Combining avalanche nowcast with GPS tracks and ‘in situ’ participant reports to understand decision-making in avalanche terrain

Design of interventions to reduce the number of avalanche accidents requires a detailed and holistic understanding of how recreational backcountry riders assess the risks and conditions. This understanding must be built up from several key questions. Firstly, what information riders use to plan a tour, and what kind of avalanche assessments they make during a tour? Secondly, do riders comprehend the information they use, and draw `correct' conclusions from this information? Gaining this understanding is challenging for many reasons. One important challenge is that we rarely have access to the `ground truth' since regional forecasts are not designed to evaluate single slopes. Another is that the best ground truth we can obtain here is an expert assessment, which involves domain knowledge that the regular rider may not be able to grasp, thus rendering the comparison between their assessments more challenging. In this study, we combine local ‘nowcasts’, made by a professional avalanche forecaster, with GPS tracks and a survey. We carried out the survey at Kattfjordeidet, a popular backcountry area outside of Tromsø that attracts a wide range of riders (from beginners to experts). The area has relatively well-defined ski runs that are in avalanche terrain, with substantial variation of avalanche exposure. The avalanche forecaster made nowcasts for each individual run. The paper describes the method and showcases some results emanating from the data. Specifically, we examine avalanche related information acquisition, avalanche likelihood estimates and terrain choices made by our participants. We further compare the likelihood estimates provided by the participants, with the assessments of the professional avalanche forecaster. Our results indicate that 1) almost all riders check the avalanche and weather forecast prior to venturing out into avalanche terrain and 2) riders who go into more serious terrain prepare more for their tours. However, our sample also contains 3) riders who go into very serious terrain without much information at all and without making assessments along the way. We also find that 4) most riders either make correct assessments or overestimate avalanche risk, and that type of terrain does not affect the perceived likelihood of a dangerous avalanche releasing. We discuss implications of our findings and elaborate on the merits and challenges of our approach, and what we have learned during this three year project.