Copulas, a novel approach to model spatial and spatio-temporal dependence

Gräler Benedikt, Kazianka Hannes, deEspindola Giovana

Research article (journal)

Abstract

Copulas are a statistical concept which allows for a novel approachto model dependencies of spatial and spatio-temporal variables. Theycapture the dependence structure of a multivariate distribution over itswhole range detached from its specific margins. In contrast to a single measureof association, this allows for varying strength of dependence throughoutthe multivariate distribution. Thus, copulas are capable of capturing manydifferent (i.e. non-Gaussian) dependence structures and allow for asymmetricdependencies which can be found in many natural processes. We appliedthis approach to data from the deforestation survey of the Brazilian Amazonin order to capture the dependence of deforestation on a selection of variables.

Details zur Publikation

Pages: 6
Release year: 2010
Language in which the publication is written: English