Modelling river and floodplain evolution using physics-based and reduced-complexity approaches

Abstract

Models of river-floodplain evolution have previously been classified as adopting either physics-based or reduced-complexity approaches, with the latter presumed to afford a weaker representation of fluvial processes and dynamics. Results are presented herein that enable the first direct comparison of two such approaches within a single fluvial morphodynamic modelling framework. This is achieved using a new morphodynamic model that can be implemented using two alternative hydrodynamic solvers, while all other model components are unchanged. The two solvers are a momentum conserving (MC) Godunov-type finite volume scheme; and an inertial formulation (IF) of the shallow water equations, which neglects momentum transport. Simulations reported herein demonstrate that the two modelling approaches can produce channels characterised by very similar morphology and process rates. Moreover, both solvers exhibit consistent behaviour that illustrates the key role of lateral dynamics (driven by both bank erosion and other mechanisms of floodplain reworking) as a control river channel pattern. Overall, the IF solver is characterised by greater sensitivity to changes in model parameter values and, for some parameterisations, may promote channels with unrealistic planform morphology. Moreover, the neglect of momentum transport by the IF solver restricts its capacity to represent flow acceleration over bar tops, which has implications for its ability to simulate bar-driven braiding and low sinuosity braided anabranches. Despite this, the simpler IF solver is capable of simulating the evolution of meandering channels (and their floodplains) using coarse model grids, due to its representation of streamline curvature effects on sediment transport direction. The resulting reduction in computational cost associated with implementing the IF solver suggests that such reduced-complexity approaches may be particularly suitable for use in simulating the long-term (millennial) evolution of meandering river floodplains, perhaps more so than braided channels, which have often been the focus of reduced-complexity modelling studies in the past.