Breadcrumb

Roadshow 3.1 Flood inundation modelling for Cooper Creek floodplain

Flood inundation modelling for Cooper Creek floodplain 

Click here to open transcript

Hello, everyone. I am Cherry Mateo, research scientist and hydrodynamic modeller who worked hard on this project. On behalf of my colleagues, Jai, Shaun, Steve, Cat, Bill, John, Russell, and Kate, I am honoured to present to you our work on the flood inundation modelling for Cooper Creek floodplain. So why do we need hydrodynamic model for the Cooper Creek? The Cooper Creek is characterised by a network of intermittent rivers that flood frequently. This floods are necessary to replenish hundreds of waterholes, lakes, and wetlands which are important ecological assets. On the other hand, extreme floods can be catastrophic to gas resource development and agricultural production in the region.

Detailed hydrodynamic model is necessary to understand the propagation of floods under current development. The set model can then be used to investigate and quantify the impacts of future development on the flood regime and connectivity of important ecological assets in the region. This sounds like a usual hydrodynamic modelling exercise. However, developing a hydrodynamic model for the Cooper Creek is quite challenging. The modelling area is huge, about 32,000 square kilometres spanning across Queensland and South Australia. The region is characterised by an extremely complex terrain with anastomosing channels and hundreds of waterholes and wetlands. The terrain is very flat, and so the representation of topography in existing SRTM maps is poor. The region is also sparsely observed with only a limited number of streamflow gauging stations with good quality flow data.

To overcome these challenges, we modelled the region in two separate sections, the Queensland and South Australia sections. This allows us to capture more details for each section and use the observed streamflow in Stonehenge, Retreat and Nappa Merrie as boundary conditions to the models. We also used the 2D MIKE 21 Flexible Mesh model. Unlike 2D regular grid models, Flexible Mesh model allows us to assign fine resolution in areas of importance and coarse everywhere.

In this zoomed-in section of the SA modelling zone, we could see that such technique allows us to use very fine mesh elements in rivers, in waterholes, and other flooded areas. This allows us to limit the mesh elements to about 7.4 million in Queensland and 4.75 million in South Australia. The 2D MIKE 21 Flexible Mesh model also allows us to use parallel GPUs for much faster computations.

For the model inputs, we tried to incorporate as much detail as is necessary. We extracted the elevation data from one metre LiDAR DEM, which was further refined by stitching the waterhole bathymetry information which were gathered through a survey. Observed streamflow data were used as infill boundary conditions where they exist. And for ungauged flow sources outputs from the Sacramento model were used. Precipitation and potential evaporation from AWAP were used for the climate data, for the bed-resistance published Manning's values corresponding through distributed land use data were used. Soil characteristics from the AWRA model were used to create the infiltration layers and information about water extent from available LandSat and modelling results were used to set the initial conditions for each event. LandSat images were also gathered for validation purposes.

We focused our efforts on estimating more frequent events which are important for replenishing water in the ecological assets. We used two events one, one in two-year and one, one in five-year events to calibrate each section. In this slide, sample results for one in two years in SA, and one in five years in Queensland are shown. Results show good agreement between the models, simulated images to the left, and LandSat inundation extents the images to the right for all the events.

The models were also used to validate one in 10-year events, these events flood most of the floodbank for each section. Reasonably good agreements were also seen, which gives us some confidence that the models could be used for analysing a wide range of events and future development scenarios.

What outputs are available or possible? Several outputs were made available in the downloadable dataset that we have prepared. The total water depth is made available as Raster tiff files. The other data such as surface elevation, U and V components of the velocity, precipitation, evaporation, saturation, infiltrated volume, and discharge could be extracted from the MIKE modelling results that we have provided. If desired, the model could be rerun to output other variables of interest which are listed in the grey box.

Now can the user on their own scenarios? Yes, while the uploaded results represent the current development, the model could be further refined or modified to analyse other scenarios. So to analyse climate change scenarios or other events of interest the climate data could be modified. To analyse future development scenarios which involves structural or terrain changes information about the structures could be included in the model or the DEM could be modified to reflect the changes in the terrain. To analyse impacts of land-use changes, the Manning's layer could be modified to reflect the corresponding modifications to the surface roughness.

In summary, MIKE 21 Flexible Mesh models were developed for the Queensland and South Australia sections of the Cooper Creek floodplain. The models were used to analyse the current development and were found to have reasonably good agreement with LandSat flood extents. With appropriate modifications, the models could be used to analyse future climate and development scenarios. And in this presentation, I would like to show a couple of videos showing the propagation of one in 10-year floods in Queensland and South Australia. So 2010 event for South Australia and 2004 event in Queensland. Thanks for listening.

Related factsheets 

LiDAR Survey Factsheet

This factsheet explains what Light detection and ranging (LiDAR) measures and describes the collection of the Cooper Creek floodplain LiDAR dataset for the Cooper GBA region

13. Flood inundation modelling for Cooper Creek floodplain

A 32,000 km2 hydrodynamic flood inundation model was developed for the Cooper Creek and its floodplain to investigate whether an increase in surface infrastructure associated with gas resource development could impact the flood regime of ecologically important waterholes and wetlands along the Cooper Creek.

 

View all GBA factsheets

About the presenter 

Dr Cherry Mateo 

CSIRO

Cherry is a Research Scientist, a hydrologist who has been involved in several hydrodynamic and hydrological modelling projects. She has been involved in several flood modelling projects with MDBA and in the improvement of the AWRA landscape model for continental Australia. 

Last updated:
15 November 2021