Geostatistical modelling of environmental data

Summary Many studies in environmental research include the analysis of spatially distributed measurements. The spatial representativity of the data can be quantified by modelling its autocorrelation structure with geostatistical methods based on the theory of spatial stochastic processes. The estimation of variogram functions — a measure to describe spatial variability — is illustrated by using precipitation and deposition data. The main charactristics of the variogram are discussed and interpreted. Kriging, a weighted moving average interpolation technique with weights depending on the specified variogram model, takes into account the spatial autocorrelation of the data. In addition kriging provides information about the accuracy of interpolation. That's why kriging and variogram analysis are useful tools for data interpolation and for network design in environmental studies.