Suspended sediment load prediction using artificial intelligence techniques: comparison between four state-of-the-art artificial neural network techniques

Accurate prediction of suspended sediment (SS) concentration is a difficult task for water resource projects. In recent years, methodologies such as artificial intelligence (AI) algorithms have been applied load estimation and these models provided efficient results. The present study investigates the abilities four distinct AI approaches estimating monthly SS in Roodak station on Jajrood River, one longest waterways north Iran, using combinations antecedent river flow data. This aims to compare predictive ability neural network (ANN), adaptive neuro-fuzzy inference systems (ANFIS), group method data handling (GMDH), least square support vector machines (LS-SVM) predict load. To develop models, average 50 years were obtained from Tehran regional authority. Data separated into three subsets (training, validation, testing) was predicted where reliability utilized assessed by statistical criterion including correlation coefficient (R), mean absolute error (MAE), root (RMSE). A comparison developed revealed that use able enhance precision concentration. results indicate LS-SVM model generated superior than other terms criteria, showing reasonably observed values.