An improved subgrid channel model with upwind-form artificial diffusion for river hydrodynamics and floodplain inundation simulation

Abstract

An accurate estimation of river channel conveyance capacity and the water exchange at the river–floodplain interfaces is pivotal for flood modelling. However, in large-scale models limited grid resolution often means that small-scale river channel features cannot be well-represented in traditional 1D and 2D schemes. As a result instability over river and floodplain boundaries can occur, and flow connectivity, which has a strong control on the floodplain hydraulics, is not well-approximated. A subgrid channel (SGC) model based on the local inertial form of the shallow water equations, which allows utilization of approximated subgrid-scale bathymetric information while performing very efficient computations, has been proposed as a solution, and it has been widely applied to calculate the wetting and drying dynamics in river–floodplain systems at regional scales. Unfortunately, SGC approaches to date have not included the latest developments in numerical solutions of the local inertial equations, and the original solution scheme was reported to suffer from numerical instability in low-friction regions such as urban areas. In this paper, for the first time, we implement a newly developed diffusion and explicit adaptive weighting factor in the SGC model. Adaptive artificial diffusion is explicitly included in the form of an upwind solution scheme based on the local flow status to improve the numerical flux estimation. A structured sequence of numerical experiments is performed, and the results confirm that the new SGC model improved the model performance in terms of water level and inundation extent, especially in urban areas where the Manning parameter is less than 0.03 m-1/3 s. By not compromising computational efficiency, this improved SGC model is a compelling alternative for river–floodplain modelling, particularly in large-scale applications.