Meteorological forecasting variables associated with skier-triggered dry slab avalanches

A variety of stability, snowpack and meteorological variables, as well as p revious avalanche activity are typically used to forecast the potential for skier-triggered avalanches. However, the relative importance of, and the i nteraction between, the various variables used to forecast for skier-trigge red avalanches have received little attention. This study analyzes the stat istical influence of 16 simple meteorological variables, 14 calculated or e laborated variables, and 2 variables for previous skier-triggered avalanche activity at a helicopter skiing operation in the Columbia Mountains of Bri tish Columbia, Canada. Forecasting variables are individually assessed using rank correlations to identify the variables most relevant for forecasting the potential for skie r-triggered slab avalanches on the regional scale. The variables showing th e strongest forecasting potential include: the largest size class of skier- triggered avalanche over the previous one and two days, the 24-h snowfall, the 24-h precipitation, the cumulative storm snow, the height of the snowpa ck, and the number of days since December 1. The physical processes that re late these variables to skier-triggered avalanches are discussed. The predictive potential of combined forecasting variables is assessed usin g a multi-variate classification tree model. This model is verified using t he last two years of data that was excluded from development of the tree mo del. The model correctly predicts relatively large avalanches approximately two-thirds of the days for the last two years of the dataset. (C) 2001 Els evier Science B.V. All rights reserved.