Estimation and prediction of oceanic Eddy fields

Abstract

A system for the optimal estimation and forecasting synoptic/mesoscale ocean currents is presented consisting of an observational system, a statistical model and a dynamical model. The methodology is developed for the case when the dynamical model is a baroclinic, quasigeostrophic open-ocean model. Issues involved include the synoptic and regular gridding of asynoptic non-uniform data, the construction of composite fields from several data types, and the initialization and verification of the dynamical model. Models for the error fields required for weighted combinations of independent field estimates are obtained from the analysis of an observational model; a barotropic example is presented. The statistical model is applied to the POLYMODE Synoptic-Dynamic Experiment (SDE) XBT and current meter data which is successfully assimilated in the dynamical model. Dynamical forecasting experiments are carried out and the difference fields between analysis fields and forecast fields are studied and attributed to: error sources in the analysis, in initial and boundary data, and to the dynamical adjustment of those fields by the model. Quantitative results indicate that useful accuracies can be efficiently obtained.