Satellite remote sensing of snow avalanches in a continental snow climate using synthetic aperture radar

Abstract

Knowing where and when snow avalanches occur is essential for avalanche forecasting and avalanche hazard mapping. Avalanche hazard, difficult access to terrain, weather conditions and the large size of regional forecasting zones make conventional manual field observations of avalanches difficult, resulting in a misrepresented avalanche occurrence dataset. In recent years, satellite based synthetic aperture radar (SAR) remote sensing has shown promising results for the detection of snow avalanches as it is not affected by environmental or daylight conditions.

This bachelor’s thesis explores to what extent can current open-source SAR data, such as Sentinel-1, facilitate near real time automated avalanche detection. The thesis examines also the detection challenges specific to continental snow climate and prospects of automated avalanche detection using SAR.

The thesis presents that currently the temporal resolution of using Sentinel-1 is not sufficient for near-real time automated avalanche detection. In addition, dry snow avalanches that are typical for continental climate, are difficult to detect using conventional backscatter change detection methods with C-band SAR. However, following the launch of new L-band constellations in a few years, revisit times of satellites improve, and near-real time avalanche detection becomes possible. Machine learning and deep learning methods can improve automated detection under changing snow conditions. Furthermore, interferometric methods using L-band can potentially provide methods for determining snow water equivalent and enhance dry snow avalanche detection.