oceans are in antiphase with the ACC at 700 m. Much the same pattern holds in these regions at 2000 m. The South Pacific and South Atlantic vary in phase with the ACC at both depths, but an unexpectedly strong signal (positive) is apparent in the western Equatorial Pacific. At 700 m the strongest signals are in the Indian Ocean, but at 2000 m this distinction is held by the South Pacific and the North Atlantic. It is obvious that the ocean mode of variation discussed here is truly global in extent.
The rms variability of the salinity field from the WW and RR runs is shown in Figure 12 for depths of 75 m. The RR run produces far more variance in the salinity field, especially in the deep ocean (e.g., 2000 m, not shown) where rms values of 0.06 to 0.08 psu are common in higher latitudes. The spatial distribution of the variability between the two runs clearly shows the importance of large-scale air-sea interactions in forcing the model. If the RR run is at all realistic, then the coupling between the two media needs to be taken into account in studies of interdecadal variability. It is interesting that the spatial structure of the
modal response was similar between all the runs, even if the levels of variance were not. This suggests the response is a leading "eigenmode" of the global ocean model that is easily excited by a wide range of forcing.
A reasonably sophisticated, realistic OGCM has been forced with annual cycles of wind stress, temperature, and fresh-water flux, and also with anomalies of fresh-water flux. The latter anomaly model simulations range from forcing that is white in both space and time to a model that is red in both domains and also incorporates feedback between the fresh-water flux and local SST.
The results of these simulations show a richly structured response. One prominent mode is that discovered by Mikolajewicz and Maier-Reimer (1990, 1991). We found that mode is driven principally by anomalous fresh-water flux in the higher latitudes of the Southern Hemisphere. The spatial structure and details of the forcing are not as important to the mode as the magnitude of the forcing. Monthly anomalies greater than 1 mm/day will excite the mode, while values below 0.5 mm/day do not. Atmospheric-model results and NMC analyses suggest that realistic values of anomalous fresh-water flux in the high latitudes are of the order