Remotely sensed determination of flood surface gradients for hydrological modelling of semi-arid floodplains

The wetland and riparian communities associated with most inland Australian rivers rely on periodic flooding and drying cycles for their healthy maintenance and ecological functioning. Community composition, structure and location are determined primarily by elevation and hence access to flood waters. As most inland rivers are heavily regulated, determination of the timing and size of environmental flow releases from water storages is essential for the proper management of these areas. Modelling of flood surface gradients across large floodplains is one means used to determine which areas of the floodplain and consequently which vegetation communities will be flooded for a given environmental flow release. Traditionally, these gradients have been determined by cell-based analysis using aerial photography and contour maps. Various Landsat TM derived water detection algorithms and L-band SAR backscatter thresholding techniques were explored to detect floodwater boundaries during flooding of a large wetland area on the Macquarie River in central western New South Wales, Australia. These were used in conjunction with a Digital Elevation Model of the area in a GIS environment to model the flood surface gradient and determine the relationships between elevation, flooding and distance from flood entry point. Results showed that automated determination of the flood surface gradient from the satellite imagery achieved better results than manual cell-based analysis and facilitated more accurate modelling of the effects of flow releases. The development of such techniques will permit better use of water resources for sustainable management of important wetland and riparian communities in the future.