Title: Precipitation modeling via spatial vine copulas
Authors: Benedikt Graeler - Ruhr University Bochum (Germany) [presenting]
Abstract: Spatial vine copulas, where the bivariate copula building blocks are parameterized by the separating distance between locations, have proven useful to model skewed spatial random fields. Modeling and interpolating rainfall adds further challenges, as the marginal distribution is not only skewed, but also contains a majority of zeros. Furthermore, the dependence of a rainfall field may strongly differ between rainfall events ranging from extreme local rainfall to widespread continuous rainfall. These challenges are addressed with a spatial vine copula approach to model and interpolate precipitation.