Anastomosing Rivers are Disequilibrium Patterns

Anastomosing rivers have multiple interconnected channels that enclose floodbasins. Various theories have been proposed to explain this pattern, including an increased discharge conveyance and sediment transport capacity of multiple channels, or, alternatively, a tendency to avulse due to upstream sediment overloading. The former implies an equilibrium pattern whereas the latter implies a temporary increase of the number of channels. Our objective was to test these hypotheses on a well-documented case: the upper Columbia River. We combined a geological approach with physics-based morphological modelling. Three geological sections across the entire valley were constructed based on hand auger data, covering a depth up to 8.6 m corresponding to the last 4000 years, corroborated by 14C dating. The sections were integrated with surface mapping to derive proportions of channel and floodplain sediment volumes and sedimentation rates along the river. Existing sediment transport measurements at a location upstream and downstream of the study area were re-analysed and time-integrated for comparison. A network model was built based on gradually varied flow equations, sediment transport prediction, mass conservation and detailed transverse slope and spiral meander flow effects at the bifurcations. The geological sections show a clear downstream trend of decreasing number of channels and decreasing bed sediment deposition in channels and crevasses, indicating bed sediment overloading from upstream and subsequent avulsions. The measured suspended sediment is much larger than the amount captured in the floodplains, indicating that this is not limiting the aggradation in the valley. Extensive crevasse splays in the upstream section and an increased bed elevation and gradient demonstrate a tendency to avulse due to overloading of bed sediment. The measured bedload transport indeed indicates bed material overloading. The 14C dating confirms that long-term average floodplain sedimentation rates decrease significantly in downstream direction. The overloading led to an anastomosing river pattern by in-channel aggradation and avulsions, followed by slow channel fill so that multiple channels remain open for a long time. Although multiple channels convey flood discharge, only one channel transports the majority of the sediment so that morphodynamically this is a single-channel system. The 1D network model indicates that the multi-channel system evolves towards a single-channel system within centuries because symmetric channel bifurcations are inherently unstable whilst confluenced channels attract the flow due to relative decreased friction. Moreover, the long profile of the valley could only be reproduced by overfeeding the model with a sediment pulse four times larger than the transport capacity for a century and then reduce the feed to capacity in agreement with the measurements. We conclude that this anastomosing river pattern is the result of a temporary pulse of sediment and is not an equilibrium feature