Accuracy Assessment of Sentinel-1 SAR-Derived DEMs: Comparative Analysis with SRTM and ALOS References Ochirkhuyag Lkhamjav1,3,4, Fuan Tsai 1, 2*
1 Department of Civil Engineering, National Central University, Taoyuan City, Taiwan 320317
2 Center for Space and Remote Sensing Research, National Central University, Taoyuan City, Taiwan 320317
3 Institute of Geography and Geoecology, Mongolian Academy of Sciences, Ulaanbaatar 15170, Mongolia
4 Mongolian Geospatial Association, Ulaanbaatar 15141, Mongolia
*olkhamjav[at]g.ncu.edu.tw- ftsai[at]csrsr.ncu.edu.tw
Abstract
Accurate topographic mapping is vital for scientific, environmental, and engineering applications. Digital Elevation Model (DEM), including Digital Terrain Model (DTM) and Digital Surface Model (DSM), are essential for these purposes. Synthetic Aperture Radar (SAR) interferometry, particularly from Sentinel-1, offers significant potential for generating and updating DEMs due to its all-weather, high-resolution capabilities. This study conducts a comprehensive validation of Sentinel-1 SAR-derived DEMs by comparing them with established reference datasets, the Shuttle Radar Topography Mission (SRTM) and Advanced Land Observing Satellite (ALOS) DEMs, across diverse terrains (600-1,500 m elevation).
Using statistical metrics such as Pearson correlation, Root Mean Square Error (RMSE), and descriptive statistics, the research evaluates geometric fidelity and absolute elevation accuracy. Results indicate exceptional relative accuracy, with correlation coefficients (r > 0.999) for both reference datasets, confirming Sentinel-1^s ability to preserve topographic structure. However, a systematic elevation bias of 38-39 m was observed, with RMSE values of approximately 38-39 m and low standard deviations (2.94-3.24 m), indicating high precision despite absolute offsets. These findings suggest that while Sentinel-1 interferometry excels in relative elevation mapping, calibration is critical for absolute accuracy.
The study highlights Sentinel-1^s potential for supplementing global DEMs, particularly in low-vegetation environments. Terrain-specific bias corrections, advanced processing techniques like Persistent Scattered Interferometry, and sensor fusion with laser altimeter or multi-frequency SAR are recommended for enhanced accuracy. These insights provide a robust framework for operational topographic mapping, supporting applications in geosciences and environmental monitoring.
