The bar instability revisited

TL;DR

This paper revisits river bar instability using a hydrodynamical model, revealing the stabilizing role of sediment transport relaxation and the importance of the saturation number in pattern formation.

Contribution

It introduces a more reliable hydrodynamical model that challenges previous shallow water approximations and clarifies the role of sediment transport saturation in river pattern instabilities.

Findings

The Saint-Venant model predicts artefacts not present in the more accurate model.

The saturation number controls the transition from ripples to bars.

The wavelength of patterns scales with water depth and channel length in the large Lsat/H regime.

Abstract

The river bar instability is revisited, using a hydrodynamical model based on Reynolds averaged Navier-Stokes equations. The results are contrasted with the standard analysis based on shallow water Saint-Venant equations. We first show that the stability of both transverse modes (ripples) and of small wavelength inclined modes (bars) predicted by the Saint-Venant approach are artefacts of this hydrodynamical approximation. When using a more reliable hydrodynamical model, the dispersion relation does not present any maximum of the growth rate when the sediment transport is assumed to be locally saturated. The analysis therefore reveals the fundamental importance of the relaxation of sediment transport towards equilibrium as it it is responsible for the stabilisation of small wavelength modes. This dynamical mechanism is characterised by the saturation number, defined as the ratio of the saturation length to the water depth Lsat/H. This dimensionless number controls the transition from ripples (transverse patterns) at small Lsat/H to bars (inclined patterns) at large Lsat/H. At a given value of the saturation number, the instability presents a threshold and a convective-absolute transition, both controlled by the channel aspect ratio {\beta}. We have investigated the characteristics of the most unstable mode as a function of the main parameters, Lsat/H, {\beta} and of a subdominant parameter controlling the relative influence of drag and gravity on sediment transport. As previously found, the transition from alternate bars to multiple bars is mostly controlled by the river aspect ratio {\beta}. By contrast, in the alternate bar regime (large Lsat/H), the selected wavelength does not depend much on {\beta} and approximately scales as H^(2/3)L^(1/3)/C, where C is the Ch\'ezy number.