Tue, 27 Jul 2010

It is depressing that the New Scientist follows parts of the blogosphere, and some other sections of the press, in asserting that of the three independent investigations into Climategate "none looked into the quality of the science itself" (Editorial, 17 July 2010, page 3). Our hope was that New Scientist would have a more informed understanding of the method of science research.

Lord Oxburgh writes, in the first paragraph of his report, that his panel was asked to address criticism "that climate data had been dishonestly selected, manipulated and/or presented to arrive at pre-determined conclusions that were not compatible with a fair interpretation of the original data". In the second paragraph he wrote "The Panel was not concerned with the questions of whether the conclusions of the published research were correct…….rather…..whether as far as could be determined the conclusions represented an honest and scientifically justified interpretation of the data" (our italics).

The Oxburgh Panel operated, and wrote their report, entirely independently and so we cannot answer for the precise form of words used, but it does seem entirely consistent with the way science works. New Scientist, when do science conclusions become "correct"? Science conclusions remain provisional, becoming more or less provisional over time, until/unless they are replaced by scientifically likelier conclusions, or unless they reach the elevated status of "fact". In the observational sciences, that process develops through the honestly and scientifically justified interpretation of data.

The compilation of a hemispheric or global land surface data time series from irregularly distributed (in time and space) historical thermometer observations can never be "correct" in an absolute sense. There will always be uncertainty, as there will be greater relative uncertainty in our knowledge of past temperatures from "proxy indicators" such as tree-rings. The discovery, or utilisation, of more or better proxy records might improve our understanding of the Mediaeval Warm Period. Developing analytical techniques may also change our understanding; hence the provisionality of scientific conclusions.

When one understands how science research progresses it is clear that the Oxburgh approach is entirely consistent with – as his report says – an investigation into whether CRU's research conclusions were scientifically justified. Moreover, as many others  have pointed out, there is a similarity with the conclusions on the character of climate change from many other groups around the world using a variety of variables and indicators.

Although the main focus of the Muir Russell Review was not the science, as the report made clear, a reading of the document clearly shows that a number of the questions put to CRU did address science. Examples include "Does not the problem of divergence for the late 20th century record invalidate the deduction of tree ring palaeotemperatures for the period prior to the instrumental record?", and "How has this choice (of data stations worldwide) been tested as appropriate in generating a global or hemispheric mean temperature (both instrumental and proxy data)?" Certainly the answers provided to these, and other, questions were couched in scientific terms.

We would urge readers of New Scientist, and others,  to read the Oxburgh and Muir Russell reports as well asCRU's submissions to Muir Russell to judge for themselves the adequacy or inadequacy of the examinations of CRU science. They may also like to bear in mind that CRU's research has been published in the top peer-reviewed international journals (unlike almost all of the criticism). Although peer review - as Muir Russell pointed out - is not perfect, the New Scientist, more than many other printed media, will understand the significance of this point.

A proper reading of this material will also demonstrate that CRU did not regard its "assembly of 160 years of global thermometer data" as "private property", as New Scientist claims. Muir Russell concluded "On the allegation of withholding temperature data, we find that CRU was not in a position to withhold access to such data or tamper with it". Some data were subject to non-publication agreements but the vast majority were already freely available; ready for anyone (as Muir Russell demonstrated) to reconstruct temperature time series virtually identical to those of CRU (and others). 

The "judgemental decisions" about which Oxburgh made comment, and to the New Scientist refers, were not about thermometer data, as the Editorial implies, but related to tree-ring data. Oxburgh and Muir Russell both emphasised that, within CRU's scientific publications, the uncertainties around tree-ring temperature reconstructions were fully explained.

The stolen emails demanded serious independent  investigation. We instigated two ourselves and, in response to a letter to the Chair of the Parliamentary Science and Technology Committee, our Vice-Chancellor made clear his enthusiasm for cooperating with a Parliamentary Inquiry, should that be the decision. Although we have suggested that New Scientist readers do examine the Oxburgh and Muir Russell reports, and our submissions to Muir Russell, we recognise that many will not and will depend, at least in part, on the  Editorial comment to form their opinion. We fear that the Editorial's  imprecision will help to perpetuate the distortion to debate produced by the controversy of Climategate, much of which was contrived.

Not all was contrived. We agree that open-ness in sharing data, "even with… critics",  is a legal requirement. Largely, we have met that requirement although we have accepted that, in some instances, we should have been more helpful, pro-actively and absolutely. A particular instance to which Muir Russell brought attention was meta-data, not observational data, but still data which were pertinent to an interpretation, by others, of some of CRU's analyses. We are addressing this.

It would have been helpful if the New Scientist had drawn a distinction between the disclosure of data (including meta-data)  and analysis methods, and the disclosure of  email conversations, including those which were clearly regarded as confidential communications by sender and recipients. Others are recognising that there areimportant differences, and there are implications for the way that research is conducted in the UK. It is important that we do have candid conversations about these matters, and we think that the New Scientist is in a good position to help in the understanding of these implications.