Multivariate time-series model for suspended sediment concentration

Abstract

Multivariate time-series models of suspended sediment concentration are presented. The suspended sediment dynamics are considered as an unknown system to be identified. The suspended sediment concentration profile and its relationship with current velocity profile are investigated. The model unknown parameter matrices of the models are identified based on an on-line recursive least-squares identification method. The simulation results based on the real data collected from the Rufiji Delta, Tanzania, show that this kind of model is well suited to predicting suspended sediment concentration and transportation of sediment.