The data assimilation system in the fine scale forecast model in use at the Air Force Weather Agency (AFWA) is a height-wind based multivariate optimum interpolation (MVOI) model. The MVOI system is currently in production in the Pennsylvania State University (PSU)/National Center for Atmospheric Research (NCAR) Mesoscale Model (MM5). The MVOI system was chosen because of its availablity. The MVOI was previously during the 1990s in the Relocatable Window Model (RWM). It performed well in that model, it was relocatable, and it used many the obsevation databases at available at AFWA. The MVOI for MM5 is also a relocatable analysis system, and it has been upgraded to ingest some of the new meteorological data currently available at AFWA.

The modified design of the data assimilation system is such that it lended itself fairly easily to parallelization. Due to the time constraints imposed in running the analysis system for multiple theaters, and the requirement to run the data assimilation system in a Four Dimensional Data Assimilation (FDDA) the analysis system has been parallelized.

This paper discusses the method used in the data assimilation process and the data used in the MVOI at AFWA Because parallelization was needed to ensure timely delivery of the analyses to multiple forecast domains, there will be a description of the parallelization techniques employed.