Probabilistic analysis of recent snow avalanche activity and weather in the French Alps

Abstract

The characterization of statistical relationships between snow avalanche occurrence and climate can be useful for avalanche prediction. We investigated the relationship between avalanche occurrences between 1978 and 2003 and meteorological parameters for 576 avalanche events from 12 avalanche tracks in the Valloire valley in the French Alps. Probabilities of avalanche occurrence based on logistic regression analyses were calculated at a daily and yearly time scale, by differentiating high- and low-frequency avalanche tracks. For high-frequency avalanche tracks, the daily probability of avalanche depends on the precipitation (water equivalent, i.e. WE) the day before a given avalanche event, along with the mean air temperature the day of the event. For low-frequency avalanche tracks, on the other hand, it depends on precipitation (WE) the day preceding a given event, only. We also tested the relationship between various meteorological parameters and the type of avalanche. The occurrence of dry snow avalanches is related to total precipitation (WE) on the day of and the day before a given event, whereas that of wet snow avalanches depends on precipitation (WE) the day of a given event, and maximum air temperature during the event. Our results show that for high-frequency avalanche tracks, annual probabilities of high avalanche activity depend on the occurrences of successive days (≥ 3 days) with high precipitation in winter and above-average air temperature (mean ± 1 S.D.). For low-frequency avalanche tracks, probabilities of high avalanche activity depend on the occurrences of successive days (≥ 3 days) with high precipitation in winter. The sensitivity of these models was tested through bootstrap analyses. We also discuss the role of meteorological parameters highlighted in these models.