Bruce Jamieson Bio:
Bruce Jamieson started avalanche work at the Fernie ski area in 1980. As a professor with the University of Calgary from 1997 to 2015, he managed field studies of snow and avalanches. Now, when not sliding on snow or riding a two-wheeler on dirt trails, he works as an avalanche consultant and educator.
Scott Thumlert Bio:
Scott has a balance of both field and analytical experience with snow avalanches. After graduating with his PhD in civil engineering specializing in avalanche mechanics from the University of Calgary, Scott worked as a post-doctoral fellow at Simon Fraser University on avalanche risk management projects. In addition to his work with Alpine Solutions, Scott is an ACMG certified ski guide and works as a helicopter guide with Canadian Mountain Holidays. As a Professional Engineer he has project experience in high arctic field operations and snow avalanche risk management.
Communicating Likelihood and Probability of Snow Avalanches
Communicating avalanche hazard requires that practitioners express their judgment of avalanche likelihood using expressions of probability (e.g. Unlikely). However, these descriptors of probability are interpreted in different ways by different people and differ widely for the same people in different contexts. We asked 75 avalanche practitioners from around the world to put a percentage probability beside each of the following words for avalanche likelihood (Unlikely, Possible, Likely, Very Likely, Almost Certain). The practitioner estimates of probability varied greatly.
Based on strategies developed in the climate science, intelligence and meteorology fields, we propose some ideas for improving communication of avalanche practitioner estimates of avalanche likelihood: 1) a definition for Likelihood of Avalanches that includes a reference class that is independent of spatial scale and that translates probabilities into relevant frequency descriptions; 2) defining numerical probability ranges that are useful for making decisions regarding avalanche terrain; and 3) matching likelihood terms with numerical probabilities that are intuitive and consistent with well-established ranges.