08 February 2010

IPCC Mystery Graph: Solved!

The most important thing to come out of last week's debate at the Royal Institution of Great Britain was final resolution on the issues associated with the so-called "mystery graph" (above) that I have been discussing. I'll have more to say on the debate itself, but this post focuses on the "mystery graph" which is a mystery no more. (Audio of the debate can be found here in .mp3 and several first-hand perspectives can be found here.)

Robert Muir-Wood explained that he had created that graph and he did so "informally." (It does not appear in the literature, then or since.) Muir-Wood was asked directly if he thought that the IPCC should have included the graph and he said (quotes taken from the audio),
"Personally, I think that it should not have been there."
At the debate, Bob Ward said that he finds the IPCC actions "difficult to defend" referring specifically to the IPCC making up false claims about my views, and more generally that there is "no excuse" for some of its actions on this subject that I reported on.

So, we now have a contributing author of the IPCC AR4 agreeing with my claim about the inclusion of misleading information, about which which an expert reviewer suggested removing because it "can mislead." Yet, the IPCC says there is nothing to this issue (PDF):
Recent media interest has drawn attention to two so-called errors in the Fourth Assessment Report (AR4) of the IPCC, the first dealing with losses from disasters and the second on the subject of Amazon forests. The leadership of the IPCC has looked into both these instances and concluded that the challenges are without foundations. In neither case, did we find any basis for making changes in the wording of the report. We are convinced that there has been no error on those issues on the part of the IPCC.
I will follow up this post with an email to Professor Christopher Field, who leads Working Group II, to ask for a formal correction. Substantively, this matter is resolved -- the mysteries are solved, unambiguously. Thus, while the facts are perfectly clear, the IPCC has yet to acknowledge them. It is not too late.


  1. It seems you have flogged this dead horse about as far as it will go. Why not lead a shadow panel that will do the work that should have been done without the pseudo-science, misrepresentations, selective indifference to competing science, and fibs?

  2. Great job tracking this down and sticking with it. Of course the elephant in the room is the question of just how many other instances of this sort of thing are there in the report that contributors such as yourself remain unaware of? And who will correct those misrepresentations?

  3. Roger, I requested a digital version of Muir-Wood's data and haven't received any acknowledgment. Three attempts so far, Steve Mc

  4. Roger:

    I am surprised anyone, particularly a Muir-Wood, would let some information (in this case, the mystery graph) "loose" without a "not for publication" notice. In fairness, if the graph was "informally" presented, did IPCC have a duty to inquire to know the difference. The IPCC Report has been out for several years without apparent public comment from Muir-Wood until now?

  5. I'm still curious about Muir-Wood's role as a contributing author. When this particular storm broke (so to speak), he suggested that the IPCC's use of his work was inappropriate. Why, as a contributing author, did he not speak up at the time?

    As for the graph, did he produce that graph at the behest of the "Writing Team"? In which case, what did he expect them to use it for, if not publication in the report?

    Just wondering...

  6. Roger, In the Q&A of the mp3 you say "believe it or not, the world has been decarbonizing for over a century". I guess you meant to add 'per capita' or am I mistaken?

  7. -6-Robin

    That statement refers to CO2/GDP, both at the global level. It is discussed in some depth in Chapter 4 of my forthcoming book ;-)

  8. But if the IPCC chooses not to state that they are in error for posting an informal graph (and it doesn't seem that they are in "error" for doing so) than this case is solved on their end and you have "honest brokered" yourself to the fullest extent possible......to the point of beating a dead horse.

  9. Chris Field - isn't he one of the "deniers?" He certainly denies that anything went wrong. If anything, Field stands with Patchuri as one of the arch-denialists.

  10. Now that we at last know how the graph came about, we may actually look at it.

    We see two trending series. There is an obvious correlation. Do they cointegrate? I would be surprised. This is misleading, so.

  11. Roger,

    I think the inclusion of the Muir-Wood graph is not quite as indefensible as you make out. As I pointed out during the debate at the RI, the paper that was published by Miller et al. (2008) does find some statistical evidence for a link between the rise in global average temperature and a rise in normalized disaster losses. This is what the paper actually states on page 238:

    “We caution that our model does not capture the possibility that there are no underlying factors that are common to years of high (low) temperature variation that would cause us to falsely attribute the trend in disaster losses to climatic reasons. Results suggest that the temperature anomaly is highly significant (at 1%) for normalized losses (r2=0.22) irrespective of the survey period (Table 12.A2 in the Appendix). Results for Australia, the Philippines, and India are again significant (at the same levels as in the first model) with a negative coefficient. The rise is equivalent to an increase in normalized catastrophe losses of US$4.8 billion (post-1950) and US$6.6 billion (post-1970) for each 0.1˚C rise in global temperatures. For details, please refer to table 12.1A in the Appendix. However, these results are highly dependent upon recent US hurricane losses during 2004 and 2005. When the regression is rerun without these losses, the results are no longer statistically significant (Table 12.2).”

    I know you tend to gloss over this part of the paper and instead quote from the ‘Condensed summary’ at the beginning, we provides this assessment: “We find insufficient evidence to claim a statistical relationship between global temperature increase and normalized catastrophe losses.” The authors doe not state how they jumped from the analysis in the body of the report to the summary. The conclusions on page 240 give yet another slightly different assessment than the summary of the results: “In sum, we found limited statistical evidence of an upward trend in normalized losses from 1970 through 2005 and insufficient evidence to claim a firm link (my italics) between global warming and disaster losses.”

    In essence, this seems to depend on whether one should exclude US hurricane losses from 2004 and 2005 – I guess this is a matter of judgment about whether they are considered outliers, although this will not become apparent until we have a much longer time series of data to analyse.

    So it seems to me that the graph that Robert provided for the Supplementary Material for Chapter 1 of working group II just provides an accurate portrayal of the findings and data described in both Muir-Wood et al. (2006) and Miller et al. (2008). The fact that neither publication actually contains this graph is, I think, merely a quibble – there are plenty of other graphs in the IPCC report that have not been reproduced exactly from the literature, but which are still accurate representations. I know that Robert didn’t want to defend the use of his graph even though he willingly provided it, but I think that is more of a sign that he, like many contributors to the IPCC report probably now feel, would rather not become targets for those who have criticisms of the IPCC process. There are many researchers out there now who I believe are intimidated by the current bout of trials by media.

    In summary, Roger, I congratulate you on your persuasive presentation at the RI and you clearly convinced a large part of the audience of your case. I agree with your objections to the misrepresentation of your views during the IPCC review process, and there are many broad points on which I agree with you. However, I remain unconvinced by your main arguments about the Muir-Wood et al. and Miller et al. papers.

  12. Thanks Bob. It was a fun event, with a lively and interesting crowd. I hope that it was enjoyable and educational for those who attended.

    On your substantive points, just two quick responses:

    1. On the increase from 1970 to 2005 ... yes it is there and yes it is a function of the start and end dates, with the 2004 and 2005 hurricane seasons responsible for the upward post-normalization trend.

    This would lead to a further question -- in the jargon of the IPCC, does this trend represent the "detection" of a change in the underlying statistics? Fortunately, we have a longer time series of US hurricane losses available with which to answer this question, and the answer is -- no. The period 1970 to 2005 is well within the historical variability of damage. In fact, from 1900 to 2005 (or 2009) there is no trend in US hurricane losses, which makes sense, given that there are also no long-term trends in hurricane landfall frequencies or intensities.

    So the bottom line is that the trend from 1970 to 2005 is not evidence of a signal of warming. Miller et al. were correct in their judgment.

    2. As far as including the graph, Robert and I agree that, in his words,

    "Personally, I think that it should not have been there."

    And as you know, my view is that the best defense to trial by media, is simply to be accurate and honest.

    I look forward to when we next cross paths. Thanks again.

  13. I'll correct that report for a fraction of the IPCC's budget with nothing more than a bic lighter and a well ventilated area.

  14. "...some statistical evidence for a link between the rise in global average temperature and a rise in normalized disaster losses."

    I would like to know where these figures came from, and if they were adjusted for inflation?