Towards Operational Ocean State Estimation

An ocean data assimilation system is being implemented so as to routinely estimate the time-evolving, global, three-dimensional state of ocean circulation. Satellite remote sensing, such as TOPEX/POSEIDON, has been providing tantalizing observations of changes in ocean circulation associated with El Nino/La Nina and possibly other longer-term climate changes (e.g., Pacific Decadal Oscillation). The assimilation system aims to extend such surface monitoring capabilities of satellites to estimate the entire three-dimensional state of the ocean by combining satellite measurements with in situ observations using ocean models. A dual assimilation scheme based on an approximate Kalman filter and the adjoint method are implemented with a global ocean general circulation model. The Kalman filter is implemented to conduct near real-time analyses of the oceanic state while the adjoint assimilation will be conducted periodically for reanalyses. The Kalman filter, based on a new approximation (a hierarchical reduced-state Kalman filter), will also provide quantitative error estimates that will be used in defining the weights in the adjoint optimization. The adjoint model is generated by an automatic differentiation software, the Tangent Linear and Adjoint Model Compiler. The adjoint is also used to study sensitivity of diagnostic quantities to various controls (initial condition, surface fluxes, and model parameters). Initial experiments focus on the tropical Pacific Ocean aimed at diagnosing processes underlying the 1997-1999 El Nino/La Nina event. Preliminary analyses will be presented in addition to progress in developing the assimilation system.

This research effort is part of a NOPP node, "Estimating the Circulation and Climate of the Ocean" (ECCO). The ECCO consortium aims to advance ocean state estimation from an experimental status to a practical quasi-operational tool for studying large-scale ocean dynamics.