Friday, 5 August 2005: 9:45 AM
Empire Ballroom (Omni Shoreham Hotel Washington D.C.)
The efficient extension to the whole sphere of a numerical gridpoint model originally formulated for a limited rectangular area poses a number of challenges associated with the topological necessity that such an extension possesses either singularities of a single grid, or nonconformities at the junction or overlap where two or more grids are employed. The direct application of the latitude and longitude grid framework is usually accompanied by the need for zonal Fourier filtering that progressively reduces the effective number of degrees of dynamical freedom per latitude circle towards the poles. This choice also incurs some inefficiency in the application of both the "dynamics" and the "physics" parameterizations at these over-resolved locations. Other global single-grid topologies, such as the increasingly popular cubic grid arrangement, while more equitable in their areal coverage, do not completely avoid the problems of grid singularies which, while weaker, become also more numerous. An alternative strategy is to cover the globe with two overlapping grids. This can be done with relatively small overlap, in proportion to the total area, by adopting the "Yin-Yang" configuration of two equivalent cylindrical grid projections whose axes are mutually orthogonal. This arrangement, originally proposed and employed by Kageyama and Sato for deep earth modeling, eliminates the problems of grid singularities and pronounced anisotropies, but at the cost of requiring frequent grid-to-grid interpolations to keep both grid solutions coordinated. In this paper we shall discuss our ongoing work to adopt this arrangement in a global extension of the NCEP Nonhydrostatic Mesoscale Model (NMM). We describe numerical techniques for progressively blending the two grid solutions in a consistent manner, we assess the potential computational efficiency of this approach to global modeling, and we report on our progress with the construction and testing of such a system based on the NMM.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner