Computing extreme avalanches

Increasing the reliability and accuracy of avalanche zoning method is of primary importance in heavily populated areas of the Alps. This usually involves computing the characteristics of large return-period avalanches. Current tools (avalanche-dynamics or statistical models) cannot achieve this objective properly. A new generation of models has emerged, which, by combining statistical and deterministic viewpoints, can reduce a number of shortcomings in the original tools. This paper explores the possibility of updating the Salm–Burkard–Gubler method by adding a probabilistic procedure for fitting parameters to field data and using it in Monte Carlo simulations. The resulting model is intended to provide more accurate results of extreme avalanches. It has been applied to a series of close avalanche paths in the French Alps. Despite their similarities, these paths reveal some differences in their avalanche activity in terms of frequency and maximum run-out distance. Therefore, in order to provide a fair description of the intensity/frequency relationship, the model parameters must be fitted for each path, which leads to thinking that not all the important physical processes have been taken into account in this model.