THEORETICAL CLIMATE DYNAMICS - DAMEE-NAB - IEM3
THEORETICAL CLIMATE DYNAMICS GROUP
UCLA's IEM 3 Abstract

TCD
home		 DA DAMEE IEM4
Model		 IEM4
DA

Model Evaluation Experiments and Parameter Estimation for the North Atlantic Basin Using the GFDL MOM

Michael Ghil(*,**), Chung-Chun Ma(*), Kayo Ide(*,**),
C. Roberto Mechoso(*), Arthur Or(*,**)

(*) Department of Atmospheric Sciences and
(**) Institute of Geophysics and Planetary Physics,
University of California, Los Angeles, CA 90095-1565, U.S.A.

Our ultimate goal is to build a comprehensive data assimilation system to nowcast and forecast meso-scale oceanic motions on the synoptic time scale in a coupled atmosphere-ocean model. This requires modeling the oceanic and coupled general circulations, designing a sophisticated but efficient data assimilation procedure including the observing system, and studying the phenomena that play a major role in the North Atlantic Basin's coupled climate system. All the components of this comprehensive system must be carefully examined individually and then optimally combined for successful now- and forecasting.

In the UCLA analysis-and-forecast system, the oceanic circulation is modeled using the GFDL Modular Ocean Model (MOM). For the Data Assimilation and Model Evaluation Experiments--North Atlantic Basin (DAMEE-NAB) with 0.5 deg. X 0.5 deg. horizontal resolution, we use 15 layers in the vertical, of which 7 layers are in the first 300m to resolve the turbulent mixing layer, using the Mellor and Yamada (1982) parameterization. Two computational domains (Figure), 9--47 deg. N with 3 deg. wide buffer zones at the northern and southern boundaries and 3--61 deg. N, with similar exterior buffers, are chosen to examine the effect of the domain size on Gulf-Stream behavior. The smaller domain agrees with the DAMEE-NAB specifications, as do the bathymetry and surface forcing. To cope with the incompleteness of the temperature-salinity data sets for the initialization and buffer-zone boundary conditions, the GDEM/Lozier (1995) data are injected into a complete Levitus (1982) data set wherever available. The model requires 1.8 and 2.2 MW in memory on C90 and a 10-year run takes 50 and 70 CPU hours for the smaller and larger domain respectively.

Results will be shown as E-W and N-S temperature and salinity sections, at 27 deg. N and 55 deg. W, respectively. Other measures of performance, according to the criteria agreed upon during the Information Exchange Meeting (IEM) No. 2, will be presented as time permits.

Data assimilation methods have been tested in a rectangular box version of the model, 15 deg. longitude by 25 deg. latitude, with 10 layers in the vertical, at 0.25 deg. X 0.25 deg. horizontal resolution. Preliminary results for both state and parameter estimation (Hao and Ghil 1995) will be discussed, based on an iterative succession correction method (Daley 1991).
Figure Caption The two computational domains for the North Atlantic Basin used in the UCLA integrations.

[ TCD home | DA | DAMEE | IEM4 Model | IEM4 DA ]
April 1, 1997