The new global ocean data assimilation system (GODAS) at NCEP was developed using the Geophysical Fluid Dynamics Laboratory's 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 (referred as RA6 hereafter), the major changes include 1) the extension from the Pacific basin to the quasi-global domain for 75^oS-65^oN, 2) the model change from MOM.v1 to MOM.v3 that contains more vertical levels, an explicit free surface, the Gent-McWilliams mixing scheme and an improved boundary layer mixing scheme, 3) the forcing change from momentum flux forcing only to momentum flux, heat flux and fresh water flux forcings of the reanalysis 2, most importantly, 4) the data input change from temperature only to temperature and synthetic salinity that is constructed from temperature and local T-S climatology. The temperature data includes those from XBTs, profiling floats and TAO moorings. The evaluation of the GODAS is focused on the comparison with RA6 since the goal is to implement the GODAS operationally at the end of 2003. The temperature fields from GODAS and RA6 are compared with the TAO mooring data to estimate how well the analyses fit to observations. Then the quality of the analyses is evaluated with independent data sets such as the sea level observations at tide gauges and from the TOPEX/Poseidon and Jason satellite altimeters, the current data from the TAO Acoustic Doppler Current Profiles and the satellite-derived current analysis.

Preliminary resutls suggest that the temperature field in GODAS is closer to observations than that in RA6, and the poor salinity field in RA6 is dramatically improved. Although this version of GODAS does not assimilate satellite sea level as RA6 does, GODAS does as well as or better than RA6 in comparisons with TOPEX altimetry and tide gauge sea level records. The current field in GODAS is generally better than that in RA6, but the surface currents in both GODAS and RA6 have large discrepancies (30 cm/s) in the western equatorial Pacific when compared with the TAO and satellite-derived currents. It is suspected that the errors in the current field are related to the errors in the density field and wind forcings. Since both sea surface salinity (SSS) and subsurface salinity have significant impacts on the density field and dynamic height in the western Pacific, and possibly influence the onset of ENSO through the barrier layer physics, we will estimate the errors in SSS and subsurface salinity by comparing them with the observed salinity and reconstructed salinity using the covariant EOF modes of temperature and salinity. We will also conduct a detailed analysis of the annual and interannual variability of SST, SSS, surface current, wind stress, heat flux and fresh water flux in the warm pool. The results will give an estimate of the quality of the analysis in the warm pool physics that is believed to play a critical role in the diagnosis and prediction of ENSO.