code). Sites that are currently open are indicated by a black dot inside the triangle
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 4-4 Monthly flow distribution at Floraville on the Leichhardt River, based on the observed flow data for 1984
to 2023
4.1.4 FLOODING
Intense seasonal rains from monsoonal bursts and tropical cyclones in the period of November to
March create flooding in parts of the Southern Gulf catchments and inundate large areas of
floodplains, mostly in the downstream reaches between the Nicholson, Gregory and Leichhardt
rivers. The floodplains along the Alexandra and Albert rivers are also heavily flooded (Figure 2-1).
Flooding is common in the Burketown area, which is located on a remnant of the main channel of
the Albert River and represents the most eastward extent of a very flat ridgeline that provides the
highest ground (~5 m above sea level) on the western bank of the river. Burketown is susceptible
to flooding from the Albert River floodplain, as well as from overland flow paths within the town
area. Flooding of the Albert River could also occur from floodwater breakout from the Nicholson
or Gregory rivers. Rivers in the Southern Gulf catchments are unregulated, and their overbank
flow is generally governed by the topography of the floodplain. Characterising these flood events
is important for a number of reasons. Flooding can be catastrophic to agricultural production in
terms of loss of stock, fodder and topsoil, and damage to crops and infrastructure. In addition, it
can isolate properties and disrupt vehicle traffic providing goods and services to people in the
catchment. However, flood events also provide the opportunity for offstream wetlands to be
connected to the main river channel. The high biodiversity found in many unregulated floodplain
systems in northern Australia is thought to largely depend on seasonal flood pulses, which allow
for biophysical exchanges to occur between rivers and offstream wetlands.
Flooding is widespread in parts of Lawn Hill Creek and the Gregory, Leichhardt and Albert rivers
near Burketown. In the last 40 years (1984 to 2023), there have been 41 floods ranging from small
to large in the catchments. This was based on an overbank threshold of 709 m3/second, which was
estimated by obtaining daily streamflow at Floraville on the Leichhardt River (913007) and
comparing it against floodplain inundation on the available satellite imagery. For flood events with
an AEP of 1 in 2, 1 in 5 and 1 in 10, the peak discharge at the Floraville gauging station on the
For more information on this figure, please contact CSIRO on enquiries@csiro.au
Leichhardt River was 1220, 3350 and 7140 m3/second, respectively. Figure 4-5 shows the annual
maximum daily discharge at Floraville for 1953 to 2023. While floods can occur in any month from
October to May, approximately 90% of historical floods have occurred between December and
March, with the maximum in January (33.3%) (Figure 4-6). Of the ten largest flood peak discharge
rates at Floraville, four events occurred during January, four in March, one in February and one in
December. Flood peaks typically take about 2 days to travel from Gregory to Burketown, with a
mean speed of 3.2 km/hour.
Figure 4-5 Annual maximum daily discharge at Floraville on the Leichhardt River, based on the observed flow data
for 1984 to 2023
Figure 4-6 Monthly flood frequency in the Southern Gulf catchments (floods defined as an AEP of ≥1 in 1, based on
flow data for 1984 to 2023) at Floraville on the Leichhardt River
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
4.2 Model configuration
The hydrodynamic model was configured for the proportion of the Southern Gulf catchments
encompassing the floodplains of the major rivers and the tidal flats at the mouth of the Nicholson,
Albert and Leichhardt rivers and Gin Arm Creek (Figure 4-7). The modelling domain included areas
downstream of Doomadgee on the Nicholson River, Gregory on the Gregory River and Lorraine on
the Leichhardt River, encompassing an area of 17,130 km2. The model domain included six inflow
boundaries across the river network and one water-level boundary off the coast (Figure 4-7). The
model domain included 45 subcatchment outlets, incorporating local runoff, as described in
Section 4.3.4.
Figure 4-7 Hydrodynamic model configuration of the Southern Gulf catchments, showing the river network, model
boundaries, and local runoff points in the model domain
For more information on this figure please contact CSIRO on enquiries@csiro.au
4.3 Model input
4.3.1 TOPOGRAPHY
Land surface topography data are essential for flood inundation modelling, in addition to hydro-
meteorological inputs (Sanders, 2007). The vertical accuracy of these data is critical, since the
inundation depths and extents simulated by hydrodynamic models are highly sensitive to vertical
errors of topographic data, especially on low-gradient floodplains (Horritt and Bates, 2002).
A digital elevation model (DEM) was produced for the Southern Gulf catchments MIKE 21 FM
model domain based on a LiDAR 5-m DEM (±0.3 m horizontal and ±0.1 m vertical accuracy)
patched with a 30 m Forest And Buildings removed Copernicus) digital elevation model (FABDEM).
The highest resolution publicly available topographic information that encompasses the whole of
the Southern Gulf catchments is based on a 1-second (i.e. ~30 m) Shuttle Radar Topography
Mission (SRTM) DEM (Gallant et al., 2011) and a 1-second FABDEM (Hawker et al., 2022). These
two global DEMs were compared, and FAB DEM was chosen due to its superior vertical accuracy in
the Assessment area (Meadows et al., 2024). The FABDEM was used for the hydrodynamic
modelling domain to cover areas for which LiDAR data were not available. The final combined
DEM was created by resampling the FABDEM to 5 m (to match the original LiDAR resolution) and
merging it with the LiDAR data using methods described by Gallant (2019). The Gallant method
attenuates the difference between the resampled coarse DEM and the LiDAR data at the interface
to zero over a distance beyond the LiDAR extent. The LiDAR elevations remain intact, while the
coarse DEM elevations are modified by the attenuated difference, resulting in a ‘seamless’
combination while retaining hydrological connectivity. The final product of the Southern Gulf DEM
is a 5 m–grid raster file of size 7 GB.
The LiDAR data covers the major proportion of the Gregory and Albert rivers and only a small
proportion of the Nicholson River (Figure 4-8). The hydrodynamic modelling domain encompasses
an area of 17,130 km2, and the area covered by LiDAR is 7490 km2 (i.e. ~43.7% of the model
domain).
Figure 4-8 LiDAR data coverage in the hydrodynamic model domain of the Southern Gulf catchments
For more information on this figure please contact CSIRO on enquiries@csiro.au
4.3.2 SURFACE ROUGHNESS
The hydraulic roughness coefficients of the land surface were derived from DEA Land Cover, which
is a collection of annual land-cover maps for Australia for the period 1988 to 2020 (Owers et al.,
2021) in which Australia’s landscapes are classified into six basic land-cover categories (Table 4-1).
The categories were represented in the model by Manning’s roughness coefficient (n), as
estimated in the published literature (Arcement and Schneider, 1989; Chow, 1959; LWA, 2009)
(Table 4-1), and a roughness map was produced.
Table 4-1 Manning’s roughness coefficient (n) for various types of land cover occurring in the Southern Gulf
catchments
For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au
4.3.3 INFLOW AND OUTFLOW BOUNDARY CONDITIONS
The hydrodynamic model consisted of four inflow boundaries, but gauge data were available for
only one boundary (i.e. Gregory River at Gregory). The upstream river boundary conditions of the
hydrodynamic models were obtained from the Australian Water Resource Assessment – River
model (AWRA-R) simulations (see the companion technical report on river model calibration for
the Southern Gulf catchments, Gibbs et al., 2024a). The river model discharge gauge nodes used in
the hydrodynamic models are shown in Table 4-2. In addition to these, the tidal sea-level data was
used for the downstream water-level boundary. The Karumba station was used, which is the
closest to the Southern Gulf catchments. The remaining inflows to the boundary of the
hydrodynamic model domain that were not captured by the river model were simulated using the
Sacramento rainfall-runoff model based on the residual reach parameters from the AWRA-R
calibrations (Gibbs et al., 2024b).
Table 4-2 List of stream gauges that were used for the Southern Gulf hydrodynamic model configuration and
calibration
For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au
4.3.4 LOCALLY GENERATED STREAMFLOW
The hydrodynamic model domain was subdivided into 53 subcatchments based on the topography
and stream network (Figure 4-7). The Sacramento-based gridded runoff was averaged for each
subcatchment by assigning Scientific Information for Land Owners (SILO) cells to the
subcatchments in accordance with the intersecting cells. Streamflow at the outlet (also called the
‘pour point’) of each subcatchment was calculated by multiplying the runoff by the subcatchment
area. This locally generated runoff was added into the hydrodynamic model as a point source of
water in the system. Subcatchment boundaries and pour points were generated using ArcGIS
tools. The locations of some pour points were manually changed to ensure all subcatchments
delivered the local flow directly into a river or creek (Figure 4-7).
4.4 Flood frequency and selected events for model calibration
The magnitudes of flood events in the Southern Gulf catchments were calculated based on the
observed flow data at Floraville on the Leichhardt River (913007) using a flood frequency analysis.
A flood event was defined as the occurrence of overbank flows that could be identified through
satellite imagery. Based on these observations, a threshold flow rate of 709 m3/second was used
to identify the overbank flow as the starting point for flood events. If the flow rate dropped below
the overbank flow threshold for five consecutive days, the event was considered to have ended.
To determine the number of times each event was exceeded events with higher peak discharge
and larger total event volume were counted.
Flood frequency analysis (FFA) was performed in the Southern Gulf catchments to establish
streamflow thresholds above which a flood event would occur. Flood frequency was estimated for
the three major rivers, the Nicholson River at Connolly’s Hole (912107), the Gregory River at
Riversleigh (912105) and the Leichhardt River at Floraville (913007). Traditionally, flood
frequencies are estimated based on maximum discharge for an individual event. However, in this
Assessment, to help determine the true magnitude of the events, the FFA took into account both
the total flow volume and the peak discharge for each event. This was motivated by the
knowledge that not only the maximum discharge but also the duration of an event can have a
great impact on the inundated area. Figure 4-9 displays the relationship between peak flow and
AEP for the three gauges, one on the Nicholson River, one on the Gregory River and the other on
the Leichhardt River. However, the Leichhardt River gauge at Floraville is the only gauge within the
hydrodynamic model domain that has 30 years or more of observed data.
Figure 4-9 Peak flood discharge and annual exceedance probability at gauge: (a) 912107 (Nicholson River at
Connolly’s Hole, (b) 912105 (Gregory River at Riversleigh) and (c) 913007 (Leichhardt River on Floraville)
Flood events were selected for the model calibration based on the availability of cloud-free
Landsat satellite imagery, while ensuring the magnitude of the flood events spanned the AEPs of
interest to ecologists and the land suitability analysis. Five events were chosen, with flood
magnitudes ranging from an AEP of 1 in 2 to 1 in 38, based on the flow data at Floraville on the
Leichhardt River (Table 4-3). The requirement for a large number of simulations involving all
combinations of scenarios and calibration events necessitated the use of only two flood events (in
2016 and 2023) for scenario modelling.
Table 4-3 Flood events used for calibration
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
4.5 Hydrodynamic model simulation and outputs
The simulations for hydrodynamic models in the calibration and scenario modelling were
undertaken with a 1-second time step to satisfy the numerical stability criteria for the biggest
flood event in the analysis. Each event was simulated for 30 days (longer than the flood durations)
irrespective of the time of flood recession, to ensure the entire rising and falling limbs were
included in the simulation. The models were run using GPU machines consisting of 16 CPU cores
and 3 GPU cards (each with 4 P100 Nvidia GPUs). For each run, it took about 2 days of computer
time to simulate a 30-day flood event. At the hydrodynamic model boundaries, daily discharges
were specified in all inflow boundaries, and hourly tide levels were specified at the seaside
boundary. The model used an inbuilt interpolation technique to derive flow variables at each
computational time step. The model outputs included the water surface elevation, depth and
velocity for each mesh element. While the model has the option of producing output at any time
interval, all outputs in this analysis were recorded at 6-hour time intervals. A separate model was
configured and simulated for each flood event.
Total water depths and water velocities were obtained from each model run and converted to
three-column XYZ format American Standard Code for Information Interchange (ASCII) data. In the
post-processing, two-dimensional triangular flexible-mesh data were converted to 5 m × 5 m
gridded data via an inverse distance weighted interpolation algorithm.
4.6 Hydrodynamic model calibration
The calibration process was constrained by the excessive time required by the hydrodynamic
model simulation, which limited the number of iterations possible for tuning. However, the
simulations were checked to ensure correct activation of the anabranches and connectivity to the
various floodplain water bodies. The duration of the floodplain inundation was tuned by adjusting
the surface roughness coefficient and the infiltration rates to ensure inundation patterns were
consistent with the satellite imagery. In addition to evaluations against the satellite imagery,
simulated stage height outputs were compared with observed stage heights at the Floraville
gauging station on the Leichhardt River (913007), which is the only gauge with recent data records
in the hydrodynamic model domain.
The evaluation of hydrodynamic modelled inundation extent was performed using available and
suitable Landsat and MODIS inundation maps for the selected flood events (including at least one
image for each flood event). To evaluate the performance of the hydrodynamic model using the
remotely sensed flood extent observations, two evaluation methods were employed:
• A visual comparison of the spatial inundation areas indicated by the satellite image and the
model simulation was undertaken to assess how the main inundation patterns were
represented by the hydrodynamic model.
• A quantitative assessment of the spatial inundation metrics (whether the model correctly
simulated inundated pixels or not) was undertaken to assess how well the hydrodynamic model
captured the overall inundation extent.
Both evaluation methods were performed by resampling of the remotely sensed inundation maps
(Landsat 30 m and MODIS 500 m) at 5 m–grid pixel horizontal resolution, to be consistent with the
hydrodynamic model gridded output. The satellite inundation maps were masked so as to only
include the modelled area within the hydrodynamic model domain.
In many cases, satellite-derived inundation maps (even for a sensor like MODIS, with a twice-daily
satellite overpass frequency) will only capture portions of the hydrodynamic model domain due to
the persistent cloud cover that can occur during flood events. Thus, only pixels in the satellite
images showing ‘inundated’ or ‘non-inundated’ were considered; this meant that all ‘cloudy’ or
‘no data’ pixels were removed from the analysis.
4.6.1 CATEGORICAL STATISTICS
Detection metrics were computed for each adjusted hydrodynamic model domain and for each
grid cell. Every satellite inundation map and modelled inundation extent was classified as a hit (H,
observed inundation by the satellite correctly detected by the model), miss (M, observed
inundation not detected by the model), or false alarm (F, inundation detected by the model but
none observed by the satellite) using a contingency table, following Ebert et al. (2007) (Figure
4-10).
For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au
Figure 4-10 Classification at the grid cell level using a contingency table
The following statistics were computed from the contingency table (see Collaboration for
Australian Weather and Climate Research (CAWCR)):
• The Probability Of Detection, POD = H/(H + M), gives the fraction of inundated pixels correctly
detected (range 0 to 1, 1 indicating a perfect score). It is sensitive to hits, but ignores false
alarms, and it should be used in conjunction with the False Alarm Ratio (FAR; see next bullet
point).
• FAR = F/(H + F), gives the fraction of wrongly detected inundated pixels (range 0 to 1, 0
indicating a perfect score). It is sensitive to false alarms, but ignores misses.
• The Frequency Bias, FB = (H + F)/(H + M), gives the ratio of the simulated to observed inundated
pixels frequency (range 0 to ∞, 1 indicating a perfect score). It measures the ratio of the
frequency of modelled inundated pixels to the frequency of the satellite imagery inundated
pixels, and it indicates whether the hydrodynamic model has tended to underestimate (FB < 1)
or overestimate (FB > 1) events. It does not measure how well the modelled inundation extent
corresponds to the satellite inundation extents, as it only measures relative frequencies.
• The equitable threat score (ETS), used as an overall performance metric, gives the fraction of the
inundated pixels that were correctly detected, adjusted for correct detections (He) that would
be expected due to random chance: ETS = (H – He)/(H + M + F – He), where
He = (H + M)(H + F)/N and N = the total number of estimates (range –1/3 to 1, 1 indicating a
perfect score and 0 indicating no skill). It is sensitive to hits. Because it penalises both misses and
false alarms in the same way, it does not distinguish the sources of error.
Although the detection statistics described above are well constrained, there are issues to
consider in the interpretation of the results, such as:
• Satellite inundation images may show inundated areas that remain in the landscape as ponded
areas in between flood events, because of the gentle topography and the low infiltration rates.
• The results of the cell to cell comparison between the hydrodynamic model output and the
satellite imagery will be inherently poor where the satellite images (from MODIS in particular)
are of a lower spatial resolution than the river channel widths, the river morphology and the
resulting inundation dynamics.
The two approaches mentioned above (visual comparison and detection statistics) complement
each other – visual comparisons, although labour-intensive and subjective, highlight the sources or
nature of the errors and provide diagnostic information regarding necessary changes to the inputs
or hydrodynamic model set-up. On the other hand, statistical metrics provide an objective and
comparable metric for assessing overall model performance. The calibration is considered
successful if the two approaches assess the performance of the model output as reasonable (in
terms of overall inundation patterns captured and detection statistics).
4.7 Results and discussion
4.7.1 STAGE HEIGHT
Figure 4-11 shows a typical comparison between simulated and observed stage heights for the
various flood events at Floraville on the Leichhardt River (913007B). In general, the simulated
flood peaks match well with the observed data, except for the 2005 flood event, for which the
model overpredicted stage height. While a good match was obtained for the flood peak, there
were differences in stage heights for the rising and falling limbs of the flood hydrograph. Also,
there were differences in the timing of the flood peak. For most of the events, the simulated stage
heights for the receding floods were smaller than the observed heights. In addition, the
hydrodynamic model failed to simulate stage heights between the two peaks correctly for flood
events with two or more peaks (e.g. 2018 flood).
The possible reasons for the discrepancies between simulated and observed stage heights include:
(i) coarse topography data for the major proportion of the Leichhardt River, (ii) lack of good-
quality bathymetry data, (iii) lack of observed river flow data, and (iv) poor representation of some
river channels. In addition, there are uncertainties in the river model–simulated inflows to the
hydrodynamic model domain.
For more information on this figure, please contact CSIRO on enquiries@csiro.au
Figure 4-11 Comparison of the model-simulated stage height and the observed stage height at Floraville (913007B)
on the Leichhardt River in the Southern Gulf catchments
4.7.2 SATELLITE-BASED FLOOD MAPS
The available Landsat, MODIS and Sentinel images were processed for the Southern Gulf
hydrodynamic model domain for the calibration periods (Table 4-4). The availability of images and
the proportion of cloud/null effects were reported. Due to the large number of MODIS images,
only those images containing 80% or less clouds/nulls were processed to obtain inundation maps.
Also, to exclude images with little or no flooding, only images with 5% or more observed
inundated area were used for comparison with the hydrodynamic model results.
Table 4-4 Flood event dates and number of satellite images (Landsat, MODIS and Sentinel) processed for the
Southern Gulf catchments hydrodynamic model calibration
START DATE
END DATE
FLOOD
MAGNITUDE
NUMBER OF
LANDSAT IMAGES
NUMBER OF
SENTINEL -1 IMAGES
NUMBER OF
SENTINEL-2 IMAGES
NUMBER OF
MODIS IMAGES
8/01/2005
18/01/2005
1 in 2 AEP
3
0
0
4
29/12/2015
10/01/2016
1 in 3 AEP
3
0
1
9
4/03/2018
21/03/2018
1 in 5 AEP
2
1
7
11
2/02/2019
20/02/2019
1 in 10 AEP
4
1
8
15
15/02/2023
1/04/2023
1 in 38 AEP
9
12
8
22
AEP = annual exceedance probability
4.7.3 INUNDATION EXTENT
The results of the hydrodynamic model simulation of inundation across the Southern Gulf
catchments were compared with satellite inundation maps through a direct visual comparison and
detection metrics. In the first instance, images that passed the first cloud cover filtering were
further scrutinised to identify flood patterns that could be used to inform the calibration of the
hydrodynamic modelling. The images within the modelling simulation period (see Table 4-4) that
showed limited inundation, or images with large cloud cover over the inundated areas, were
omitted from further analysis. Generally, at least one image was retained for each flood event to
assist in the calibration. Figure 4-12 shows an example of Landsat, MODIS and Sentinel inundation
maps with corresponding hydrodynamic model inundation maps for the same date.
Table 4-5 presents the detection statistics for selected images considered in the analysis, including
images from all five flood events (see Table 4-4 for the flood events). The POD, which gives the
fraction of inundated pixels correctly detected, varies within a range of 0.08 to 0.56 (range 0 to 1,
1 indicating a perfect score), with a mean of 0.28. The FAR, which gives the fraction of wrongly
detected inundated pixels, varied within a range of 0.33 to 0.96 (range 0 to 1, 0 indicating a
perfect score), with a mean of 0.67. This indicates over estimation of the inundation area by the
model compared with the satellite imagery. The ETS varied within the range of 0.03 to 0.39 (1
indicating a perfect score and 0 indicating no skill), with a mean of 0.14. This indicates poor overall
matching between the simulated and the observed inundation. In general, the simulated
inundation was greater than that identified in the satellite images for all flood events (FB > 1). The
FB varied within a range of 0.14 to 4.24 (range 0 to ∞, 1 indicating a perfect score), with a mean of
1.42, that there was generally overprediction of inundation by the model, although there was
some underestimation as well.
Locations of poor fit generally coincided with complex anabranching networks. Closer inspection
of the satellite imagery for these locations revealed that they often did not display flooding of
these anabranches. The inability of MODIS to capture inundation in narrow floodplains has been
reported for the Fitzroy catchment in WA (Karim et al., 2011) and for other catchments in
northern Australia (Ticehurst et al., 2013). Furthermore, MODIS regularly falsely identifies cloud
shadow as inundation, which is particularly an issue when using imagery with high (up to 80%)
cloud cover.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 4-12 Comparison of (MODIS and Sentinel) satellite–based inundation maps with hydrodynamic model results
for the Southern Gulf catchments
Table 4-5 Detection statistics for the Landsat, MODIS and Sentinel images considered in the analysis for the
Southern Gulf hydrodynamic model calibration
Percentage available refers to pixels other than cloud/null (including inundated and non-inundated pixels) within the
hydrodynamic model domain. ETS stands for Equivalent Threat Score (a measure of overall performance), POD for the
Probability Of Detection, FAR for the False Alarm Ratio and FB for Frequency Bias.
SATELLITE
DATE
% AVAILABLE
ETS
POD
FAR
FB
MODIS
9/01/2016
84
0.07
0.32
0.91
3.61
MODIS
10/01/2016
26
0.06
0.32
0.92
4.24
MODIS
11/03/2018
9
0.25
0.34
0.47
0.65
MODIS
12/03/2018
64
0.05
0.08
0.74
0.32
MODIS
13/03/2018
91
0.06
0.11
0.84
0.73
MODIS
16/03/2018
90
0.06
0.12
0.86
0.84
MODIS
18/03/2018
88
0.03
0.11
0.96
2.57
MODIS
9/02/2019
60
0.14
0.18
0.49
0.36
MODIS
10/02/2019
87
0.21
0.31
0.49
0.60
MODIS
11/02/2019
92
0.16
0.22
0.50
0.44
Landsat
14/02/2019
56
0.07
0.12
0.34
0.17
MODIS
15/02/2019
92
0.08
0.17
0.85
1.16
MODIS
16/02/2019
80
0.05
0.15
0.91
1.71
MODIS
19/02/2019
50
0.04
0.13
0.91
1.38
Sentinel-1
1/03/2023
54
0.06
0.08
0.38
0.14
Sentinel-1
6/03/2023
62
0.25
0.54
0.60
1.34
MODIS
14/03/2023
52
0.39
0.56
0.33
0.84
MODIS
15/03/2023
59
0.29
0.49
0.52
1.03
MODIS
16/03/2023
91
0.27
0.40
0.44
0.72
MODIS
19/03/2023
19
0.21
0.40
0.64
1.11
MODIS
25/03/2023
90
0.08
0.43
0.90
4.24
MODIS
26/03/2023
93
0.10
0.44
0.87
3.39
Landsat
28/03/2023
45
0.21
0.38
0.62
1.00
4.8 Summary
Intense seasonal rains from monsoonal bursts and tropical cyclones in the period of November to
March create flooding in parts of the Southern Gulf catchments and inundate large areas of
floodplains, mostly in the downstream reaches between the Nicholson, Gregory and Leichhardt
rivers. The floodplains along the Alexandra River are also heavily flooded. During large events (e.g.
2023), flooding is widespread on both sides of the Albert River, including at Burketown. The Lawn
Hill Creek and Gregory Rivers burst their banks, and the inundation area is large, especially at the
junction. There are large tidal floods in between the Nicolson and the Albert rivers, as well as
between the Albert and the Leichhardt rivers, where there is regular inundation by tides and a
combination of tidal and river flow (Figure 2-1 and Figure 4-12).
Flood inundation maps were produced using Landsat, MODIS and Sentinel imagery. MODIS
imagery of 500 m resolution at1-day intervals was acquired from November 2001 to March 2023
and processed using the OWL algorithm. Landsat imagery of 30 m resolution at 16-day intervals
was acquired from 2001 to 2023 and processed using the NDWI algorithm. A total of 21 Landsat
images, 61 MODIS images, 14 Sentinel-1 images and 24 Sentinel-2 images were selected for
processing for the hydrodynamic model calibration. Event-based inundation maps were produced
for individual floods, and composite flood maps were produced by combining all images. Cloud
cover was a challenge when producing event-based inundation maps. Both Landsat and MODIS
show inconsistencies in the spatial flood extent due to the limited number of cloud-free
observations. In addition, the inability of MODIS to capture inundation in narrow floodplains has
been reported for the Fitzroy catchment (Karim et al., 2011) and for other catchments in Australia
(Ticehurst et al., 2013).
Inundations on the floodplains of the Gregory, Nicholson, Leichhardt and Albert rivers and their
tributaries, covering an area of 17,130 km2, were modelled for five flood events ranging from an
AEP of 1 in 2 to an AEP of 1 in 38. Inundation is primarily driven by high flows down the Nicholson,
Gregory and Leichhardt rivers and to a lesser extent down the Alexandra River and Lawn Hill
Creek. Flooding is widespread in the downstream reaches of the major rivers near Burketown and
at the junction of Gregory River and Lawn Hill Creek. Most rivers in the Southern Gulf catchments
consist of a network of braided channels that produce large inundation during floods.
A two-dimensional hydrodynamic model (MIKE 21 FM) was used to simulate flood inundation. The
model was calibrated for the 2005, 2016, 2018, 2019 and 2023 flood events using inundation maps
derived from satellite imagery and gauged stage height data. The models were calibrated primarily
by adjusting the roughness coefficient and the infiltration rate.
Compared with the Landsat and MODIS inundation maps, the hydrodynamic model captured the
overall inundation patterns better along the main river channels and their tributaries. However,
the detection statistics showed that the cell-to-cell matching of the model-generated data against
the observed satellite data was overall poor, largely due to the inability of MODIS to detect
inundation of narrow floodplains. The detection metrics suggest that there is overestimation of
inundation area by the hydrodynamic model, especially during receding floods, as well as a
general misalignment of inundation patterns. The locations of poor fit generally coincided with
complex anabranching rivers. Closer inspection of the satellite imagery in these locations revealed
that it often does not display flooding of these anabranches. The inability of MODIS to capture
inundation in narrow floodplains has been reported for the Fitzroy catchment in WA (Karim et al.,
2011) and for other catchments in northern Australia (Ticehurst et al., 2013). Furthermore, MODIS
regularly falsely identifies cloud shadow as inundation, which is particularly an issue when using
imagery with high (up to 80%) cloud cover. The hydrodynamic model has some limitations.
However, lack of good-quality satellite images and gauged data, which restricts rigorous
calibration of the model results. Moreover, there are uncertainties in the river model simulations
for inflows to the hydrodynamic model domain.
5 Flood modelling under future climate and
development scenarios
5.1 Introduction
Rising global air temperatures are likely to be accompanied by changes in the intensity and
patterns of rainfall in Australia. The Australian Academy of Science released a report (Australian
Academy of Science, 2021) stating that current emissions trajectories will likely result in Australia
experiencing a 3 °C temperature increase by 2100. McJannet et al. (2023) found that GCMs
indicated changes in rainfall across the Victoria, Roper and Southern Gulf catchments. These
changes in rainfall are usually amplified in runoff. Consequently, increases in global temperatures
may be accompanied by changes in the extent and patterns of flood inundation.
In addition, the development of the surface water resources for irrigated agriculture in the highly
seasonal streamflow regime prevailing in the Southern Gulf catchments is likely to require some
degree of storage and river regulation. Surface water storage options have been evaluated at
several hypothetical dam locations in the Southern Gulf catchments (Yang et al., 2024). Flood
waters stored during the wet season and their gradual release during the dry season will modify
the timing and magnitude of floods and the subsequent inundation of floodplains. The impacts of
water harvesting during high-flow events were also evaluated during the flood study.
To explore how flood characteristics may change under projected future climate and hypothetical
development scenarios, a series of simulation experiments or scenarios were devised. Due to the
long run times of the hydrodynamic model, it was only possible to explore a limited number of
scenarios. Hence, scenarios were selected to enable general conclusions about likely impacts on
floodplain inundation. A summary of the future climate and development scenarios undertaken in
the Assessment area is presented in Table 5-1.
Table 5-1 Summary of selected future climate and development scenarios
For more information on this figure, table or equation please contact CSIRO on enquiries@csiro.au
5.2 Future climate scenarios
GCMs are an important tool for simulating global and regional climate. To assess the level of
uncertainty in the range of future runoff projections, future climate projections from a large range
of archived GCM simulations were downloaded from the Coupled Model Intercomparison Project
6 (CMIP6) website (https://pcmdi.llnl.gov/CMIP6/). Of the 92 available GCMs, 32 included the
rainfall, temperature, solar radiation, and humidity data required for the Australian Water
Resource Assessment Landscape model (AWRA-L) and AWRA-R hydrological modelling. The IPCC in
its Sixth Assessment Report (AR6) presented five different climate scenarios based on a SSP for the
future (IPCC, 2022). For the Assessment, SSP2-4.5 was used to investigate the sensitivity to
changes in rainfall and PE of streamflow at approximately the year 2060 (McJannet et al., 2023).
Under SSP2-4.5, emissions rise slightly before declining after 2050, but they do not reach net zero
by 2100. At approximately 2060, SSP2-4.5 is representative of a 1.6 °C temperature rise relative to
a time slice centred around 1990.
GCMs provide information at a resolution that is too coarse to be used directly in catchment-scale
hydrological modelling. Hence, an intermediate step is generally performed: the broad-scale GCM
outputs are transformed to catchment-scale variables. For this reason, and due to the scale of the
catchments being assessed (which makes it resource-intensive to undertake dynamic or statistical
downscaling), a simple scaling technique – the pattern scaling (PS) method (Chiew et al., 2009) –
was adopted. The seasonal PS method employed used output from the 32 GCMs to scale the 133-
year historical daily rainfall, temperature, radiation and humidity sequences (i.e. SILO climate
data) to construct the 32 by 133-year sequences of future daily rainfall, temperature, radiation
and humidity. The method is described in the companion technical report on future climate across
the Victoria, Roper and Southern Gulf catchments (McJannet et al., 2023).
The resulting percentage changes in rainfall and PE spatially averaged across the Southern Gulf
catchments under SSP2-4.5 at approximately 2060 for each GCM are shown in Figure 5-1. As
outlined by McJannet et al. (2023), scenarios Cwet and Cdry were selected to represent the range
of projections from the 32 GCMs shown in Figure 5-1. They were selected based on the 10th and
90th percentile exceedance changes in rainfall. That is, the global climate model – pattern scaling
(GCM-PS) time series derived from the GISS-E2-1-G (3rd-ranked GCM) and CMCC-ESM2 GCMs
(29th-ranked GCM) were used for the Cdry and Cwet scenarios, respectively. The calibrated river
model was used to simulate the river flow at the boundary of the Southern Gulf hydrodynamic
model domain under Cdry and Cwet scenarios (Gibbs et al., 2024b).
For more information on this figure, please contact CSIRO on enquiries@csiro.au
Figure 5-1 Percentage change in mean annual rainfall and potential evaporation under Scenario C relative to
Scenario A
Simple scaling of rainfall and potential evaporation have been applied to global climate model output. GCMs are
ranked by increasing rainfall.
Figure 5-2 shows simulated monthly streamflow (catchment mean runoff, used as an input to the
hydrodynamic model) under scenarios A, Cdry and Cwet from 2000 to 2023 (the hydrodynamic
model was calibrated for flood events within this period). Under scenario Cwet, the mean annual
catchment streamflow increased by 25%, 21% and 19% for the Nicholson, Gregory and Leichhardt
rivers, respectively, and for the wet season (December to March) these increases were 24%, 21%
and 17%, respectively. Under Scenario Cdry, the mean annual streamflow decreased by 33%, 26%
and 31% for the Nicholson, Gregory and Leichhardt rivers, respectively, and for the wet season
(December to March) these reductions were 32%, 25% and 29%, respectively.
For more information on this figure, please contact CSIRO on enquiries@csiro.au
Figure 5-2 River flow under Cdry and Cwet scenarios relative to Scenario A (Baseline) (for 2000 to 2023) at the
boundary of Southern Gulf catchments hydrodynamic model, gauge 91210104 on the Gregory River at Gregory,
91210703 on the Nicholson River at Connolly’s Hole and 91390002 on the Leichhardt River at Lorraine
Figure 5-3 compares the river flow under the current and future climate for the major rivers in the
Southern Gulf catchments, used as inflows to the hydrodynamic model for flood simulation
(Section 4.4). It shows relatively large changes in peak and total discharge under scenarios Cwet
and Cdry relative to Scenario A for each event. Under Scenario Cdry, the peak streamflow
decreased by 21%, 43% and 17% for the Gregory, Nicholson and Leichhardt rivers, respectively, for
the 2016 flood, and the respective reductions were 34%, 31% and 38% for the 2023 flood. Under
Cwet, the peak streamflow increased by 13%, 40% and 11% for the Gregory, Nicholson and
Leichhardt rivers, respectively, for the 2016 flood, and the respective increases were 46%, 28%
and 33% for the 2023 flood. For all three river systems, the increases and decreases in total event
flow were similar to the maximum flow, but slightly less in magnitude.
For more information on this figure, please contact CSIRO on enquiries@csiro.au
Figure 5-3 Simulated aggregated streamflow used as inflows in the hydrodynamic model (91210104 on the Gregory
River, 91210703 on the Nicholson River and 91390002 on the Leichhardt River) for two different flood events – 2016
(AEP of 1 in 3) and 2023 (AEP of 1 in 38) – under scenarios A (Baseline), Cdry (future dry climate) and Cwet (future
wet climate)
5.3 Potential development scenarios
5.3.1 INSTREAM DAMS
Potential dams identified in the companion technical report on surface water storage (Yang et al.,
2024) were located outside the hydrodynamic model domain. Hence, to explore the impact of
instream dams and water harvesting on flood inundation downstream, river system models were
configured for various scenarios (refer to the companion technical report on river model
calibration and scenario analysis, Gibbs et al., 2024b), and the resulting river system model output
provided for the upstream boundary of the hydrodynamic model domains. Only streamflows at
the upstream boundaries of the hydrodynamic model domains were updated; the remaining input
datasets and boundary conditions in the calibrated hydrodynamic models remained unchanged.
Several potential dam sites were investigated in the Southern Gulf catchments for irrigation and
hydro-electric power generation (Table 5-2). However, only three of the potential dams (Dam 1,
Dam 3 and Dam 28) were considered for inundation impact assessment. The capacities of these
dams at full supply level are 441, 1403 and 716 GL, respectively. At the beginning of each flood
event, the dams were set to 50% full.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-4 Locations of existing water users under scenarios A and C, and additional hypothetical options considered
under scenarios B and D
AMTD = Adopted Middle Thread Distance. FSL = full supply level. Qld = Queensland.
Table 5-2 Surface area and reservoir capacity at full supply level (FSL) of the short-listed hypothetical dams in the
Southern Gulf catchments
NAME
LOCATION
FSL
(mEGM96)
MODEL NODE
RESERVOIR
SURFACE
AREA
(ha)
RESERVOIR
CAPACITY
(GL)
Dam 1
Gregory River at AMTD 174 km
145
9121050
7,090
441
Dam 3
Nicholson River AMTD 198 km
108
9121070
12,417
1403
Dam 28
Gunpowder Creek AMTD 66 km
186
9130030
4,021
716
Dam 165
Mistake Creek AMTD 60 km
149
9130080
2,320
158
Dam 206
Gold Creek AMTD 58 km
84
9121097
756
119
Dam 275
Ewen Creek AMTD 6 km
217
9130040
2,515
245
†AMTD = Adopted Middle Thread Distance.
The Southern Gulf catchments feature a highly seasonal climate, with the majority of streamflow
occurring in the months December to March (Gibbs et al., 2024a). The effects of instream dams
were simulated in the catchment at various locations for 133 years (1890 to 2022). One
characteristic of the dams used for simulated irrigation supply was the annual cycle of filling across
the wet season and of emptying across the dry season, due to irrigation diversion and evaporation
from the dams. This pattern can be seen in the plot of mean monthly dam storage at three sites in
the Southern Gulf catchments (Figure 5-5), noting that there can be substantial variability in the
storage level from year to year.
For more information on this figure, please contact CSIRO on enquiries@csiro.au
Figure 5-5 Mean monthly dam storage in different months at three dam sites in the Southern Gulf catchments
As a percentage of total dam capacity, dam storage reaches its peak towards the end of the wet
season in April and empties across the dry season with irrigation diversion. With regard to floods,
this means that, generally, there would be reduced capacity for mitigation of flood events later in
the wet season, particularly in March and April; maximum flood mitigation would be expected to
be possible for events in the late dry/early wet season (November and December).
These patterns of storage have consequences for hydrodynamic simulation of instream dam
scenarios. In particular, simulated flood events later in the wet season are less likely to generate
substantial change in the estimated flooded area due to the antecedent storage. Ideally,
hydrodynamic models would be run for the entire 133-year period, as for the Southern Gulf
catchment model. This would give a better understanding of the effects of dams on flood
mitigation. However, due to the very high computational demand of hydrodynamic models, only
selected flood events can be simulated, which are affected by the particular antecedent conditions
prior to those events. To counter this, two approaches were taken. First, for each simulated event,
the reservoir storage was reduced to 50% immediately prior to the event. Second, a regression
between river flow at multiple locations and the estimated flooded area was derived. This was
then used to estimate the total flooded area for various scenarios across the entire time series of
the river model. This was denoted the flood ‘emulator’ and allowed for a more balanced
assessment of the relative effects of each scenario on flooded area, since it enabled a daily
estimate of the flooded area that took into account any antecedent effects. More detailed
information on the emulator can be found in Section 5.5.
5.3.2 WATER HARVESTING
Water harvesting was implemented at five locations for an annual limit of 150 GL withdrawal, with
a pump rate of 600 ML/day (Gibbs et al., 2024b). For the Water Harvesting scenario, extractions
were delayed so as to commence at the start of the hydrodynamic simulation at all five nodes. The
pumps were operated assuming a pump start threshold of 600 ML/day, selected to prevent
impacts to downstream users. A system target across all five nodes was set at 150 GL, and the
pumping rate was adjusted to enable this target volume to be extracted in 20 days. The volume
and pump capacity were selected, because they could be physically supported with an annual
reliability of 75%, taking into account soil and water limitations, mitigating impacts on existing
users, and providing an annual DCFR that protects early-season flows (Gibbs et al., 2024b).
5.4 Floodplain inundation scenario analysis
Evaluation of hydrodynamic modelled scenarios was undertaken by comparing future
development (Scenario B), future climate (Scenario C) and future climate and development
(Scenario D) scenarios with the baseline simulation (Scenario A). Two types of evaluations were
performed across the hydrodynamic model domains for each scenario (Table 5-1):
• a spatial comparison using maps of percentage inundation frequency (the ratio of the number of
times a pixel was inundated to the entire duration of the simulation), maximum inundation
extent, and inundated depth at maximum inundation extent
• a time-series comparison of the inundated area.
5.4.1 SCENARIO B CURRENT CLIMATE AND INSTREAM DAMS
Figure 5-6 shows the maximum inundation extent as well as the spatial variation in inundation
frequencies for 2016 (AEP of 1 in 3) and 2023 (AEP of 1 in 38) flood events. The dams decreased
the inundation extent and frequency, but the effects were relatively small in the model domain.
Similarly to inundation frequency, the effects on spatial inundation depth were also small (Figure
5-7). However, changes in inundation area due to the dams were noticeable for both the 2016 and
2023 flood events (Figure 5-8). The maximum inundated area for the 2016 event (AEP of 1 in 3)
was 999.3 km2 under Scenario A (Baseline) and 662.0 km2 under Scenario B (3-dams). This
represents a decrease in inundated area of approximately 33.8%. The maximum inundated area
for the 2023 event (AEP of 1 in 38) was 5983.3 km2 under Scenario A (Baseline) and 5677.6 km2
under Scenario B (3-dams), representing a decrease of approximately 5.1%. The smaller volume of
floodwater in the 1 in 3 AEP event during 2016 caused a larger relative impact. The larger relative
impact may also have been due to the difference in the antecedent conditions at the beginning of
the two events.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-6 Percentage inundation frequency in the Southern Gulf hydrodynamic model domain under scenarios A
(Baseline) and B (3-dams)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-7 Depth at maximum inundation extent in the Southern Gulf hydrodynamic model domain under scenarios
A (Baseline) and B (Dam)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-8 Comparison of inundated area (in square kilometres) under scenarios A (Baseline) and B (Dam) in the
Southern Gulf hydrodynamic model
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
5.4.2 SCENARIO B CURRENT CLIMATE AND WATER HARVESTING
Figure 5-9 shows the maximum inundation extent as well as the spatial variation in inundation
frequencies for the 2016 (AEP of 1 in 3) and 2023 (AEP of 1 in 38) flood events. Water extraction
reduced the flow in the river and produced less inundation. In general, the effect of the water
harvesting was small for the maximum inundation extent and the inundation frequency. As for
inundation frequency, the effect on inundation depth was very small for both the 2016 and 2023
events (Figure 5-10). The changes in inundation areas due to water harvesting were noticeable for
the 2016 flood event, but for the 2023 event the effects were minimal (Figure 5-11). The impacts
of water harvesting on flood characteristics over the hydrodynamic model domain were larger for
the smaller event than those for the larger event. The maximum inundated area for the 2016
event (AEP of 1 in 3) under Scenario A was 999.3 km2 and 945.3 km2 under Scenario B (Water
Harvesting of 150 GL). This represents a decrease in inundated area of approximately 5.4%. The
maximum inundated area under Scenario A for the 2023 event (AEP of 1 in 18) was 5983.3 km2
and 5934.1 km2 under Scenario B (Water Harvesting of 150 GL), representing a decrease of only
approximately 0.8%. As expected, the impacts were relatively large for the smaller flood event,
given the same amount of water was extracted for both flood events.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-9 Percentage inundation frequency in the Southern Gulf catchments under scenarios A (Baseline) and B
(Water Harvesting of 150 GL)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-10 Depth at maximum inundation extent in the Southern Gulf hydrodynamic model domain under
scenarios A (Baseline) and B (Water Harvesting of 150 GL)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-11 Comparison of inundated area (in square kilometres) in the Southern Gulf hydrodynamic model domain
under scenarios A (Baseline) and B (Water Harvesting of 150 GL)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
5.4.3 SCENARIO C FUTURE CLIMATE SCENARIOS
Figure 5-12 shows the difference between scenarios A, Cdry and Cwet in terms of percentage
inundated frequency and maximum inundation extent. The results show decreases in inundation
frequency and inundation extent under Scenario Cdry relative to Scenario A. Similarly, a significant
increase can be seen under Scenario Cwet relative to Scenario A. Similar changes are noticed for
spatial inundation depth (Figure 5-13). The maximum inundated area under Scenario Cdry and
Cwet for the 2016 event (AEP of 1 in 3) were 689.1 km2 and 1186.7 km2, respectively, a reduction
of approximately 31.0% under Cdry and an increase of approximately 18.8% under Cwet. For the
2023 event (AEP of 1 in 38), the maximum inundated area under scenarios Cdry and Cwet were
4790.0 km2 and 6927.4 km2, respectively, indicating a reduction of approximately 19.9% under
Scenario Cdry and an increase of approximately 15.8% under Scenario Cwet. Although the relative
increase in inundation area for the 2016 event (~18.8%) was higher than for the 2023 event
(~15.8%), the absolute increase in inundation area was higher for the 2023 flood (944.1 km2)
compared with the 2016 flood (187.4 km2).
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-12 Percentage inundated frequency in the Southern Gulf hydrodynamic model domain under scenarios A
(Baseline) and C (Future Climate)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-13 Depth at maximum inundation extent in the Southern Gulf hydrodynamic model domain under
scenarios A (Baseline) and C (Future climate)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
The time series in Figure 5-14 shows the large differences between the climate scenarios in the
inundated area that occurred under scenarios Cdry and Cwet relative to Scenario A for both
events. The differences were largest at the times of peak inundation under both the Cdry and
Cwet scenarios. Under Cdry, the peak in inundated area decreased by approximately 29.6% and
approximately 19.8% for the 2016 and 2023 flood events, respectively. Under Scenario Cwet, the
peak in inundated area increased by approximately 17.9% and approximately 15.6% for the 2016
and 2023 flood events, respectively. Table 5-3 summarises the inundated area under Scenario C
and the changes in maximum and mean inundation areas relative to Scenario A.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-14 Comparison of inundated area (in square kilometres) (left) in the Southern Gulf hydrodynamic model
domain under scenarios A (Baseline) and C (Future Climate)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
Table 5-3 Comparison of the inundated area and associated changes under Scenario C (Future Climate) relative to
Scenario A (Baseline)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
* decrease.
2016 FLOOD
SCENARIO A
2023 FLOOD
SCENARIO A
2016 FLOOD
Cdry
2023 FLOOD
Cdry
2016 FLOOD
Cwet
2023 FLOOD
Cwet
Maximum inundated
area (km2)
999
5983
689
4790
1187
6927
% change in maximum
inundated area
–
–
31.0*
19.9*
18.8
15.8
Mean inundated area
(km2)
648
2858
486
2288
767
3342
% change in mean
inundated area
–
–
25.0*
19.9*
18.4
16.9
5.4.4 SCENARIO D DRY CLIMATE AND INSTREAM DAMS
The maps of percentage inundated frequency (Figure 5-15) and depth at maximum inundation
(Figure 5-16) show that the dams combined with future dry climate decreased inundation
(frequency, extent and depth) for both the 2016 (AEP of 1 in 3) and 2023 (AEP of 1 in 38) flood
events. As expected, the combined impacts of the future dry climate and 3-dams on the
inundation extent and frequency were significant, although the climate appears to have been the
main driver of the differences. The maximum inundated area under scenarios A (Baseline) and D
(Dry Climate and Dam) for the 2016 event was 999.3 km2 and 375.7 km2, respectively (Figure
5-17). This represents a decrease in inundated area of approximately 62.4%. The maximum
inundated area under scenarios A (Baseline) and D (Dry Climate and Dam) for the 2023 event was
5983.3 km2 and 3999.6 km2, respectively, representing a decrease of approximately 33.2%.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-15 Percentage inundation frequency in the Southern Gulf hydrodynamic model domain under scenarios A
(Baseline) and D (Dry Climate and Dam)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-16 Depth at maximum inundation extent in the Southern Gulf hydrodynamic model domain under
scenarios A (Baseline) and D (Dry Climate and Dam)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-17 Comparison of inundated area (in square kilometres) in the Southern Gulf catchments under scenarios A
(Baseline) and D (Dry Climate and Dam)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
5.4.5 SCENARIO D DRY CLIMATE AND WATER HARVESTING
Maps of percentage inundated frequency (Figure 5-18) and depth at maximum inundation (Figure
5-19) show that water harvesting under a future dry climate scenario substantially decreased
inundation for the 2016 (AEP of 1 in 3) and 2023 (AEP of 1 in 38) events. The maximum inundated
area under scenarios A (Baseline) and D (Dry Climate and Water Harvesting of 150 GL) for the 2016
event was 999.3 km2 and 678.3.0 km2, respectively. This represents a decrease in inundated area
of approximately 32.1%. The maximum inundated area for the 2023 event was 5983.3 km2 and
4747.7 km2 under scenarios A (Baseline) and D (Dry Climate and Water Harvesting of 150 GL),
respectively, representing a decrease of approximately 20.7%. The mean inundations under
Scenario D were 481.2 km2 and 2269.1 km2 for the 2016 and 2023 floods, respectively, and
respectively 648.1 km2 and 2858.6 km2 under Scenario A, representing an approximately 25.8%
reduction for the 2016 flood and an approximately 20.6% reduction for the 2023 flood. As seen
earlier, the relative impacts were higher for smaller events (Figure 5-20).
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-18 Percentage inundation frequency in the Southern Gulf hydrodynamic model domain under scenarios A
(Baseline) and D (Dry Climate and Water Harvesting)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-19 Depth at maximum inundation extent in the Southern Gulf hydrodynamic model domain under
scenarios A (Baseline) and D (Dry Climate and Water Harvesting)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-20 Comparison of inundated area (in square kilometres) in the Southern Gulf catchments under scenarios A
(Baseline) and D (Dry Climate and Water Harvesting)
The 2016 flood event had an AEP of 1 in 3, and the 2023 flood event had an AEP of 1 in 38.
5.5 Floodplain inundation emulator
5.5.1 DEVELOPMENT OF EMULATOR
An emulator was developed for the Assessment area by relating hydrodynamic model–simulated
inundation area to flood discharge through a regression model. Inflows to the hydrodynamic
model domain through the three major rivers (Gauge 912116 on the Nicholson River, 912101 on
the Gregory River and 913900 on the Leichhardt River) were aggregated in order to produce a
time series of flood discharge. Two measures of flow data were compared with the flooding extent
data. The first was total volume over the event period, and the second was peak flow during the
event period. These were graphed against maximum inundation area. In both cases, there was a
clear relationship between flow and inundation area. Of the two measures, peak flow during the
event period was determined to give the best relationship and chosen as the measure for use in
the regression model. To ensure the calibration data covered a large range, all five calibration
events, plus Scenario B (Water Harvesting), Scenario B (3-dams), Scenario Cdry (Future Dry
Climate) and Scenario Cwet (Future Wet Climate) estimates of the flooded area for the two events
were utilised.
A power curve in the form of ð´ð´=ð‘ð‘ð‘„ð‘„ð‘ð‘ was tested and found to suitable for the Southern Gulf
catchments. The parameters were optimised using the following objective function:
ð‘‚ð‘‚ð‘‚ð‘‚= Σ|ð´ð´ô€· ð‘–ð‘–−ð´ð´ð‘–ð‘–|ð‘›ð‘›ð‘–ð‘–
ð‘›ð‘› (1)
where ð´ð´Ì‚
ð‘–ð‘– is the estimated flooded area, ð´ð´ð‘–ð‘– is the hydrodynamic model–simulated flooded area
and ð‘›ð‘› is the number of events.
The calibrated emulator had the following relationship between flow and inundation area:
ð‘¨ð‘¨ô€·¡
= 8.6776∗(ð‘¸ð‘¸1+ð‘¸ð‘¸2+ð‘¸ð‘¸3)0.6633 (2)
where ð‘¨ð‘¨ ô€·¡is the estimated time series of the flooded area (km2), and ð‘¸ð‘¸1, ð‘¸ð‘¸2 and ð‘¸ð‘¸3are the time
series of flow (m3/second) at nodes 9121160, 9121010 and 9139000.
Overall, the emulator produced a good estimate of the hydrodynamic model–simulated
inundation area, with an ð‘…ð‘…2 value of 0.97 (Figure 5-21). It can be seen that the calibration data
covers a large range of data, although there are no data points for the area range of 2000 to
4000 km2, and estimates in this range will be more uncertain.
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-21 Relationship between flood discharge and inundation area for the Southern Gulf catchments
5.5.2 ESTIMATION OF INUNDATION AREA USING THE EMULATOR
The flood emulator was applied to time-series outputs from the river model for a range of
scenarios. Figure 5-22 shows the distribution of annual maximum inundation area in the period
1890 to 2022 (133 years) for various climate and development scenarios. Scenario B (Water
Harvesting) has little effect on maximum inundation area relative to Scenario A (Baseline), because
this method relies on pumps to extract water from the river, which even with an assumed high
capacity will be far lower than peak flows in most instances. The effects of dams on inundation are
higher relative to water harvesting, because dams reduce and delay the peak by storing water
during the high flows, depending upon the antecedent conditions. It is important to note that dam
volumes, and thus their capacities to reduce peak flows, vary depending on the timing of the wet
season and how dry the catchment has been in the preceding years. The flooded area reductions
are obvious under the Cdry and combined Cdry and dams scenarios, since inflows will be reduced.
Scenario D (Cdry-Dam) for dry climate and the three instream dams has the lowest distribution of
the flooded area estimates of all scenarios, essentially combining the effects of the dams and of
lower streamflow (Table 5-4).
For more information on this figure please contact CSIRO on enquiries@csiro.au
Figure 5-22 Estimated annual maximum flooded area for the various climate and development scenarios for the
Southern Gulf catchments
Table 5-4 Emulator estimates of the flooded area for 133 years of simulation
Scenario
Mean annual maximum flooded area
(km2)
Maximum flooded area
(km2)
A (Baseline)
1132
3659
B (Dam)
785
2968
B (Water Harvesting)
1086
3589
Cdry (Dry Climate)
911
3108
Cwet (Wet Climate)
1286
4041
D (Cdry-Dam)
615
2578
6 Summary
This part of the Assessment had two major components: calibration of a two-dimensional flexible-
mesh hydrodynamic model (MIKE 21 FM), followed by scenario modelling under projected dry and
wet future climates (Cdry and Cwet) and hypothetical developments (three instream dams and
Water Harvesting). The outputs from the hydrodynamic modelling were used to:
• identify areas susceptible to seasonal flooding under the historical climate and current
development
• predict changes in inundation across the floodplains under future dry and wet climate scenarios
• predict how dam storages and water harvesting would alter the inundation dynamics across the
floodplain
• assess the combined effects of future climate and development scenarios (Dams and Water
Harvesting) on inundation extent and depth.
Observed discharge and stage height data were obtained from the Water Monitoring Information
portal of the Queensland Government, and tide data were obtained from the Bureau of
Meteorology. A flexible-mesh floodplain hydrodynamic model was configured for the middle and
lower reaches of the Nicholson, Gregory and Leichhardt rivers and their tributaries. Sacramento
rainfall-runoff simulations and discharge data from AWRA-R simulations were used as input at the
hydrodynamic model boundaries. Flood inundation maps for individual flood events were
produced using satellite (Landsat, MODIS, Sentinel-1 and Sentinel-2) imagery. Composite flood
maps were also produced by combining all images to delineate the maximum flood extent in the
catchment. These maps and the observed water level at Floraville on the Leichhardt River were
used to calibrate the Southern Gulf hydrodynamic model. The calibrated hydrodynamic model was
used to simulate the impacts of future climate and future developments on inundation extent,
frequency and depth.
The hydrodynamic model was calibrated for the 2005 (AEP of 1 in 2), 2016 (AEP of 1 in 3), 2018
(AEP of 1 in 5), 2019 (AEP of 1 in 10) and 2023 (AEP of 1 in 38) flood events, and two of these flood
events (2016 and 2023) were used for scenario modelling. The model was calibrated primarily by
adjusting the roughness coefficient and the infiltration rate. While a good match was attained for
the flood peaks, there were differences in the rising and falling limbs of the flood hydrograph. In
general, model predictions were found to be more accurate for large floods.
Comparison with the Landsat, MODIS and Sentinel inundation maps revealed that the
hydrodynamic model captured overall inundation patterns along the Nicholson, Gregory,
Leichhardt and Albert rivers. However, the detection statistics showed that the cell-to-cell
matching against observed satellite data was poor, largely due to the inability of MODIS to detect
inundation of narrow floodplains. The model overpredicted inundation area, especially during a
receding flood. Locations of poor fit generally coincided with complex anabranching, for example
along the Gregory River and Beames Brook. Closer inspection of satellite imagery in these
locations revealed that it often does not display flooding of these anabranches. The inability of
MODIS to capture inundation in narrow floodplains has been reported in the Fitzroy catchment in
WA (Karim et al., 2011) and in other catchments in northern Australia (Ticehurst et al., 2013).
Furthermore, MODIS regularly falsely identifies cloud shadow as inundation, which is particularly
an issue when using imagery with high (up to 80%) cloud cover. The hydrodynamic model has
some limitations, and lack of good-quality satellite images restricts rigorous calibration of the
model results. Moreover, there are uncertainties in the river model simulations for inflow
boundaries and locally generated runoff.
Future climate scenario modelling showed marked changes in inundation areas relative to those
under Scenario A, with the major differences being observed under scenarios Cdry and Cwet. For
example, under the Cdry scenario, the maximum inundation area decreased by 31.0% and 20.6%
for the 2016 (AEP of 1 in 3) and 2023 (AEP of 1 in 38) events, respectively. For the Cwet scenario,
the maximum inundation extent increased by 18.8% and 14.8% for the 2016 and 2023 events,
respectively.
The reduction in modelled maximum inundation extent under Scenario B (Dam) is significant. The
reductions in maximum inundation area were 33.8% and 5.9% for the 2016 and 2023 events,
respectively. However, under the Water Harvesting scenario, the changes were very small relative
to other scenarios. Under Scenario B (Water Harvesting), the reduction in maximum inundation
extent for the 2016 event (AEP of 1 in 3) was 5.4%, and it was only 0.9% for the 2023 event (AEP of
1 in 38). The reductions were much higher for the combined Dry Climate and Development
scenarios. For example, the reductions were 62.4% and 32.1% for the 2016 event for Cdry-Dam
and Cdry-Water Harvesting, respectively. The Development scenario modelling demonstrated that
these reductions in maximum inundation extent were largely dependent on the timing of the
events and the water storage capacity of the dams. It is important to note that dam volumes, and
thus their capturing capacities, vary depending on the timing of the wet season and how dry the
catchment has been in the preceding years.
A flood inundation emulator was developed by investigating the relationship between the
hydrodynamic model–simulated inundation area and the flood discharge. The emulator produced
a good estimate of the hydrodynamic model–simulated inundation area, except for a few outliers.
The flood emulator was applied to time-series outputs from the river model for various climate
and development scenarios. Under Scenario B Water Harvesting, there was little change in the
maximum inundation area relative to Scenario A, because even very high pump capacities would
be small relative to the rates of streamflow during floods. The effect of dams on inundation was
higher relative to Water Harvesting, because the dams reduce and delay the flood peak by storing
water during the peak flow. The reductions were much higher under the Cdry and the combined
Cdry and Dam scenarios.
References
Arcement GJ and Schneider VR (1989) Guide for selecting Manning’s roughness coefficients for
natural channels and floodplain. US Geological Survey, Virginia.
Arnell NW and Gosling SN (2016) The impacts of climate change on river flood risk at the global
scale. Climatic Change 134(3), 387–401. DOI: 10.1007/s10584-014-1084-5.
Arthington AH, Godfrey PC, Pearson RG, Karim F and Wallace J (2015) Biodiversity values of
remnant freshwater floodplain lagoons in agricultural catchments: evidence for fish of the
Wet Tropics bioregion, northern Australia. Aquatic Conservation: Marine and Freshwater
Ecosystems 25(3), 336–352. DOI: 10.1071/MF12251.
Australian Academy of Science (2021) The risks to Australia of a 3°C warmer world. Australian
Academy of Science. Available online:
<https://www.science.org.au/files/userfiles/support/reports-and-plans/2021/risks-australia-
three-deg-warmer-world-report.pdf
Bayley PB (1991) The flood pulse advantage and the restoration of river–floodplain systems.
Regulated Rivers: Research and Management 6, 75–86. DOI: 10.1002/rrr.3450060203.
Bayley PB (1995) Understanding large river floodplain ecosystems. Bioscience 45(3), 153–158. DOI:
10.2307/1312554.
Beighley RE, Eggert KG, Dunne T, He Y, Gummadi V and Verdin KL (2009) Simulating hydrologic and
hydraulic processes throughout the Amazon River Basin. Hydrological Processes 23(8), 1221–
1235. DOI: 10.1002/hyp.7252.
Bomers A, Schielen RMJ and Hulscher SJMH (2019) The influence of grid shape and grid size on
hydraulic river modelling performance. Environmental Fluid Mechanics 19(5), 1273–1294.
DOI: 10.1007/s10652-019-09670-4.
Budget Strategy and Outlook (2024) Budget 2024-25. A future made in Australia. Viewed 11
September 2024, https://budget.gov.au/content/factsheets/download/factsheet-fmia.pdf.
Bulti DT and Abebe BG (2020) A review of flood modeling methods for urban pluvial flood
application. Modeling Earth Systems and Environment 6(3), 1293–1302. DOI:
10.1007/s40808-020-00803-z.
Bunn SE, Thoms MC, Hamilton SK and Capon SJ (2006) Flow variability in dryland rivers: boom,
bust and the bits in between. River Research and Applications 22(2), 179–186. DOI:
10.1002/rra.904.
Charles S, Petheram C, Berthet A, Browning G, Hodgson G, Wheeler M, Yang A, Gallant S, Marshall
A, Hendon H, Kuleshov Y, Dowdy A, Reid P, Read A, Feikema P, Hapuarachchi P, Smith T,
Gregory P and Shi L (2016) Climate data and their characterisation for hydrological and
agricultural scenario modelling across the Fitzroy, Darwin and Mitchell catchments. A
technical report from the CSIRO Northern Australia Water Resource Assessment to the
Government of Australia. CSIRO, Australia
Chiew FHS, Teng J, Vaze J, Post DA, Perraud JM, Kirono DGC and Viney NR (2009) Estimating
climate change impact on runoff across southeast Australia: method, results, and
implications of the modeling method. Water Resources Research 45. DOI:
10.1029/2008wr007338.
Chormanski J, Miroslaw-Swiatek D and Michalowski R (2009) A hydrodynamic model coupled with
GIS for flood characteristics analysis in the Biebrza riparian wetland. Oceanological and
Hydrobiological Studies 38(1), 65–73. DOI: 10.2478/v10009-009-0004-x.
Chow VT (1959) Open channel hydraulics. McGraw-Hill International Edition, Singapore.
Department of State Development, Manufacturing, Infrastructure and Planning (2019) North west
Queensland economic diversification strategy 2019. Viewed 12 September 2024,
https://www.statedevelopment.qld.gov.au/regions/regional-priorities/a-strong-and-
prosperous-north-west-queensland/north-west-queensland-economic-diversification-
strategy.
DHI (2012) MIKE 21 Flow Model: scientific documentation. Danish Hydraulic Institute, Denmark.
DHI (2016) MIKE 21 flow model FM, Hydrodynamic Module, User Guide. Danish Hydraulic Institute
Water and Environment Pty Ltd, Hörsholm, Denmark.
<https://manuals.mikepoweredbydhi.help/2017/Coast_and_Sea/MIKE_FM_HD_2D.pdf>.
Dhu T, Dunn B, Lewis B, Lymburner L, Mueller N, Telfer E, Lewis A, McIntyre A, Minchin S and
Phillips C (2017) Digital earth Australia – unlocking new value from Earth observation data.
Big Earth Data, 1(1–2), 64–74. DOI: 10.1080/20964471.2017.1402490.
Doble RC, Crosbie RS, Smerdon BD, Peeters L and Cook FJ (2012) Groundwater recharge from
overbank floods. Water Resources Research 48(9). DOI: 10.1029/2011wr011441.
Dottori F, Szewczyk W, Ciscar JC, Zhao F, Alfieri L, Hirabayashi Y, Bianchi A, Mongelli I, Frieler K,
Betts RA and Feyen L (2018) Increased human and economic losses from river flooding with
anthropogenic warming. Nature Climate Change 8(9), 781–786. DOI: 10.1038/s41558-018-
0257-z.
Ebert EE, Janowiak JE and Kidd C (2007) Comparison of near-real-time precipitation estimates from
satellite observations and numerical models. Bulletin of the American Meteorological
Society 88(1), 47–68. DOI: 10.1175/BAMS-88-1-47.
Engeny (2020) Burketown flood risk management study. A report to Burke Shire Council from
Engeny Water Management. Burke Shire Council, Brisbane.
https://www.burke.qld.gov.au/downloads/file/514/burke-shire-planning-scheme-2020-pdf.
Frazier P and Page K (2009) A reach-scale remote sensing technique to relate wetland inundation
to river flow. River Research and Applications 25, 836–849. DOI: 10.1002/rra.1183.
Frazier P, Page K, Louis J, Briggs S and Robertson AI (2003) Relating wetland inundation to river
flow using Landsat TM data. International Journal of Remote Sensing 24(19), 3755–3770.
DOI: 10.1080/0143116021000023916.
Gallant J (2019) Merging lidar with coarser DEMs for hydrodynamic modelling over large areas. In:
23rd International Congress on Modelling and Simulation. Modelling and Simulation Society
of Australia and New Zealand, 1161–1166.
Gallant J, Wilson N, Dowling T, Read A and Inskeep C (2011) SRTM-derived 1 second Digital
Elevation Models Version 1.0 dataset. Geoscience Australia, Canberra.
https://data.gov.au/dataset/ds-ga-aac46307-fce8-449d-e044-00144fdd4fa6/details?q=>.
Gallardo B, Gascon S, Gonzalez-Sanchis M, Cabezas A and Comin FA (2009) Modelling the response
of floodplain aquatic assemblages across the lateral hydrological connectivity gradient.
Marine and Freshwater Research 60(9), 924–935. DOI: 10.1071/mf08277.
Gibbs M, Hughes J and Yang A (2024a) River model calibration for the Southern Gulf catchments. A
technical report from the CSIRO Southern Gulf Water Resource Assessment for the National
Water Grid. CSIRO, Australia.
Gibbs M, Hughes J, Yang A, Wang B, Marvanek S and Petheram C (2024b) River model scenario
analysis for the Southern Gulf catchments. A technical report from the CSIRO Southern Gulf
Water Resource Assessment for the National Water Grid. CSIRO, Australia.
Gladkova I, Grossberg MD, Shahriar F, Bonev G and Romanov P (2012) Quantitative Restoration for
MODIS Band 6 on Aqua. IEEE Transactions on Geoscience and Remote Sensing 50(6), 2409–
2416. DOI: 10.1109/Tgrs.2011.2173499.
Guerschman JP, Warren G, Byrne G, Lymburner L, Mueller N and Van-Dijk A (2011) MODIS-based
standing water detection for flood and large reservoir mapping: algorithm development and
applications for the Australian continent. CSIRO Water for a Healthy Country National
Research Flagship Report, Canberra.
Hawker L, Uhe P, Paulo L, Sosa J, Savage J, Sampson C and Neal J ( 2022) A 30 m global map of
elevation with forests and buildings removed. Environmental Research Letters 17(2) 024016.
DOI: 10.1088/1748-9326/ac4d4f.
Heiler G, Hein T and Schiemer F (1995) Hydrological connectivity and flood pulses as the central
aspects for the integrity of a river–floodplain system. Regulated Rivers: Research and
Management 11, 351–361. DOI: 10.1002/rrr.3450110309.
Horritt MS and Bates PD (2002) Evaluation of 1D and 2D numerical models for predicting river
flood inundation. Journal of Hydrology 268(1–4), 87–99.
IPCC (2022) Climate Change 2022: Impacts, adaptation, and vulnerability. Contribution of Working
Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change.
[Pörtner HO, Roberts DC, Tignor M, Poloczanska ES, Mintenbeck K, AlegrÃa A, Craig M,
Langsdorf S, Löschke S, Möller V, Okem A and Rama B (eds)] Cambridge University Press,
Cambridge, UK and New York, NY, USA.
https://report.ipcc.ch/ar6/wg2/IPCC_AR6_WGII_FullReport.pdf.
Junk WJ, Bayley PB and Sparks RE (1989) The flood pulse concept in river–floodplain systems. In:
Proceedings of the International Large River Symposium. Canadian Special Publications of
Fisheries and Aquatic Sciences 106, 110–127.
https://publications.gc.ca/site/eng/9.816457/publication.html
Karim F, Dutta D, Marvanek S, Petheram C, Ticehurst C, Lerata J, Kim S and Yang A (2015) Assessing
the impacts of climate change and dams on floodplain inundation and wetland connectivity
in the wet–dry tropics of northern Australia. Journal of Hydrology 522, 80–94. DOI:
10.1016/j.jhydrol.2014.12.005.
Karim F, Kinsey-Henderson A, Wallace J, Arthington AH and Pearson RG (2012) Modelling wetland
connectivity during overbank flooding in a tropical floodplain in north Queensland, Australia.
Hydrological Processes 26, 2710–2723. DOI: 10.1002/hyp.8364.
Karim F, Petheram C, Marvanek S, Ticehurst C, Wallace J and Gouweleeuw B (2011) The use of
hydrodynamic modelling and remote sensing to estimate floodplain inundation and flood
discharge in a large tropical catchment. In: Chan F, Marinova D and Anderssen RS (eds), 19th
International Congress on Modelling and Simulation. Modelling and Simulation Society of
Australia and New Zealand, Perth.
Kim B, Sanders BF, Schubert JE and Famiglietti JS (2014) Mesh type tradeoffs in 2D hydrodynamic
modeling of flooding with a Godunov-based flow solver. Advances in Water Resources 68,
42–61. DOI: 10.1016/j.advwatres.2014.02.013.
Kron W (2015) Flood disasters – a global perspective. Water Policy 17(S1), 6–24. DOI:
10.2166/wp.2015.001.
Kumar V, Sharma KV, Caloiero T, Mehta DJ and Singh K (2023) Comprehensive overview of flood
modeling approaches: a review of recent advances. Hydrology 10(7), 141. DOI:
10.3390/hydrology10070141.
Kvocka D, Falconer RA and Bray M (2015) Appropriate model use for predicting elevations and
inundation extent for extreme flood events. Natural Hazards 79(3), 1791–1808. DOI:
10.1007/s11069-015-1926-0.
Liu Q, Qin Y, Zhang Y and Li ZW (2015) A coupled 1D–2D hydrodynamic model for flood simulation
in flood detention basin. Natural Hazards 75(2), 1303–1325. DOI: 10.1007/s11069-014-1373-
3.
LWA (2009) An Australian handbook of stream roughness coefficients. Land and Water Australia,
Canberra.
Mackay C, Suter S, Albert N, Morton S and Yamagata K (2015) Large scale flexible mesh 2D
modelling of the Lower Namoi Valley. In: Floodplain Management Association National
Conference. Floodplain Management Australia, 1–14.
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 River and Southern Gulf Water Resource
Assessments for the National Water Grid. CSIRO, Australia.
Meadows M, Jones S and Reinke K (2024) Vertical accuracy assessment of freely available global
DEMs (FABDEM, Copernicus DEM, NASADEM, AW3D30 and SRTM) in flood-prone
environments. International Journal of Digital Earth 17(1). DOI:
10.1080/17538947.2024.2308734.
Middleton BA (2002) The flood pulse concept in wetland restoration. In: Middleton BA (ed.) Flood
pulsing in wetlands: restoring the natural hydrological balance. John Wiley & Sons, Inc, USA.
Neal J, Villanueva I, Wright N, Willis T, Fewtrell T and Bates P (2012) How much physical
complexity is needed to model flood inundation? Hydrological Processes 26(15), 2264–2282.
DOI: 10.1002/hyp.8339.
Nicholas AP and Mitchell CA (2003) Numerical simulation of overbank processes in topographically
complex floodplain environments. Hydrological Processes 17(4), 727–746. DOI:
10.1002/hyp.1162.
Nobre AD, Cuartas LA, Hodnett M, Rennó CD, Rodrigues G, Silveira A, Waterloo M and Saleska S
(2011) Height Above the Nearest Drainage – a hydrologically relevant new terrain model.
Journal of Hydrology 404(1–2), 13–29. DOI: 10.1016/j.jhydrol.2011.03.051.
NT Government (2023) Territory Water Plan. A plan to deliver water security for all Territorians,
now and into the future. Viewed 6 September 2024,
https://watersecurity.nt.gov.au/__data/assets/pdf_file/0003/1247520/territory-water-
plan.pdf.
Ogden R and Thoms M (2002) The importance of inundation to floodplain soil fertility in a large
semi-arid river. International Association of Theoretical and Applied Limnology, SIL
Proceedings, 28(2), 744–749. DOI: 10.1080/03680770.2001.11901813.
Opperman JJ, Galloway GE, Fargione J, Mount JF, Richter BD and Secchi S (2009) Land use.
Sustainable floodplains through large-scale reconnection to rivers. Science 326(5959), 1487–
1488. DOI: 10.1126/science.1178256.
Overton IC (2005) Modelling floodplain inundation on a regulated river: integrating GIS, remote
sensing and hydrological models. River Research and Applications 21(9), 991–1001. DOI:
10.1002/Rra.867.
Owers CJ, Lucas RM, Clewley D, Planque C, Punalekar S, Tissott B, Chua SMT, Bunting P, Mueller N
and Metternicht G (2021) Implementing national standardised land cover classification
systems for Earth Observation in support of sustainable development. Big Earth Data 5(3),
368–390. DOI: 10.1080/20964471.2021.1948179.
Peake P, Fitzsimons J, Frood D, Mitchell M, Withers N, White M and Webster R (2011) A new
approach to determining environmental flow requirements: sustaining the natural values of
floodplains of the southern Murray–Darling Basin. Ecological Management & Restoration
12(2), 128–137. DOI: 10.1111/j.1442-8903.2011.00581.x.
Phelps QE, Tripp SJ, Herzog DP and Garvey JE (2015) Temporary connectivity: the relative benefits
of large river floodplain inundation in the lower Mississippi River. Restoration Ecology 23(1),
53–56. DOI: 10.1111/rec.12119.
Pinos J and Timbe L (2019) Performance assessment of two-dimensional hydraulic models for
generation of flood inundation maps in mountain river basins. Water Science and
Engineering 12(1), 11–18. DOI: 10.1016/j.wse.2019.03.001.
Queensland Government (2023) Queensland Water Strategy. Water. Our life resource. Viewed 11
September 2024, https://www.rdmw.qld.gov.au/qld-water-strategy/strategic-direction.
Queensland Reconstruction Authority (2012) Report on flood investigation for Gregory Downs.
Department of Natural Resources and Mines, Queensland Government, Brisbane.
Queensland Reconstruction Authority (2013) Doomadgee flood mapping study. Department of
Natural Resources and Mines, Queensland Government, Brisbane.
https://www.data.qld.gov.au/dataset/queensland-flood-mapping-program-2013-
series/resource/9e35a262-a9b2-4a4c-89cc-7393b7bb6372.
Rice M, Hughes L, Steffen W, Bradshaw S, Bambrick H, Hutley N, Arndt D, Dean A and Morgan W
(2022) A supercharged climate: rain bombs, flash flooding and destruction. Canberra.
https://www.climatecouncil.org.au/wp-content/uploads/2022/03/Final_Embargoed-
Copy_Flooding-A-Supercharged-Climate_Climate-Council_ILedit_220310.pdf.
Sanders BF (2007) Evaluation of on-line DEMs for flood inundation modeling. Advances in Water
Resources 30(8), 1831–1843. DOI: 10.1016/j.advwatres.2007.02.005.
Schumann G, Bates PD, Horritt MS, Matgen P and Pappenberger F (2009) Progress in integration of
remote sensing–derived flood extent and stage data and hydraulic models. Reviews of
Geophysics 47(4). DOI: 10.1029/2008rg000274.
Shaikh M, Green D and Cross H (2001) A remote sensing approach to determine environmental
flows for wetlands of the Lower Darling River, New South Wales, Australia. International
Journal of Remote Sensing 22(9), 1737–1751. DOI: 10.1080/01431160118063.
Sims N, Anstee J, Barron O, Botha E, Lehmann E, Li L, McVicar T, Paget M, Ticehurst C, Van-Niel T
and Warren G (2016) Earth observation remote sensing: a technical report to the Australian
Government from the CSIRO Northern Australia Water Resource Assessment. CSIRO,
Canberra.
Symonds AM, Vijverberg T, Post S, van der Spek B, Henrotte J and Sokolewicz M (2016)
Comparison between Mike 21 FM, Delft3D and Delft3D FM flow models of Western Port
Bay, Australia. Lynett P (ed.) Proceedings of the 35th International Conference on Coastal
Engineering. ICCE, Turkey.
Tabari H (2020) Climate change impact on flood and extreme precipitation increases with water
availability. Scientific Reports 10(1), 13,768. DOI: 10.1038/s41598-020-70816-2.
Teng J, Jakeman AJ, Vaze J, Croke BFW, Dutta D and Kim S (2017) Flood inundation modelling: a
review of methods, recent advances and uncertainty analysis. Environmental Modelling &
Software 90, 201–216. DOI: 10.1016/j.envsoft.2017.01.006.
Thomas M, Philip S, Zund P, Stockmann U, Hill J, Gregory L, Watson I and Thomas E (2024) Soils
and land suitability for the Southern Gulf catchments. A technical report from the CSIRO
Southern Gulf Water Resource Assessment for the National Water Grid. CSIRO, Australia.
Thoms MC (2003) Floodplain–river ecosystems: lateral connections and the implications of human
interference. Geomorphology 56(3–4), 335–349. DOI: 10.1016/S0169-555x(03)00160-0.
Ticehurst C, Dutta D, Karim F, Petheram C and Guerschman JP (2015) Improving the accuracy of
daily MODIS OWL flood inundation mapping using hydrodynamic modelling. Natural Hazards
78(2), 803–820. DOI: 10.1007/s11069-015-1743-5.
Ticehurst CJ, Chen Y, Karim F, Dutta D and Gouweleeuw B (2013) Using MODIS for mapping flood
events for use in hydrological and hydrodynamic models: experiences so far. In:
International Congress on Modelling and Simulation. Modelling and Simulation Society of
Australia and New Zealand, 1721–1727.
Tockner K, Bunn SE, Quinn G, Naiman R, Stanford JA and Gordon C (2008) Floodplains: critically
threatened ecosystems. In: Polunin NC (ed.) Aquatic ecosystems. Cambridge University
Press, Cambridge, UK, 45–61.
Tockner K, Lorang MS and Stanford JA (2010) River flood plains are model ecosystems to test
general hydrogeomorphic and ecological concepts. River Research and Applications 26(1),
76–86. DOI: 10.1002/rra.1328.
Townsend PA and Walsh SJ (1998) Modeling floodplain inundation using an integrated GIS with
radar and optical remote sensing. Geomorphology 21(3–4), 295–312.
Tuteja NK and Shaikh M (2009) Hydraulic modelling of the spatio-temporal flood inundation
patterns of the Koondrook Perricoota Forest Wetlands – The Living Murray. In: 18th World
IMACS, MODSIM Congress. Modelling and Simulation Society of Australia and New Zealand,
4248–4254.
Ulubasoglu MA, Rahman MH, Onder YK, Chen Y and Rajabifard A (2019) Floods, bushfires and
sectoral economic output in Australia, 1978–2014. Economic Record 95(308), 58–80. DOI:
10.1111/1475-4932.12446.
Wang ZC and Gao ZQ (2022) Dynamic monitoring of flood disaster based on remote sensing data
cube. Natural Hazards 114(3), 3123–3138. DOI: 10.1007/s11069-022-05508-3.
Xu HQ (2006) Modification of normalised difference water index (NDWI) to enhance open water
features in remotely sensed imagery. International Journal of Remote Sensing 27(14), 3025–
3033. DOI: 10.1080/01431160600589179.
Yang A, Petheram C, Marvanek S, Baynes F, Rogers L, Ponce-Reyes R, Zund P, Seo L, Hughes J,
Gibbs M, Wilson P, Philip S and Barber M (2024) Assessment of surface water storage
options in the Victoria and Southern Gulf catchment. A technical report from the CSIRO
Victoria River and Southern Gulf Water Resource Assessments for the National Water Grid.
CSIRO, Australia.
Yu Q, Wang YY and Li N (2022) Extreme flood disasters: comprehensive impact and assessment.
Water 14(8), 1211. DOI: 10.3390/w14081211.
As Australia’snational scienceagency and innovation catalyst,
CSIRO is solving the greatestchallenges through innovativescience and technology.
CSIRO. Unlocking a better futurefor everyone.
Contact us
1300 363 400+61 3 9545 2176csiroenquiries@csiro.aucsiro.au
For further informationEnvironment
Dr ChrisChilcott+61 8 8944 8422chris.chilcott@csiro.au
Environment
Dr Cuan Petheram+61 467 816 558cuan.petheram@csiro.au
Agriculture andFood
Dr Ian Watson+61 7 4753 8606
Ian.watson@csiro.au