Ocean data assimilation using intermittent analyses and continuous model error correction

A new data insertion approach is applied to the Derber and Rosati ocean data assimilation (ODA) system, a system that uses a variational scheme to analyze ocean temperature and provide ocean model corrections continuously. Utilizing the same analysis component as the original system, the new approach conducts analyses to derive model corrections intermittently at once-daily intervals. A technique similar to the Incremental Analysis Update (IAU) method of Bloom et al. is applied to incorporate the corrections into the model gradually and continuously. This approach is computationally more economical than the original. A 13-year global ocean analysis from 1986 to 1998 is produced using this new approach and compared with an analysis based on the original one. An examination of both analyses in the tropical Pacific Ocean shows that they have qualitatively similar annual and interannual temperature variability. However, the new approach produces smoother monthly analyses. Moreover, compared to the independent observations from current meters, the new equatorial currents are significantly better than the original analyses, not only in maintaining the mean state but also in capturing the annual and interannual variations.