Evaluating extreme snow avalanches in long term forecasting
Résumé
Contents of the presentation : A few words about A few words about snow avalanches. Avalanche risk in the (French) Alpine space. Avalanche risk mitigation. Reference hazards in the snow and avalanche field. Are we using EVT for snow avalanches? The alternative: statistical-numerical (physical) modelling. Numerical modelling of avalanche flows. Building a statistical-numerical multivariate POT model. Simulation: joint distribution of model variables. Runout distance and return periods. Joint distribution P(v,h,μ.. I xstop > xstopT). Flow properties and impact pressure. Bayesian inference for the magnitude model. Generic principle of MCMC algorithms. Posterior distributions of magnitude model parameters. Latent variables and posterior correlation. Bayesian prediction of high runout distance percentiles. Back to EVT : Avantages and limitations. Validation of model predictions? Asymptotic properties of avalanches simulations. Response to climate change and stationarity.