Item: HOW AND WHY ARE SNOWPACK MODELS USED OPERATIONALLY?
-
-
Title: HOW AND WHY ARE SNOWPACK MODELS USED OPERATIONALLY?
Proceedings: Proceedings, International Snow Science Workshop, Whistler, BC, Canada, 2026
Authors:
- Zachary Miller [ Sawtooth Avalanche Center, Ketchum, ID, USA ]
- Simon Horton [ Avalanche Canada, Revelstoke, BC, Canada ] [ Avalanche Research Program, Simon Fraser University, Burnaby, BC, Canada ]
- Stan Nowak [ Avalanche Canada, Revelstoke, BC, Canada ] [ Avalanche Research Program, Simon Fraser University, Burnaby, BC, Canada ]
- Christine Donnelly [ Flathead Avalanche Center, Hungry Horse, MT, USA (volunteer) ]
Date: 2026-09-28
Abstract: Public avalanche forecasters in North America rely on sparse weather station and field observation networks, coupled with numerical weather models, to predict avalanche danger—a challenge complicated by the large size of forecast regions (often exceeding 2500 km²). SNOWPACK models offer a more spatially comprehensive view of avalanche hazards than the sparse traditional data sources afford, yet remain underutilized operationally. This study examines how and why forecasters use SNOWPACK, focusing on two factors driving adoption: organizational familiarity/training and trust in model outputs. We surveyed public avalanche forecasters in Canada and the United States (U.S.) during the 2025–2026 winter (n = 80, 6 forecast centers) to identify how SNOWPACK was integrated into their workflows. Results suggest that SNOWPACK frequently has an indirect influence on forecaster assessments of danger. Forecasters with longer-term exposure and more structured training reported more frequent integration, while those with more recent or limited exposure described more exploratory use. Specifically, Canadian forecasters, who have more experience and training with SNOWPACK, reported that it influenced their decisions 77% of the time compared to only 57% for U.S. forecasters. The survey also characterized forecaster trust in SNOWPACK outputs. Lower trust was reported following atmospheric rivers, during persistent weak layer formation, and wetting front evolution. To contextualize this self-reported trust, we compared SNOWPACK-derived hazard outputs with forecasted avalanche danger and problems. General regional and seasonal trends aligned well, but problem-specific agreement varied, contributing to periods of lower perceived accuracy. SNOWPACK model outputs and interactive dashboards are an ever-improving toolset for avalanche forecasters. The results of the survey and contextual support of the hazard analysis indicate that SNOWPACK model utility for public avalanche forecasting depends on a combination of familiarity and training as well as trust developed from applied use. This research supports more widespread operational adoption of models in two ways; first, by informing developments that would make SNOWPACK tools more accurate and useful, and second, by suggesting operational training and practice strategies for working with these tools.
Object ID: ISSW2026_O9.4.pdf
DOI: https://doi.org/10.15788/1790098692
Language of Article: English
Presenter(s): Zachary Miller
Keywords: avalanche forecasting, snowpack modeling, large-scale operational model chain, spatial variability, avalanche problems
Page Number(s): 1291 - 1297
-