B1688
Title: Effective probability distribution approximation for non stationary non Gaussian random fields
Authors: Anastassia Baxevani - University of Cyprus (Cyprus) [presenting]
Dinissios Hristopoulos - Technical University of Crete (Greece)
Christos Andreou - University of Cyprus (Cyprus)
Abstract: Most environmental data, like precipitation wind, are usually modelled in terms of stochastic fields. These fields need to possess not only complex spatial and temporal dependence structures but also to be non-stationary. Some of the non-stationarities may be due to the dynamic nature of the phenomena, and some others due to the fact that phenomena possess different properties depending on the location and the time of the year. Moreover, while environmental data often exhibit significant deviations from Gaussian behaviour, only a few non-Gaussian joint probability density functions admit explicit expressions. In addition, random field models are computationally costly for big datasets. An effective distribution approach is proposed, which is based on the product of univariate conditional probability density functions modified by local interactions. The effective densities involve local parameters that are estimated either by means of kernel regression or using kriging equations. The prediction of missing data is based on the median value from an ensemble of simulated states generated from the effective distribution model. The latter can capture non-Gaussian dependence and is applicable to large spatial datasets since it does not require the storage and inversion of large covariance matrices. It is concluded with an application of the methodology to precipitation data.
