Marcott et al 2013 has received lots of publicity, mainly because of its supposed vindication of the Stick. A number of commenters have observed that they are unable to figure out how Marcott got the Stick portion of his graph from his data set. Add me to that group.
The uptick occurs in the final plot-point of his graphic (1940) and is a singleton. I wrote to Marcott asking him for further details of how he actually obtained the uptick, noting that the enormous 1920-to-1940 uptick is not characteristic of the underlying data. Marcott’s response was unhelpful: instead of explaining how he got the result, Marcott stated that they had “clearly” stated that the 1890-on portion of their reconstruction was “not robust”. I agree that the 20th century portion of their reconstruction is “not robust”, but do not feel that merely describing the recent portion as “not robust” does full justice to the issues. Nor does it provide an explanation.
The uptick problems are even more pronounced in the zonal temperature reconstructions (NH and SH extratropics on which I will be posting) and in the reconstructions from individual proxies (alkenone, Mg/Ca). In today’s post, I’ll illustrate the uptick problem from the alkenone temperature stack on the premise that there are fewer moving parts in the alkenone reconstruction and therefore this will facilitate diagnosis of the uptick mystery.
The first figure below shows the Marcott alkenone reconstruction. Alkenones are the largest single class of Marcott proxies (31 of 73). According to the alkenone reconstruction, temperatures in 1920 were at their lowest value in 11,300 years (Marcott’s selection of 11,300 years begs the question of what happened just before the left frame of the graphic – an interesting issue that I’ll return to on another occasion.)
Between 1920 and 1940, Marcott’s alkenone reconstruction increased by a remarkable 1.4 deg C, returning to values of the start of the reconstruction as shown. Marcott’s 1940 value (0.39 deg C reference 1961-90) was considerably above the 1961-90 reference zero and higher than the 1990-2010 average, shown as red “+” for reference. Interesting as this figure may be, I don’t wish to dwell at present on what it might or might not show about temperature history (a topic that I’ll return to on another day.)
Today’s post is merely about the mysterious uptick from 1920 to 1940, an uptick that is not observed in any of the actual alkenone data.
Despite the recent popularity of alkenone series over the Holocene, there are vanishingly few high-resolution alkenone series over the last two millenia for comparison. I keep my eye out for such series and their absence is very frustrating. Only one of Marcott’s 31 alkenone series (MD01-2421; KR02-06 St.A GC; KR02-06 St.A MC’s 31 series) has a published date after 1950. Two others have published end dates of “0 BP”. This conventional date can be problematic, but again this is a story for another day. There is negligible change in any of three series: overall, there is an average decrease of one-hundredth of a degree, with none of the individual series changing as much as a tenth. In alkenone series with published values in both 1920 and 1940, nothing happens to yield Marcott’s 1.4 deg C uptick.
Marcott (in my opinion, unwisely) re-dated the modern portion of many cores, including numerous alkenone series. (I will write a separate post on this small and unwise adventure.) I don’t believe that this re-dating is at the root of today’s mystery (but neither can I completely exclude it.) As re-dated by Marcott, there are now five cores with alkenone measurements in both 1920 and 1940, again with negligible change between periods. On average, there is a decline of 0.002 deg C. (Interestingly, only one core is in common between the two cases.)
In small samples with inconsistent proxies, the dropping out of a highly negative or highly positive proxy can produce seeming changes to a mean. To consider this possibility, here are the values for proxies (after mid-Holocene centering) with “Marcott” values in either 1920 or 1940. Marcott redated the MD01-2421 series (which is strongly negative in the 20th century); after redating, it drops out between 1920 and 1940, thereby producing an increase in the mean of available proxies. Whereas the average of the five proxies with values in both periods has no change, the average of available values for the seven proxies shown below increases by 0.36 deg C – a (spurious) increase that is still far short of Marcott’s 1.4 deg C.
I don’t believe that the baffling uptick has anything to do with splicing of instrumental or Mannian series, as speculated elsewhere. (One of the reasons for working on the alkenone version was to minimize this moving part.) However, I am at present totally baffled. I’m disappointed that Marcott refused to explain the phenomenon when asked politely. It is clear that the uptick is, at a minimum, “not robust”. My own surmise is that the situation is worse than that: that the uptick in the alkenone (and other reconstructions) is an artifact, but of what I don’t know.
As alluded to in this post, many other things about this article concern me. I’ll try to cover them in future posts.