intermediate-water formation, which is very difficult, as Aagaard and Carmack noted. But if you truly want realism, you need to put in intermediate-water formation, the Greenland-Scotland Ridge, and the lack of convection to the bottom of the Labrador Sea.
SWIFT: As Bob mentioned, what affects the overflows may be more intermediate-water formation than deep-water. I'm not at all sure that there ever was deep convection in the Iceland Sea, at least in the present-day climate. My impression is that the GSA caused a freshening of the waters there involved in intermediate-depth convection, which meant that the Denmark Strait overflow freshened. Meanwhile, the fresh pulse rounded the tip of Greenland and mixed into the central Labrador Sea Water. The Iceland-Scotland Overflow Water probably took longer to respond to the fresh pulse, because a good deal of warm, salty Atlantic Water is resident at the sills and mixes with the outflow. The deep freshening signal is the sum of these components, each with a slightly different time of freshening.
MYSAK: Bob, is it true that convection was suppressed in the region of the Icelandic Sea, but not in the Greenland Sea, by the GSA?
DICKSON: Yes, it is. There is a tendency to call that area a gyre, but actually they are horses of a different feather. The Greenland, Icelandic, and Norwegian Seas are quite distinct, almost isolated. When the GSA caused salinity to drop below 34.7 in the area north of Iceland, suppressing deep convection, the peak convection in the Greenland Sea was still going like a train. I think Aagaard and Swift are right in saying that the Arctic intermediate water from the Icelandic Sea rarely moves over the convective center of the Greenland Sea.
ROOTH: I'd like to switch horses here, and ask Andrew a question. It's been proposed that for salt accumulation in subtropical gyres there is a critical temperature-salinity vertical-gradient ratio that allows 'salt fingers' (double diffusion) to preferentially transport salt down to intermediate- or deep-water levels. Do you know whether you are exceeding those salt-flux situations? And do you know whether anyone has used that as a constraint?
WEAVER: I must admit I've never checked that myself; I don't know whether anyone's looked into it.
CANE: I'd like to raise the issue of coarse resolution. It seems to me that it forces you to use—at least with the current numerical procedures—very high diffusivity characteristics, which will affect circulation stability. It's hard to say what that would mean when you can't get to a parameter range that resembles the real ocean.
Coarse resolution will generally suppress eddies, though Kirk's heat-transport results suggest that the resulting noise reduction might be a good thing. But then eddies are tied to mean circulation in ways that noise is not.
What worries me the most, though, is the necessity for simplifying the topography. Things move so slowly in the ocean when they don't have some slopes to rub up against.
The other thing I'd like to comment on is the question of coupling to the atmosphere. It seems to me that we might make one small step forward by replacing the ocean with a swamp-mixed layer and putting an atmospheric boundary layer on top of the ocean. It's difficult to calculate changes in the wind that way, or to do anything very convincing about precipitation changes, but you can get variations in heat flux and evaporation that way.
WEAVER: A Ph.D. student at the University of Victoria is taking a sector model, putting a simple thermodynamic ice model and a simple energy-balance model on top of it, and specifying some ad hoc hydrological cycle within it, as a first step toward the coupling you'd like to see. I agree with you entirely about the eddies and topography, but most of us just haven't the computing resources needed.
MCWILLIAMS: Let me add to the list of model-credibility questions the plausibility of the flushing modes. I think it might be worth looking at using some non-hydrostatic and, if you will, more defensible parameterizations of small-scale vertical fluxes.