Deterministic and Stochastic Modeling of Aquifer Stratigraphy, South Carolina

Abstract

AbstractDeterministic and stochastic methods of three‐dimensional hydrogeologic modeling are applied to characterization of contaminated Eocene aquifers at the Savannah River Site, South Carolina. The results address several important issues, including the use of multiple types of data in creating high‐resolution aquifer models and the application of sequence‐stratigraphic constraints. Specific procedures used include defining grid architecture stratigraphically, upscaling, modeling lithologic properties, and creating multiple equiprobable realizations of aquifer stratigraphy. An important question answered by the study is how to incorporate gamma‐ray borehole‐geophysical data in areas of anomalous log response, which occurs commonly in aquifers and confining units of the Atlantic Coastal Plain and other areas. To overcome this problem, gamma‐ray models were conditioned to grain‐size and lithofacies realizations. The investigation contributes to identifying potential pathways for downward migration of contaminants, which have been detected in confined aquifers at the modeling site.The approach followed in this investigation produces quantitative, stratigraphically constrained, geocellular models that incorporate multiple types of data from borehole‐geophysical logs and continuous cores. The use of core‐based stochastic realizations in conditioning deterministic models provides the advantage of incorporating lithologic information based on direct observations of cores rather than using only indirect measurements from geophysical logs. The high resolution of the models is demonstrated by the representation of thin, discontinuous clay beds that act as local barriers to flow. The models are effective in depicting the contrasts in geometry and heterogeneity between sheet‐like nearshore‐transgressive sands and laterally discontinuous sands of complex shoreline environments.