Title: Modelling extreme European windstorms with functional peaks-over-threshold analysis
Authors: Raphael de Fondeville - EPFL (Switzerland) [presenting]
Anthony Davison - EPFL (Switzerland)
Abstract: Estimating the risk of extreme windstorms has become important in recent decades, but up until now it has been largely limited to re-using catalogues of historical events, which usually do not exceed 40 to 50 years in length, and to numerical models, which require heavy computation but are often unreliable for extrapolation. Extreme value theory provides statistical methods for estimating the frequency of past extreme events as well as for extrapolating beyond observed severities, but natural hazards cannot be modelled using only univariate results. We present a statistical methodology based on functional peaks-over-threshold analysis which allows one to define complex extreme events as special types of exceedances, and then obtain their limit distribution for increasingly high thresholds, namely the generalized $r$-Pareto process. We describe a model based on log-Gaussian functions, which enables to use classical Gaussian covariance structures to model extremal dependence and then develop a stochastic weather generator for extreme windstorms over Europe, capable of quantifying the recurrence of past events as well as generating completely new ones.