Skip to main content
ARS Home » Midwest Area » West Lafayette, Indiana » National Soil Erosion Research Laboratory » Research » Publications at this Location » Publication #331962
Title: Evaluating the effects of watershed size on SWAT calibration

Author
item WALLACE, CARLINGTON - Purdue University
item Flanagan, Dennis
item ENGEL, BERNARD - Purdue University

Submitted to: Water
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/2/2018
Publication Date: 7/6/2018
Citation: Wallace, C.W., Flanagan, D.C., Engel, B.A. 2018. Evaluating the effects of watershed size on SWAT calibration. Water. 10(7):898. https://doi.org/10.3390/w10070898.
DOI: https://doi.org/10.3390/w10070898

Interpretive Summary: Losses of water and pollutants from agricultural production areas are of increasing concern, particularly in the Lake Erie Basin of the United States, where soluble and total Phosphorus (P) transported into Lake Erie are responsible for large harmful algal blooms and eutrophication. Computer simulation models are often used to determine the effects of climate, soils, topography, and land management practices on runoff, stream flow, and pollutant losses, due to their relative ease of application and low resource requirements. In this study we applied the SWAT (Soil and Water Assessment Tool) model to four agricultural watersheds of different sizes located in northeastern Indiana. In addition to the climate, soils, and land management information for these watersheds, we also had measured values for stream flow and pollutant losses, which allowed us to improve the model input parameters (calibrate) in order to get the best comparisons between the modeled and observed values. We found that in terms of predicting the amount of water flow coming out of the watersheds, it was possible to use the observed data at any of them to arrive at calibrated parameters, and then apply these at the other sized watersheds and obtain satisfactory model prediction results. For the pollutants of most concern in Lake Erie (total P and soluble P), we found that using calibrated parameters based on data measured at the smaller sized watersheds improved the comparisons between predicted and observed soluble and total P losses at all watershed sizes. These results impact scientists, university faculty, students, conservation agency personnel and others involved in natural resource assessment and/or management and use of the SWAT model. Limitations in observed data availability are very common, so these results that indicate that it is possible to successfully apply calibrated parameters from one watershed size (which may be more heavily instrumented and monitored) to another (which may not be monitored) could be extremely important.

Technical Abstract: The Soil and Water Assessment Tool (SWAT) has been calibrated in many watersheds of various sizes and with varying physiographic features. However, the relationship between model calibration parameters and the watershed size at which SWAT was calibrated is not clearly understood. Understanding the influence of watershed size on SWAT model calibration parameters will allow users to determine the validity of using such an assessment tool for various sized watersheds. Additionally, this will allow users with limited data to determine whether it is appropriate to calibrate SWAT in one watershed and implement the optimized parameters in a different watershed with similar physiographic features. In this study, we investigated the influence of watershed size on SWAT model calibration on four watersheds (CCW = 680 km2, F34 = 183 km2, AXL = 42 km2 and ALG = 20 km2) located in northeastern Indiana. The results show that calibrating SWAT at one size and applying the optimized parameters at different watershed sizes of similar physiographic features provided satisfactory simulation results. The size at which the model was calibrated had little effect on streamflow predictions. Soluble nitrogen loss estimates were improved when calibration was performed at the larger CCW watershed, while calibrating SWAT at the smaller AXL and ALG watersheds produced improved statistical indicator (NSE, R2 and PBIAS) values for soluble P and total P when applied to the larger CCW and F34 watersheds. Due to the physical characteristics of the F34 watershed, SWAT parameters optimized there did not always provide satisfactory results when applied to the CCW, AXL and ALG watershed configurations.