with a less comprehensive data set when we go to longer time scales?

Because over two-thirds of the globe is covered by ocean, we also need to ask how confident we are about the SST corrections that have been applied to the data.

Dr. Jones mentioned the ice-core isotope-temperature relationships. He indicated that they explain only about 25 percent of the annual temperature variance in the observed record. The question here is: Is that a high enough proportion for these records to be useful indicators of larger-scale changes? Perhaps it is not if we are looking at annual scales, but we are really interested in century scales here.

Another issue is the sensitivity of the tree rings and ice cores to climate change. For instance, if climate is changing and the moisture source regions that contribute to tree growth or ice depth change, the signal may be contaminated. If they are responding to climate change, can they still adequately reflect what happened several hundred years in the past?

Dr. Jones also showed that there is little coherence among the several 1000-year tree-ring time series that have been assembled. That is very interesting; if it is true, putting together a global-scale multi-century time history of temperature change will be very challenging indeed! Another question about the analyses relates to the fact that the tree rings represent only certain portions of the year. For example, although the analyses all related to the summer, each was done for different months during the summer. Could part of the difference in the time series be due to the fact that they were not all measured in exactly the same seasons?

My last question is whether we can prioritize the work that needs to be done to improve our ability to estimate global temperature change on the century time scale. Should we even attempt to do this? Or should we let science work from the bottom up?

Discussion

SOCCI: Do the isotopes used to measure temperature account for only 25 percent of the variability because a large part of the signal is an ice-volume signal?

JONES: What I've done is to take an isotope record from the coastal portions of Greenland and the Antarctic peninsula and to compare the 18O or deuterium values with year-to-year temperature values at a coastal site. That's where the 25 percent comes from. Of course, you might get more variance on longer time scales.

MYSAK: What's the state of the art in extracting, say, precipitation or runoff from tree-ring data?

JONES: Most of the tree-ring work we've dealt with is from Scandinavia, where the response is clearly to temperature. In southern Europe or the southern part of the United States you can get a good reconstruction of river flow or precipitation from trees, because there the trees are responding to moisture. In between, as in England or northern Germany, the trees are responding to a mix of the two, and it's hard to unravel the climate signal.

GROOTES: Remember that the isotopes are a measure only of the conditions when snow falls, and the season of greatest accumulation may vary considerably. The value of Phil's tree-ring records is that they have been carefully selected for their response to climate to get at the margins of the trees' limits.

BERGMAN: Would it be possible to take information from mountain-valley glaciers and deconvolve some of the problems with ice dynamics to get something useful?

JONES: Well, yes, but you wouldn't get the year-to-year reconstructions I can get from the tree-ring records. And remember that different source locations will give different response times, and changes to the source area can make dramatic differences to ice cores, as the formation of the Weddell Sea polynya did to Antarctic cores in the late 1970s.

LEHMAN: The Greenland isotope data suggests that higher elevations on an ice sheet are less likely to be contaminated by source variation effects. On a different topic, I was disturbed by how short the correlation decay lengths were for your Greenland records. It seems to me that they would have to be much longer for the larger temperature changes that characterize the deglaciation.

JONES: Yes, the ones I showed were calculated on decadal time scales from 90 years of data. But I don't think it has yet been shown, for the longer time scales, that you do have a much larger-scale signal. We also need to assess how representative our sites are of the time and space scales of interest.

KEELING: I was wondering how you took care of the heat-island effect. Also, those coastal Greenland stations you used for comparison are subject to very strong temperature inversions in the winter, and thus have very high variability. They would not be totally representative of the conditions on the top of the ice sheet.

JONES: We had only one station with a heat-island problem, Toronto.

RIND: We should remember that isotopes are an integrator over the path between the moisture source and the ice core, which will extend the length scale of a perturbation by comparison with a station record.

PARKER: You could estimate the representativeness of the



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement