HR: 1330h
AN: A42B-0750    [PDF]
TI: Development of Adaptive Mesh Refinement Techniques for an Ocean General Circulation Model
AU: * Herrnstein, A
EM: aherrnst@llnl.gov
AF: Institute For Scientific Computing Research, Lawrence Livermore National Laboratory, P.O. Box 808,, Livermore, CA 94551 United States
AU: * Herrnstein, A
EM: aherrnst@llnl.gov
AF: Department of Applied Science, University of California at Davis, 3001 Engineering III 1 Shields Avenue, Davis, CA 95616-8254 United States
AB: A problem faced by all computational physicists is the trade off between run time and grid resolution. The number of time steps and cost per time step drastically increases as finer and finer grids are used. In effort to combat this dilemma, research areas such as fluid dynamics use a modeling technique known as Adaptive Mesh Refinement (AMR). This involves increasing grid resolution in localized areas of the domain while preserving the structured nature of the original grid. The concept of AMR has been around for nearly two decades, but only in recent years have software packages become available to aid in many of the details. Atmospheric and ocean models simulate phenomena with motion derived from fluid dynamics. This suggests AMR is feasible for global climate modeling. The success of nested regional models also supports this hypothesis. In this research, techniques are developed which will allow AMR to perform on an existing ocean model known as the Modular Ocean Model (MOM). Of particular interest is the ability to use AMR on a staggered grid with centered differencing numerics. These are two characteristics of MOM which reduce computational noise and run time respectively. To aid AMR implementation, the software package SAMRAI (Structured Adaptive Mesh Refinement Application Infrastructure) is used. Refinement is based on solution gradients and possibly other criteria as needed. In addition, regions of critical topography might be refined if necessary. Current tools include time integrators which implement Leap Frog, Predictor Corrector, and Runge Kutta numerics. Velocities are defined at nodes and all other quantities at cell centers. A bench mark simulation of a barotropic modon will be presented.
DE: 3200 MATHEMATICAL GEOPHYSICS (New field)
DE: 3210 Modeling
DE: 3230 Numerical solutions
DE: 4200 OCEANOGRAPHY: GENERAL
DE: 4255 Numerical modeling
SC: Atmospheric Sciences [A]
MN: 2003 Fall Meeting