Evaluating single and dual-station calibration approaches for streamflow prediction in ungauged watersheds using SWAT: A case study of the Upper Blue Nile Basin

Upper Beles watershed, Upper Blue Nile Basin, Ethiopia. Accurate estimation of surface water is crucial for effective resource management in regions like Ethiopia, as the performance of the SWAT model largely depends on the calibration strategies employed. This study examines the effect of single-station and dual-station calibration approaches on streamflow prediction in ungauged watersheds within the Upper Blue Nile Basin, Ethiopia. The study employed SWAT to simulate hydrological processes in the Upper Beles watershed and its sub-watersheds, using historical climate and streamflow data (1997–2012) for model calibration (1999–2006) and validation (2007–2010). Results indicate that single-station calibration at respective watershed outlets provided reliable predictions, with R² > 0.75, NSE > 0.69, and PBIAS < 19 %. The model also performed acceptably during the validation period, except for the Gilgel Beles watershed (NSE=-1.34 and PBIAS=91 %), calibrated at the Main Beles station. In contrast, dual-station calibration enhanced model performance by better capturing spatial variability for the Main and Gilgel Beles watersheds. Furthermore, water balance analysis revealed variations in its components under different calibration scenarios. The hydrograph of the dual-station calibration prediction was also better aligned with the observed data than the single-station predictions. The findings underscore the necessity of multi-station calibration for improving the reliability of hydrological models, enhancing parameter transferability, reducing calibration uncertainty, and improving streamflow simulation accuracy in data-scarce regions. • The study assessed dual-station calibration for streamflow simulation using SWAT. • Dual-station improved spatial variability and prediction in ungauged areas. • Single-station caused errors in Gilgel Beles streamflow prediction. • Calibration choice significantly affected water balance components. • The study offers insights to improve modeling in data-scarce basins.