Summary: Most standard methods of geostatistical analysis are built upon the basic assumption of isotropy. However, spatial data in most cases are not isotropic. In practice typically we find data with zonal anisotropy. Zonal anisotropy is more complex than geometric; what is more, most software requires anisotropy to be geometric. In this paper an overview of models of zonal anisotropy is presented. Models of variograms with zonal anisotropy were fitted to the salinity data using R, the system for statistical computations and analysis.