Sediment load prediction by combined fuzzy logic-wavelet method

Sediment load prediction is very important in planning, operation, and maintenance of water structures located on rivers. The sediment loads exhibit random characteristics due to the uncertain nature of sediment transportation in rivers. To predict suspended sediment load, stochastic processes, regression methods, neural network models, and fuzzy logic have been used in the literature so far. The purpose of this study was to develop a model that can make accurate predictions of suspended sediment loads. Here, a combination of wavelet and fuzzy logic techniques (WFL) is proposed as a new technique to model the behavior of sediment load. While the wavelet method is able to decompose the original series into its sub-bands, fuzzy logic method can be used as a predictive model for each sub-band. It is possible to detect significant power at specific intervals from average wavelet spectra. The WFL was compared with the stand-alone fuzzy logic approach based on the Nash-Sutcliffe Sufficiency Score (NSSS), mean absolute error (MAE), and square of correlation coefficient (R) used as performance indicators. The results of the study show that the WFL provides a considerable improvement over the stand-alone fuzzy logic approach in the prediction of sediment load.