Application of models with different types of modelling methodologies for river flow forecasting

Abstract

In the present study, a conceptual watershed model, a distributed watershed model, and an artificial neural network (ANN) have been applied to river flow forecasting in the Kalu River upper catchment in Sri Lanka. The Xinanjiang watershed model has been used as a conceptual watershed model and the SWAT model (Neitsch, 2000) has been used with spatial data as a distributed model. Two types of ANN architectures, namely multi-layer perceptron network (MLP) and radial basis function network (RBF) have been implemented as "black box" type modelling methodology. Based on the application results, it seems that the conceptual watershed model could perform slightly better than the distributed model and the ANN for this watershed. It was clearly noted that the performance of distributed models strictly depends on the quality of input data (Arnold et al., 1998) whereas the performance of conceptual models depends on the calibration (Duan et al., 1992, 1993).