Merging lidar with coarser dems for hydrodynamic modelling over large areas

Abstract

Hydrodynamic modelling of flood flows requires accurate topographic data, which in most cases means lidar or photogrammetric data with proper separation of ground and non-ground features like vegetation. The entire modelling domain must be represented but the area of focus is often a small part of the entire domain, so it is often cost-effective to use expensive and detailed elevation data in the focus area and cheaper, less detailed data elsewhere. This leads to the need for joining the two digital elevation models (DEMs) seamlessly so that there are no abrupt changes in height or slope at the transition. In the Fitzroy River of Western Australia, an area of 37000 km2 was targeted for hydrodynamic modelling. Lidar data was collected in the floodplain area covering about 5800 km2 and the remaining area was covered by SRTM-derived DEM-H at 1 arcsecond resolution. A method for adjusting the DEM-H to match the lidar data was developed to remove abrupt steps at the boundary to ensure the combined data was suitable for hydrodynamic modelling. The two main steps in the process were (1) removal of systematic vertical errors and (2) adjusting the less reliable DEM-H to match the lidar at the boundary. The method successfully removed local steps and produced a result suitable for hydrodynamic modelling. There is scope for extending the adjustment process to account for broader scale differences.