Climate Models on Parallel Computers

As part of the CHAMMP program, researchers in the Mathematics and Computer Science Division at Argonne National Laboratory are developing the parallel algorithms and methods required to execute climate models on massively parallel computers. Results of this work include PCCM2, a parallel implementation of the Community Climate Model; new parallel algorithms for the spectral transform method; and novel load balancing algorithms for climate models. See, for example, our study of load balancing in CCM . In support of this work, we have also developed a number of AVS graphics modules.

The STSWM and PSTSWM testbed codes are sequential and parallel programs, respectively, that solve the shallow water equations using the spectral transform method. They are useful for both pedagogical and benchmarking purposes.

This work is performed in collaboration with researchers at the National Center for Atmospheric Research (NCAR) and Oak Ridge National Laboratory (ORNL). A group at Los Alamos National Laboratory is also involved in the CHAMMP project.

In a related project with the University of Wisconsin, we have used PCCM2 to create a coupled ocean-atmosphere model suitable for very long integrations called FOAM.

We also work on Parallel Weather Models.

A number of technical reports and preprints are available on-line.