Assimilation of Observations into a NAVGEM Global Analysis and Multiple COAMPS Limited Area Analyses by a Single Data Assimilation Process

This presentation describes an experimental data assimilation system that assimilates observations into a global analysis and multiple regional analyses in different limited area regions by a single data assimilation process. The system is based on the NAVGEM global model, the COAMPS limited area model and the NAVDAS data assimilation system of the U.S. Navy. In particular, NAVDAS is applied to a global model state vector that is informed by a lower resolution state vector of NAVGEM and the higher resolution state vectors of COAMPS for the different limited area regions. Then, the lower resolution global analysis and the higher resolution limited area analyses are obtained by interpolation of the resulting NAVDAS analysis. The motivation to design such a system is threefold: (1) in the COAMPS regions, the NAVGEM analysis can consider information about the atmospheric state that is higher resolution than its own representation of the state, (2) the COAMPS analyses can consider information from NAVGEM about the global state from beyond the lateral boundaries, and (3) when there are many COAMPS regions, as in the operational models suite of the U.S. Navy, preparing a single analysis rather than a global analysis and many limited area analyses can save significant computer time. The talk presents both results of experiments that have been carried out with different resolution versions of NAVGEM, in order to study the role of scale interactions in the evolution of the initial condition errors in the ensuing forecasts; and preliminary verification results for a test version of the system that includes two limited area domains: one over the continental U.S. and another in the Mediterranean.