Evaluation of the Accuracy of Open-Source DEMs using GPS Data

Abstract

The DEM is often required for flooding or drainage modeling, land use studies, and geological or other applications. A variety of DEMs, such as the Advanced Space Thermal Emission Radiometer (ASTER GDEM) and the Radar Topography Mission (SRTM) are currently available to the public free of charge. However, these DEMs remain a representation of reality, which requires, before any application, their evaluation using high-precision reference data. The objective of this study is to assess the quality of the two DEMs, ASTER GDEM and SRTM, in Morocco. The methodology adopted for this validation is based on two approaches: internal validation and external validation. The first validation consists of a comparison of the versions of each product with each other. This processing showed that version 2 and version 3 (1 arcsecond) are the best versions of ASTER and SRTM respectively. For the second validation, the vertical precision of DEMs is evaluated by 3551 GPS control points. The overall vertical accuracy attests to a ″ RMSE ″ mean square error of 8.17 and 10.15 respectively for ASTER and SRTM, compared to the GPS elevation points. These ″ RMSE ″ values are the lowest compared to the other versions of these two products, which confirms the quality of version 2 of ASTER GDEM and that of version 3 (1 arcsecond) of SRTM. An in-depth analysis of SRTM version 3 data (1 arcsecond) and ASTER GDEM v2, shows the presence of outliers associated with the slope, which shows the degradation of the performance of these DEMs with the increase of the slope.