Climate Research Projects and Collaborations
From 1991 - 1997 a collaborative project between Oak
Ridge National Laboratory,
the National Center
for Atmospheric Research
addressed the use of massively parallel computers
for climate modeling. Besides a range of research projects involving numerical
methods and parallel algorithms for climate modeling, the NCAR Community
Climate Model, CCM2, was implemented on the Intel Paragon, the Thinking
Machines CM-5 and the IBM SP-2. This collaboration was supported by the
CHAMMP program of the Department
of Energy, Environmental Sciences Division of the Office of Biological
and Environmental Research.
The development of new climate models that can take advantage of present
supercomputers continues to be pursued under the auspices of the Climate
Change Prediction Program (CCPP) of DOE/OBER. Computational platforms of
interest are the IBM SP and Compaq systems at ORNL, the SGI SMP clusters
at LANL, the Cray T3E and IBM SP at NERSC and other inovative systems.
PROJECTS and SOFTWARE
PROPOSED RESEARCH PROJECTS
Spectral Transform Shallow Water Model, Version 4.0, January 1995.
Authored by P. H. Worley and I. T. Foster.
Community Climate Model, Version 2.1, September 1995.
Parallel Climate Model,
in a Distributed Computing Environment. Adapted PCCM3.2 for use with the
PCM for a 2-D Decompostion in collaboration with researchers at NCAR.
Avante Garde, To design and implement a high performance version
of the Climate System Model. This is in collaboration with ANL, LANL, LBL,
LLNL, NCAR, and NASA collaborators.
COWPOKE - Climate,
Ocean, Weather Parallel Object KErnels, A parallel MPI/OpenMP collection
of F90 modules encapsulating distributed data structures (with transposition
between structures) and mathematical operations defined on fields and grids.
for Coupled Atmosphere-Ocean General Circulation Models.'' Single source
management of climate models will require the use of source translation
tools and parallel programming conventions to achieve good performance
on a variety of computational platforms. This proposal develops the necessary
tools and techniques to realize high performance on massively parallel
and vector supercomputers.
for Global Atmospheric Dynamics on Parallel Computers.'' Three novel
numerical methods are being developed for the dynamical core of a General
Circulation Model (GCM). Two involve the semi-Lagrangian transport method.
The other involves a high order local spectral approximaiton.
``Long Term Climate
Simulations Using an Adaptable Virtual Machine.''
This seeks to support
the development of climate models in general computing environments like
HARNESS. Many of the required tools used by a climate model are viewed
as "plug-ins" to the virtual machine. Checkpointing and restarting, fault-tolerance
as well as coupling mechanisms with other climate components can be treated
this way. The result will be a capability for long running climate simulations
in a dynmacally changing computing environment.
John B. Drake / (email@example.com)
Last Modified February 5, 1998