I mentioned yesterday in the press release thread:
So here’s the question, the press release below mentions sediments. Place your bets now on whether the Tiljander sediment series remains inverted or not.
Peer review doesn’t seem to catch the problem of using inverted data. That’s a good question for science and the peer reviewers. I suggest those who have contact put the question to them, because the results will look different when the data is used properly. In case anyone doubts this. The inversion was confirmed by the principal researcher that gathered the data, Tiljander, who confirmed this in an email to Steve McIntyre. – Anthony
Yet another Upside Down Mann out
Science published today yet-another-Mann-et-al-reconstruction:
Michael E. Mann, Zhihua Zhang, Scott Rutherford, Raymond S. Bradley, Malcolm K. Hughes, Drew Shindell, Caspar Ammann, Greg Faluvegi, and Fenbiao N: Global Signatures and Dynamical Origins of the Little Ice Age and Medieval Climate Anomaly, Science 326 (5957), 1256. [DOI: 10.1126/science.1177303].
Seems to me that Mann has re-discovered the Medieval Warm Period.
I had a quick look at the paper, SI, and the code. What seems to be done this time is that the proxy network of Mann et al (2008) is processed with a slightly modified screening of Mann et al (2008), and then the reconstruction is done with a slightly modified RegEM CFR of Mann et al (2007)! Now to answer the question that seems to be on everyone’s lips: yes, Tiljander series are still used as inverted. This can be seen from the positive screening correlation values reported in the file 1209proxynames.xls. In fact, going quickly through the screening code, it seemed to me that they have really “moved on” from the screening employed in Mann et al (2008): only “two-sided test” is used!
%% below is for selecting full/screened/1856-1925 screened/1926-1995 screened proxy-network
%% replacing "abs(z(4,i))>=0.165"/"abs(z(5,i))>=0.513" in line 75/84 with the followings for your expected proxy-network
%% abs(z(4,i))>=0 / abs(z(5,i))>=0 (full proxy-network)
%% abs(z(4,i))>=0.162 / abs(z(5,i))>=0.496 (screening over 1850-1995)
%% abs(z(6,i))>=0.195 / abs(z(7,i))>=0.602 (screening over 1896-1995)
This means that if a proxy has a strong inverted correlation to the (two-pick?) local temperature, it gets picked – no matter what the physical interpretation is! Since RegEM doesn’t care about the sign, it is now really so that the sign does not matter to them anymore. Anything goes!