MODELING DECADAL CHANGES IN SANDBAR VOLUME USING SUB-DAILY RECORDS OF FLOW AND SAND CONCENTRATION DOWNSTREAM FROM A LARGE DAM

We coupled continuous records of flow and suspended-sand concentration with a simple morphodynamic model to predict more than a decade of sandbar volume change in eddies formed downstream from debris flow constrictions in the Grand Canyon. The model builds on the physically-based model of Andrews and Vincent (2007), but is adapted to include erosion and focus on the subaerial portions of bars that are exposed during normal dam operations. These high elevation parts of bars are particularly important for recreation and terrestrial habitat. In the model, sand is advected into the eddy using an eddy exchange coefficient, which is the fraction of the eddy volume exchanged with the main channel per second. Sand concentration and discharge in the main channel are assumed equal to values measured at gaging stations. Sand advected into the eddy is then deposited as a function of sand grain size, sand concentration and settling velocity. The model simulates sandbars as a triangular wedge, in which only part of the bar may be submerged, depending on the bar volume and stage elevation of flow. Erosion is modeled at each time step along the bar-water interface. Measurements of sandbar volume at a suite of dynamic and vegetation-free sites are used to constrain two free parameters in the model – the eddy exchange coefficient and an erosion rate parameter. These parameters were fit using a Nelder-Mead approach that minimized the least-squares error of the modeled versus measured bar volume. Results demonstrate that optimized values of the eddy exchange coefficient and erosion rates are with the range of those measured in independent studies, but there can be considerable variation amongst model runs depending on the selected parameterization. Refinement of erosion modeling, particularly accounting for rapid mass wasting events, may be necessary for better model performance. We show that coupling a simple morphodynamic model with flow and sediment concentration data can reasonably predict (R²~0.85) measured bar volumes. The model results have important implications for the potential to predict the long-term behavior of alluvial sandbars for a variety of flow and sediment management scenarios in canyon rivers.