RIND: Well, for solar variability we could look at sunspot cycle length and variations in the sun's radius. During the last solar cycle people actually did try to measure this, but they couldn't relate it to the observed solar radiance variations. As for volcanos, we could claim that the acidity in ice cores indicates a change in eruption frequency. The forcing is not well constrained in this area, so it's very difficult to confirm or discount an apparent correlation.
MCWILLIAMS: In your categorization, you apologized for declaring the ocean variability to be external. How about declaring the greenhouse gases to be external variability too? Aren't they both really part of the internal variability?
RIND: Yes, in the sense that you can change nothing in the model and still get those fluctuations in the global surface air temperature. That's internal to the model rather than to the system, though.
SHUKLA: In a paper several years ago Lorenz showed that with his smallest GCM he could reproduce 100-year records simply as a result of the internal dynamics of the system. On what basis can we distinguish between his model and our complex GCMs, and what criteria can we use to decide how much to believe the results?
RIND: With something like the NSF/NOAA ARRCC program, which has both a modeling and an observational component, it's hard to tell whether the forcing is wrong or the model is poor when the results don't agree with the observations. A complicated 3-D model can at least show geographic distributions you can relate to what happened.
LEHMAN: One thing that seems to be diagnostic of your ocean cooling experiment is the appearance of warming over the Eurasian Arctic. I think that might be in part an artifact of the limitations of the 18-ka climate data set, which you take a fifth of an increment of to apply to your model. There is no resolution of SST variability between the Norwegian Sea and the Eurasian Arctic in the data set. Looking at deglaciation and the Holocene, we find that if the import of ocean deep water into the Norwegian Sea is cut back by just a small amount the Barents Sea fills it rapidly. The downstream effects of cooling over the northeast Atlantic might be offset by local cooling of the sea surface in the Barents Sea. Now I might be tempted to use this warming as diagnostic, but I wanted to introduce this reservation about the climate data set.
RIND: Have you any data set that will tell you what happened to the Norwegian Sea during the Little Ice Age?
LEHMAN: Yes, we see evidence for a reduction in the inflow of warm Atlantic water. Our SST proxy records are not very sensitive, though, so we can't put a number of degrees on the amount of cooling that occurred.
RIND: Would you relate this to actual changes in North Atlantic Deep-Water production at that time?
LEHMAN: We can find records with resolution sufficient to capture these sorts of changes in the trajectory of surface currents, but when we look for similar records of the changes in abyssal circulation we run into the problem of ocean-sediment deposition rates—4 cm for 1000 years. We have not yet found a suitable location for making such measurements.
LINDZEN: One comment on Jim Hansen's volcano simulation: Remember that was superimposed on his nominal greenhouse warming. I think our results are actually pretty close. Also, a propose of the forcing, the interannual variabilities turn out to be so huge that I wonder whether we shouldn't ask how the system remains as stable as it does, rather than looking for the very small forcings we call external.
RIND: Indeed, if the interannual variability and net forcing are as large in the models, we should try to diagnose why it remains stable.