Modeling spatial extremes using normal mean-variance mixtures

Classical models for multivariate or spatial extremes are mainly based upon the asymptotically justified max-stable generalized Pareto processes. These suitable when asymptotic dependence is present, i.e., joint tail decays at same rate as marginal tail. However, recent environmental data applications suggest that independence equally important and, unfortunately, existing in this setting both flexible and can be fitted efficiently scarce. Here, we propose a new copula model on hyperbolic distribution, which specific normal mean-variance mixture very popular financial modeling. The properties of distribution have been studied literature, but with contradictory results. It turns out proofs from literature contain mistakes. We here give corrected theoretical description its structure then exploit to analyze simulated dataset inverted Brown–Resnick process, hindcast significant wave height North Sea, wind gust state Oklahoma, USA. demonstrate our proposed enough capture not only also bulk.