Flooding is a major hazard in both rural and urban areas worldwide, but it is in urban areas that the risks to people and the economic impacts are most severe. Urban flood modelling is still at an early stage, and to date most models have been applied in rural areas. Flood modelling in urban areas is more complicated partly because flows interact with the built environment, necessitating at least a 2D treatment of the surface flow hydraulics. Urban flood models need considerable data for their parameterisation, including high resolution LiDAR data, estimates of channel and floodplain friction factors, and estimates of building coverage in each model grid cell. In rural areas, 2D flood models have been successfully calibrated using flood extents determined from SAR data (typically ERS and ASAR) in order to estimate channel and floodplain friction factors. SAR data are used because of their all-weather day-night capability. While ERS-2 and ASAR have too low a resolution for use in urban areas, TerraSAR-X should be suitable for urban flood detection because of its high resolution in stripmap/spotlight modes. This project is assessing the degree to which a TerraSAR-X image can detect flood water in an urban area, and the degree to which the detected regions can constrain the parameterisation of an example 2D urban flood model. The method is being tested on a 1-in-100 year flood on the Severn near Tewkesbury, UK, in 2007, for which near-simultaneous aerial photography exists for validation of the TerraSAR-X flood extent. Due to the side-looking nature of SAR, substantial areas of ground surface are not visible in urban areas due to shadowing and layover caused by buildings and vegetation. The DLR SETES SAR simulator has been applied to high resolution airborne LiDAR data of Tewkesbury to identify regions not affected by buildings, in which flood water may be detected in the TerraSAR-X image. An algorithm is being developed to extract disconnected regions of water in urban areas automatically. This uses information on the backscatter characteristics of water determined from the flood extent in adjacent rural areas obtained using a region-growing algorithm seeded using the un-flooded river channel. A 2D hydrodynamic model is also being constructed. Progress on the project will be described.
|Translated title of the contribution||Using TerraSAR-X data for improved urban flood model validation|
|Title of host publication||3rd Science Team Meeting on TerraSAR-X at DLR|
|Publication status||Published - Nov 2008|