2022 Volume 78 Issue 2 Pages I_61-I_66
FS3M, a numerical model that can compute dynamic interaction between waves and topographic change, was applied to local scouring around piles to investigate the influence of three sediment transport parameters. Machine learning models, XGBoost and LightGBM, were trained using numerical results and the influence of the parameters was also evaluated from the machine learning. Results showed that the final scour depth increased with an increase in a coefficient of the pick-up function Cp, which had the largest contribution to the final scour depth. Furthermore, the height of the tangential flow velocity zvel from the surface of the sediment substrate used to compute the friction velocity and the vertical cell height Δzsurf near the surface of the sediment substrate had the smaller contribution than Cp. Based on the results, a method for determining the parameter values was proposed.
