Satellites and snow models help predict avalanches across Nordic mountains
Data-driven Prediction of Satellite-observed Avalanche Activity from Snowpack Simulations
Machine Learning
Summary
Predicting avalanches is difficult because there are only a few places to check snow and avalanche conditions by hand. The paper’s authors combined computer simulations of snowpack layers with satellite images that detect avalanches in Norway and Sweden. They used a machine learning model to predict the amount of avalanche activity a day ahead based on five days of snow simulation data. The model could guess broad trends over large areas but had trouble predicting exact times or the biggest avalanches. This suggests snow simulations carry useful information about avalanche risk over wide mountain regions.
What this means in practice
- •For mountain safety teams: Predict regional avalanche activity using combined snowpack models and satellite data to guide safety alerts across large mountain areas.
- •For remote sensing analysts: Improve interpretation of avalanche activity detected by radar satellites by integrating snow simulation data and machine learning predictions.
Tested on one dataset.
Authors
Jakob Grahn, Filippo Maria Bianchi, Bert Kruyt, Karsten Müller
Abstract
Avalanche forecasting requires knowledge of snowpack conditions and recent avalanche activity, but field observations are sparse across large mountain regions. We explore whether SNOWPACK simulations can predict avalanche activity mapped by synthetic aperture radar (SAR). We compiled five winters of Sentinel-1 avalanche detections across Norway and parts of Sweden, alongside SNOWPACK simulations forced by numerical weather predictions on a 20 x 20 km grid at different elevations and predefined slope angles. A transformer used five days of SNOWPACK outputs to predict the following day's SAR-detected Avalanche Activity Index (SAR-AAI). This index weights larger debris more heavily, spreads detections across possible occurrence dates and normalises by modelled runout area. The model was trained on four winters and evaluated on one validation winter. Regional mean predicted and reference SAR-AAI correlated at r = 0.803 after averaging over complete six-day periods. The model followed broad changes in time and space but produced smoother predictions and underestimated the strongest activity. Agreement at the 20 km cell scale was weaker (r = 0.549) after the same averaging. These results come from a single training run without evaluation on an untouched winter. Satellite observations also contain missed and false detections and uncertain timing. The results therefore do not establish operational forecast skill, but suggest that regional SNOWPACK simulations contain information about broad variations in satellite-observed avalanche activity.