NORTH: The warming pattern John mentioned, with the leading in the middle of the continental areas and the lagging in the ocean, seems to be a very robust feature. It seems to hold for transient runs for all models, and even toy models have it. I think it's something that needs to be put into the filter when you're trying to do the signal processing I mentioned earlier. You get an increase on the order of 30 percent in the signal-to-noise ratio.
This same land-leading-ocean feature will appear when you have a large-scale forcing. But when it's natural variability at very low frequencies, the ocean tends to lead—for example, with Mt. Pinatubo. It would be interesting to look at the phase relations during the 1940s, when the cooling was so dramatic in the Northern Hemisphere, to see whether the sea ice was leading.
DELWORTH: In the GFDL model for natural variability, for time scales less than 50 years, the continental-region temperatures actually correlate most with the hemispheric means. It's complex, but I think you have to go to longer time scales to get the deep ocean involved.
SHUKLA: I wonder about the apparent paradox of a higher snow amount's being succeeded by warming and reduction in area in the following season.
WALSH: If the snow is disappearing earlier because of the greater springtime warmth, the perturbation of soil moisture due to the extra snow may just be swamped. The additional solar radiation available may be a larger effect than the soil moisture. Also, we don't know that the time scales of the soil perturbations in the model agree with the real world's.
ROBINSON: You might also have different effects deep within the pack and near the periphery.
RASMUSSON: One of the contributors to the recent warming over Eurasia was the exceptionally warm winter of 1982-1983. We were mystified by it, since it was not supposed to be associated with the 1982-1983 El Niño event, but Alan Robock's work on volcanic eruptions suggests that it could have been a response to El Chichon. Also, Dave, I wondered whether you had been able to clear up the snow-cover data problems you were having.
ROBINSON: Yes; I found that two different land masks had been used in the 1970s and 1980s, both of which were wrong. Using GIS techniques, I've come up with a land mask and a procedure that yield consistent and reasonably accurate numbers.
RIND: One caveat: The model run you looked at did not have one additional forcing, the reduction in ozone in the lower stratosphere. That is most effective at polar regions, and would also induce a cooling trend at the surface.
BRYAN: It seems to me that the resemblance to a greenhouse effect that appeared in the slide you showed for the 30-year average may be at least partly an artifact of the averaging—a combination of a very cold event at the beginning of the period and warming over land masses toward the end. I think it could be a little misleading to compare it with the slow warming observed in the greenhouse model experiment.
WALSH: I agree with your comment on the land, but the changes were actually negative over the high-latitude oceans during that period, so I don't think the start can have been unrepresentative.
NORTH: The Hanson-Lebedeff paper of 1988 or so, which was for a longer period leading up to the 1940s, showed the same kind of pattern.
TRENBERTH: I feel obliged to mention that the Southern Hemisphere has snow cover and sea ice too. I know Claire Parkinson has done some work there; can you comment, John?
WALSH: There's essentially no snow there outside the Antarctic continent. The one thing we know from the 20-year satellite record of the South Orkneys is that those data are not representative of the larger-scale changes.
ROBINSON: Ken Dewey and Randy Cerveny are doing a little work with snow over Patagonia and other areas.
JONES: Returning to an earlier point, I wanted to say that the temperature variations at the two Antarctic inland stations and Vostok over the past 35 years are similar to what we've seen around the coast. Since Greenland is a much smaller ice sheet than Antarctica, we may be able to infer that the temperature changes in its interior would also be similar to those measured around the coast.
KUSHNIR: Just a correction to an earlier suggestion: The coldest temperatures in the Atlantic were in the 1970s, not the 1960s, so John may indeed be picking up a trend.
GROISMAN: I'd like to note that it's important that John found the changes in sea ice in summer. Wintertime ice can go back to its old limits, but the new ice is most easily melted off in the spring.