The alluvial basin of northern Santa Clara Valley in central California has undergone extensive ground water development since early this century. Typical of many alluvial basins in the Western United States, the basin consists of coalesced fans deposits with little lateral continuity among individual sand and gravel zones.
The discontinuous lithology of Santa Clara Valley Basin, coupled with the sometimes minimalist lithologic logs provided by water well drillers, frustrate our attempts to use geology as a surrogate for defining the hydrostratigraphy and resultant model layers. Our alternative approach uses well completion data, supplemented by geologic data, to define depths of water-bearing zones. This method of defining hydrostratigraphy assumes that water well drillers typically recognize and successfully locate preferential depths for producing ground water.
We grouped well completion data from approximately 400 wells into basin subareas, based on general basin geology. Within each subarea, well screen abundance was graphed by depth. The screen-interval abundance graphs, supplemented with existing geologic interpretations, provided the basis of our hydrostratigraphic interpretation.
Screen-interval abundance suggested that the previously accepted three layer conceptual model of the Santa Clara Valley Basin could be divided into six hydrogeologic layers. We interpolated the six layers identified in each subarea to form basinwide layers, primarily based on ground water flow patterns and depth, with secondary consideration to geology. Geologic data and aquifer test data provided the basis for zoned hydrologic parameters in each layer. This method of estimating hydrostratigraphy, and subsequent model layers, results in a bias for placing well screens in individual model layers. This bias aids model calibration by associating historical water levels with individual model layers.
This six layer conceptual model provided a basis for a numerical flow model that successfully simulated ground water flow conditions in the Santa Clara Valley. Existing ground water maps showed important flow features that previous model attempts, incorporating simpler conceptual models, could not simulate. By reconciling the simpler conceptual models with our new conceptual model, we demonstrated that the important flow features were simulated, even though no single model layer reproduced the flow features observed on the existing ground water maps.