Project Overview
Snowpack models used in avalanche forecasting rely heavily on numerical weather prediction (NWP) models. This reliance can lead to errors in the simulated snowpack that compound over the season, particularly when forecast precipitation is incorrect. Real-time evaluation of model output is critical for building trust in operational tools and identifying when corrections are needed.
This project focused on the idea that snow depth, a routinely observed and easily interpreted variable, can be used to assess model reliability. We developed methods to combine snow depth data from automated weather stations, manual study plots, and practitioner field summaries into regional estimates that could be directly compared with model output. These comparisons were integrated into Avalanche Canada’s operational systems, providing daily feedback on model performance.
We also tested methods for correcting major snowfall errors in real time. These methods were tested over multiple seasons and demonstrated improved alignment between modelled and observed snow depth. The approach supports improved trust in snowpack model output and lays the groundwork for future data assimilation efforts.
Involved Researchers
Partner Organizations
Project Funding
Relevant Publications
