1:30 PM-2:30 PM
Room: Texas Ballroom B
Chair: Robert L. Higdon, Oregon State University
This plenary talk will be presented in a joint session of the Geosciences and the Parallel Processing conferences.
In order for climate simulations to be both realistic and useful, they must have enough spatial resolution to correctly represent transports of heat and moisture by the atmosphere and oceans as well as the natural variability of the climate system. Since long integrations and ensemble studies are also involved, access to highest performance supercomputers is required. With US computing platforms moving toward clusters of multi-cpu microprocessors, mathematical formulations and computational designs of climate models must adapt to the new technology. This talk describes the above issues in overall terms, and it then illustrates a successful adaptation to microprocessor-based computers that has been carried out at the National Center for Atmospheric Research. Results from this Parallel Climate Model are shown for equilibrium climate, El Niño, and C02-induced change.
Albert J. Semtner
Naval Postgraduate School, Monterey