Evaluating snowfall patterns pertinent to regional-scale avalanche forecasters with data from stations, field reports and models

Abstract:

Snowfall is a primary driver of avalanche hazard, but forecasters regularly face uncertainty about the timing, location and quantity of snowfall in avalanche terrain. Available data streams include automated stations, field observations and weather model output, but reliable assessments require continuous synthesis of these data sources grounded in an in-depth understanding of their respective strengths and weaknesses. Developing this meta understanding is difficult because data are presented in different formats and represent different terrain scales. To address this issue, we developed tools that allow forecasters to compare regional-scale snowfall patterns from different data streams consistently so they can strengthen their understanding of their optimal use.

To better understand how snowfall patterns relate to avalanche hazard, we convert snow forecasts and observations from western Canada into a common format. Regional-scale maps of snow depth and daily snowfall amounts were produced from three sources: weather stations, field reports from professional avalanche observers, and the SNOWPACK model forced with weather model data on a 2.5 km grid. To create a consistent link to avalanche terrain, all data were converted to typical values at treeline elevation by applying lapse rate corrections and spatial interpolations. The resulting maps were on a coarse grid that smoothed local topography and highlighted regional-scale patterns. Maps produced from each data stream were compared with a distance measure that quantified how their agreement varied across space and time. The combination of maps and distance measure highlighted situations where the data sources showed consistent patterns and situations where disagreements arose from either model errors or gaps in the observation networks.

In addition to offering avalanche forecasters with a consistent perspective of their data streams, the regional-scale snowfall maps also provide insights about optimal forcings for snow cover models (which are highly sensitive to precipitation inputs) and create opportunities to verify weather models in data sparse terrain by leveraging the knowledge of avalanche professionals and their unique observation networks.