The ocean component of the coupled model, MICOM, has 2 x cos(lat) deg. resolution on a near-global domain (69S-66N) with 15 isopycnal layers (sigma_0) topped by a Kraus-Turner mixed layer. The atmospheric component, NCAR/CCM3, is configured at two horizontal resolutions: T42, which translates to 2.8x2.8 degree grid size; and T21, whose resolution is half that of T42. Both configurations have 18 levels in the vertical.

Because the atmospheric and oceanic grid points are not coincident, data must be interpolated during information exchange. Our 'flux coupler' works as follows: the only variable that is interpolated from the ocean to the atmosphere is SST (outside the MICOM domain, SST forcing for the atmosphere is from the GISST monthly climatology); air-sea interface fluxes such as heat, fresh water and momentum flux are only calculated by the CCM3; these fluxes are then interpolated back to the oceanic grid points to provide surface forcing. The interpolation algorithm assumes that fields are spatially constant in each grid cell and preserves the global mean.

The coupled simulations use neither separate spin up runs nor flux corrections, except that the ocean is initialized with Levitus climatology and driven for one year by COADS monthly climatology to establish reasonable mixed layer depths. The atmosphere is started from an instantaneous state representing January 15. Ocean and atmosphere exchange information every ocean time step, that is, every hour. The atmospheric time step is 20 minutes.

Results from T21 MICOM-CCM3 Simulation

There does not appear to be one when looking at the time evolution of zonal annual mean sea surface temperature and surface wind stress amplitude (see Fig. 1). However, there are trends in the deep ocean, judging from global mass balance (see Fig. 2 ), even though the amplitude is moderate.

How similar are the surface forcing fields in the coupled model to their observational counterparts, and to those obtained by running the CCM3 in stand-alone mode?

We looked at meridional profiles of annual mean wind stress, heat and fresh water fluxes at the air-sea interface (see Fig. 3). There clearly are discrepancies between the AGCM simulations and the observations. The differences between coupled and uncoupled simulations are particularly obvious in the equatorial region.

Does the model capture the basic structure of observed mean climate?

The answer to this question is yes after examining the meridional overturning streamfunction in the ocean (see Fig. 4) and the vertical structure of the zonal wind component in the atmopshere (see Fig. 5).

Based on these results and others not shown here, we conclude that the coarse resolution coupled MICOM-CCM3 is doing a reasonably good job in simulating the current climate. We stopped the T21 MICOM-CCM3 run at year 50 and moved on to higher resolution (T42).

Results from T42 MICOM-CCM3 Simulation

Fig. 1 shows EOFs of winter mean sea surface temperature between the equator and 68N, 100W and 20E. (These bear resemblance to observations and other modeling studies.)

Fig. 2 shows the North Atlantic Oscillation (NAO) index and its spectral characteristics. (There is a statistically significant peak at a period of about 12 years.)

Fig. 3 shows the "Nino3" index (SST averaged between 5S-5N, 150W-90W) with interannual variability comparable to observations.

Fig. 4 shows the maximum value of the meridional overturning stream function in the Atlantic against time. (This time series has a pronounced 30-year oscillation -- the period is not strictly regular -- in the coupled run but not in the uncoupled MICOM simulation.)

For further information, contact W. Cheng.

Return to MICOM home page