In a recent post, Keith Kloor worried about what would happen when a “bad paper spotlighted on a popular climate blog” is not debunked. In fact, even when a “bad paper spotlighted on a popular climate blog” is debunked, that doesn’t prevent subsequent reliance in policy discussions, as we see in EPA citation of the Mann et al 2008 no-dendro reconstruction. This reconstruction, as CA readers know, used the contaminated Tiljander sediments upside down. EPA described this as a “more sophisticated” methodology than averaging of upside-up proxies as done in Loehle and McCulloch.
Various petitions for re-consideration cited Climategate emails as evidence against the Hockey Stick reconstructions used by IPCC:
Petitioners also claim that the Medieval Warming Period may have been warmer than present temperatures, undermining the conclusion that greenhouse gases are a primary cause of current warming. Issues involving the Medieval Warming Period were addressed during the public comment period (see Response 2–62 of the RTC document) Petitioners raise this issue again because of their assertion that the CRU e-mails indicate that the current temperature record may be faulty, which to them gives the Medieval Warming Period new significance.
Response 2-62 in the RTC document made extensive citations from the 2006 NAS Panel and the 2007 IPCC reports. The EPA continued:
In making their case, petitioners cite a temperature reconstruction without tree rings, notably a study that could have been submitted during the public comment period. [24 - Loehle and McCulloch] However, that paper uses an improper methodology, a straight average of 18 proxies, apparently without weighting them to account for geographic distribution or the strength of the data to detect temperature changes. In contrast, another study using a more sophisticated methodology [25 - Mann et al 2088] found that recent Northern Hemispheric warmth was anomalous regardless of whether tree ring data were included.
EPA’s observation that Loehle and McCulloch “could have been submitted during the public comment period” is a procedural point relating to the petitions for reconsideration. As I understand it, a petition for reconsideration requires evidence unavailable in the public comment period. The Climategate dossier was released after the closing of the comment period and was new material for EPA purposes, while Loehle and McCulloch, as EPA observes, could have been cited by commenters in the public comment period.
Even within the standards of the IPCC community, EPA had no basis for stating that Loehle and McCulloch used an “improper methodology”.
EPA criticized Loehle and McCulloch for not “weighting [proxies] to account for geographic distribution or the strength of the data to detect temperature changes”. However, they had no basis for criticizing Loehle and McCulloch on this point for a couple of reasons.
First, prominent reconstructions relied upon by IPCC did not weight proxies to account for geographic distribution, despite claims to the contrary. The most familiar example for CA readers is Mann et al 1998-99. As CA readers know, MBH methodology (particularly using Mannian PCs, but even using centered PCs) did not weight proxies to account for geographic distribution, but, instead, overweighted strip bark bristlecone and foxtail chronologies in southwest US. Mann et al 2008 RegEM methodology also does not weight proxies to account for geographic distribution.
In fact, the even weighting of Loehle and McCulloch prevents the very over-weighting observed in Mann et al. Nor does weighting procedure of Loehle and McCulloch warrant EPA disapproval relative to most reconstructions in the IPCC spaghetti graph, as they also use more or less similar weighting systems: e.g. Jones et al 1998, Hegerl et al 2007, Moberg et al 2005 and Esper et al 2002. (Actually Esper et al 2002 used two nearby strip bark chronologies as separate proxies.) Further some of the IPCC reconstructions using relatively even weighting for the proxies shown had used the biased Mannian PC1 which overweighted strip bark bristlecones e.g. Hegerl et al 2007, Mann and Jones 2003. So there is no valid reason for EPA to disapprove of Loehle and McCulloch on this count. Quite the opposite.
Their endorsement of the Mann et al 2008 no-dendro reconstruction is equally questionable. This has been discussed in numerous CA posts. It was observed early on that they used the Tiljander sediments upside-down from the orientation of the original authors and that the modern portion of these sediments was contaminated, preventing calibration by correlation to instrumental records.
The “more sophisticated” method endorsed by EPA was to ignore these problems. Regardless of whether the original reconstruction was an oversight, the problem was reported in early 2009 at PNAS and denied by the authors. Much later, the failure of the no-dendro reconstruction to calibrate before 1500 was conceded by Mann et al in the supplementary information to a different article, without a retraction or corrigendum being issued at PNAS or disclosed at the “grey” Penn State website for Mann et al 2008.
Indeed, to my knowledge, there was no public awareness that the SI to Mann et al 2009 withdrew the main claims of Mann et al 2008 until early August 2010, when it was mentioned by Gavin Schmidt inline at realclimate, immediately provoking commentary at CA and elsewhere. This was only a few days before the EPA published its response to the petitions on August 13, 2010.
Had Mann and his coauthors properly retracted the untrue claims in Mann et al 2008 about their no-dendro reconstruction (as they should have), EPA would not have relied on these untrue claims in their Response to Reconsideration to say:
another study using a more sophisticated methodology [25 - Mann et al 2088] found that recent Northern Hemispheric warmth was anomalous regardless of whether tree ring data were included.