Research Objectives
The coupled atmosphere, ocean, and sea ice model being developed (now designated the Parallel Climate Model, or PCM) is a multi-institutional effort involving NCAR, the Naval Postgraduate School (NPS), and the Los Alamos National Laboratory (LANL), under the sponsorship of the Department of Energy's Computer Hardware, Advanced Mathematics, and Model Physics (CHAMMP) Program on Climate Research. This coupled model will be used for sensitivity studies on century-long time scales to address questions related to the climatic impacts of greenhouse gas and other radiative forcings.
Computational Approach
The ocean component is the Parallel Ocean Program (POP) developed by Smith, Malone and Dukowicz (LANL). The model uses second-order finite differencing in space and leapfrog finite differencing in time. An implicit technique is used for the free-surface calculation. The model uses a generalized curvilinear coordinate system which permits displacing the North Pole to a location such that the convergence of meridians does not place an excessive restriction on the allowable time step. POP has recently been reformulated in terms of data structures and the use of memory and cache and completely rewritten for enhanced performance in a message-passing environment.
The sea ice component was derived from an eddy-resolving Arctic Ocean model by Zhang (NPS) and optimized for MPP architecture by Craig (NCAR). It uses the Zhang and Hibler ice dynamics with line relaxation for solving the viscous-plastic ice rheology. The thermodynamics are from the Semtner and Parkinson-Washington models. The grid is transformed such that the resolution is constant, thus avoiding the problem of convergence near the pole as on a latitude-longitude grid. For the coupled model, an ice model grid with 27 km resolution has been designed that includes all of the present day ice-coveraged areas in both hemispheres, minimizing the grid space required.
The atmospheric component for the coupled model is the massively parallel version of the NCAR Community Climate Model version 3 (CCM3), which is based on the spectral transform technique to use spherical harmonics as basis functions for the spatial discretization. Leapfrog finite differencing is used in time, and semi-Lagrangian transport is used for advection. This model includes the latest versions of radiation, boundary physics, and precipitation physics.
The method of tying the components together and allowing the exchange of fluxes and state variables is the flux coupler concept developed at NCAR for the Climate System Model. Since the grids of the various components are quite different, a conservative interpolation scheme developed by Jones (LANL) is being used for passing information between components.
Accomplishments
Four codes (CCM3, old POP, the prototype PCM, and new POP) were made operational on the T3E at NERSC. Several performance bottlenecks were identified and corrected. The PCM version of new POP was optimized by Wayland (NCAR) for the NERSC Cray T3E and demonstrated excellent scalability using 4, 8, 16, 32, 64, 128, and 256 PEs.
Significance
The scientific purpose of the production runs will be to assess the impact of greenhouse and other radiative forcings on decade to century time scales.
Publications
Washington, W. M., J. W. Weatherly, and R. M. Chervin. 1998. A new DOE coupled parallel climate model with high resolution ocean and sea ice. Proc. Ninth Symposium on Global Change Studies. Boston: American Meteorological Society, in press.
Snapshot of ocean temperature.