| dc.description.abstract | Previous landslides can be important indicators for where to expect future landslide activity and under which conditions. Unfortunately, information concerning previous landslides is often insufficient and landslide databases are incomplete. Only a subset of landslides is registered, in most cases those who interact or affect human life and infrastructure. However, landslides often appear in remote areas, and several landslides, therefore, remain unknown and unnoticed. This study proposes the use of free accessible optical satellite data, produced by moderate resolution sensors in combination with vegetation indices to map and detect previous landslides. The focus is change detection analysis using images from the Sentinel 2 and Landsat 8 optical satellites, which are post-processed to calculate various vegetation indices. Primarily two vegetation indices are used, Atmospherically Resistant Vegetation Index (ARVI) and Normalized Difference Vegetation Index (NDVI). The mapping of landslides is performed in ArcMap, where two different mapping approaches are undertaken. First, a manual mapping where the difference between pre-and post-image is studied. Secondly, a semi-automatic mapping approach in the Raster Calculator. These mapping approaches are applied to Jølster municipality for detecting landslides after the Jølster landslide event in 2019. After the event, 18 landslides were defined within Jølster municipality and registered in the NVEs landslide database. In this study, a total of 108 landslides were identified. A comparative analysis in ArcMap, between ARVI and NDVI, reveals that ARVI detected changes caused by landslides better than NDVI. To verify the method, it has been applied to a second study site; Oso located in the state of Washington, USA. At this case site as well, ARVI maps more accurately than NDVI for landslide recognition. This study suggests that moderate resolution, optical satellite images used in the study of detecting previous landslides increases the number of detected landslides. Further, the vegetation index ARVI should be used when detecting landslides in humid climates. Throughout the research conducted, this study contributes to an improved understanding of the Jølster event, as well as a comparison of different methods for remote landslide detection. The research contributes supplementary landslide information to the database of already known landslides. Furthermore, testing and comparison of methods contribute towards an understanding of best-practice. | eng |