What is COAMPS?

COAMPS is a acronym for "The Coupled Ocean/Atmosphere Mesoscale Prediction System" and it represents an analysis-nowcast and short-term (up to 48 hours) forecast tool applicable for any given region of the earth. COAMPS includes an atmospheric data assimilation system comprised of data quality control, analysis, initialization, and nonhydrostatic atmospheric model components and a choice of two hydrostatic ocean models. The atmospheric component of COAMPS can be used for real-data simulations, the analysis can use global fields from the Navy Operational Global Atmospheric Prediction System (NOGAPS) or the most recent COAMPS forecast as the first-guess. Observations from aircraft, rawinsondes, ships, and satellites are blended with the first-guess fields to generate the current analysis. For the idealized experiments, the initial fields are specified using an analytic function and/or empirical data (such as a single sounding) to study the atmosphere in a more controlled and simplified setting. The atmospheric model uses nested grids to achieve high-resolution for a given area and contains parameterizations for subgrid scale mixing, cumulus parameterization, radiation, and explicit moist physics. Typical mesoscale phenomena that COAMPS has been applied to includes mountain waves, land-sea breezes, terrain-induced circulations, tropical cyclones, mesoscale convective system, coastal rainbands, and frontal systems.

The COAMPS model domain typically covers a limited area on the earth. The model grid size, usually referred to as grid resolution, can range from a few hundred kilometers (synoptic scale) down to approximately one meter when using the large-scale eddy (LES) mode. The actual dimensions used depend on the scale of the phenomena the user is interested in simulating. The model dimensions can be set so as to produce any rectilinear pattern and can also be rotated to align with any surface feature, such as the terrain or a coastline. COAMPS can be run with any number of nested grids, with the grid resolution in any mesh one-third that of the next coarser mesh.

COAMPS also contains an option to utilize either the Modular Ocean Model (MOM) or the Princeton Ocean Model (POM). In a fully-coupled mode, the atmospheric and ocean models can be integrated simultaneously so that the precipitation and the surface fluxes of heat, moisture, and momentum are exchanged across the air-ocean interface every time step. Optionally, the atmospheric model or either of the ocean models can be used as a stand-alone system.

Restriction for COAMPS usage

COAMPS is developed to support Tactical Naval operations and R&D usage. In order to get a copy of COAMPS source code, organizations other than those within the Department of Defense (DOD), are required to sign NRL's Memorandum of Agreement.

What kind of experience is needed to use COAMPS

The users should have basic knowledge of numerical weather prediction, an understanding of atmospheric science in the area of numerical weather prediction, some knowledge of UNIX operating system and Fortran 77 programming language.

Hardware and software requirement for COAMPS

COAMPS is a portable atmospheric modeling system (the ocean model coupling is under development, not for release now), that can currently be run on most of the major vendor machines such as Cray, SGI, DEC ALPHA, SUN and HP workstations. Future releases will include the ability to run on the mpp type machines.

The COAMPS analysis uses less memory than the forecast. As a first-cut estimation, one can use the Fortran program memory.f to obtain the COAMPS memory requirement for your grid set up. More accurate memory requirement can be obtained by actually running COAMPS.
cpu time (per domain) = CF * number of grid points * number of model time steps
    CF is the computation factor, this number is about 1.6e-5 second on 
    one processor CRAY C90 and 6.7e-5 second on one processor SGI R10000 
     number of grids = (number of grid points in X direction) x 
                       (number of grid points in Y direction) x 
                       (number of grid points in Z direction)
     number of model time steps = forecast time in second / 
                                  model time steps
     usually, the coarse domain model time steps in seconds is about 3 
     times the coarse domain grid size in kilometers. The nested domain
     time step is 3 times smaller than the next coarser mesh (for example 
     if the time step used in coarse domain is 360 second, the 
     nested domain 1 has a time step of 120 second, the nested domain 2
     has a time step of 40 second ... etc.).  

Chen (chen@nrlmry.navy.mil) Last Updated: Nov 20 1998 

| First Page | Prev Page | Next Page | Back to Text |