Assessment of the new NASA/GMAO Ocean Reanalysis

Robin M Kovach1, Eric C Hackert2, Andrea Molod3, Anna Borovikov1 and Jelena Marshak4, (1)Science Systems and Applications, Inc., NASA/GMAO, Lanham, MD, United States, (2)NASA Goddard Space Flight Center, Global Modeling and Assimilation Office, Greenbelt, MD, United States, (3)Global Modeling and Assimilation Office, NASA GSFC, Greenbelt, United States, (4)NASA Goddard Space Flight Center, GMAO, Greenbelt, MD, United States
Recently NASA’s Global Modeling and Assimilation Office (GMAO) has developed a new Subseasonal to Seasonal Prediction system Version 3 (GEOS-S2S-3). This upgrade replaces the GEOS-S2S-2 which is NASA’s current contribution to the North American Multi-Model Experiment seasonal prediction project (Kirtman et al., 2014). The main improvements for our GEOS-S2S-3 system include 1) a higher resolution MOM5 (Griffies et al., 2005) ocean model (now 0.25o x 0.25o x 50 layers), 2) an improved atmospheric/ocean interface layer (Akella and Suarez, 2018), and 3) assimilation of along-track satellite salinity into the ocean model (Hackert et al, 2019). Atmospheric forcing is provided by the NASA MERRA-2 reanalysis (Gelaro et al., 2017). Initialization for the ocean relies on the GMAO ocean reanalysis system which assimilates all available in situ temperature and salinity, satellite sea surface salinity, and sea level using the Local Ensemble Transform Kalman Filter (LETKF) implementation of (Penny et al., 2013) on a 5 day assimilation cycle with 20 fixed ensemble members.

In this presentation, we will authenticate our new GEOS-S2S-3 ocean reanalysis using standard GODAE validation metrics. For example, we will compare gridded fields of mean and standard deviation of the ocean reanalysis versus observed fields. We will show correlation/RMS of model versus observations and temperature and salinity mean profiles for the various basins and latitude bands. Basin-scale volume transports, such as the Atlantic Meridional Overturning Circulation and the Indonesian Throughflow will be validated. Equatorial ocean waves will be compared by decomposing sea level into Kelvin and Rossby components. For each of these metrics, we plan to validate the results and then compare our new GEOS-S2S-3 against the current production version, GEOS-S2S-2. Finally, we will compare 9-month seasonal forecasts initialized from these two systems for the tropical Pacific NINO3.4 region over the period 1981-present.