Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Assessment of Model Configuration Effect by Alternative Evapotranspiration, Runoff, and Water Routing Functions on Watershed Modeling Using SWAT

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan

Citation:  Transactions of the ASABE. 58(2): 393-404. (doi: 10.13031/trans.58.10901) @2015
Authors:   Haw Yen, Jaehak Jeong, Xiuying Wang, Shenglan Lu, Min-Kyeong Kim, Yu-Wen Su
Keywords:   IPEAT, Model calibration, Optimization, SWAT, Uncertainty analysis.

Abstract. The choice of different model structures and the subsequent parameter identification are relevant and usually have a strong subjective component. The impacts of various decisions associated with the internal configuration of a given watershed model need greater exploration to ensure confidence in hydrologic and water quality modeling results, yet they are often overlooked. In this study, a non-subjective approach to selecting alternative methods is developed. The goal is to assess the impact of different configurations of the Soil and Water Assessment Tool (SWAT) model on consequential hydrologic and water quality behavior. A total of 12 SWAT configurations, each containing a unique combination of alternative algorithms in estimating surface runoff, potential evapotranspiration, and water routing, were calibrated against flow and nutrient data using the Dynamically Dimensioned Search (DDS) optimization algorithm within the Integrated Parameter Estimation and Uncertainty Analysis Tool (IPEAT). The assessment ensures that non-subjective decisions are formed by fully exploring different model configurations and consequences in the optimization process. The accuracy of calibrated outputs was sensitive to the choice of alternative methods, especially for ammonia-N predictions: NSE values varied from 0.23 to 0.67 for streamflow, from 0.59 to 0.82 for sediment, and from -0.15 to 0.66 for ammonia-N. Similarly, uncertainty in the calibrated outputs varied among alternative methods: inclusion rate varied from 31% to 58% for streamflow, from 46% to 71% sediment, and from 25% to 87% for ammonia-N. The results highlight the significance of non-subjectivity in selecting alternative methods in the calibration of complex watershed models. The potential impact of selecting alternative methods should be fully explored in advance before further applications of complex watershed simulation models.

(Download PDF)    (Export to EndNotes)