Hydrostratigraphic interpretation using indicator geostatistics

In this paper we describe an approach that uses indicator geostatistics to interpret qualitative borehole logs and compute experimental variograms for complex alluvial sediments. Borehole descriptions are first transformed into binary indicator values based on inferred relative permeability from the borehole descriptions. The resulting indicator data can then be used to compute variograms and construct three‐dimensional variogram models. The ranges of computed indicator variograms for a groundwater contamination site in Santa Clara Valley, California, are very sensitive to the orientation of the search plane. These variograms are consistent with known stratigraphie features and describe the spatial structure of deposits from different depositional environments. Indicator kriging weighs all the available data on the basis of a three‐dimensional, anisotropic variogram model and provides an estimate of uncertainty in the hydrostratigraphic correlation. Kriged indicator values represent probabilities that sediments at a specific location fall into one of two indicator categories. The location of the 0.5 indicator contour is approximately the boundary between high‐ and low‐permeability sediments that might be constructed in a geologic cross section.