EVALUATION OF THE GLOBAL OCEAN DATA ASSIMILATION SYSTEM AT NCEP: THE PACIFIC OCEAN

The El Nino-Southern Oscillation (ENSO) phenomenon has been forecast reasonably well by both numerical and statistical models at lead times of 6-12 months. As the atmospheric weather system behaves in a chaotic manner beyond timescales of days, it is mainly the ocean that carries the potential for predictability skill up to a year in advance. Therefore, a large amount of effort has been devoted to intialize accurately the oceanic components of coupled oceanatmospheric models using ocean data assimilation systems (ODAS). The ocean analyses generated by ODAS are being used not only in initializations for coupled ocean-atmospheric models but also in diagnostic studies and real time monitoring of the subsurface oceanic conditions in support of climate prediction activities (e.g. “ENSO Diagnostic Discussion” at Climate Prediction Center of NOAA). The ODAS developed at the National Centers for Environmental Prediction in 1995 was configured for the Pacific Ocean only. Recently, a new global ocean data assimilation system (GODAS) has been developed at NCEP for initialization of the new global coupled oceanatmospheric model that will be used for operational ENSO prediction in the future. It is imperative for us to evaluate the GODAS against independent data sets, and compare it with the operational ODAS for the Pacific Ocean, and, eventually, implement the GODAS in the real time monitoring products at CPC. The GODAS at NCEP was developed using the Geophysical Fluid Dynamics Laboratory Modular Ocean Model version 3 (MOM.v3) and a three-dimensional variational data assimilation scheme. Compared with the operational ODAS developed for the Pacific Ocean (Behringer et al. 1998; referred as RA6 hereafter), the major changes include 1) the extension from the Pacific basin to the quasi-global domain for 75S-65N,