The Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS) developed by the Naval Research Laboratory (NRL) is a complete mesoscale prediction system consisting of atmospheric and ocean data assimilation (including data quality control), analysis, initialization, and a nonhydrostatic atmospheric forecast model coupled to a hydrostatic ocean model. The atmospheric system has been used for operational mesoscale forecasting since 1996, providing products to the meteorological community from both a supercomputer central site (Fleet Numerical Meteorology and Oceanography Center) as well as regional sites using workstations (Naval centers, Universities, Government Agencies, etc.).

This talk will present an overview of recent developments and improvements to the atmospheric and ocean model components. The most visible change in the system has been a restructuring to run on massively parallel machines. COAMPS was originally coded for vector machines, but has recently been recoded to be able to run on either vector or parallel machines through inclusion of MPI and OPEN-MP directives. A summary of the efficiency and portability of the new system will be given.

An option in the atmospheric model to allow for moving inner nests has been implemented into the newly coded system. This feature is particularly useful for naval operations as well as forecasting features such as tropical cyclones or squall lines. The user has flexibility in specifying how the nests will be moved, to allow for pre-determined movement with a naval battle group for instance, or to follow the maximum vorticity, surface pressure, precipitation maximum, etc. of the feature. Results showing sensitivity of model forecasts to the method in which surface fields (such as topography and surface characteristics) are derived in "moved areas" will be presented

The atmospheric component of COAMPS is also coupled to an aerosol model in which the species are treated as passive scalars. Any number of size bins and species may be used. This aerosol model uses COAMPS fields at each time step, and thus it is considered tightly coupled. Simulations will be presented showing the sensitivity of aerosol transport to surface characteristics determined by high-resolution (1-km) land surface databases.

The ocean data assimilation system currently being developed at NRL uses an analysis based on a 3-D multi-variate optimum interpolation (MVOI) scheme and the Navy Coastal Ocean Model (NCOM) forecast model with a hybrid sigma-z coordinate system. Recent high-resolution real-data simulations will be presented.