RSS Feed GEOSTROPHIC TURBULENCE
Jim McWilliams
Scientific Significance:
As part of an HPCC Grand Challenge project in Geophysical and Astrophysical Turbulence, McWilliams and his colleagues are investigating large-scale turbulent flow patterns of the Earth and other planets. Their investigations pertain to turbulent flows such as the jet stream and ocean currents, and their research aims to accurately characterize the roles of this large-scale turbulence in general circulation models of global climate. One series of calculations has focused on the regime of rotating, stably-stratified flows, which is an idealization for large regions of either the atmosphere or ocean. Their high-resolution numerical simulations on the C90 show significant discrepancies from the long-standing theoretical prediction of isotropy. These discrepancies are associated with self-organization of the flow into a large population of coherent vortices. The chaotic interactions among these vortices govern the subsequent evolution of the flow toward a final configuration that is nonturbulent. These findings give quantitative confirmation to ideas that have emerged during the last 10 years: that turbulence is far more ordered than previously believed. In the future, they are planning to focus on equilibrium, turbulent, jet flows (an artic circumpolar current, jet-stream, Jupiter's bands) for which lengthy intervals of computation will be needed.
Numerical Approach and Performance:
Our immediate purpose is to adapt one of our primary codes, QGMG (quasi-geostrophic multigrid method) to the Cray T3D. This method is a fully implicit multigrid solver for the quasi-geostrophic equations in a three-dimensional rectilinear geometry. It is used to compute turbulent flows in various parameter regimes, including f-plane fully isotropic and beta-plane Eckman drag with both periodic and solid boundary conditions. We have recently parallelized this code for the Cray T3D using PVM and are currently benchmarking and tuning it. We are planning to first employ this parallel multigrid code for equilibration problems with climatic forcing. These require long-time integration with relatively little I/O, and are therefore appropriate for the T3D.
Back to Contents Page