On Extreme Value Analysis of a Spatial Process

Abstract

One common way to deal with extreme value analysis in spatial statistics is by using the max-stable process. By employing a representation of simple max-stable processes in de Haan and Ferreira ([3]), we propose a stationary max-stable process as a model of the dependence structure in two-dimensional spatial problems. We calculate its two-dimensional marginal distributions, which creates the opportunity to estimate the dependence parameter. The model is used in Buishand, de Haan and Zhou ([1]) for a spatial rainfall problem.