Bayesian multiscale modelling of spatial extremes
Conference
65th ISI World Statistics Congress 2025
Format: IPS Abstract - WSC 2025
Keywords: bayesian approach, extreme value theory, mcmc simulation, spatial statistics
Session: IPS 754 - Advances in High-Dimensional Extreme Value Statistics
Wednesday 8 October 10:50 a.m. - 12:30 p.m. (Europe/Amsterdam)
Abstract
Relevant data and quantities of interest in extremes often refer to different spatial scales. Environmental data might, for instance, be available as point data from gauging stations or low-resolution gridded data from models, while some spatial aggregate of a high-resolution field is required. In this talk, we present a full Bayesian model framework that allows for down- and upscaling, e.g. by (conditional) simulation or by aggregation, based on spatial statistical models for extremes. All the necessary parameters and latent variables can be sampled via Markov Chain Monte Carlo methods.