全球变化知识资源中心

Over the past decades, climate change has exerted pronounced influences on the cryo-hydrologic environments in High Mountain Asia (HMA). Digital elevation models (DEMs) are one of the critical datasets required in many hydrology- and cryosphere-related studies in HMA, which include estimations of glacier mass balances and lake water storage changes and hydrological runoff modelling. Several global or near-global coverage DEMs (e.g. SRTM and ASTER missions) have been widely utilized for such studies. However, a comprehensive assessment of the quality of these DEMs, particularly over the high-relief HMA area, is still lacking. This study investigated the performances of seven public freely-accessed DEM datasets (ASTER GDEM V2, SRTM-3 V4.1 DEM, SRTM-1 DEM, AW3D30 DEM, VFP-DEM, MERIT DEM, Seamless SRTM-1 DEM) over the HMA region by referring to high-accuracy elevation data from ICESat altimetry. We also provided an additional detailed assessment for typical lake basins and extremely rugged areas in the Himalayas by using available global positioning system (GPS) measurements and gridded SETSM DEM raster data. Results show that AW3D30 exhibits the highest accuracy in HMA land areas among these datasets. The accuracy of all studied DEMs demonstrates high spatial variation, which shows the strongest relationship with slope among several topographical factors, such as elevation and aspect. Over these rugged terrains with a large slope, such as the Hengduan Mountains and Himalayas, the vertical accuracy is relatively low in all investigated DEMs. We established empirical models for fitting the relationship between DEM vertical error and mean slope, which can be used to estimate the average elevation errors for any given local or basin-scale DEM grids. In addition, this study identified regions with anomalous elevation data in each dataset caused by the imperfect void-filling processing scheme. In sum, the 3-arc-second MERIT DEM is judged to be relatively dependable for HMA which not only achieves good performances in statistical error values but also retains few void-filling anomalies. The 1-arc-second AW3D30 could be the most promising product over HMA with its high spatial resolution and accuracy, although some void-filling anomalies need to be noted in some regions. (C) 2019 Elsevier B.V. All rights reserved.