|Zhang, Xuesong - Pacific Northwest National Laboratory|
|Srinivasan, Raghavan - Texas Agrilife|
|Izaurralde, R - Pacific Northwest National Laboratory|
|Bosch, David - Dave|
Submitted to: Hydrological Processes
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/11/2011
Publication Date: 7/1/2011
Citation: Zhang, X., Srinivasan, R., Arnold, J.G., Izaurralde, R.C., Bosch, D.D. 2011. Simultaneous calibration of surface flow and baseflow simulations: A revisit of the SWAT model calibration framework. Hydrological Processes. 25(14):2313-2320.
Interpretive Summary: The Soil and Water Assessment Tool (SWAT) is currently being used to determine the impact of land management and conservation practices on sediment and pollutant loadings into lakes, bays, and estuaries across the U.S. USDA is using SWAT and other computer models for national conservation planning, and USEPA is using SWAT for national environmental planning. It is important that these models accurately simulate water, sediment, and contaminant transport across the landscape and through rivers and reservoirs. In this study, an optimization method was developed to calibrate SWAT and ensure that the model is accurately simulating streamflow and the movement of water through the landscape. The model can then be used by policy makers with confidence that the results are accurate and the impacts of conservation practices are realistically simulated.
Technical Abstract: Accurate analysis of water flow pathways from rainfall to streams is critical for simulating water use, climate change impact, and contaminant transport. In this study, we developed a new scheme to simultaneously calibrate surface flow (SF) and baseflow (BF) simulations of Soil and Water Assessment Tool (SWAT) by combing evolutionary multi-objective optimization (EMO) and BF separation techniques. The application of this scheme demonstrated pronounced trade-off of SWAT’s performance on SF and BF simulations. The simulated major water fluxes and storages variables (e.g. soil moisture, evapotranspiration, and groundwater) using the multiple parameters from EMO span wide ranges. Uncertainty analysis was conducted by Bayesian model averaging of the Pareto optimal solutions. The 90 percent confidence interval (CI) estimated using all streamflows substantially overestimate the uncertainty of low flows on BF days while underestimating the uncertainty of high flows on SF days. Despite using statistical criteria calculated based on streamflow for model selection, it is important to conduct diagnostic analysis of the agreement of SWAT behaviour and actual watershed dynamics. The new calibration technique can serve as a useful tool to explore the tradeoff between SF and BF simulations and provide candidates for further diagnostic assessment and model identification.