HR: 16:45h
AN: A52F-03    [PDF]
TI: Increasing Simulation Speed in a Coupled Ocean/Atmosphere Model by Using Mixed-Resolution, Parallel Components
AU: * Jacob, R L
EM: jacob@mcs.anl.gov
AF: Argonne National Laboratory Mathematics and Computer Science Division, 9700 S. Cass Ave., Argonne, IL 60439 United States
AU: Vavrus, S J
EM: sjvavrus@wisc.edu
AF: University of Wisconsin-Madison Center for Climatic Research, 1225 W. Dayton St, Madison, WI 53706 United States
AU: Poulsen, C J
EM: poulsen@umich.edu
AF: University of Michigan Department of Geological Sciences, 425 E. University Ave, Ann Arbor, MI 48109 United States
AU: Liu, Z
EM: zliu3@facstaff.wisc.edu
AF: University of Wisconsin-Madison Center for Climatic Research, 1225 W. Dayton St, Madison, WI 53706 United States
AU: Pierrehumbert, R T
EM: rtp1@geosci.uchicago.edu
AF: University of Chicago Department of the Geological Sciences, 5734 S. Ellis Ave, Chicago, IL 60637 United States
AB: The traditional goal of a coupled ocean/atmosphere model is to accurately simulate the current climate and the primary application of the resulting model is to the prediction of future climate. The constant push for higher resolution and more and better physical parameterizations in these models has kept their simulation speed, simulated years per wallclock day, nearly constant despite a decade of growth in computer power. The Fast Ocean Atmosphere Model is a coupled ocean/atmosphere general circulation model designed to take advantage of Moore's Law and maximize simulation speed. The fundamental design choice in FOAM is the use of mixed-resolution components: the spectral atmosphere model is kept fixed at the relatively low resolution of R15 while the finite-difference ocean model employs a medium resolution grid with an average spacing of less than 2 degrees. The resulting model has an output of nearly 100 years per day on a single IBM Regatta node. FOAM is also well suited for the 16 and 32 processor Linux clusters that are becoming common equipment for small research groups and individual researchers. Despite the low resolution, FOAM is able to simulate much of the observed variability of the climate system, including ENSO and mid-latitude decadal variability, without flux corrections. FOAM can also reproduce the major effects of CO2 increase seen in higher resolution models. FOAM has also found many applications in paleoclimate, performing simulations of the Neo-Proterozoic and early Holocene.
UR: http://www.foamodel.org
DE: 3337 Numerical modeling and data assimilation
DE: 3344 Paleoclimatology
DE: 4255 Numerical modeling
SC: Atmospheric Sciences [A]
MN: 2003 Fall Meeting