Public avalanche forecasting is the practice of assessing current and future avalanche hazards to inform the decision-making of those at risk and traveling in avalanche-prone mountainous terrain. Avalanche forecasters utilize weather forecasts, field observations, and their own expert judgement to produce their forecasts. Of critical importance is an understanding of how layers of snow transform over the course of the season and produce instabilities that may result in avalanche events. Relatively recently, researchers have started to develop simulations of these layers by utilizing weather forecast models to recreate the atmospheric conditions that drive the creation and transformation of the snowpack. These have largeley remained in the realm of academia and have been under-utilized by operational avalanche forecasters. The reasons for this include but are not limited to the fact that: The models output large amounts of data that are intractable in their raw form, model accuracy and biases are not well-understood and apparent to most forecasters, and it has not been clear how such models fit into the standard everyday workflow of operational avalanche forecasting.

Taking on these issues are Simon Horton and his supervisor Pascal Haegeli at the SFU Avalanche Research Program in the department of Resource and Evironmental Management. These researchers identified that visualization and visual analytics could start to address some of the challenges for operationalizing snowpack simulation models. We have been engaged in a several years long collaboration that has been focused on creating visual analytics solutions that make snowpack simulation models tractable, fit into everyday operational avalanceh forecasting, and set the foundation for future efforts that aim to clarify model accuracy/performance and improve it.


This project has resulted in snowpack simulations being increasingly used in operational settings. In the 2019-2020 forecast season, visualizations developed in collaboration with VIVA were used by forecasters to aid avalanche hazard assessments and in turn inform the production of public risk communications by Avalanche Canada at avalanche.ca. Due to the impacts of COVID-19 and the potential for public avalanche forecasters having to work with limited data gathered from field observations, the snowpack models have the potential to become a critical source of information for the upcoming winter (2020-2021).

Part of this project involved the design of new color palettes for displaying various snow "grain types". These new color palettes were designed to address perceptual issues with the previous palettes and to draw attention to areas of concern and risk. They are currently used as the primary color palette in operational snowpack models as well as publically available snowpack simulation visualizations at http://arfi.avalanche.ca/

This collaboration has resulted in one journal publication and one conference publication that also included a presentation:


The visualization below is a dashboard that is designed to match how avalanche forecasters think about and assess avalanche hazards. The visualizations support various analytic tasks that comprise the widely used Conceptual Model of Avalanche Hazard . The dashboard supports the identification of problematic layers, their prevalance, their potential destructive impact, and their distribution in various terrain classes. For a richer description please refer to the above paper "Enhancing the operational value of snowpack models with visualization design principles."