Major doubts have been raised about the reliability of the historical temperature record by two US statisticians in a paper that has yet to be published but has already provoked a strong response from leading US climate scientists.
The paper re-opens the controversy over the so-called hockey stick graph, made famous by Al Gore in “An Inconvenient Truth”, and may re-energise climate change sceptics in the wake of “climategate” and on the eve of the United Nations Cancun climate conference.
The hockey-stick graph, constructed mainly from tree ring “proxy” data by US climate scientist Michael Mann, shows historic temperatures remaining fairly steady and then rising sharply since the industrial revolution. But statisticians Blakeley McShane of Northwestern University in the US and Abraham Wyner of the University of Pennsylvania in the US write in a new paper that “we conclude unequivocally that the evidence for a ”long-handled” hockey stick (where the shaft of the hockey stick extends to the year 1000 AD) is lacking in the data”.
Climate scientists "greatly underestimated the uncertainty"
Temperature records reconstructed from tree rings, ice cores and other so called proxies that give indications of ancient temperatures “do not predict temperature significantly better than random series generated independently of temperature”, McShane and Wyner state in their paper which has been accepted for publication in the peer-reviewed journal Annals of Applied Statistics. The two statisticians conclude that climate scientists “have greatly underestimated the uncertainty of proxy based reconstructions and hence have been overconfident in their models”. They comment that the long flat handle of the hockey stick is best understood to be a feature of the data processing and “less a reflection of our knowledge of the truth”.
“The fundamental problem is that there is a limited amount of proxy data which dates back to 1000 AD; what is available is weakly predictive of global annual temperature,” McShane and Wyner comment in their paper which is called “A Statistical Analysis Of Multiple Temperature Proxies: Are Reconstructions Of Surface Temperatures Over The Last 1000 Years Reliable?".
The paper is due to be published early in 2011 but has already met a robust response from climate scientists including hockey stick graph creator Michael Mann, from Pennsylvania State University in the US, and Gavin Schmidt of NASA's Goddard Institute for Space Studies. Schmidt and Mann have, with Scott Rutherford of Roger Williams University in the US, written a comment paper, accepted for publication by the same journal, in which they contest the methods used by McShane and Wyner. In it Schmidt and colleagues say “the absence of both proper data quality control and appropriate “pseudoproxy” tests” by McShane and Wyner “to assess the performance of their methods invalidate their main conclusions.”
The McShane and Wyner paper is being published as a “discussion paper” and the editor of Annals of Applied Statistics has invited contributions commenting on this work. Reporting Climate Science . Com understands that there are 13 papers “in press” (that is, still to be published) including the piece by Schmidt, Mann and Rutherford, that comment on the original paper.
McShane and Wyner are preparing a response to all the issues raised by these various comment papers. Reporting Climate Science . Com understands that all the papers – the discussion paper, the comments and the response - will be published together in the first 2011 issue of Annals of Applied Statistics. Typically a discussion paper in an academic journal will have three or four other papers commenting on it – the fact that this paper has attracted 13 is a measure of its importance.
McShane and Wyner have told Reporting Climate Science .Com that they did not want to respond to the points raised in the various comment papers on the record until their formal response has been finalised for publication.
A scientific controversy first erupted over the hockey stick in 2003 when Canadian mathematician Stephen McIntyre and Canadian economist Ross McKitrick, from the University of Guelph, questioned the statistical methods used by Michael Mann and colleagues. This was vigorously rebutted by Mann.
Reconstructing past temperatures is difficult
It is a complicated process to reconstruct a record of ancient temperatures. There were no thermometer measurements until around 150 years ago. This means that scientists must look at so called proxies such as tree rings, ice cores, corals, plant remains, or sediment cores from lakes, among many others. These all exhibit features that are affected by the temperature and can help scientists to infer what temperatures were in the past.
Scientists must extract the data from the proxy and then convert it into a record of surface temperatures – this enables them to “backcast” or ”reconstruct” historical temperature records. This involves significant statistical analysis and it is this part of the process that McShane and Wyner have criticised. “Our paper is an effort to apply some modern statistical methods to these problems,” they state. “While our results agree with the climate scientists findings in some respects, our methods of estimating model uncertainty and accuracy are in sharp disagreement.“
Blake McShane is a faculty member in the Marketing Department of the Kellogg School of Management at Northwestern University. He is a statistical methodologist, and has developed models for problems in a variety of fields.
Abraham Wyner is Associate Professor, Department of Statistics at The Wharton School, University of Pennsylvania.
“A Statistical Analysis Of Multiple Temperature Proxies: Are Reconstructions Of Surface Temperatures Over The Last 1000 Years Reliable?” by Blakeley B. McShane and Abraham J. Wyner
"A Comment On “A Statistical Analysis Of Multiple Temperature Proxies: Are Reconstructions Of Surface Temperatures Over The Last 1000 Years Reliable?” By McShane and Wyner" by Gavin A. Schmidt, Michael E. Mann and Scott D. Rutherford
Both papers are to be published in Annals of Applied Statistics - see here