A number of troubling weaknesses have been found in the ocean monitoring tools used by the CPC. Precipitated by the international Ocean Reanalysis Intercomparison Project (ORA-IP) and the Observing System Experiments for evaluating the TAO array, Co-PI Xue identified large systematic errors in salinity and velocity fields in the operational Global Ocean Data Assimilation System (GODAS) and Climate Forecast System Reanalysis (CFSR), as well as a serious issue in fitting to observations too strongly in both GODAS and CFSR. The ocean-alone GODAS is forced by the NCEP Reanalysis 2, has an outdated ocean model (MOM3), a low resolution (1ox1o), atmospheric fluxes from an old atmospheric reanalysis, and lack of a sea-ice model. The coupled CFSR suffered a serious issue of climatology shift around 1999 due to assimilation of ATOV satellite observations. Both GODAS and CFSR had large departures from the ensemble mean of multiple international ocean reanalysis products during a large data gap in the TAO array in late 2012, most likely due to the outdated data assimilation scheme (3DVar).
As part of a previous CPO MAPP project, a Hybrid 3DVar/EnKF Global Ocean Data Assimilation System (Hybrid-GODAS) was implemented, and was evaluated using real data for a 21-year reanalysis. PI Penny showed that the Hybrid-GODAS produced significantly improved thermohaline structure compared to the 3DVar-based GODAS. Comparing GODAS and CFSR with the new Hybrid-GODAS, Co-PI Xue has found significant improvements in the temperature, salinity, sea surface height and velocity analysis against observations.
We propose to transition the Hybrid-GODAS to TRL 8, ready for evaluation and implementation into operations at NCEP. We use the recently released GFDL Modular Ocean Model version 6 (MOM6) at 1⁄4ox1⁄4o horizontal resolution and Sea-Ice Simulator (SIS2), which is the ocean component of GFDL’s CM4 earth system model for the next series of Coupled Model Intercomparison Project (CMIP6) experiments. We will implement near real-time monitoring of the ocean/sea-ice state by assimilating oceanic variables such as in situ temperature and salinity profiles, surface drifter data, and along-track satellite measurements of sea surface height, temperature, and salinity. A companion project (PI: Carton) will prepare sea-ice data feeds and perform specialized assimilation of observed sea-ice data into the ice model using the Local Ensemble Transform Kalman Filter (LETKF). Pending positive results from Carton’s team, the sea-ice data can be included into the operational Hybrid-GODAS with minimal effort. The proposed upgrades will significantly improve our capability in monitoring and understanding not only temperature but also salinity and velocity variability that play a critical role in the evolution of El Niño / Southern Oscillation (ENSO) and therefore seasonal climate forecasts.