A Kalman Filter for Ocean Monitoring

Abstract

The feasibility of global ocean state estimation by sequential data assimilation is demonstrated. The model componenet of the assimilator is the GROB version of the MPIMET ocean circulation model HOPE. Assimilation uses the Fokker-Planck representation of the Kalman Filter. This approach determines the temporal evolution of error statistics by integration of the Fokker-Planck Equation. Phase space advection and diffusion are obtained from histogram techniques considering the model as a black box. For efficiency, the estimate combines nudging and Kalman Filtering. The ocean state is estimated for the El Nino year 1997 by dynamical extrapolation of observed sea-surface temperatures and TAO/Triton subsurface temperatures. The model-data combination yields improved estimates of the ocean's mean state and a realistic record of El Nino related variability. The assimilator proves as an efficient, viable and thus practical approach to operational global ocean state estimation.