Modeling daily suspended sediments of a hyper-concentrated river in Malaysia

Estimation of suspended sediments in hyper-concentrated rivers is prime important as it is highly desired in design and operation of hydraulic structures. In this study the application of Multiquadric basis function neural network for prediction of suspended sediment of a hyper-concentrated river was investigated. Five years daily time series data of discharge and suspended sediments from 1992 - 1996 at Bidor River in Perak, Malaysia was used to develop the prediction model. Several trials were made to investigate the appropriate number of hidden neurons. Performance of the model was evaluated by comparing the observed and predicted sediments with perfect line of agreement. Furthermore, root mean square error and coefficient of efficiency were also used as performance statistical measures for the model. The results showed that the model successfully predicted the suspended sediments with minimum error of (RMSE = 9.06, MAE = 6.0) and highest efficiency of (CE = 0.94). The performance of the model with previous models was also comparable. The results suggested the suitability of Multiquadric basis function neural network for modelling suspended sediments of hyper-concentrated river. © 2006-2016 Asian Research Publishing Network (ARPN).