Konrad Gajewski of the University of Ottawa wrote a letter to the Hill Times saying:
Further, in a report published this year in The Journal of Geophysical Research, we showed that the general trend was correct, using a completely different methodology and data-source and that the current warming is unusual with respect to the past 10,000 years …We have better ways to waste taxpayer money than to dredge up this dead issue.
However, this claim appears nowhere in his article, which includes an illustration showing otherwise.
Gajewski’s study (Viau et al, JGR 2006) applies pollen diagrams in a large sampling of sites in North America, seeking the closest modern analogue. They describe one of their checks as follows:
Next, we evaluated if any particular region was overly influential by computing regional means. Since climate change at these scales can be of different magnitude and direction, we evaluated if the millennial-scale variations were present in most or all regions of North America. All regions demonstrated millennial-scale climate variability to some degree, although the magnitude and direction of change in some cases differed from region to region (Figure 5). In spite of the small number of sites in some regions, millennial-scale variability is visible in all of the regional temperature time series.
Figure 5 of Viau et al shows the 6 regional components of the Gajewski reconstruction. The above text seems a rather laconic description of the information contained in Figure 5, shown below. One of the 6 regional reconstructions (SW – where the bristlecones live) follows a completely opposite course to the rest of North America. The right hand side of this graphic shows the closing stages of the last glacial period. While the rest of the continent got rapidly warmer as North America emerged from the Ice Age, Gajewski shows that the southwest U.S. reached its maximum cold spell at exactly the same time that the U.S. Midwest reached its maximum warmth. Is the above paragraph a good description of the Figure below?
Original Caption:Figure 5. Regional mean July temperature time series reconstructed from pollen diagrams. NQ, northern QueⳢec (181 sites); CANPR, Canadian prairies (141 sites); MIDW, American Midwest (121 sites); NW: north west (84 sites); SE: eastern United States (162 sites); and SW, southwestern United States (36 sites). Values were rescaled to range between 0 and 1 for visual purposes using (x-min)/(max-min),where x is temperature value (C), min is minimum value of series, and max is maximum value of series.
This result is very puzzling. Now although the first part of the caption says “Regional mean July temperature time series reconstructed from pollen diagrams”, do not be deceived by this. This seemingly straightforward description of the figure is undone later in the figure by the phrase:
Values were rescaled to range between 0 and 1 for visual purposes using (x-min)/(max-min),where x is temperature value (C), min is minimum value of series, and max is maximum value of series.
This is an unusual form of standardization that I’ve seen previously in Crowley and Lowery. Unlike Crowley, Gajewski does not appear to have lost his data and archived here the pre-standardization temperature estimates – something that meets the description of the above caption. Here is a re-plot using the temperature estimates centering each series on its average (an anomaly method if you will.) This shows the puzzling cooling in the SW USA as North America emerged from the Ice Age, with the southwestern USA being warmer in the Last Glacial Maximum than at present, an idea that, to my knowledge, is held by no other paleoclimatologist and, if true, deserves a lot of separate commentary in this article.
Now here’s something even stranger – here’s a plot of the temperatures: as noted by a reader below. The temperatures in the US Southwest are lower than in northern Quebec. On the other hand, their Figure 4 showing a regional map doesn’t show odd results in SW USA, so it’s hard to say what’s going on, but there’s something wrong with the data shown here.
Next here is an excerpt of their reconstruction from their Figure 7. Does this figure show that present warming is “unusual with respect to the last 10,000 years”. As I interpret the graphic, it shows modern pollen-estimated temperatures as lower than many intervals in the past, including the MWP at about 1000 BP.
They go on to say:
Our reconstructed continental-scale temperature variations on century to millennial-scales are remarkably comparable to those estimated for northern hemispheric temperature variations of the last 2000 years using annually resolved proxy-climate records [Mann and Jones, 2003; Esper et al., 2002; Moberg et al., 2005] (Figures 6 and 8). The Little Ice Age (LIA) and Medieval Warm Period (MWP) are the most recent global manifestations of millennial-scale temperature variations, which are observed at the continental scale of North America.
Now I am getting very tired of paleoclimate authors saying that two time series are “remarkably similar” to one another without providing any statistical analysis to support their claim. Here’s a plot of two versions of the Gajewski reconstruction (slightly inconsistent versions are archived – I didn’t notice an explanation of the inconsistency). I think that one could also say that the shape of this curve is “remarkably similar” to the IPCC 1990 diagram, showing a MWP warmer than the present and a colder LIA (but with an attenuated scale here.)
Replot of two Gajewski versions. Black – version in Figure 2; red – version in Figure 7. Blue – difference between versions.
Gajewski created time series for many different pollen cores, which are dated by radiocarbon methods. WDCP has a large archive of pollen data which appears to include most if not all of the sites studied by Gajewski. But this pollen data is different from the site time series and the site time series as calculated by Gajewski are unarchived. While modern sediments are all precisely synchronized, past sediments have considerable dating uncertainty. The average sample in Gajewski’s data is also said to cover about 200 years. Craig Loehle has observed that chronologies from series with dating uncertainty are necessarily dampened since maxima and minima can get out of phase. With uncertain dating, it is impossible to distinguish regional variation from dating error.
A. E. Viau, K. Gajewski, M. C. Sawada, and P. Fines. 2006. Millennial-scale temperature variations in North America during the Holocene. Journal of Geophysical Research Vol. 111, D09102, doi:10.1029/2005JD006031, May 2006. http://www.ncdc.noaa.gov/paleo/pubs/viau2006/viau2006.html