The Som Kamla Amba Catchment Sub-watershed’s Artificial Neural Networks (ANN) and Water Assessment Tool (SWAT) Modelling

Abstract

The hydraulic structures, soil conservation structures, water harvesting structures, and flood mitigation studies, and other related projects, rainfall-runoff modelling for unregulated river basins or catchments is useful. The Som Kamla Amba catchment area, which is located in the Rajasthani districts of Dungarpur and Udaipur, is the subject of the current study. Standard statistical indices like r, R2, RMSE, MAE, IA, VE, and NSE were used to evaluate the performance of monthly stream flow and rainfall forecasts created with the Soil and Water Assessment Tool (SWAT) and Artificial Neural Networks (ANN). For the Som Kamla Amba watershed, 27 years (1995–2021) of input data, including daily rainfall, maximum and minimum temperatures, relative humidity, sun radiation, and wind speed, and stream flow, were gathered. The Som Kamla Ambamba catchment's sub watersheds (W1 to W9) had their morphometric parameters examined using ASTER Dem at a spatial resolution of 30 m × 30 m in ArcGIS 10.4.1 software. It was discovered that the ANN model was more accurate and realistic at predicting rainfall. The SWAT model was then discovered to be accurate in forecasting stream flow. Hence, for catchments, basins, or watersheds with comparable hydrological characteristics, ANN and SWAT can be used for rainfall forecasting and stream modelling.