Towards Navier-Stokes simulations of the ocean

Sean Vitousek, Stanford University

We review current practices in ocean modeling and prospects for increasing simulation capabilities through the application of nonhydrostatic ocean models. Most current generation simulations of ocean dynamics are computed with the primitive (hydrostatic) equations. Many physical processes, however, contain appreciable nonhydrostatic characteristics and thus may require nonhydrostatic simulations to be well-resolved. The hydrostatic approximation has been a cornerstone of ocean modeling since its practice began in the 1960s. Yet, this approximation breaks down for small-scale processes and high-resolution simulations. The main limitation to further utilization of nonhydrostatic ocean models is their computation cost, which can be an order of magnitude larger than hydrostatic models of equivalent resolution. As computer power and thus model resolution capabilities continue to grow, so will the utilization of nonhydrostatic models to maintain accuracy and physical consistency. We consider techniques to improve computational efficiency and accuracy such as Adaptive Mesh Refinement and Large Eddy Simulation, respectively, among others. We conclude that a truly resolved Navier-Stokes simulation of regional ocean dynamics may be just within reach.

Abstract Author(s): Sean Vitousek, Oliver B. Fringer