carried out: The first was a continuation of the climatology run for an additional 1500 years. Results from this run have the designator C. Each of the three stochastic, anomalous fresh-water flux models was mated to the OGCM and run for 1500 years. The first 500 years of each integration were discarded, leaving the remaining 1000 years for analysis. The designators for these runs are WW, RW, or RR depending on which atmospheric model was used. All three anomaly runs included the same seasonal cycle, fresh-water flux, wind stress, and pseudo-heat flux forcing used in the climatology run.
This section briefly summarizes some of the more interesting results of the four experiments described above. Since this research effort is in the very early stages, these results are presented in descriptive form.
The model-simulated transport through the Drake Passage is shown in Figure 2 for several of the experiments. The C run shows nearly constant transport close to that observed, with annual variations of less than 1 percent (no illustration shown). The WW run, on the other hand, for s = 2 mm/day shows large variability, wherein the transport can change by 50 percent (Figure 2a). The time scale for this fluctuation is of the order of 300 years; it seems to be the mode of variation previously found by MMR (see below). By contrast, the RW run (Figure 2b) shows a step-like jump but otherwise low interdecadal variability. The feedback or RR run (not shown) is much like the RW run but with almost no high-frequency variability.
We deduced from the above results that the spatial correlation in the fresh-water forcing was not particularly important to the model response. Similarly, we concluded that it is the magnitude of the forcing in the high latitudes (only) that really affects the model behavior. We confirmed these conclusions by rerunning the RW case, but with the magnitude of the forcing increased by a factor of 5 for latitudes above 40°, where the RW atmospheric model was deficient in energy. This increased forcing brought the magnitude of the flux up to the order of I mm/day and corrected for the model's low variability in that region. The resulting transport through the Drake Passage (Figure 3) now resembles that found in the WW run with regard to both magnitude and time scale. In addition, similar experiments showed that it was the E - P flux south of 40°S that generated the MMR mode, while the flux in the northern ocean was of little significance other than local.