|
Abstract
|
Accurate prediction of suspended sediment concentration (SSC) in vegetated flows remains a great challenge due to the complexity of vegetation-induced turbulence. This study develops a novel turbulence-based analytical model that includes vegetation density ( a ) and stem-scale turbulence (α) in sediment diffusivity, leading to an explicit SSC profile with minimal calibration. The model, validated with flume experiments (dimensionless vegetation density parameter ah = 0.1−0.5, where a is the frontal area per unit volume ( cm−1) and h is vegetation height, flow velocity 1.7–6.0 cm/s), reduces root mean square error (RMSE) by 43% and 35% compared to Nezu–Nakagawa and Stone–Shen models, respectively, for sparse vegetation (ah = 0.1), and 49% and 46% for dense vegetation (ah = 0.5), where ah represents the dimensionless vegetation density parameter. Nash–Sutcliffe efficiency is greater than 0.93 in all scenarios, suggesting improved predictive capability. Sensitivity analysis identifies vegetation density (a) as the overarching control (normalized sensitivity index Si =1.87 near-bed), with turbulence coefficient α linearly related to stem Reynolds number (Red = Ud∕v , R2 = 0.89). Soft computing methods (ANN, SVR) provide slightly better accuracy (RMSE = 0.68–0.71%) but without mechanistic interpretability. The new framework combines theoretical rigor and practical applicability, enabling sediment concentration to be estimated within a 10% error margin for the regions of mid-depth (z > 5 cm) and providing a tool unprecedented for ecohydraulic design and sediment management in vegetated aquatic environments.
|