SWAT Literature Database for Peer-Reviewed Journal Articles

Title:Examining model performances and parameter uncertainty for streamflow and suspended sediment regime simulation: Comparison of three calibration methods 
Authors:Ranjan, R. and A. Mishra 
Journal:Journal of Hydrology 
Volume (Issue):612(Part C) 
Article ID:128304 
URL (non-DOI journals): 
Broad Application Category:hydrologic and pollutant 
Primary Application Category:calibration, sensitivity, and/or uncertainty analysis 
Secondary Application Category:sediment loss and transport 
Watershed Description:20,235 km^2 Kantamal River, a tributary of the Mahanadi River located in the State of Odisha in east central India. 
Calibration Summary: 
Validation Summary: 
General Comments:This study investigated three operational calibration procedures, namely SQN, SML, and SQN_SML, to model streamflow and suspended sediment using the SWAT model. The performance of calibration approaches is examined in terms of simulation accuracy of streamflow and SSL at the basin outlet during the calibration and validation period. Flow and sediment duration curves were segmented into high, mid and low regions as per probability of exceedance to assess the calibration performance under three regimes. Finally, parameter uncertainty and model equifinality were assessed using the analysis of behavioural solutions. When comparing the goodness of fit scores for streamflow across calibration schemes for the calibration and validation period, SQN_SML calibration had the least level of bias (PBAIS = 1.7, −4.2), the highest NSE (0.91,0.92), KGE (0.95, 0.94), and R2 (0.95, 0.94). According to segmented flow duration curve analysis, SQN _SML performed the best among the three techniques under all three flow regimes but as per segmented sediment duration curve analysis, SQN_SML outperformed SQN only in the mid and low SSL regimes, whereas all three calibration approaches performed comparably in the high SSL regime. The SML technique produced minimal parameter uncertainty, followed by SQN_SML and SQN. Also, the SQN_SML method yielded the highest P-factor for sediment simulation, suggesting the least model error for sediment. SQN_SML generated the least equifinal solution, whereas SQN produced the most equifinal solution. While the findings of this research are encouraging, the proposed solution must be supported by further studies using different hydrological models and hydro-climatic environments. 
Keywords:Multi-variable calibrationUncertaintySimultaneous calibrationExtreme events simulationSuspended sediment