OPTICAL MONITORING OF AVALANCHE REALEASE ZONES IN BESSANS, HAUTE MAURIENNE VALLEY, FRANCE, WITH A REMOTE AND ENERGY SELF-SUFFICIENT CAMERA SYSTEM
Résumé
The slopes between the Pointes du Ch âtelard and the Pointe de Claret in the Vanoise National Park, Haute-Maurienne in France are located right above the village of Bessans, its main access road and ski lifts. In recent years, several significant avalanches came down these slopes, potentially threatening the village. Notably, several bungalows of the local camping ground were partially destroyed by a large powder avalanche impact in 2004. Avalanche safety in the valley is managed by a committee consisting of road authorities, local officials, and independent avalanche experts. Due to the topography, most slopes are not visible from the valley, hence limited information about snow coverage and the avalanche situation of the upper part of the slopes is available. This causes a major challenge for the decision-makers who take actions such as avalanche control or road closing. Being the only access road to the valley, minimizing road closure is critical for its 6000 inhabitants. To address this challenge, an energy self-sufficient camera system was installed at almost 3000 masl on a ridge on the opposite side of the valley, to automatically take photographs of the avalanche slopes. The camera was modified to enhance contrast of the snow cover and avoid pixel saturation due to high snow albedo. Photographs can be recorded every 15 minutes at 42 MP resolution and in a high dynamic range. A specific high dynamic range mode was developed for night conditions so that valuable information is available 24/7 for the avalanche safety committee. Preliminary results of image analysis tests performed using both digital image correlation and deep learning algorithms demonstrate the potential of this quantitative method to support decision-making. Digital image correlation is used to measure local accelerations in the snow cover that could potentially lead to glide-snow avalanches, while deep learning algorithms classify events. During the first winter of application, the 2023/2024 season, camera images played a crucial part in the decision-making process of the avalanche safety committee and can be used by the avalanche experts to further evaluate critical remote locations of the region known to be prone to similar conditions and where no data is available until now.
Origine | Fichiers produits par l'(les) auteur(s) |
---|