B0817
Title: Modeling multivariate space-time extreme-event episodes with r-Pareto processes
Authors: Thomas Opitz - BioSP-INRAE (France) [presenting]
Abstract: Recent advances in peaks-over-threshold modelling of extreme-event episodes are reviewed using the class of r-Pareto processes that arise asymptotically when conditioning on increasingly large threshold exceedances of certain risk functionals. A large class of models related to log-Gaussian spectral processes allows the use of classical geostatistical tools such as variograms or Gaussian likelihoods. In a simulation study, fast and reliable estimation of parameters characterizing the extremal dependence is shown for exact and domain-of-attraction settings. The practical utility of the approach is illustrated through an application to multivariate space-time episodes of extreme weather events in the different administrative regions of France.