2013 Volume 61 Pages 31-48
The present study focuses on the recent update of the global ocean data assimilation system in the Meteorological Research Institute (MRI). The model domain of the system is extended to cover the Arctic Ocean with advanced physical processes incorporated (e.g., sea ice). Compared with data-free simulation, data assimilation improves reproduction of the circulation field in the Arctic Ocean, as well as temperature and salinity which are directly modified by data insertion. The system also effectively reproduces the observed sea-ice field. In addition to extending the domain, we have sophisticated statistical information in the global ocean for data assimilation, including coupled temperature-salinity empirical orthogonal function modes in a background error covariance matrix, by incorporating new observational datasets. This improves variability, especially in the salinity field in the tropical Pacific and Indian Oceans, even during the period before the recent increase of salinity observations. The updated system effectively reproduces the oceanic structure associated with the 1997-1998 El Niño. These results will likely contribute to more accurate seasonal predictions using a coupled atmosphere-ocean model.