GENERATION OF MULTI-TEMPORAL DEMS FROM SENTINEL-1 FOR ASSESSING GEOMORPHOLOGICAL CHANGES IN THE HÍTARDALUR VALLEY, WESTERN ICELAND

Multi-temporal digital elevation models (DEMs) provide valuable information for investigating geomorphological changes related to landslides or other landscape shaping processes, especially when in-situ measurements are difficult to gather. The free availability of Sentinel-1 synthetic aperture radar (SAR) data from the European Union's Earth Observation Programme Copernicus opened a new era for generating such multi-temporal topographic datasets, which allow for regular mapping and monitoring of land surface changes. The main objectives of this study are to assess the applicability of Sentinel-1 data for the generation of multi-temporal DEMs and to evaluate their quality for studying geomorphological changes. We focus on the Hítardalur landslide, which happened on July 07, 2018, in western Iceland. This landslide led to significant landscape changes since it dammed a river and led to the formation of a dammed lake. We created six post-event DEMs using Sentinel-1 data (i.e. data acquired after the landslide happened) using Interferometric SAR (InSAR) techniques available in the ESA SNAP toolbox. The quality of the DEMs was evaluated using spatial statistics such as spatial autocorrelation and the root mean square error. As a reference, a post-event DEM with 5 m resolution of the landslide area provided by the Iceland Meteorological Office was used. Moreover, we assessed the influence of the quality and spatial resolution of the external DEM for topographic correction during InSAR analysis, using (a) the GETASSE30 DEM (30 arc second resolution) accessible in the SNAP toolbox, and (b) the ArcticDEM (2 m resolution) provided by the Polar Geospatial Center. The results indicate substantial quality improvement for the DEMs derived from Sentinel-1 when using the external ArcticDEM during the InSAR topographic correction. The best Sentinel-1 DEM was then used for landslide change analysis and volume estimation. Usually, high-resolution multi-temporal DEMs are rarely available, especially in remote areas. Thus, DEMs derived from freely available data such as Sentinel-1 are a potential alternative for studying and monitoring geomorphological landscape changes, e.g. caused by landslides, if carefully derived to reach a sufficient quality for the specific application case.