Modeling the hydrological response of a snow-fed river in the Kashmir Himalayas through SWAT and Artificial Neural Network

Accurate streamflow data and its appropriate treatment are of paramount importance for water resource management. With the growing role of computational models such as soil and water assessment tool (SWAT) and artificial neural network (ANN) in hydrological assessments, we conducted an evaluation of the accuracy of these models for the streamflow simulation of Sindh River. In this study, we utilized monthly time-series data to assess the accuracy of the ANN and SWAT models. A comparative analysis based on prediction accuracy was conducted, and the results indicated that the ANN model demonstrated excellent performance in forecasting peak flow, whereas the SWAT model showed better outcomes when simulating low-flow values. Our findings reveal that the SWAT model will achive the highest Nash-Sutcliffe efficiency (NSE) and coefficient of determination (R2) values during calibration and validation stages. The study showed that ANN-based modeling is time and computationally efficient, does not require extensive studies, is not limited by the nature or quantity of inputs, and produces results equivalent to those generated by process-based models. On the contrary, process-based models necessitate the collection of comprehensive data, as well as regular field inspections and monitoring. Therefore, our study has the potential to be utilized in the management of freshwater resources in the Indian Himalayan Region (IHR).