Item: PROTOTYPE OF PROBABILISTIC AVALANCHE HAZARD MAP REFLECTING UNCERTAINTIES OF SNOW COVER CONDITIONS DURING WINTER
-
-
Title: PROTOTYPE OF PROBABILISTIC AVALANCHE HAZARD MAP REFLECTING UNCERTAINTIES OF SNOW COVER CONDITIONS DURING WINTER
Proceedings: International Snow Science Workshop 2024, Tromsø, Norway
Authors:
- Takahiro Tanabe [ National Research Institute For Earth Science and Disaster Resilience ]
- Sojiro Sunako [ National Research Institute for Earth Science and Disaster Resilience ]
- Kouichi Nishimura [ Professor emeritus of Nagoya University ]
Date: 2024-09-23
Abstract: Snow avalanches are natural disasters that occur in snowy mountainous regions, threatening settlements and their inhabitants. Hence, a hazard map is a means of mitigating their impact. Avalanche flow models have been employed to indicate avalanche runout areas. These models require inputs, such as the digital elevation model, release area, initial conditions, and model parameters. Although the dynamics of the models depend on the input values, they are generally uncertain. In this study, we considered three uncertain input variables: the initial flow thickness, friction coefficient, and drag coefficient, whereas the other input variables were set to constant values. Additionally, we assumed that the uncertainties in the input variables could be represented by probability density distributions and combined them with flow models to generate probabilistic hazard maps. In this study, probability refers to the exceedance probability that the output of the flow model, such as the maximum flow thickness, exceeds a given threshold value. To evaluate the uncertainties, we used the Polynomial Chaos Quadrature method and prototyped probabilistic avalanche hazard maps using the following procedure: Potential Release Areas (PRAs) were identified on a specific slope in Japan. Assuming lower and upper bounds for each uncertain input variable of the flow model, we created a probabilistic hazard map for avalanches at a specific PRA. Two scenarios for uncertain input distributions are considered: uniform and Gaussian distributions. The uniform distribution case indicates superimposing potential avalanches on the PRA, and the Gaussian distributions were determined to reflect actual snow conditions, such as snow depth and snow wetness, which update the uncertain distributions. The necessity of reflecting snow conditions was confirmed by comparing the hazard maps. Additionally, by reflecting snow conditions, the runout areas of past avalanche events can be reproduced. These probabilistic hazard assessments led to quantitative risk evaluations, which will be the focus of future work.
Object ID: ISSW2024_P7.4.pdf
Language of Article: English
Presenter(s): Takahiro Tanabe
Keywords: Probabilistic hazard map, Uncertainty quantification, Hazard assessment, Risk assessment
Page Number(s): 897 - 904
-