Japan Geoscience Union Meeting 2022

Presentation information

[J] Poster

M (Multidisciplinary and Interdisciplinary) » M-IS Intersection

[M-IS16] Geophysical particulate gravity current

Tue. May 31, 2022 11:00 AM - 1:00 PM Online Poster Zoom Room (30) (Ch.30)

convener:Hajime Naruse(Department of Geology and Mineralogy, Graduate School of Science, Kyoto University), convener:Yuichi Sakai(Graduate School of Science, Kyoto University), Hiroyuki A. Shimizu(National Research Institute for Earth Science and Disaster Resilience), convener:Takahiro Tanabe(National Research Institute for Earth Science and Disaster Resilience), Chairperson:Yuichi Sakai(Faculty of Agriculture, Utsunomiya University), Hiroyuki A. Shimizu(National Research Institute for Earth Science and Disaster Resilience)

11:00 AM - 1:00 PM

[MIS16-P01] A study for hazard mapping method with an uncertainty input for depth-averaged avalanche model

*Takahiro Tanabe1 (1.National Research Institute for Earth Science and Disaster Resilience)

Keywords:Avalanch, Hazard map, Numerical model

Avalanche model with shallow-water equation are powerful tools to predict the runout distance. In the input of avalanche model, there include various kinds of uncertainties, such as friction coefficient, fluid viscosity, and initial conditions. These input uncertainties are propagated to outputs via the model, which is called uncertainty propagation. In order to obtain useful hazard map with the numerical model, this uncertainty propagation should be taken into account.
In this presentation, three methods are employed to evaluate the uncertainties of inputs and draw hazard maps, Monte Carlo (MC), Latin Hypercube Sampling (LHS), and Polynomial Chaos Quadrature (PCQ). Both MC and LHS are sampling-base method, and PCQ consists of polynomial chaos expansion and Gaussian quadrature, i.e. a polynomial approximation of outputs. We quantify an uncertainty of initial avalanche volume in the model with each method, then draw resultant hazard maps. Comparison among methods revealed that PCQ showed the better result in the view of computational cost and the quality. In addition, sampling-base method became better with increasing sampling and simulations, on the other hand, PCQ had a proper number of simulations above which the quality of hazard map did not become better.