Sensitivity of simulated flow fields and bathymetries in meandering channels to the choice of a morphodynamic model

Morphodynamic models are used by river practitioners and scientists to simulate geomorphic change in natural and artificial river channels. It has long been recognized that these models are sensitive to the choice of parameter values, and proper calibration is now common practice. This paper investigates the less recognized impact of the choice of the model itself. All morphodynamic models purport to simulate the same flow and sediment dynamics, often relying on the same governing equations. Yet in solving these equations, the models have different underlying assumptions, for example regarding spatial discretization, turbulence, sediment inflow, lateral friction, and bed load transport. These differences are not always considered by the average model user, who might expect similar predictions from calibrated models. Here, a series of numerical simulations in meandering channels was undertaken to test whether six morphodynamic codes (BASEMENT, CCHE-2D, NAYS, SSIIM-1, TELEMAC-2D and TELEMAC-3D) would yield significantly different equilibrium bathymetries if subjected to identical, initial flow conditions. We found that, despite producing moderately similar velocity patterns on a fixed-flat bed (regression coefficient r of 0.77 ± 0.20), the codes disagree substantially with respect to simulated bathymetries (r = 0.49 ± 0.31). We relate these discrepancies to differences in the codes' assumptions. Results were configuration specific, i.e. codes that perform well for a given channel configuration do not necessarily perform well with higher or lower sinuosity configurations. Finally, limited correlation is found between accuracy and code complexity; the inclusion of algorithms that explicitly account for the effects of local bed slope and channel curvature effects on transport magnitude and direction does not guarantee accuracy. The range of solutions obtained from the evaluated codes emphasises the need for carefully considering the choice of code. We recommend the creation of a central repository providing universal validation cases and documentation of recognized fluvial codes in commonly studied fluvial settings. This article is protected by copyright. All rights reserved.