Mapping impact of landslides on glacier dynamics based on medium-resolution satellite data

Global climate warming leads to rapid glacier recession and increased activity of other geomorphological processes, like landslides, rockfalls, and debris flows. The question is how these two groups of processes (i.e., glacier dynamics and mass movement activities) are related to each other, and to what extent can they threaten the human population and infrastructure? This study focuses on glacier-landslide interactions trying to develop a workflow for the use of medium-resolution satellite data (Landsat, Sentinel, Aster) for systematic quantification of the response of selected glaciers to additional debris load delivered by landslides on the ice surface. The workflow includes eight main steps: (1) Downloading and verification of glacier outlines from global glaciers databases: Randolph Glacier, GLIMS (Global Land Ice Measurements from Space); WGI (The World Glacier Inventory); (2) Filtering of glacier outlines based on topographic properties of their vicinity: Digital Elevation Models (DEMs) were used to select glaciers located in mountain areas and with slopes situated next to the glacier surface. This step excluded ice caps and ice fields without valley outlets as they are unlikely to be affected by landslides; (3) Identification of available images in Google Earth Engine; (4) masking of clouds and surface water bodies using NDWI (normalised difference water index); (5) Classification of debris-covered and clean ice surface using NDSI (normalised difference snow index); (6) Identification of a time step when the debris arrived onto the glacier surface; (7) quantification of changes in the area covered by debris through time; (8) calculation of displacement of characteristic points. The presented workflow was tested in several glaciers located in different geographic settings where the occurrence of large landslides was identified in previous studies: Svinafelljökull and Morsarjökull (Iceland); Tyndall and Lamplugh glaciers (Alaska); Siachen and Baltoro glaciers, and Barun River and Imja River areas (Himalaya), Zaalai Range (Pamir), Glaciar Leones (Patagonia). The results indicated that based on medium-resolution satellite data, it was possible to identify the presence of landslide deposits and, in most cases, to estimate the approximate period of the landslide event. However, the quantification of displacement of landslide deposits was possible only when there was a distinct difference in reflectance characteristics between glacier surface and debris.

This research was funded by National Science Centre, Poland, project number 2021/42/E/ST10/00186