SWAT and HBV Models' response to streamflow estimation in the Upper Blue Nile Basin, Ethiopia

Abstract The study aimed to evaluate the SWAT and HBV model’s responses to estimate streamflow in three typical selected watersheds in the upper Blue Nile Basin, Ethiopia. The evaluation was carried out through simulation of runoff in three watersheds, namely: Gilgel Abay, Anger, and Maybar located in the upper Blue Nile Basin, Ethiopia. The same calibration and validation period were applied for each watershed model with their respective historical records. The objective function indicates the optimized range of model parameters during the calibration and validation procedure, and it gives a good model response result, which is determined by the recommended performance indicators and hydrograph plots. The HBV and SWAT model’s calibration responds with (NSE = 0.81, NSE = 0.48 and NSE = 0.46) and (NSE = 0.81, NSE = 0.63, NSE = 0.61), respectively for Gilgel Abay, Anger, and Maybar watersheds. The optimized parameters were applied to validate the model using independent historical data. The SWAT and HBV model responses for NSE during the validation period were 0.8, 0.55, and 0.54 and 0.63, 0.45, and 0.34, respectively, for Gilgel Abay, Anger, and Maybe watersheds. An alternative was observed over and under estimation runoff during stimulation periods in extreme wet and dry periods. Generally, the calibrated and validated hydrograph output result indicates that the model’s responses to too wet and dry conditions were not satisfactory. Therefore, further model refinements for extreme events, including season-based model evaluations that could improve the process representation of rainfall-runoff events, are recommended.