A three-dimensional numerical model of groundwater flow was constructed for the Columbia Plateau Regional Aquifer System (CPRAS), Idaho, Oregon, and Washington, to evaluate and test the conceptual model of the system and to evaluate groundwater availability. The model described in this report can be used as a tool by water-resource managers and other stakeholders to quantitatively evaluate proposed alternative management strategies and assess the long‑term availability of groundwater. The numerical simulation of groundwater flow in the CPRAS was completed with support from the Groundwater Resources Program of the U.S. Geological Survey Office of Groundwater.

The model was constructed using the U.S. Geological Survey modular three-dimensional finite-difference groundwater-flow model, MODFLOW-NWT. The model uses 3-kilometer (9,842.5 feet) grid cells that subdivide the model domain by 126 rows and 131 columns. Vertically, the model domain was subdivided into six geologic model units. From youngest to oldest, the units are the Overburden, the Saddle Mountains Basalt, the Mabton Interbed, the Wanapum Basalt, the Vantage Interbed, and the Grande Ronde Basalt.

Natural recharge was estimated using gridded historical estimates of annual precipitation for the period 1895–2007. Pre-development recharge was estimated to be the average natural recharge for this period. Irrigation recharge and irrigation pumping were estimated using a remote-sensing based soil-water balance model for the period 1985–2007. Pre-1985 irrigation recharge and pumping were estimated using previously published compilation maps and the history of large-scale irrigation projects. Pumping estimates for municipal, industrial, rural, residential, and all other uses were estimated using reported values and census data. Pumping was assumed to be negligible prior to 1920.

Two models were constructed to simulate groundwater flow in the CPRAS: a steady-state predevelopment model representing conditions before large-scale pumping and irrigation altered the system, and a transient model representing the period 1900–2007. Automated parameter-estimation techniques (steady-state predevelopment model) and traditional trial-and-error (transient model) methods were used for calibration. To calibrate the steady-state and transient models, 10,525 and 46,460 water level measurements, respectively, and 50 base-flow estimates were used.

The steady-state model simulated the shape, slope, and trends of a potentiometric surface that was generally consistent with mapped water levels. For the transient model, the mean and median difference between simulated and measured hydraulic heads is -10 and 4 ft, respectively, with a standard deviation of 164 ft over a 5,648 ft range of measured heads. The residuals for the simulation period show that 52 percent of the simulated heads exceeded measured heads with a median residual value of 43 ft, and 48 percent were less than measured heads with a median residual value of -76 ft.

The CPRAS model was constructed to derive components of the groundwater budget and help understand the interactions of stresses, such as recharge, groundwater pumping, and commingling wells on the groundwater and surface-water system. Through these applications, the model can be used to identify trends in groundwater storage and use, and quantify groundwater availability. The annual groundwater budgets showed several patterns of change over the simulation period. Groundwater pumping was negligible until the 1950s and began to increase significantly during the 1970s and 1980s. Recharge was highly variable due to the interannual variability of precipitation, but began to increase in the late 1940s due to the increase in surface-water irrigation projects. Groundwater contributions to streamflow (base flow) followed recharge closely. However, in areas of significant groundwater-level decline, base flow is reduced.

Groundwater pumping had the greatest effect on water levels, followed by irrigation enhanced recharge. Commingling was a larger factor in structurally complex upland areas where hydraulic-head gradients are naturally high.

Groundwater pumping has increased substantially over the past 40–50 years; this increase resulted in declining water levels at depth and decreased base flows over much of the study area. The effects of pumping are mitigated somewhat by the increase of surface-water irrigation, especially in the shallow Overburden unit, and commingling wells in some areas. During dry to average years, groundwater pumping causes a net loss of groundwater in storage and current condition (2000–2007) groundwater pumping exceeds recharge in all but the wettest of years.