The detailed 3-D hydrogeologic model was based on some 600 lithologic well logs in the Hays area. Private wells are abundant throughout the study area (Fig. 7) logs of which were assembled by RSKERC staff into a single database. These include 366 wells summarized by Perez (1986). Additional published data was derived from a USGS drilling effort, undertaken to improve water supplies at a local Army Air Force base during World War II [Latta, 1948]. An unpublished consulting report also described efforts in the early 1970's to explore the Big Creek aquifer at greater distances from Hays [Nuzman, 1974]. Remedial investigations relating to the PCE plume also provided some well test data [Black & Veatch Waste Science, 1995]. A problem in interpreting these data sources is that the sand-gravel aquifer tends to form the basal Quaternary deposits in the valley, and driller's logs tend to neglect isolated sand-gravel bodies above the main aquifer, as well as silt-clay zones within the aquifer. This tendency represents an unresolvable bias in the data, and therefore in the results described below.
As discussed above, the hydrostratigraphy at Hays can be adequately described by three layers, where the middle layer (Qal) is the most productive unit (Fig. 5). Qualitative and anecdotal information suggests hydraulic conductivity in the Qal sand-gravel aquifer is an order of magnitude higher than in the overlying Qt silt-clay deposits, which is at least an order of magnitude higher than in the bedrock Kc layer. The distribution of Qal has long been held to be the principal control on well productivity in the area, and wells intercepting the greatest thickness of Qal consistently show the highest yield. Initial and sustained yields show a positive, roughly linear, relationship with Qal thickness encountered in city wells [Latta, 1948].
The hydrogeologic model was derived by lateral interpolation of borehole intercepts of hydrostratigraphic unit boundaries. Smoothed representations of the upper bounding surfaces of Kc, Qal, and Qt and time-averaged water table were determined in this fashion. Volume representations of these units are depicted in the 3-D model accompanying this report. Discontinuity of the Quaternary units is immediately obvious (Fig. 8). Since the Quaternary deposits were formed in a meandering stream environment, the Qal lenses and areas of greatest thickness are quite variable, abruptly terminating and reappearing in a pattern typical of scour and fill environments. As a result, aquifer thickness as determined from the RSKERC Hays well database is very poorly correlated in space. Efforts to calculate a useful semi-variogram from thickness or contact elevation data were unsuccessful, and indicate spatial correlation over distances less than 100 m, smaller than the spacing between all but a few of the wells in the database. Correlation was slightly better parallel to the trend of Big Creek valley, with the second principal axis perpendicular to this trend. Lack of useful spatial correlation indicates that a deterministic 3-D hydrogeologic model will be required describe transport potential at Hays.