Delineation of debris-covered glaciers with multi-temporal UAV images (Gössnitzkees, Schober Group /Austria)

The relative share of debris-covered glaciers strongly increased in recent decades
due to climate change and amplified rock production, but despite their relevance as climate indicators, accurate demarcation of respective glaciers is challenging. Remote sensing is an important tool for glacier mapping, but most existing studies apply medium resolution sensors, which are not suitable for small, rock covered glaciers, or use semi-automatic approaches. We present a simple methodology to automatically derive debris-covered glacier areas by using multitemporal, high resolution UAV images. Thereby, standard products, such as elevation differences calculated from digital surface models and isplacement rasters, combined with statistical or error thresholds, provide the basis to automatically delineate glacier areas. The comparison of the automatically derived debris-covered glacier area to the geodetically determined glacier snout showed lowest errors for the elevation-based method using yearly data (total error of 0.32 m or 8.6% of the yearly glacier retreat) and higher errors for four-year intervals (0.97 m, 34% of the yearly glacier retreat) or displacement-based methods (0.51 m, 13.6% of the yearly glacier retreat with yearly epochs). Visual evaluation also showed strong errors of the displacement-based method with many areas wrongly identified as debris-covered glacier area. We conclude that the elevation-based method allows for accurate delineation of debris-covered glaciers and pro-glacial areas, providing increased standardization of glacier monitoring using remote sensing.