Physics-based modeling of large braided sand-bed rivers: Bar pattern formation, dynamics, and sensitivity

Braided rivers have complicated and dynamic bar patterns, which are challenging to fully understand and to predict both qualitatively and quantitatively. Linear theory ignores nonlinear processes that dominate fully developed bars, whereas natural river patterns are determined by the combined effects of boundary conditions, initial conditions such as planimetric forcing by fixed banks and the physical processes. Here we determine the capability of a state-of-the-art physics-based morphological model to reproduce morphology and dynamics characteristic of braided rivers and determine the model sensitivity to generally used constitutive relations for flow and sediment transport. We use the 2-D depth-averaged morphodynamic model Delft3D, which includes the necessary spiral flow and bed slope effects on morphology. We present idealized scenarios with the smallest possible number of enforced details in the planform and boundary conditions in order to allow free development of bars driven by the physical processes in the model. We analyze bar and channel shapes and dynamics quantified by a number of complementary metrics and compare these with imagery, field data captured in empirical relations, flume experiments, and predictions by linear analyses. The results show that the chosen set of boundary conditions and physics in the numerical model is sufficient to produce many morphological characteristics and dynamics of a braided river but insufficient for long-term modeling. Initially, braiding intensity with low-amplitude bars is high in agreement with linear analysis. In a second stage when bars merge, split, and increase amplitude up to the water surface, the shape, size, and dynamics of individual bars compare well to those in natural rivers. However, long-term modeling results in a reduction of bar and channel dynamics and formation of exaggerated bar height and length. This suggests that additional processes, such as physics-based bank erosion, or enforced fluctuations in boundary conditions, such as spatial-temporal discharge variation, are necessary for the simulation of a dynamic equilibrium river. The most important outcome is that the modeled pattern of bars and channels is highly sensitive to the constitutive relation for bed slope effects that is used in many morphological models. Regardless of this sensitivity and present model limitations of many models, this study shows that physics-based modeling of sand-bed braided improves our understanding and prediction of morphological patterns and dynamics in sand-bed braided rivers.