Tuesday, 16 January 2007: 9:00 AM
The Global Ocean Data Assimilation System (GODAS) at NCEP
212B (Henry B. Gonzalez Convention Center)
The Global Ocean Data Assimilation System (GODAS) is a 3D variational assimilation system that has as its primary purpose to provide ocean initial conditions for seasonal to interannual (S/I) forecasting with the NCEP Coupled Forecast System (CFS). This presentation will provide a general description of the current operational version of GODAS and of recent developments that will become part of future versions. There will be particular emphasis on the importance of constraining salinity as part of the assimilation and on the impact of assimilating altimetry data. Because S/I prediction has focused on the ENSO phenomenon, ocean observations in the tropical upper ocean have been most important to GODAS. Also, because S/I forecasts must be calibrated by a long series of hindcasts, a data set that spans 20+ years is required. Given these criteria, GODAS relies heavily on ocean profiles of temperature (XBT, TAO/TRITON, Argo) and on sea surface temperature (SST). The stability of the TAO/TRITON mooring array since its establishment in the early 1990s and the recent growth of the Argo autonomous drifter array mean that in situ profiles will retain their central importance. However, if only temperature observations are assimilated, the result is a significant degradation of the tropical salinity field. Historically, ocean observations of salinity have been too sparse to be of practical use, but constraining GODAS salinity by the climatological relationship between temperature and salinity has proven to be a workable solution. In the future a better solution will be provided by the Argo array which produces, in addition to temperature profiles, the first extensive global set of salinity profiles. Initial experiments show that when the GODAS is adapted to use Argo salinity profiles, there is a measurable improvement in the GODAS salinity field and some improvement in tropical surface currents. The Argo array has only approached full global coverage during the last 2 years and several more years will be needed to fully gauge the impact of these data. A consistent, uninterrupted altimetry data set has been available since late 1992 from the TOPEX and Jason-1 (T/J-1) satellites. Experiments demonstrate that, in the equatorial Pacific, the assimilation of the TAO/TRITON mooring data already leads to a good representation of tropical SSH in the operational GODAS. In the Pacific, the additional assimilation of T/J-1has slight impact in the tropics but improves GODAS SSH beyond the bounds of the TAO array and well into the subtropics. In the Atlantic and Indian Oceans, where there are no exact equivalents of the TAO/TRITON array, the assimilation of T/J1 greatly improves the GODAS SSH. The presentation will finish with a brief description of the impact of deep data assimilation (down to 2500m) and of a multivariate constraint on geostrophic velocity. An important conclusion to be drawn from the presentation is that careful attention to the use of available assimilation data sets and to technical details of the assimilation system can lead to significant improvements in the GODAS representation of the ocean state.