Suspended Load Modeling of River Using Soft Computing Techniques

The phenomenon of sediment transport has always affected many river and civil structures. Not knowing the exact amount of sediment, causes much damage. Correct estimation of river sediment concentration is essential for planning and managing water resources projects and environmental issues. For this, you can use the artificial intelligence method, which has high flexibility. In this research, adaptive neuro-fuzzy models (ANFIS), gene expression programming (GEP), support vector regression (SVR), Group Method of Data Handling (GMDH), and the classical method of sediment rating curve (SRC) were used to model and prediction. For this purpose, the daily data of temperature, rainfall, sediment, and discharge of the Jalair station located in the Markazi province of Iran were used. The results obtained from these five methods were compared with each other and with the measured data. To evaluate the methods used, correlation coefficient, root mean square error, mean absolute error, and Taylor diagram were used. The results show the acceptable performance of data mining methods compared to the Sediment rating curve. Also, the model's superiority (GEP) was shown with the highest coefficient of determination R2 with a value of 0.98 and the lowest root mean square error RMSE in terms of tons per day with a value of 3721. The efficiency of the ANFIS and GMDH model with R2 values of 0.93, 0.98, and RMSE values of 16556, and 18638 was somewhat better than the SVR model with an R2 value of 0.90 and RMSE value of 35158.