1
An approximate Kalman filter for coupled
ocean-atmosphere assimilation A NASA-MAP
Project
I.Fukumori, T.Lee (JPL), M.Ghil (ENS UCLA),
D.Kondrashov (UCLA), A.Rosati, S.J.Lin (GFDL),
G.Philander (Princeton), C.Sun (GSFC/UMBC),
D.Neelin (UCLA), R.Atlas (AOML)
ECCO meeting, 31 Jan 2006
2
Reduced-State Kalman filter
Gist
Employ approximation of model state with much
smaller degrees of freedom than the model itself
to derive approximate model state error
covariance matrix.
ECCO B coarse horizontal grid, vertical dynamic
modes, diabatic heaving motion
3
Strategy for atmosphere QTCM
Quasi-Equilibrium Tropical Circulation Model
A Galerkin representation of vertical structure
based on quasi-equilibrium convective closures.
temperature
moisture
horiz. vel.
vert. vel.
Zeng, Neelin, Chou (2000, JAS)
4
Example QTCM Precipitation Anomaly
Zeng, Neelin, Chou (2000, JAS)
5
Example QTCM Precipitation Anomaly
model precipitation along Eq
Xie-Arkin obs. along Eq
Zeng, Neelin, Chou (2000, JAS)
6
Other Elements
Model
GFDL Coupled Climate Model, CM2.1 Atmosphere
(AM2.1, 2.5ºx2ºx24), ocean (OM3.1,
0.3º1ºx1ºx50), sea ice (SIS, 3-layer)
Control
Fluxes between ocean atmosphere, Parameters
defining the exchange.
Sequential Parameter Estimation
Augmented KF taking advantage of the low-rank
modification (Sherman-Morrison-Woodbury formula.)