Next-generation modeling for buoyancy driven coastal ocean flows
Next-generation modeling for buoyancy driven coastal ocean flows
Abstract:
Numerical simulation of buoyancy driven coastal and estuarine flows poses several challenges: Models must be able to capture sharp gradients between fresh and oceanic water masses in the presence of complex topography and (often) strong tidal currents. The flow field is also often affected by either local or large-scale wind patterns. Relevant length scales vary from 10 m in the estuary to 100 km in the coastal sea, and time scales vary from minutes to years. Capturing all the scales related to the river-to-ocean continuum typically require an unstructured mesh and a hybrid time integration scheme. Many existing unstructured-grid circulation models, however, have proven out to be too diffusive in practice, therefore failing to capture fronts and other essential small scale features of coastal flows. Here, we demonstrate the performance of a novel coastal ocean circulation model using a cascade of test cases. The test cases vary in complexity, ranging from simple lock-exchange studies to full river-plume simulations. We experiment with several (commonly-used and state-of-the-art) numerical schemes and demonstrate that the accuracy of the scheme can significantly alter the effective model diffusivity, having a profound impact on the simulated physics. The circulation model is implemented on an automated finite element framework, which allows unprecedented flexibility in terms of numerical methods without sacrificing computational efficiency. With a unified benchmarking platform, we are able to compare the accuracy and cost of different methods in a rigorous manner, potentially paving the way for next-generation coastal ocean modeling.