Skip to main content
ARS Home » Research » Publications at this Location » Publication #172910
Title: SUITABILITY OF SWAT FOR THE CONSERVATION EFFECTS ASSESSMENT PROJECT: A COMPARISON ON USDA ARS WATERSHEDS

Author
item Van Liew, Michael
item Veith, Tameria - Tamie
item Bosch, David - Dave
item Arnold, Jeffrey

Submitted to: Journal Hydrologic Engineering
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/12/2006
Publication Date: 3/1/2007
Citation: Van Liew, M.W., Veith, T.L., Bosch, D.D., Arnold, J.G. 2007. Suitability of SWAT for the Conservation Effects Assessment Project: A comparison on USDA ARS watersheds. Journal Hydrologic Engineering. 12(2):173-189.

Interpretive Summary: Hydrologic simulation models are valuable tools for studying the beneficial effects of conservation practices on pollution levels in streams and rivers. Models used for these kinds of studies contain parameters that describe watershed properties such as vegetative cover, soil characteristics, and landscape features. These parameters must be given watershed specific values to accurately simulate runoff from that watershed. This process of assigning appropriate values for a particular watershed is referred to as model calibration. In recent years the increasing complexity of simulation models has led to the development of automatic calibration techniques that use high-speed computers and mathematical expressions for matching model simulations to observed data. To better understand the strengths and limitations of an automated approach to model calibration, a study was conducted to assess a newly developed autocalibration tool in the Soil and Water Assessment Tool (SWAT) watershed scale model. Performance of the model was tested on five USDA ARS watersheds that included Mahantango Creek Experimental Watershed in Pennsylvania, Little River Experimental Watershed in Georgia, Little Washita River Experimental Watershed in Oklahoma, Reynolds Creek Experimental Watershed in Idaho, and Walnut Gulch Experimental Watershed in Arizona. Model simulations were made on a total of 30 data sets that were obtained from a long record of multi-gage climatic and streamflow data on each of the watersheds. Eleven parameters that control surface and subsurface response in the model were calibrated on the three southern watersheds, and an additional five parameters that control the accumulation of snow and snowmelt runoff processes were calibrated on the two northern watersheds. Simulation results suggest that SWAT generally performed better on watersheds in more humid climates than in desert or semi-desert climates. Findings from this study show the importance of selecting appropriate starting values for the parameters that are used in the autocalibration so that hydrologic simulations are representative of field conditions. This investigation demonstrates that the newly developed autocalibration tool in SWAT is a labor saving tool that can be used for a wide range of watershed conditions.

Technical Abstract: Practitioners of sophisticated hydrologic models often face a daunting task in manually calibrating a multitude of parameters suitable for simulating streamflow response for a specific watershed. The various shortcomings associated with manual calibration methods have led to the development of automatic calibration techniques that utilize high-speed computers and efficient algorithms for matching model response to observed data. In this investigation, the strengths and weaknesses of the newly developed autocalibration tool in the Soil and Water Assessment Tool (SWAT 2003) were evaluated. Performance of the model was tested on five USDA ARS watersheds that included Mahantango Creek Experimental Watershed in Pennsylvania, Little River Experimental Watershed in Georgia, Little Washita River Experimental Watershed in Oklahoma, Reynolds Creek Experimental Watershed in Idaho, and Walnut Gulch Experimental Watershed in Arizona. Model simulations were performed on a total of 30 calibration and validation data sets that were obtained from a long record of multi-gage climatic and streamflow data on each of the watersheds. Eleven parameters that govern surface and subsurface response in the model were calibrated on the three southern watersheds, and an additional five parameters that govern the accumulation of snow and snowmelt runoff processes were calibrated on the two northern watersheds. Simulation results suggest that SWAT generally performed better on watersheds in more humid climates than in desert or semi-desert climates. Based on monthly Nash Sutcliffe coefficient of efficiency (NSE) values, model simulations were considered 'good' (NSE>0.75) for 86%, 25%, 11%, and 17% of the data sets that correspond to average annual precipitation amounts of >1000 mm, 701 to 1000 mm, 401 to 700 mm, and <400 mm, respectively. Based on daily NSE values, model simulations were considered 'adequate' (0.361000 mm, 701 to 1000 mm, 401 to 700 mm, and <400 mm, respectively. Findings from this study accentuate the importance of selecting appropriate initial lower and upper bounds for certain model parameters in SWAT such as the curve number, surface runoff lag time, and ground water delay factor so that hydrologic simulations are representative of field observations. This investigation demonstrates that the newly developed autocalibration tool in SWAT is a labor saving tool that can be used for a diverse range of watershed conditions.