Integration of watershed and groundwater models with application to Lower Republican River basin

A documented model code is presented that links simulations of watershed hydrology and a stream-aquifer system, and is demonstrated by application to the Lower Republican River basin in Kansas. The code links SWAT, a daily soil water budget simulator for a river basin, with MODFLOW, which simulates saturated ground water flow and stream-aquifer interaction. The linkage implements a conceptual model for spatial heterogeneity in the unsaturated zone and hydrologic connections to the stream-aquifer system. Based on spatially variable factors of soil type, land use, and geomorphology, a watershed is decomposed into hydrologically similar components known as hydrologic response units (HRUs). These are simulated separately by SWAT and then averaged. These results are used to specify flux conditions for MODFLOW's solution, including ground water recharge, tributary inflow, pumping rates from both ground and surface water diversions, and evaporation from shallow ground water.; In addition to a basic conceptual model of spatial heterogeneity (or HRU scheme) that considers land use and soil type as independent factors, a second HRU scheme disaggregates the alluvial valley underlain by an aquifer from upland areas underlain by bedrock. A third scheme further disaggregates deep and shallow ground water, and simulates capillary uptake from shallow ground water into the unsaturated zone. This coupling is achieved with separate execution of SWAT and MODFLOW by use of successive approximation.; The basin model was calibrated to observed ground water levels, stream yield, and reported irrigation water use. Of these, simulated stream yield showed the greatest residual variance, attributed to not only model error but also uncertainty in measured precipitation. Tributary flow is the dominant component of simulated stream yield. For the study period 1977--1994, simulated baseflow ranges from negligible to negative, contributing significantly to ground water. The overall watershed hydrologic budget showed a relatively small mass balance error. Reducing irrigation water use produces corresponding increases in simulated baseflow and stream yield. Eliminating irrigation water use reduced the simulated frequency of insufficient streamflows by about 25 percent during the drought period 1988--1991.