Water users rely on surface water and ground water to irrigate crops and to maintain lakes on wildlife refuges in the 2,200-square-mile Mud Lake study area. Ground-water development between the late 1970's and 1989 increased withdrawals from about 240,000 acre-feet in 1983 to about 370,000 acre-feet in 1990. Concurrent with ground-water development, change from subirrigation to sprinkler irrigation was predicted to reduce recharge by 95,000 acre-feet, according to an independent study. Of the 660,000 acre-feet total estimated recharge from precipitation and irrigation in the study area in 1980, half was in the area in which irrigation methods were changed. Water managers need the ability to evaluate the effects of water-use changes on the future supply of surface water and ground water.

Basalt and rhyolite predominate on the surface and in the subsurface of the study area. Total basalt thickness is less than 4,000 feet; total sediment thickness (clay, silt, sand, and gravel) is less than 1,000 feet. Basalt and sediment interbeds contribute to confined ground-water conditions and affect movement and supply of water in parts of the aquifer system.

Estimated losses from and gains to perennial streams and lakes in 1980 were each about 110,000 acre-feet. Water-table altitudes ranged from about 4,500 to 6,200 feet above sea level, and water-table gradients were 3 to 120 feet per mile. Underflow from basins tributary to the study area was estimated to be about 450,000 acre-feet in 1980; measured discharge from flowing wells was about 10,000 acre-feet.

A five-layer, three-dimensional, finite-difference, numerical ground-water flow model was calibrated by trial-and-error to assumed 1980 steady-state hydrologic conditions to obtain a better understanding of the geohydrology and provide a tool to evaluate water-use alternatives. Water-level gradients simulated by the model were similar to gradients measured in 1980. Simulated underflow across model boundaries for 1980 was 932,000 acre-feet. Simulated losses from and gains to most streams and lakes were within 2 percent of estimated values. Simulated discharge from flowing wells matched measurements for 1980. An attempt to calibrate the numerical model to transient hydrologic conditions in monthly increments from 1981 to 1990 was discontinued because available data did not justify changes that were indicated by model simulations.