An overview of the MPR OCEAN project

Computational challenges in oceanography

Our interest in ocean flows is motivated mainly by their climatological relevance. The global ocean circulation is a major link in the climate system and its variability and (perhaps) loss of stability have a direct and major impact on possible climate changes.

Ocean dynamics are essentially governed by mesoscale eddies (or large-scale turbulence) which have sizes of, say, 10 to 50 km. The trend in computer simulations is to resolve these eddies which requires grid sizes of 10 km or less (comparable to large-eddy simulations in other CFD fields). Together with a vertical resolution of 20 to 30 layers, this leads to millions of grid points for the world ocean. Time steps will be of the order of 1/2 day. Simulations have to be done on time scales of at least several hundred years. Even longer simulations (thousands of years) are of interest for climate studies. These might be performed using lower-resolution models.

Fluid dynamics of ocean flows can be characterized as 3D boundary-layer flows with essential influence of rotation, driven by surface forcing (wind, heat fluxes, evaporation, and precipitation). Transport of heat and salinity constitutes an essential factor in determining the density distribution which is the internal driving mechanism. This requires accurate solutions of advection-diffusion type equations. The flow is governed by the Navier-Stokes equations but it is predominantly 2D in horizontal planes, with properties resembling those of shallow-water flows.

Some widely used numerical codes are available, one of which will be selected to serve as a framework for the present project. Computational kernels will be replaced using modern parallizable techniques.

We have some typical applications in mind which (a) can serve to illustrate the parallel performance, (b) are interesting from an oceanographic point of view, and (c) fit into ongoing research at IMAU:

1. Shedding of warm eddies from the Agulhas Current near the tip of South Africa into the Atlantic Ocean is a poorly understood mechanism, which however may be of major relevance to the global ocean circulation. Numerical simulations with increasingly finer grids may help bringing out the mechanism, complementary to satellite observations and data-assimilation studies.
2. Convection cells in the North-Atlantic region form another major link in the global ocean circulation. Again, the effect of increasing grid resolution on the occurrence and scale of convective regions will give interesting information on the reliability of numerical simulations.

   The real interest is in coupled atmosphere-ocean-ice models. However, for this project we have chosen to treat the ocean-only as an essential factor in the coupled system.

Last modified Friday 23 February 2007 13:45:43 UT by Kees van Reeuwijk.