PROBLEMS AND POTENTIAL OF AUTOCALIBRATING A HYDROLOGIC MODEL

Abstract

An investigation was conducted to evaluate strengths and limitations of manual calibration and the existingautocalibration tool in the watershed-scale model referred to as the Soil and Water Assessment Tool (SWAT). Performanceof the model was tested on the Little River Experimental Watershed in Georgia and the Little Washita River ExperimentalWatershed in Oklahoma, both USDA-ARS watersheds. A long record of multi-gauge streamflow data on each of thewatersheds was used for model calibration and validation. Model performance of the streamflow response in SWAT wasassessed using a six-parameter manual calibration based on daily mass balance and visual inspection of hydrographs andduration of daily flow curves, a six-parameter autocalibration method based on the daily sum of squares of the residuals afterranking objective function (referred to as SSQRauto6), a six-parameter method based on the daily sum of squares of residuals(SSQauto6), and an eleven-parameter method based on the daily sum of square of residuals (SSQauto11). Results show thatfor both watersheds, manual calibration generally outperformed the autocalibration methods based on percent bias (PBIAS)and simulation of the range in magnitude of daily flows. For the calibration period on Little River subwatershed F, PBIASwas 0.0%, -24.0%, -21.5%, and +29.0% for the manual, SSQRauto6, SSQauto6, and SSQauto11 methods, respectively.Based on the coefficient of efficiency (NSE), the SSQauto6 and SSQauto11 methods gave substantially better results thanmanual calibration on the Little River watershed. On the Little Washita watershed, however, the manual approach generallyoutperformed the automated methods, based on the NSE error statistic. Results of this study suggest that the autocalibrationoption in SWAT provides a powerful, labor-saving tool that can be used to substantially reduce the frustration and uncertaintythat often characterize manual calibrations. If used in combination with a manual approach, the autocalibration tool showspromising results in providing initial estimates for model parameters. To maintain mass balance and adequately representthe range in magnitude of output variables, manual adjustments may be necessary following autocalibration. Caution mustalso be exercised in utilizing the autocalibration tool so that the selection of initial lower and upper ranges in the parametersresults in calibrated values that are representative of watershed conditions.