Modelling of Geometric Anisotropic Spatial Variation

Abstract

Isotropic processes form an inadequate basis in modelling many spa- tially distributed data. In particular environmental phenomena often have strong anisotropic spatial variation, especially when the regions monitored are very large (see 1). Among different forms of spatial anisotropy a geometric anisotropy is most common (see 4). Geometric anisotropy, which provides the most common genera- tion of isotropy within stationarity, is typically dealt with by simple transformations of coordinates. For modelling spatial processes, we propose a rich class of stationary geometric anisotropic variograms. Objective of our investigation is to select and identify optimal models of isotropic variograms for different direction regions, using R, a system for statistical computa- tion and graphics 2. Spatial data was used for realization of the proposed modelling procedure. General form of geometric anisotropic semivariogram for salinity data was obtained.