Abstract

A formal comment on a current Global Digital Elevation Model (GDEM) benchmark by Guth et al. is presented. It reproduces the original study, emphasizing and modifying the methodologies in two key aspects: (1) adding statistical testing and assessment, and (2) making the processing workflow (code) and test site tiles (data) publicly available and hence reusable. To test whether the performance of different GDEMs is statistically significant, the Kruskal–Wallis test was used. In addition, a pixel-alignment strategy to resolve pixel-shifting issues commonly observed in GDEMs was introduced. The results align with the research outcomes in Guth et al. regarding ranking FathomDEM, a restricted license product, first. However, a notable discrepancy was observed in the performance evaluation of COPDEM and GEDTM v1.2 within barren areas. Contrary to the findings of Guth et al., these results indicate no degradation in the performance of GEDTM v1.2 versus COPDEM. Furthermore, statistical comparison of the quality of GDEMs across various land surface parameters indicates no significant differences among GDEMs. The results of Guth et al. are potentially affected by the high sensitivity of the proposed FUV approach and the limited sample size (representing parts of the USA and western Europe). All code used to produce the results and all input datasets analyzed are provided, allowing for full reusability and reproducibility of these results.

IPC Classification

Keywords