Use of a precipitation-runoff model to simulate natural streamflow conditions in the Methow River basin, Washington

Management of the water resources of the Methow River Basin is changing in response to the listing of three species of fish under the Endangered Species Act and the Washington State-legislated watershed-planning process. Management options must be considered that minimize adverse effects on people but meet instream flow needs for fish. This report describes the construction and calibration of the Methow River Basin watershed model and evaluates the accuracy of the model as a predictive tool for assessing the natural instream flow conditions. The term ?natural? instream flow is stressed because surface water within the Basin is used for agricultural irrigation through an extensive system of diversions.

The USGS Modular Modeling System was used for the watershed modeling component of the Methow River Basin study. The Geographic Information System Weasel characterized the physical properties of the basin, and the Precipitation-Runoff Modeling System simulated the natural streamflow. Natural streamflow conditions in the Basin were difficult to calibrate because six of the seven streamflow gaging stations are located below irrigation diversions and few streamflow measurements exist for the study area before the diversions were present. Therefore, limited records of natural streamflow conditions were available and estimations concerning some physical processes could not be quantified.

Streamflow was simulated for water years 1992-99 to calibrate the model to measured streamflows. Simulated and measured streamflow generally showed close agreement, especially during spring runoff from snowmelt. Low-flow periods, most restrictive to fish habitation, were simulated reasonably well, yet possessed the most uncertainty. Simulations of the total annual runoff as a percentage of measured annual runoff for the 8-year calibration period at seven gaging stations ranged from -33.7 to +30.5 percent with 70 percent of the simulated values within 16 percent. Simulation of water years 1959-99 demonstrated great variability in monthly streamflow statistics. The simulated mean monthly flows for the seven streamflow-gaging stations were an average of 11.5 percent higher for the calibration period (1992-99) than for the entire simulation period (1959-99).