Sampling spatial structures in geostatistical framework

Abstract

Extreme values geostatistics make it possible to model the asymptotic behaviors of random phenomena which depends on space or time parameters. In this paper, we propose new models of the extremal coefficient within a spatial stationary fields underlied by multivariate copulas. Some models of extensions of the extremogram and the cross-extremogram are constructed in a spatial framework. Moreover, both these two geostatistcal tools are modeled using the extremal variogram which characterizes the asymptotic stochastic behavior of the phenomena.