Yi Li, Imperial College London
A regional ocean data assimilation system
Abstract:
A regional ocean data assimilation system was developed in the South Australian Sea to provide accurate initial conditions for ocean currents prediction. This system consists of the NCAR Data Assimilation Research Testbed (DART) and the Regional Ocean Modelling System (ROMS). In this study, the horizontal resolution was set at 10 km and a deterministic Kalman Filter, the Ensemble Adjustment Kalman Filter (EAKF), was implemented with 50 ensembles. By assimilating the satellite observations the system enhances the simulation of key oceanic variables including SST, SSH and surface currents compared to control run without data assimilation. However, the posterior current field was found to be noisy so we introduced three balance operators (temperature-salinity balance, hydrostatic balance and geostrophic balance) to this system. All the unobserved variables were updated through the balance operators instead of through covariance matrix. It was shown that the assimilated currents with the balance operators were smoother than the original EAKF results and showed a lower RMSE when comparing with OSCAR observation data.