Geostatistical Methods and Applications

Abstract

Geostatistical analysis techniques are used to predict the values of parameters between data points. Geostatistical methods are only valid for spatially dependent (i.e., nonrandom) data. The basic method is to first identify and quantify the spatial structure of the variables of concern and then to interpolate or estimate the values of variables from neighboring values taking into account their spatial structure. Conditioning is the incorporation of hard or soft data into a model to reduce uncertainty. Hard data, by definition, has negligible uncertainty (e.g., direct measurements of property of interest), whereas soft data (inferred properties) have significant uncertainty. Geostatistical methods have been used to obtain realizations of sedimentary facies distributions, which typically require upscaling to a groundwater model grid, and assignment of hydraulic parameters values. A promising approach is hybrid methodologies that combine facies models and other soft geological information with geostatistical methods. Geostatistical techniques, when properly applied, are data intensive and are not a substitute for detailed field investigations and hydrogeological knowledge.