global climate change

Links to Scaffetta & West papers cited in Benestad and Schmidt 09.

14 Comments

It appears I have been fallen into the vortex of BS09/SW05-09. I can’t say I struggled much.

This climate-blog-kerfuffle involves some real scientific questions, some pesky mathematical details as applied in the context of a specific analysis, questions about precisely what is shown in some specific peer reviewed paper and the level of uncertainty associated with what is shown, some bungling during the peer review process, some blind-spots in the climate science community and what amounts to a pissing contest. As everyone knows, I think all of these matter in some sense. (I say this because some people like to drop into comments and say “Well, the real point is…”. Well.. uhm… there is more than one “real” point. )

That said, going forward, I’m going to try to focus on the following:

  1. Generally: Do we know the level of uncertainty associated with the estimate of the effect solar forcing on temperature claimed in the results of in SW05 and SW06(a)&(b)? Is their method biased high?
  2. What I think BS09 managed to demonstrate and what they claimed to demonstrate. Initially, I plan to focus on results in BS09 that have nothing to do with ModelE.
  3. Various ways one might estimate the level of uncertainty when the SW05- SW06 type methods are applied to data “in the wild”. (That is, to messy data with features ‘similar’ to climate data.)
  4. Discussing other ways one might estimate the uncertainty in S&W’s method. (Some have been done; some have not.)

Obviously, the level of uncertainty in associated with the application of an analytical method to real data affects anyone’s judgment about what has been shown and how certain we are about what has been shown either in one individual paper or by a collection of papers.

I’m sure people will want to discuss the pissing contest in comments; I enjoy those as much as anyone.
I’m open to discussing other issues in comments– or over time. What I will volunteer here in the post is that, in my opinion,

  1. Many of Nicola Scaffetta’s criticisms of BS09 itself are valid. (These are posted at Roger Pielke Sr’s blog; I commented a bit here.) Scaffetta’s valid criticisms include identification of specific mistakes, as well as gavin and rasmus’s responsibility to communicate to clarify that they were, in fact, testing the method they claimed to be testing,
  2. Gavin and Rasmus’s paper (i.e. BS09) nevertheless suggests to me the uncertainties in the results reported in Scaffetta and Wests papers are likely larger than reported in SW06(a) and SW06(b). Moreover, BS09 contains a demonstration that does not depend on Model E.
  3. I’m assuming the editor at JGR did not pick Scaffetta as a reviewer; I think this was a blunder. A. Big. One. The editor deserved 20 lashes with a wet noodle.

    Why? I can’t know for sure, I’m pretty sure that, if chosen as a reviewer, Nicola would have found the mistakes he reported at Roger Pielke Sr’s blog. He would have provided a blistering review. Gavin and Rasmus would have read the review, fixed the mistakes and, probably still have demonstrated that the method in SW06 has large uncertainty intervals. They would have revised, resubmitted and a better paper would have been published. This would have spared B&S some embarrassment: that’s one of the purposes of peer review!

    And finally,

  4. I think BS09 and all the S&W papers were publication worthy based on what appears to be the standards of climate science. However, in the case of BS09, I should modify that statement to say that: the mistakes Scaffetta complains about should have been caught and remedied. Those mistakes are rudimentary and unrelated to the general tendency of papers in climate science to fly by peer review despite the fact they likely contain unrealistic uncertainty intervals.

I’m not going to be discussing these in detail in this post. The purpose of this post is to act as a place holder for links to relevant papers, which I am in the process of reading and marking up:

  1. The paper that launched the kerfuffle: Benestad and Schmidt 2009 Solar trends and global warming J. Geophys. Res., doi:10.1029/2008JD011639. The bits Scaffetta’s critique at Roger Sr.’s blog discuss appear to relate to the final half of paragraph 40, paragraph 50, paragraph 51 and those that follow immediately. Naturally, my first few blog posts are going to discuss the importance of the final two sentences in paragraph 44 and paragraph 63. 😉
  2. Scafetta, N., and B. J. West (2005), Estimated solar contribution to the global surface warming using the ACRIM TSI satellite composite, Geophys. Res. Lett., 32, L18713, doi:10.1029/2005GL023849. I’ve read this one.
  3. Scafetta, N., and B. J. West (2006b), Phenomenological solar signature in 400 years of reconstructed Northern Hemisphere temperature record, Geophys. Res. Lett., 33, L17718, doi:10.1029/2006GL027142. Much of BS09 focuses on this paper and the following one. I haven’t read them yet.
  4. Scafetta, N., and B. J. West (2006a) Phenomenological solar contribution to the 1900 – 2000 global surface warming, Geophys. Res. Lett., 33, L05708, doi:10.1029/2005GL025539.
  5. Scafetta, N., and B. J. West (2008) Is climate sensitive to solar variability? BS09 cites this. I haven’t read it.
  6. Scaffetta tells me there is another paper about to fly off the presses. Links to additional papers by Scaffetta on his home page; I assume a link will appear when that paper is available.

14 thoughts on “Links to Scaffetta & West papers cited in Benestad and Schmidt 09.”

  1. Pissing contests are usually decided by who has the biggest bladder. The trouble is, in a real pissing contest it’s fairly obvious who wins; in this case, I very much doubt either side will claim anything other than victory.

  2. Douggerel– Also, to some extent the pissing contest doesn’t matter. But some points made by either party in the pissing contest aspect of the issue do reveal some things about how science actually progresses rather than how some people might like to say science progresses. As details come out, we will also all learn snippets about how the peer review process really works, how it can go astray, how what amount to political alliances can affect decisions etc.

    Still… the issue of the level of uncertainty in Scaffetta’s analytical method (or those of anyones) ought to be assessed in several different ways. Then we can make a fair decision.

  3. I also think it worth mention that Leif Svalgaard believes that even with correct methods these papers do not provide interesting insights because they use out of date TSI reconstructions.

  4. Raven–
    I understand that. But I’m looking at the flip side: If, there was no dispute as to the TSI construction, what is the uncertainty associated with this analytical method? And how could we assess that.

    I don’t know a perfect way to assess it, but BS09 did at least try, and what they did report should be sufficient to indicate that the uncertainty intervals in SW09 are too small. This is not to say that SW09’s results are necessarily biased upward– but even that may be so.

    In my opinion, SW09’s results are still valuable, but it is important to know whether their uncertainty intervals are correct when the method is used to data found “in the wild”.

  5. Lucia,

    ” the issue of the level of uncertainty in Scaffetta’s analytical method (or those of anyones) ought to be assessed in several different ways. Then we can make a fair decision.”

    Of course you are right. But what rankles is that assessment of the level of uncertainty of the ‘anyones’ has been fiercely resisted by the climate science community since at least 1998.

  6. Dave–
    Agreed. The Scaffetta papers uncertainty estimates appear to fall well inside a historic tradition of unrealistically small uncertainty estimates. These unrealistically small estimates for uncertainty intervals appear to exist papers spanning in full spectrum of published results. They are absolutely, positively not unique to the skeptic papers.

  8. A quick layman’s question here. If the uncertainty estimates approach, equal or exceed the claimed signal or forcing, is this what is mean’t when one says that the results also confirm the “null hypothesis”?

  9. hswiseman:

    “is this what is mean’t when one says that the results also confirm the “null hypothesis”?

    We shouldn’t say it that way. What we should say is we can’t reject the null hypothesis based on that data set.

    So, for example, if we decide the “null hypothesis” is that the earth’s temperature various are utterly uncorrelated the suns, and data are very noisy, we might not be able to reject (i.e. disconfirm) that hypothesis. However, that doesn’t mean we have to accept it. One could plausibly argue that they believe the sun has an effect but that particular data set is too noisy or too limited to detect the effect. (Or, the analysis method might be suboptimal. If so, you should see if you can find an analysis method which you can show has higher statistical power given identical amounts of data and use that.)

  11. David– That’s specific to remote sensing. That is to say: The real, honest to goodness experimentalists taking data think about it. People who deploy instruments nearly always think about whether or not their measurements are wretched. Generally, they’ve had courses in that, and agonize over it.

    But those doing high-fallutin’ analyses afterwards sometimes don’t really think about it. They do something plausible, do some cursory uncertainty analyses and call that good enough. That is, until some answer they don’t believe gets proven….

    The thing about actually deploying instruments is that you are so frequently confronted with problems on first deployment that you have to think about uncertainty whether you are include to or not.

  12. The National Science Foundation just closed a solicitation on:
    Decision Making Under Uncertainty (Climate Change) Collaborative Groups (DMUU) NSF 09-544

    In addition to generating societally useful information and tools, DMUU collaborative groups should enhance basic understanding within and across the social and behavioral sciences as well as related science and engineering disciplines.

    That appears to cover the full gamut through to decision making.

    The formal references on Uncertainty are maintained by NIST Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results

    Even identifying the Type B systematic errors is often overlooked, let alone “quantifying” (“guestimating”) them.

  13. Lucia, I’ve posted up some scripts to access solar data and splice a la Scafetta-West and a wavelet decomposition function that is set at the relevant parameters to emulate the various procedures with reflection padding.

    It looks like Scafetta-West used data ending in 1999. Rahm-smoothing this data set with a 1999 end would probably lead to an interesting example.

Comments are closed.