Group Leader: Alain Recking
Modelling hydraulics and sediment transport in natural rivers is a difficult and challenging task. This is particularly true in braided rivers where, in addition to standard questions still not resolved for single tread channels (such as effects of bed form propagation, of grain sorting and partial transport in gravel bed materials, role of vegetation…), additional questions concern the definition of the channel itself, not only for the hydraulics, but also for sediment transport (active channel width). Developing prediction tools for such rivers also raises the question of validation, and consequently of our capacity to measure accurate data in such highly fluctuating environment.
Braiding rivers can extend their flow laterally, in one or several channels, which has two major consequences making the hydraulics complicated. The first direct consequence is that uncertainties on the channel width make it difficult to compute the hydraulics with the discharge. The second consequence is that these channels produce shallow flows, with low relative depths (ratio of the flow depth and grain diameter), and the law of the wall (from which were derived most standard flow resistance equations) can not be considered valid in most part of the section. Shallow flows also implies a large variance in shear stress, which is also additional difficulty to be taken into account when section averaged data are used. But overall, even in very active systems, experiments and observations suggest that the part of the wetted width really active in transporting sediments is usually small, and reduced to no more than one or two channels of the braiding network. Thus it is primordial to better understand the processes of transport in such rivers, including the role of confluences and grain sorting, and to define how spatial and time fluctuations should be considered in the modelling process.
Considering the complexity of these in-channel processes, it is questionable how the approaches used for single tread channels can be extended to braided rivers: is it possible to consider averaged hydraulic geometry equations? Are equations developed at low flows still valid during flooding (considering evidences that the braiding pattern persist for such flows)? Can we define equilibrium (for given discharge and sediment input)? Are equations developed for equilibrium still valid for aggrading/eroding conditions? This complexity also has, in some way, imposed the total flow power (product of discharge by the valley slope) as a natural candidate parameter for bedload modelling. But other solutions were proposed which consist for instance in computing the bed shear stress numerically for each part of the bed section, or in defining an appropriate pdf function describing the variance around the mean values.
A lot is still to be done for constructing equations valid in such environment. But it will necessitate first to produce accurate data: measuring the hydraulics is challenging in these highly moveable beds (where is the bottom elevation during a flood?) and direct measurements of bedload transport is a problem when only a small part of the section contribute to transport at a given time. Alternative methods such as sediment budget based on morphological changes can be used but methodological aspects are still to be solved, including a definition of appropriate time and space scales for fluctuations.