Suspended Sediment Modeling Using Neuro-Fuzzy Embedded Fuzzy c-Means Clustering Technique

The assessment of the suspended sediment (SS) amount in rivers has an importance because it specifically affects the design and operation of numerous hydraulic structures such as dams, bridges, etc. This paper proposes an adaptive neuro-fuzzy embedded fuzzy c-means clustering (ANFIS-FCM) approach for estimating SS concentration. The accuracy of ANFIS-FCM models was compared with classical ANFIS, artificial neural networks (ANNs) and sediment rating curve (SRC). Daily streamflow and SS data from two stations, Muddy Creek near Vaughn and Muddy Creek at Vaughn, operated by the United States Geological Survey were used in the study. Applied models were compared with each other based on root mean square errors and correlation coefficient. Based on comparison, ANFIS-FCM performed superior to the other two models for modeling complex non-linear behavior of the suspended sediment concentration. The ANFIS-FCM model increased the performance (RMSE) of the optimal MLP model by 10 % and 16 % in estimating SSC for the downstream and upstream stations, separately. ANFIS-FCM model provided improvements in performance and parsimonious and took lesser time in calibration than the classical ANFIS model.