Item: EVALUATION OF THE TRANSFERABILITY TO THE PYRENEES OF MACHINE LEARNING MODELS DEVELOPED FOR AVALANCHE FORECASTING IN THE SWISS ALPS
-
-
Title: EVALUATION OF THE TRANSFERABILITY TO THE PYRENEES OF MACHINE LEARNING MODELS DEVELOPED FOR AVALANCHE FORECASTING IN THE SWISS ALPS
Proceedings: International Snow Science Workshop 2024, Tromsø, Norway
Authors:
- Montse Bacardit [ Centre Lauegi Aran - Conselh Generau Aran ]
- Cristina Pérez-Guillén [ WSL Institute for Snow and Avalanche Research SLF ]
- Jordi Gavaldà Bordes [ Centre Lauegi Aran - Conselh Generau Aran ]
- Ivan Moner Seira [ Centre Lauegi Aran - Conselh Generau Aran ]
- Martin Hendrick [ WSL Institute for Snow and Avalanche Research SLF ]
- Fabiano Monti [ Alpsolut SLR ]
Date: 2024-09-23
Abstract: Avalanche forecasting is crucial for ensuring safety and mobility in snow-covered mountainous regions worldwide. While traditional avalanche forecasts relied solely on human expertise, machine-learning models have been developed in recent years, and they are already in operation for avalanche forecasting in Switzerland. Such models can rapidly assimilate large amounts of complex datasets, providing objective and high-resolution predictions on the avalanche danger level, the snow-pack instability, and wet-snow avalanche activity, assisting forecasters with a “second opinion”. These models were originally trained on a historical database containing 20 years of Swiss meteorological measurements, snow cover simulations, danger level assessments and avalanche observations. The performance of the models was comparable to the experienced-based avalanche forecasts in the Swiss Alps, although some variations were observed in different climatic regions. For the future operational application of the models, there is the challenge of addressing the spatial variation of predictive performance between different snow climate regions, and hence verifying the capability to transfer such models into other mountain regions outside of Switzerland. In this study, we test for the first time the avalanche danger and wet-snow avalanche activity models using a database recorded during the last winter seasons in Val d’Aran, a small county on the Atlantic watershed of the Spanish Central Pyrenees. The snow climate has oceanic influence, with an average annual snowfall of 5-6 meters and the snowpack reaching 1-3 meters in high altitudes during mid-winter. The Aran Avalanche Centre issues a daily public regional avalanche forecast based on EAWS standards. Additionally, for the last winter seasons, the SNOWPACK model has provided snow cover simulations for two mountain meteorological stations for the flat field and four mountain slope aspects (north, east, south, west). These data are used as input to operate the danger level and avalanche activity models. Even though Val d’Aran is a relatively small mountain region, the high-quality data collected there, including daily avalanche danger level forecasts, snow-cover simulations, snowpack stability field assessments, and avalanche activity observations, makes it a perfect study site to test and validate the predictions of the Swiss machine-learning models in a region with different snow climate conditions.
Object ID: ISSW2024_P1.27.pdf
Language of Article: English
Presenter(s): Montse Bacardit
Keywords: machine learning, avalanche forecasting, avalanche danger, avalanche activity, SNOWPACK model
Page Number(s): 241 - 247
-