A performance comparison of CCHE2D model with empirical methods to study sediment and erosion in gravel-bed rivers

Abstract

Depending on environmental conditions and hydrodynamic characteristics, erosion and sediment models always have higher marginal of errors. Thus, accurate and cost-effective models should be used for modeling erosion and sediment. Unlike empirical models, numerical models have more stability and less dependence on the variable parameters due to the uniformity of the conditions studied. Therefore, in this research a numerical CCHE2D (National Center for Computation Hydro-Science and Engineering 2D tool) model was used for erosion and sediment modeling of Beshar gravel-bed river (in Yasuj, Iran). Then, the results of the model were compared with several empirical formulas including Einstein, Vilkanov, Lane–Kalinske, Brooks, Englund and Bagnold based on statistical parameters. The results showed that CCHE2D and Einstein methods had the most efficiency in estimating suspended sediment load than other applied methods. Also, the best R2 (0.97–0.99) was observed for the results of CCHE2D, Einstein and Bagnold methods. Moreover, the number of errors in the range of 0.5 < r < 2 for CCHE2D and Einstein’s methods was more than other methods. The lowest root-mean-square error and mean absolute error (RMSE = 0.4–0.7, MAE = 0.4–0.5) were observed for these two models under different discharges, as well. Using CCHE2D model as an applied method for rivers with non-uniform sediment, non-equilibrium bed, unsteady flow and morphological changes is prioritized over other methods.