Strategies for managing water supplies and groundwater quality in the Modesto region of the eastern San Joaquin Valley, California, are being formulated and evaluated by the Stanislaus and Tuolumne Rivers Groundwater Basin Association. Management issues and goals in the basin include an area in the lower part of the basin that requires drainage of the shallow water table to sustain agriculture, intra- and inter-basin migration of poor-quality groundwater, and efficient management of surface and groundwater supplies. To aid in the evaluation of water-management strategies, the U.S. Geological Survey and the Stanislaus and Tuolumne Rivers Groundwater Basin Association have developed a hydrologic model that simulates monthly groundwater and surface-water flow as governed by aquifer-system properties, annual and seasonal variations in climate, surface-water flow and availability, water use, and land use. The model was constructed by using the U.S. Geological Survey groundwater-modeling software MODFLOW-OWHM with the Farm Process.

Available measurements of groundwater pumped for municipal, irrigation, and drainage purposes are specified in the model, as are deliveries of surface water. Private irrigation pumping and recharge associated with agricultural land use were estimated by using the Farm Process in MODFLOW-OWHM, which simulates landscape processes associated with irrigated agriculture and other land uses. The distribution of hydraulic conductivity in the aquifer system was constrained by using data from more than 3,500 drillers' logs. The model was calibrated to 4,061 measured groundwater levels in 109 wells and 2,739 mean monthly surface-water flows measured at 6 streamgages during 1960-2004 by using a semi-automated method of parameter estimation.

The model fit to groundwater levels was good, with an absolute mean residual of 0.8 feet; 74 percent of simulated heads were within 10 feet of those observed. The model fit to streamflow was biased low, but reasonable overall; the absolute mean residual of streamflow was 780 cubic feet per second, and 68 percent of simulated streamflows were within 500 cubic feet per second of observed. Hydrographs both of groundwater levels and streamflow indicated overall an acceptable fit to observed trends.

Simulated private agricultural pumpage ranged from about 780,000 to 1,380,000 acre-feet per year and averaged about 1,000,000 acre-feet per year from 1960 to 2004. Simulated deep percolation, or groundwater recharge from precipitation and irrigation, varied with climate and land use from about 1,100,000 to 1,700,000 acre-feet per year, averaging 1,360,000 acre-feet per year. Key limitations of the model with respect to estimating these large components of the water budget are the uncertainty associated with actual irrigation deliveries and irrigation efficiencies and the lack of metered data for private agricultural groundwater pumping. Different assumptions with respect to irrigation deliveries and efficiencies, and other model input, would result in different estimates of private agricultural groundwater use.

The simulated exchange between groundwater and surface water was a small percentage of streamflow, typically ranging within a loss or gain of about 2 cubic feet per second per mile. The simulated exchange compared reasonably with limited independent estimates available, but substantial uncertainty is associated with these estimates.