Regional Mountain Conference

Anthropogenic climate change strongly affects the Alpine water cycle. Shifts in the timing, intensity and spatial distribution of precipitation and increasing winter temperatures alter the frequency and timing of snowfall and snowmelt, as well as the structure of the snowpack. As a result, avalanche risk and the regionally assessed avalanche hazard levels are expected to change under ongoing climate change.

In this study, we present novel climatologies for avalanche hazard levels in the Austrian Alps covering the period of 1950 – 2025.  These climatologies are based on the back-extension of a consistent multi-year observational dataset of avalanche hazard levels aggregated from regional avalanche hazard assessments. Individual mountain ranges are clustered into units that exhibit similar avalanche hazard responses to different types of weather patterns. These response units are then used to train a machine-learning algorithm to predict avalanche hazard levels based on pattern recognition and circulation analogues from meteorological ERA5 reanalysis data. Applying the algorithm to the full ERA5 dataset yields a reconstruction of past avalanche hazard levels, which are validated against long-term observations from Tyrol. The resulting climatologies reveal interannual and decadal variability as well as the possible influence of anthropogenic climate change on avalanche hazard over the past 75 years. 

When applied to the latest generation of regional climate projections, the trained algorithm will provide new insights into how future climate change may further affect avalanche hazard in the Austrian Alps.