Statistical and visual quality assessment of nearly-global and continental digital elevation models of Trentino, Italy

Nearly-global digital elevation models (DEMs)–Shuttle Radar Topography Mission DEM (SRTM1 DEM), Advanced Spaceborne Thermal Emission and Reflection Radiometer Global DEM (ASTER GDEM), and Advanced Land Observing Satellite World 3D digital surface model (AW3D30 DSM)–are widely used in geosciences. A nearly-continental European DEM (EU-DEM) is also publicly available. We assess statistical accuracy and artefact occurrence of the latest versions of these models for Trentino, Italian Alps. We used a set of 111 geodetic network points and a lidar-based reference DEM. Statistically, AW3D30 DSM has the highest accuracy, while EU-DEM is marked by the worst characteristics. To visualize the spatial distribution of errors, we calculated residuals between the reference and validated DEMs. Voids abound in SRTM1 DEM that makes problematic its application in the Alpine region. Randomly distributed, grain-like artefacts are common for ASTER GDEM. AW3D30 DSM includes artefacts caused by to the Gibbs phenomenon. Retaining defects of SRTM1 DEM and ASTER GDEM, EU-DEM has artefacts caused by its hydrological correction. Usual statistical evaluation of DEM accuracy is not capable to describe completely DEM quality. A robust spatial analysis, including interactive visual inspection, should be adopted for a quality control of DEMs.