As an intermediate step towards such seamless systems, an experimental coupled system made of a 25km (horizontal) resolution version of the NASA GEOS-5 AGCM coupled to the MOM-5 ocean GCM running at 3km-11km resolution and to a 3km-7km version of the CICE sea ice model has been implemented. The atmospheric component of the system is constrained by the NASA global modeling and assimilation office (GMAO) MERRA-2 atmospheric reanalysis while sea ice fraction, sea surface temperature, sea surface salinity from the Aquarius mission and surface chlorophyll from MODIS are assimilated into its ocean and sea ice components.
The assimilation of ocean and sea ice observations into eddy resolving global models presents several challenges usually absent from lower-resolution applications. First, the computational cost of the popular ensemble Kalman filter (EnKF) and ensemble optimal interpolation (EnOI) methods precludes their use to estimate background error covariances. To address this, a methodology named ensemble state adaptive forecast error estimation (EnSAFE) has been developed to generate an ensemble of background error estimates from the meso-scale information content of a single background state. Other resolution-dependent difficulties necessitated (1) the development of an adaptive observer operator to assimilate large scale information into the system without suppressing the model eddy variability and (2) the elaboration of a methodology to make the background error covariance model aware of (coastal) land boundaries. This talk will focus on these and other issues peculiar to high resolution coupled data assimilation applications and results pertinent to tropical cyclone modeling and prediction as well as ocean color, sea surface salinity and sea ice data assimilation will be presented and illustrated with high definition (HD) animation material.