An effort, currently underway at the Jet Propulsion Laboratory, aims to produce routine analyses of the three-dimensional state of the ocean by assimilating satellite and in situ observations into a state-of-the-art general circulation model. The scientific objectives are the study of the seasonal-to-interannual oceanic variability and of the heat and water flux exchanges with the atmosphere. This paper summarizes the status of this effort.
Satellite altimetry provides the only routine observation of a dynamic variable of the global ocean. For instance, measurements from the TOPEX/POSEIDON altimeter has provided dramatic basin-wide images of the space-time evolution of sea level associated with the '97-'98 El Nino event. However, sea level is a vertically integrated property of the oceanic state and cannot by itself fully determine the three-dimensional circulation of the ocean. To resolve the vertical structure of the circulation and to estimate properties that are otherwise difficult to observe remotely, such as subsurface current velocity and temperature, it is necessary to combine the altimetric data with in situ observations and with a general circulation model.
The oceanographic data assimilation problem is distinct from its meteorological counterpart in several respects. First the relative scarcity of oceanographic observations puts increased emphasis on the proper characterization of model and data errors. Second, crude parameterizations of the energy containing scales are required because global ocean models cannot resolve them. Third, the temporal scales of ocean circulation are much longer than in the atmosphere--- offering both the opportunity for longer term predictions and the complication of cumulative biases. Finally, the oceanographic emphasis is still one of understanding, rather than forecasting.
A dual assimilation approach is being pursued whereby an approximate Kalman filter is used for routine weekly analyses and for understanding the errors, and the adjoint method is used for periodic reanalyses. The effectiveness of the assimilation of altimetry data will be tested using withheld in-situ data (such as XBT, floats, etc.). The ultimate analyses will be based on all the available observations, however. It is anticipated that this effort will also benefit regional applications and climate prediction by providing open ocean boundary conditions and initial oceanic conditions, respectively.