Item: Optimization of Computational Snow Avalanche Simulation Tools
-
-
Title: Optimization of Computational Snow Avalanche Simulation Tools
Proceedings: International Snow Science Workshop 2014 Proceedings, Banff, Canada
Authors:
- Jan-Thomas Fischer [ Austrian Research and Centre for Forests - BFW, Department of Natural Hazards, Innsbruck, Austria ]
- Andreas Kofler [ Austrian Research and Centre for Forests - BFW, Department of Natural Hazards, Innsbruck, Austria ] [ University of Innsbruck, Institute of Infrastructure, Division of Geotechnical and Tunnel Engineering, Innsbruck, Austria ]
- Wolfgang Fellin [ University of Innsbruck, Institute of Infrastructure, Division of Geotechnical and Tunnel Engineering, Innsbruck, Austria ]
- Matthias Granig [ Snow and Avalanche Center, Avalanche and Torrent Control (WLV), Innsbruck, Austria ]
- Karl Kleemayr [ Austrian Research and Centre for Forests - BFW, Department of Natural Hazards, Innsbruck, Austria ]
Date: 2014-09-29
Abstract: Snow avalanche simulation tools are used for hazard estimations and protection planning. Initial conditions and flow model parameters have to be chosen carefully in order to gain meaningful simulation results. A depth averaged flow model is used for this investigation, where simple entrainment and friction relations are implemented in the SamosAT simulation software. The employed mass balance relation allows for the full range of entrainment mechanisms, from frontal plowing to gradual erosion. The initial snow reservoir distribution for release and entrainment is estimated by measurement and empirical observation for the entire mountain. Flow model parameters for the entrainment model and the Voellmy friction relation are systematically optimized by back calculating a documented event. The simulation results are analyzed in three-dimensional terrain with the help of a transformation into an avalanche path dependent coordinate system. Six different optimization variables are scrutinized, related to runout, affected area, velocity, deposition depth and mass growth due to entrainment. The optimization method explicitly takes the uncertainties associated with the observational variables into account. To cover the entire physically relevant parameter range a large number (104) of random flow model parameter combinations and their corresponding simulation runs are investigated. This yields posterior parameter distributions representing optimal parameter combinations, which are of fundamental interest for engineers and scientists. We demonstrate how the proposed systematic simulation analysis contributes to an objective parameter calibration and optimization.
Object ID: ISSW14_paper_P2.08.pdf
Language of Article: English
Presenter(s):
Keywords: snow avalanche, computational avalanche dynamics, simulation optimization.
Page Number(s): 708-712
-