The following HTML text is provided to enhance online
readability. Many aspects of typography translate only awkwardly to HTML.
Please use the page image
as the authoritative form to ensure accuracy.
Natural Climate Variability on Decade-to-Century Time Scales
TRENBERTH: Let me comment a moment on the trend aspect of our analysis of the MSU data. Our paper comparing it with the IPCC data set for surface temperature will be in the September issue of the Journal of Climate, by the way. Over North America, the MSU Channel 2 data has about 0.2°C per decade less trend than the surface record. My best guess is that this results from the way in which all the differences between the satellites getting the TOVS series of data are taken out.
WALLACE: That indeed is a point to be taken seriously, but I don't think we should dismiss the possibility that these differences might prove to be real. Another five years of data should settle the matter.
TRENBERTH: Yes, that's true; there are real physical differences. Over North America the correlation is better than 0.9, whereas over the rest of the world it's not that good.
SHUKLA: I had the impression from various Oort papers that a large part of the vertically integrated three-dimensional temperature for the Northern Hemisphere is the ENSO signal.
WALLACE: Yes, I think his results are consistent with what I have indicated—that the ENSO signal shows up clearly in the globally averaged temperature time series, but not in the time series for extratropical latitudes.
KAROLY: I've actually performed a correlation of SST with upper-air temperatures, stratified by latitude and elevation. It's very strong up to 25° north and south, being dominated by direct tropical SST, but between 25° and the poles you get structures that vary in sign in different latitude bands, and the correlation is negligible.
RASMUSSON: What about the differences between land and sea surface temperatures, though both seem to have been increasing in the last couple of decades?
WALLACE: The low-frequency interdecadal trends exhibit the same signature in land-based and marine data sets, as can be seen from the IPCC report's figures. But I suspect that those two data sets show quite different scatter around the trend lines for individual months or seasons, year by year. I think that if you found a point in the time series for land that was way below the trend line, the corresponding one for the marine series would be way above the trend line.
RASMUSSON: Different parts of the ocean certainly have different temporal responses; the tropical temperatures have ramped up very quickly since the mid-1970s.
LINDZEN: I think you made a crucial point, Mike, about the difference between detection and interpretation. For example, data reflecting crustal movement languished for years before the concept of plate tectonics gave them meaning. Also, when we talk about detection, we should be careful not to equate detection of CO2 increase with detection of climate change. Some of the measures of CO2 change have to do with radiative flux, which varies strongly with the time of day. Also, one thing I would very much like to see done is further work comparing measurements made at surface level with those made at 500 millibars, particularly as regards lapse-rate variability. Some of Oort's recent work suggests that surface temperature changes are amplified at 500 mb globally, while some of the MSU data indicate that the integral changes over a certain region may be reduced.
KEELING: As I remember, Oort found a substantial signal, about 6 months delayed from the tropical signal, at 200 mb, which he used as a proxy for integrated temperature.
KAROLY: There is a very strong 200-mb signal, but if you average around the hemisphere the latitudinally averaged signal in the extratropics is weak.
GHIL: I'd like to make two remarks. The first is about the spatial patterns. It's quite clear that signals such as the so-called ENSO interannual signal have associated spatial patterns. Since this signal is relatively homogeneous throughout the tropics, which is a large part of the globe, it will dominate any global average. Also, spatial patterns of various signs outside the tropics are going to cancel each other out if you do zonal averages.
My second remark relates to Dick Lindzen's point about the volcanic signal and its two-part decay. Mike showed us clusters of colder months that disappeared after a year or two, that being the fast-decay part. I wonder whether interdecadal variability might not be masking the slow-decay part.
RIND: It would be interesting to clarify the nature of the system's long response time. A volcanic eruption is such a quick signal that you might observe a much shorter response than a system with so many feedbacks would otherwise exhibit; the longer responses could be drowned in the rest of the variability.
TRENBERTH: I'd like to pick up on one of Michael Ghil's points. The Pacific is not in fact coherent with the ENSO signal; the western Pacific tends to have a sign opposite to the eastern Pacific's. The tropics are not necessarily a coherent whole either. However, the ENSO signal is so large in the eastern Pacific that it is twice as high as the standard deviation and four times or more the variance.
WALLACE: There's more on this issue in my paper, but for the purposes of this discussion, I think the tropics can be divided into the 'active' eastern Pacific, where SST is directly influenced by equatorial upwelling, and the 'passive' remainder. I think the ENSO signal is also present, with smaller amplitude and lagged by about a season, in the passive region. Furthermore, I believe that stations in that region, which includes nearly all the land stations in the DOE data set, all exhibit very similar ENSO-related time variability.