loading page

Assessment of single- and multi-site calibration approaches on SWAT model performance in watersheds with missing flow data

      Abstract

      Hydrological models are used to represent the hydrological cycle occurring in a basin. How accurately the hydrological model reflects reality is related to good calibration of the model. In this study, the SWAT hydrological model was used to simulate the streamflow of the Mudurnu River basin in the Black Sea Region of Turkiye for the period 1978-1995. The first 3 years between 1978-1980 were introduced to the model as a warm-up period. 10 years between 1981-1990 were chosen as the calibration period and 5 years between 1991-1995 as the validation period.SWAT-CUP’s SUFI-2 algorithm was used in the study for multi (Gebes and Dokurcun stations) and single-site (Gebes station) model calibration and validation for monthly time steps. The single and multi-site calibration results for Gebes station for the monthly simulation for the period 1981-1990 showed excellent model performance for flow rates with R2 of 0.91 and 0.90, respectively, with NSE values of 0.91, 0.90; PBIAS values of -4.0, 1.6 and RSR values of 0.30 and 0.32. The single and multi-site validation results for Gebes station for the monthly simulation for the 1991-1995 period are 0.87 and 0.87 R2, NSE values are 0.87, 0.87; PBIAS values are -3.6, 3.7 and RSR values are 0.36 and 0.36, respectively. When the calibration and validation results were analyzed, it was observed that there was no significant difference between single-site and multi-site calibration approaches. Based on this situation, it can be said that there are no significant differences between single- and multi-site calibration approaches in small basins, and calibration with the flow gauging station at the basin outlet will not pose a problem in terms of model reliability.