Numerical modelling of braided rivers

Group Leader: Erik Mosselman

Numerical modelling is a powerful tool for analysing and predicting the features and patterns of braided rivers. Challenges ahead regard the development of more detailed process descriptions, the upscaling of these descriptions, the coupling of processes, the potential of alternative non-mechanistic modelling approaches and the validation of the resulting numerical models. Current numerical models for braided rivers are usually based on depth-averaged hydrodynamic flow equations, empirical equilibrium sediment transport formulae, schematic representations of sediment mixtures by means of active layers and empirical corrections for hiding and exposure, and simplifying assumptions about interstitial processes and the packing of sediment grains in the bed. Strongly three-dimensional flow structures may occur in the scour holes at channel confluences. Grains of different sizes may exhibit more complex interactions, for instance when they are transported over sloping beds or when fine sediments penetrate into the pores of a matrix of coarser sediment. Vertical sorting in bedforms is not captured well by formulations based on an active layer. Several applications call for numerical models that can simulate these processes in more detail. One of the possible avenues is direct numerical simulation of the motion of discrete sediment particles. Detailed process descriptions need to be parameterized in computationally efficient applications to larger spatial and temporal scales. An example is that formulations using an active layer will probably remain necessary for this upscaling. The active layer corresponds to the layer of sediment that is reworked by erosion and deposition within a single morphological time step. This reworking results not only from migrating dunes, but also from river bed variations due to discharge variations that are filtered out in the modelling, such as variations in cross-sectional bed tilting in river bends and the generation of erosion and deposition waves at locations where water enters or leaves the channels. Similar issues of upscaling can be identified for other subtime and subgrid processes. The classical view on the modelling of braided rivers is that their morphodynamic evolution can be understood from the interplay of water and sediment. Vegetation dynamics, however, form an equally important factor. Vegetation dynamics may have little influence on the world’s largest braided rivers, but an overriding influence on smaller braided streams. Vegetation provides hydraulic resistance and alters the resistance of river banks to erosion and failure. By trapping sediment, it is often a key factor in the development of pioneer landforms such as accreting banks and new islands. In turn, vegetation dynamics are affected by the processes of water flow and the erosion, transport and deposition of sediment, including substrate development. A more complete modelling approach thus requires process descriptions of vegetation dynamics as well as the coupling of these descriptions to the equations for the motion of water and sediment. Similarly, the system can be coupled to processes of bank failure and interactions with the hyporheic zone. This coupling of processes calls for theoretical analyses to deepen the understanding of the complex phenomena that emerge from interactions. Non-mechanistic approaches provide complementary ways of modelling braided rivers numerically. They include data-driven models, such as artificial-intelligence neural networks, and cellular models that aim at capturing the essence of braid-pattern development from the simplest equations possible. The challenge of model validation lies in the development of a set of elementary test cases, along with corresponding criteria for acceptance, that is agreed by the international fluvial morphodynamics community. These may be based on either analytical solutions for idealized situations or detailed data sets from measurements in the laboratory or in the field.

Leave a Reply