A comparative study between dynamic and soft computing models for sediment forecasting

Runoff–sediment process modeling is highly variable and nonlinear in nature. For sediment yield prediction, the difficulty of rainfall–runoff–sediment yield hydrological processes remains challenging. The present study uses a simple nonlinear dynamic (NLD) model to predict daily sediment yields, taking into account the degree of daily–sediment yield in catchment areas, and its findings were compared to three widely used models including artificial neural networks (ANN), support vector machine (SVM), and gene expression programming (GEP). The daily measured discharge–sediment data for 25 years were obtained from Shakkar Watershed; Central India as in the current study. The coefficient of correlation (CC), Nash-Sutcliff (NS), and root-mean-square error (RMSE) were employed to assess the performance of the models. The results show that the NLD model was found better than ANN, SVM, and GEP model. These models had correlation coefficient (CC = 0.975, 0.887, 0.843, and 0.901), root-mean-square error (RMSE = 0.748, 1.751, 1.961, and 1.545), and Nash–Sutcliffe efficiency (0.952, 0.784, 0.673, and 0.814) correspondingly. Hence, the NLD model can be used for predicting sediment. In order to implement appropriate measures of soil conservation in the watershed to reduce the sediment load in the river, predicting the sediment yield is very necessary to maximize the life of the structure.