CONNECTING STOCHASTIC AND DETERMINISTIC HYDROSTRATIGRAPHIC MODELS OF THE QUATERNARY FORT WAYNE MORAINE GLACIAL AQUIFER SYSTEM, ANN ARBOR, MICHIGAN, USA
Natural gamma radiation logs for 77 of the wells were digitized and used to interpret the locations of aquifer and aquitard subunits within each well. Gamma radiation measurements in counts per minute established a basis for stochastic modeling. Count data collected through the underlying shale bedrock were removed from the data set. 3D vertical and horizontal omnidirectional variograms were then constructed using raw gamma counts as a function of well location and depth. Vertical variograms incorporated small lag distances (30 cm or 1 ft) to account for high data density in the vertical direction. In the horizontal dimension, lag distances ranging from 6 to 300 m (20 to 1,000 feet) were examined to determine the short and long-scale correlation structure. A number of different approaches were used to explore variogram structure. The raw data were normalized to account for variation introduced by different logging and drilling equipment. Normalized gamma values were further transformed to indicator values based on pentile thresholds and deterministic classifications. Vertical variograms displayed clear structure, while horizontal variograms displayed almost pure nugget effect. Consequently, subsequent groundwater models constructed in this site will need to rely upon both the deterministic and stochastic methods employed.