Published works

A digital elevation model (DEM) is a three-dimensional representation of the Earth surface showing mountain peaks, valley floors, rivers and lakes. DEMs are an essential input for the application of environmental and flood models and, for large catchments,these datasets are produced using data collected by instruments on-board satellite missions. However, data collected from such techniques are inevitably affected by inaccuracies which have to be accurately understood and evaluated in order to allow the implementation of any numerical model. This study evaluated the accuracy of the recently released 12 m resolution TanDEM-X (German Aerospace Center) and the widely used 30 m resolution DEM-H (Geoscience Australia). One of the most important sources of inaccuracy in satellite-derived DEMs is a dense vegetation canopy. This problem is particularly important in dryland catchments, where vegetation is not uniformly distributed across the landscape but often concentrates along the river corridors. The Condamine-Balonne catchment, a low-gradient dryland area in Queensland, offers a representative case study. Satellite-derived DEMs were compared with more than 600 ground control points, three LiDAR datasets, and over 60 ground measured river cross sections. This analysis allowed the evaluation of the capability of the satellite-derived DEMs to reproduce (i) surface elevation at the point scale, (ii) river channels and floodplain morphology, and (iii) catchment slope along the river network. TanDEM-X performed better at the point scale and in transects with sparse vegetation resulting in Root Mean Square Deviation (RMSD) values respectively 0.30 m and 0.12 m lower than the DEM-H. However, TanDEM-X was affected by large errors in densely vegetated areas. Moreover, the enforcement of hydrological connectivity is required. This analysis offers useful insight for the application of TanDEM-X and DEM-H data for environmental and flood modelling in the Condamine-Balonne catchment and in similar dryland areas.