Calibration and Validation of Semi-distributed Model of Monthly Stream Discharge Using SUFI-2 Algorithm for Shimsha Catchment, Karnataka, India

The Shimsha Catchment employed the Semi-distributed SWAT model for runoff prediction, which considered geographical features, surface vegetation, and soil characteristics. The catchment was subdivided into six sub-watersheds based on geography, natural drainage patterns, and designated discharge points. In the Hydrological Response Unit (HRU) analysis, 136 HRUs were created in SWAT model by incorporating land use and soil maps and defining HRUs with specific threshold percentages. To calibrate and validate the model, simulated values were compared with observed data from stream gauge discharge records. The calibration process utilized the SUFI-2 algorithm integrated into the SWAT-CUP model. The results demonstrated the model's strong predictive capabilities across the entire catchment, achieving calibration values of 0.87, 0.92 and 0.78 for the Nash-Sutcliffe Efficiency (NSE), coefficient of determination (R2) and index of agreement(d) respectively. Parameter selection and ranges were determined by considering the unique characteristics of the study area, recommendations from the model for new parameter ranges, and examination of a 95% probability plot. The analysis of uncertainty highlighted 14 sensitive parameters, with the curve number emerging as the most influential factor, followed by groundwater parameters. Capturing the dynamics of water flow, sediment transport and nutrient cycles to ensure reliable predictions to show the model reliability need to be assessed.