Understanding the Influence of DEM Vertical Accuracy on Sentinel-1 Geolocation Performance

Abstract. Digital Elevation Models (DEMs) are fundamental for ensuring the geolocation accuracy of Synthetic Aperture Radar (SAR) imagery, as vertical errors in elevation data propagate into horizontal displacements. This study examines the impact of DEM accuracy on the geolocation performance of Sentinel-1 Ground Range Detected (GRD) products over an urban area in Ankara, Türkiye. A comparative analysis was conducted using four global DEMs (SRTM 1″, SRTM 3″, AW3D30, Copernicus 30 m), a high- resolution national DEM (YÜKPAF), and two interferometrically derived DEMs generated from Sentinel-1 Single Look Complex (SLC) pairs. Vertical accuracy was validated against ICESat-2 ATL08 reference elevations, with Copernicus (2.2 m RMSE) and YÜKPAF (3.2 m RMSE) providing the highest reliability, while the InSAR-derived DEMs showed larger errors (>12 m) due to coherence loss and phase unwrapping inconsistencies. Horizontal accuracy was evaluated using Ground Control Points (GCPs) obtained from HGM-Küre. The results demonstrated that high-quality DEMs, particularly YÜKPAF, achieved the lowest horizontal RMSE (8.2 m), whereas InSAR-based DEMs produced the largest errors, approaching 15 m. Despite these variations, all orthorectified outputs remained below 15 m geolocation error, owing to the precise orbital information of Sentinel-1 and the flat topography of the study area. Overall, the study confirms that DEM quality is a decisive factor for SAR geolocation accuracy and offers practical guidance for dataset selection in operational SAR orthorectification workflows.