Spatial variability of geologic media, and consequently of media properties, is of key concern in ground-water transport studies, yet important information about geologic processes governing the creation of aquifer architecture has not been incorporated into hydrogeologic models. In previous modelling approaches, spatially variable aquifer properties have been inferred from the statistical characteristics of a few measured data or an assumed spatial covariance model. Typically there is a neglect of interpretive geological information, even when it is readily available. It is unlikely that existing (spatial) statistical models capture many of the essential elements of actual aquifer systems and may in fact produce a geologically "impossible" subsurface architecture. Therefore, we propose to replace the traditional stochastic simulation approach with a physically-based approach based upon synthesized geology which is generated by a sedimentary process model. We will explicitly consider the geologic processes which control aquifer evolution and which result in heterogeneity of aquifer properties. This approach combines sedimentary processes simulation, petrophysics, geostatistics, and groundwater modelling. We plan to use simulation to generate a three-dimensional synthetic aquifer (geometry and sedimentary geology), transform the geologic information at a very fine scale (i.e., centimeters) into aquifer hydraulic properties, and simulate three- dimensional groundwater transport using large scale numerical flow simulation and particle tracking. We plan to compare the geologic-processes oriented approach with the traditional stochastic/statistical approach to groundwater flow simulation. Our approach will be tested on a complex field area (Santa Clara Valley, CA) as well as applied to hypothetical aquifer systems representing a variety of sedimentary environments.
