Journal of Watershed Management Research

Accurate estimation of sediment load in rivers and reservoirs is an important issue in hydraulic engineering as it affects the design, management and operation of water resources projects. Extract of mathematical relationship in sediment transportation has special complexity. Data-driven methods can be used for Modeling of these phenomena. One of these heuristic self organization methods is Group Method of Data Handling (GMDH) that is uses as a method to detect non-linear relationships between input and output variables. In this research, based on GMDH, a model has been developed for the prediction of suspended sediment concentration in river systems. The daily stream flow and suspended sediment concentration data from two stations, Rio Valencia no Station and Quebrada Blanca Station, were used for evaluating the ability of model. The accuracy of model was evaluated using mean square error (MSE), Relative Bias (RB) and determination coefficient (R²) statistics. Comparison between the results of statistical parameters of model with other algorithms like Neural Networks, Neuro-Fuzzy and Genetic Programming indicated that GMDH has high capability for predicting and simulating of suspended sediment concentration than other models.