Hydrological Modelling using Process based and Data Driven Models

Point source pollution is relatively easily controllable and identifiable. Pollutants detected in a concentrated water source such as stream, river or lake are called non-point source pollution. Sedimentation and deteriorating water quality due to non-point source pollution are nowadays a very serious problem in reservoirs, lakes and rivers and is caused by high discharges of nutrients and sediment from upstream river basins. Minimization of the discharges and improvement of the agricultural practices are the obvious solution of the problem. The application of various hydrological models helps the planners and water resource managers to plan appropriately to implement effective measures. In the present study, process-based Soil and Water Assessment Tool (SWAT) and artificial neural network models such as Multi-Layer Perceptron (MLP) and Radial Basis Neural Network (RBNN) have been applied to simulate hydrological processes such as stream flow and sediment yield on monthly time steps in agricultural watershed in Eastern India. The model outputs are compared in terms of well laid out statistical criterion.