Extremes: spatial parametric modeling
Statistics of spatial extremes is developing very rapidly, owing to the demands of applications in the environmental sciences and the insurance and risk industries. This entry sketches the main ideas, based on classical extreme-value statistics. The two main threads of work are the fitting of max-stable models to spatial extremes, and the use of latent variable models, the latter typically in a Bayesian context.
Keywords: Bayesian hierarchical model; Brown--Resnick process; Composite likelihood; Copula; Environmental data analysis; Gaussian process; Generalised extreme-value distribution; Geostatistics; Latent variable; Max-stable process; Statistics of extremes
Record created on 2012-11-20, modified on 2016-08-09