Ben Hodges, Ran Li
Tuesday 30 june 2015
14:05 - 14:20h at Europe 1 & 2 (level 0)
Themes: (T) Water engineering, (ST) Computational methods
Parallel session: 6E. Engineering - Computational
Modeling of braided rivers and river deltas is challenging due to their convoluted flow paths, which have required both fine-scale topography and model grid resolution to accurately capture the flow dynamics. Prior to the advent of lidar, topographic scales smaller than practical model grid resolution were poorly known, so modelers relied on calibrated roughness coefficients for the complex flows in braided channels and river deltas. However, present lidar and multibeam sonar provide meter and sub-meter scale topography that are computationally demanding for hydrodynamic models. Each order of magnitude decrease in grid resolution requires three orders of magnitude increase in computational speed to maintain the same level of performance for a 2D hydrodynamic model. Although present high performance computing systems can model 2D hydrodynamics at the 1 x 1 m scale, model time steps are typically less than 1 second, and obtaining annual or multi-annual simulations for large systems is impractical. Herein, we examine different ways of handling subgrid-scale features at coarser resolution that still retains hydrodynamic fidelity to the finer scale topography. Techniques developed for image processing are shown to provide effective tools for quantifying the blockages and connectivity at coarser scales. A particular challenge for the hydrodynamic modeling is implement selective connections between coarse grid cells connected by minor braids or channels. Such channels cannot be directly resolved at a coarse grid scale, so they must be modeled as a flow feature that is driven by the coarse-grid water surface gradient.