International Snow Science Workshop Proceedings - Montana State University Library

Item: Avalanche Simulations in Forested Terrain: A Framework Towards a Bayesian Probabilistic Model Calibration

• • Title: Avalanche Simulations in Forested Terrain: A Framework Towards a Bayesian Probabilistic Model Calibration

    Proceedings: Proceedings, 2012 International Snow Science Workshop, Anchorage, Alaska

    Authors:

    • Michaela Teich  [ WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland ]  [ Planning of Landscape and Urban Systems PLUS, Swiss Federal Institute of Technology ETH, Zurich, Switzerland ]
    • Irene Vasella  [ WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland ]  [ Institute of Natural Resources Sciences, Zurich Un iversity of Applied Sciences ZHAW, Waedenswil, Switzerland ]
    • Perry Bartelt  [ WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland ]
    • Peter Bebi  [ WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland ]
    • Thomas Feistl  [ WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland ]  [ Technical University Munich (TUM), Munich, Germany ]
    • Adrienne Grêt Regamey  [ Planning of Landscape and Urban Systems PLUS, Swiss Federal Institute of Technology ETH, Zurich, Switzerland ]

    Date: 2012

    Abstract: Avalanche dynamics models are used for hazard zoning and engineering to predict runout distances and impact pressures of snow avalanche events. The effect of mountain forests as an effective biological protection measure against avalanches has rarely been addressed in this context. Avalanche runout distances of small to medium avalanches are strongly influenced by the structural conditions of forests in the avalanche path; however, this varying decelerating effect has not yet been implemented in avalanche models. Within the two-dimensional avalanche dynamics program RAMMS the standard Voellmy-Salm model can be applied to predict runout distances, flow velocities and impact pressures in complex three-dimensional terrain. Currently, the occurrence of forests is realized by increasing but constant friction parameters μ (dry-Coulomb type friction) and ξ (velocity squared friction) compared to open unforested terrain. Back-calculations of 41 well documented small avalanches which released in forests of the Swiss Alps emphasize the need for a further calibration dependent on differences in forest structure. Since the friction parameters are more conceptual than physical, they must be fitted by matching model results and recorded data which basically involves solving an inverse problem. A way of providing probabilistic statements about unobservable information is Bayesian inference. Therefore, we present a framework for a Bayesian probabilistic model calibration of the friction parameter ξ accounting for differences in forest structure in the avalanche path. Considering different forest characteristics within avalanche simulations will improve current applications for avalanche models, e.g. in mountain forest and natural hazard management.

    Object ID: issw-2012-628-632.pdf

    Language of Article: English

    Presenter(s): unknown

    Keywords: avalanche simulation, avalanche dynamics, avalanche runout, swiss alps

    Page Number(s): 628-632

    Subjects: avalanche forecasting hierarchical bayesian change point models

  • Get Full Article

Back to Home page