The Fleet Numerical Weather Central Operational Primitive-Equation Model

Abstract

The U.S. Navy Fleet Numerical Weather Central (FNWC), Monterey, Calif., five-layer, primitive-equation, atmospheric prediction model has been under development since late 1968 and became operational in September 1970. Seventy-two-hour prognoses are generated twice daily, requiring 2 hr per run. The conservation forms of the difference equations are based on the Arakawa technique and are integrated, using a 381-km space step (at 60°N) and a 10-min time step, on sigma surfaces. Realistic mountains are used. Pressure-force terms are replaced by a single geopotential gradient on pressure surfaces synthesized locally to reduce inconsistent truncation error. Lateral diffusion is performed on forecast difference fields to prevent systematic distortions of sigma-surface state parameter distributions. Stress is applied at the lowest level. Restoration boundaries and centered time differencing are used. The integrations are restarted each 6 hr with a Euler-backward step to reduce solution separation. The moisture and heat source and sink terms are modeled in a similar manner to those in the Mintz and Arakawa general circulation model. Terms representing evaporation and large-scale condensation, sensible heat exchange, a parameterization of cumulus convection and precipitation, and solar and terrestrial radiation are included. Dry convective adjustment precludes hydrostatic instability. Initialization of the model is based on FNWCs Northern Hemisphere objective analyses of the state parameter structure from the surface to 50 mb. Nondivergent initial wind fields are obtained from solution of a linear balance equation. The verification results are encouraging. The 36-hr sea-level pressure prognoses exhibit considerable skill. The model's ability to simulate the generation of new storms in the winter is noteworthy. Problems included the slow movement of small-scale features (truncation problem), need for improvement in description of the heating and friction, and a need for improved initialization.