Analyse und Detektion von Gletscherspalten in den österreichischen Alpen: Eine raumbezogene Analyse auf Basis von Fernerkundungsdaten

Kurzfassung

The present master’s thesis examines the distribution and detection of crevasses across the entire Austrian Alpine region of 446 masked glaciers. The objective of this study is to identify geophysical influencing factors and to supplement further crevasses using DTM-based methods. The present study builds on existing orthophoto labels (F1 86.2%, 20 cm resolution, [Baumhoer u. a. 2025]), topographic indices (elevation, slope, exposure, curvature) and surface flow velocities. These are spatially correlated using Sentinel-2 feature tracking. Methodologically, the work combines pixel-by-pixel extraction from a glacier crevasse dataset, revised glacier polygons by the austrian alpine club (ÖAV), a 10-metre-resolution airborne laser survey digital terrain model of Austria, and velocity. The pixels in the glacier areas were aggregated using hexagon aggregation (H3) to reduce the computational load. In order to expand the crevasse inventory, a difference model is employed in conjunction with a smoothed Digital Terrain Model (DTM). The implementation of resampling and threshold tests serves to optimize sensitivity to false positives. The results indicate an overrepresentation of glaciers and crevasses on northfacing slopes, a concentration of crevasses on moderate slopes (approximately 5–20°) and an increased crevasse density at higher flow velocities (from approximately 10 m/a, clearly at 20–30 m/a). Positive curvatures (i.e. convex steep steps) have been found to correlate most strongly with crevasse formation. The DTM method is a valuable addition to the suite of methods used for the detection of mineral deposits, however it is not immune to misclassifications resulting from the presence of water veins and the presence of snow farming artefacts. Finally, measures for filtering, temporal harmonization, and polygonal quality control are proposed to improve detection accuracy and applicability for route planning and hazard mapping