Streamflow prediction uncertainty analysis and verification of SWAT model in a tropical watershed

Appropriate implementation of Soil and Water Assessment Tool (SWAT) hydrologic model requires prediction uncertainty analysis, calibration and validation of the model against historical output records. Sequential Uncertainty Fitting-2 (SUFI-2) and Generalized Likelihood Uncertainty Estimation (GLUE) algorithms with ArcSWAT2009 and ArcGIS10.0 were used in this research to conduct uncertainty analysis, calibration and validation of the SWAT model using monthly observed streamflow data in Maybar experimental watershed, Ethiopia. The results revealed that the model was generally satisfactory as proved by the uncertainty, calibration and validation goodness of fit indicators: The goodness of fit and the degree to which the calibrated SWAT model accounted for the uncertainties assessed by: P-factor (72, 70 %) and (65, 69 %) for (calibration, validation) stages of SUFI-2 and GLUE algorithms, respectively, and R-factor (0.97, 0.90) and (0.89, 0.95) for (calibration, validation) stages of SUFI-2 and GLUE algorithms, respectively, are reached acceptable values, then the parameter uncertainties used were the desired parameter ranges. Further model evaluation statistics: Coefficient of Determination (R ² ≥ 0.76), Nash–Sutcliffe efficiency (NSE ≥ 0.63), Percent Bias (PBIAS ≤ ± 7.10 %) and Root Mean Square Error-observations Standard deviation Ratio (RSR ≤ 0.46) for both calibration and validation periods were quantified and the extent of similarity between predicted and recorded streamflow data suggests that SWAT model can adequately simulate monthly streamflow at Maybar gauged watershed.