B0976
Title: Non-stationary models for extremal dependence
Authors: Carolin Forster - University of Stuttgart (Germany) [presenting]
Marco Oesting - University of Stuttgart (Germany)
Abstract: Being asymptotically justified by limit theorems, parametric models for Pareto processes have become popular choices to model spatial extreme events defined in terms of threshold exceedances. Apart from a few exceptions, existing literature mainly focuses on models with stationary dependence structures. Therefore, a non-stationary approach is developed that can be used for Pareto processes of both Brown-Resnick and extremal-t type - two of the most popular spatial process models, by including covariates in the corresponding variogram and correlation functions, respectively. In addition, the effect of random covariates is investigated on the theoretical properties of the limit processes defined conditionally on the covariates. It is shown that this approach can result in both asymptotically dependent and asymptotically independent processes. Thus, conditional models do not suffer from the usual restrictions of classical Pareto models.