DESER: What is the mechanism for beginning a new cycle of the eastward-moving salinity anomalies?

SARACHIK: The Antarctic Circumpolar Current is a re-entrant current in our model. But of course it's hard to say how any oscillation begins; there's no particular ''start" spot in the cycle.

PHILANDER: You mentioned that you were concerned about the consequences of a 0.6°C perturbation. Could it reflect a large change in the westerlies, via Ekman drift?

SARACHIK: Perhaps. What most concerns me is the sensitivity these sorts of model seem to exhibit. I'm wondering whether that's the result of something in the idealized configuration, the re-entrant current, or what.

CANE: Eli Tziperman and colleagues have done some work recently that relates to the time constants you put into the restoring terms for temperature and salinity and whether you go to a flux condition. In some sense, I think one of the characteristics of most of these ocean-model studies with time-constant forcing is that the atmosphere is perhaps not active enough. There are various ways of calculating the sensitivity of the heat flux to a change in surface temperature; except for the truly coupled GCMs, I think that the models tend to hold the atmosphere more fixed than is appropriate for its heat capacity, which makes the time shorter. If that were loosened up, and the time lengthened, we'd get somewhat different answers for all these cases—the points at which you go from stable to unstable behavior and the like.