Prediction of river suspended sediment load using radial basis function neural network-a case study in Malaysia

Rivers contain a large amount of sediment along with flowing water. It is vital to know the sediment discharge in a river while designing different water resources engineering projects. In this study, suspended sediment discharge has been predicted using a radial basis function (RBF) neural network. Time series data of water discharge and suspended sediment discharge of Pari River, in Perak, Malaysia has been used for modeling the network. The most common radial basis function, called the Gaussian function has been used for modeling the RBF neural network. Three different statistical performance measures namely the root mean square error (RMSE), coefficient of determination (R2) and coefficient of efficiency (CE) were used as performance evaluation criterion for the model. Results obtained from the RBF model are satisfactory and was found that RBF is able to predict the nonlinear behavior of suspended sediment discharge of Pari River.