Fluvial Bank Erosion in the Meandering River Asker, UK: Insights from Computational Fluid Dynamics (CFD) Modelling

River bank erosion often significantly contributes to the catchment sediment yield. Knowledge of the rates & controls on bank erosion events is therefore important in understanding sediment flux. In recent years progress has been made in understanding processes controlling large-scale mass failure (MF) of stream banks, but less attention has been paid to the role that direct fluvial erosion (FE) plays in bank retreat. This is an important omission, not only because FE is a significant process in its own right, but because FE also often triggers mass failure. FE models are typically of the form: E = k(τ - τ _c)^b where E is the bank erosion rate, τ is the applied fluid shear stress, τ _c is the critical stress for entrainment of the bank material, k is an empirically-derived erodibility parameter, and b is an empirically-derived exponent, often assumed to be close to unity. To apply this model, accurate observations of applied fluid stresses, FE rates & bank erodibility are required. Recent developments in bank erosion monitoring technology [e.g. Lawler, 1993], and in the quantification of the bank erodibility parameters k and τ _c using jet-testing devices [e.g. Hanson and Simon, 2001; Dapporto, 2001], offer the means of determining FE rates and bank erodibility. Nevertheless, the problem of collecting the high-resolution spatially-distributed data needed to characterise near-bank fluid stresses remains. One possible solution is to use Computational Fluid Dynamics (CFD) models as a substitute for empirical data. CFD simulations potentially offer a means of acquiring near-bank, distributed, boundary shear stress data at very high spatial resolution. In contrast, empirical data sets of comparable spatial extent and resolution are very difficult to obtain, particularly during the large (competent) flows of interest here. The critical question is therefore whether CFD-derived data are sufficiently accurate for this purpose. Herein we evaluate a series of 3-dimensional CFD simulations for a (200 m long) meander loop on the River Asker at Bridport in southern England. CFD models under specific steady (peak) flow conditions were developed using FLUENT, with peak flow discharge estimates obtained from an adjacent gauging station. The geometry of each model was specified using DEMs of the channel created from high-resolution tacheometric surveys of the study reach, with water surface elevation defined using a network of crest gauges spaced at 20 m intervals along the reach. Zero slip boundary conditions were defined at all sidewall nodes and initial flow velocity vectors at all nodes at the upstream inlet were estimated with reference to 3D flow velocity data acquired using Acoustic Doppler Velocimetry (ADV) at this location. Simulated flow fields for the extent of the study reach were then evaluated by comparing simulated and observed surface velocity vectors, the latter being derived from Particle Image Velocimetry (PIV), supplemented by ADV data in selected (accessible) locations. Finally, we use the near-bank boundary shear stress data obtained from the CFD models to develop insight into the nature and effectiveness of FE processes within the study reach.