1 Near-Real Time Automatic Snow Avalanche Activity 2 Monitoring System Using Sentinel-1 SAR Data in 3 Norway 4

Knowledge of the spatio-temporal occurrence of avalanche activity is critical for 11 avalanche forecasting. We present a near-real time automatic avalanche monitoring system that 12 outputs detected avalanche polygons within roughly 10 min after Sentinel-1 SAR data download. 13 Our avalanche detection algorithm has an average probability of detection of 67.2 % with a false 14 alarm rate averaging 45.9, with maximum POD’s over 85 % and minimum FAR’s of 24.9 % 15 compared to manual detection of avalanches. The high variability in performance stems from the 16 dynamic nature of snow in the Sentinel-1 data. After tuning parameters of the detection algorithm, 17 we processed five years of Sentinel-1 images acquired over a 150 x 100 km large area in Northern 18 Norway, with the best setup. Compared to a dataset of field-observed avalanches, 77.3 % were 19 manually detectable. Using these manual detections as benchmark, the avalanche detection 20 algorithm achieved an accuracy of 79 % with high POD’s in cases of medium to large wet snow 21 avalanches. For the first time, we present a dataset of spatio-temporal avalanche activity over 22 several winters from a large region. Currently, the Norwegian Avalanche Warning Service is using 23 our processing system for pre-operational use in three regions in Norway. 24