DEM Generation Using Interferometric SAR

Abstract

Digital Elevation Models (DEMs) are critical for terrain analysis, hydrological modeling, disaster risk assessment, and land-use planning, especially in mountainous regions like Nepal. Conventional DEM generation methods such as field surveying, photogrammetry, and LiDAR are often costly and impractical in remote areas. Interferometric Synthetic Aperture Radar (InSAR) has emerged as a viable and cost-effective alternative, capable of generating high-resolution DEMs with vertical accuracies of 1–3 m RMSE. This study presents a comprehensive InSAR-based DEM generation workflow for a region covering Muglin and the surrounding central Himalayan terrain of Nepal using Sentinel-1 Single Look Complex (SLC) images acquired on May 10 and May 22, 2024, with a 12-day temporal baseline. The study area was defined by the available Sentinel-1 scene footprint, extending approximately from 84.04°E to 85.22°E longitude and 27.53°N to 29.21°N latitude, encompassing diverse topographic conditions ranging from the mid-hills around Muglin to higher Himalayan regions. The methodology involves coregistration, interferogram formation, Goldstein phase filtering, SNAPHU-based phase unwrapping, and phase-to-height conversion, producing a 10 m resolution DEM. The generated DEM was validated against ASTER DEM reference data at nine ground control points. The resulting Root Mean Square Error (RMSE) of approximately 52 m highlights the challenges posed by steep and complex Himalayan terrain, including geometric distortions such as layover, foreshortening, and shadow effects. Despite these limitations, the study establishes a reproducible InSAR processing framework applicable to geospatial and hazard-related applications in Nepal and similar mountainous environments.