Pielke Jr on Spinning Science

Roger Pielke Jr has an interesting post an objectionable press release – recall the controversy over the climateprediction.net press release last year. Some time, I’ll discuss rules on press releases that apply to stock promoters – there are things that mining promoters are not allowed to say in press releases.

As someone with experience with press releases by speculative mineral exploration companies, I never cease to be amazed at how promotional climate scientists and their institutions have become.


  1. John A
    Posted Feb 28, 2007 at 1:25 PM | Permalink

    When they tell you its not about money, then its all about money.

  2. Pat Frank
    Posted Feb 28, 2007 at 2:27 PM | Permalink

    It’s not just climate science. I had an email debate with some scientists from the Weizmann Institute, in Israel, when their press release claimed they’d made a breakthrough that would assist conversion to a solar-powered hydrogen economy. What they had in fact done, no doubt an admirable advance, was figure out how to use solar energy to drive the reaction of elemental carbon with zinc oxide, to produce metallic zinc. The zinc could be used in batteries, but there was plenty of front-loaded CO2 production in the solar process. The press release was very misleading, and didn’t mention the chemistry at all; I had to infer it. When I pointed out the misdirection in an email, their reaction was a bit abusive. We had several exchanges, and they continued to weasel right to the end.

    I think the main problem is that everyone wants to have cosmic results. Ordinary progress is no longer respected.

  3. Ken Fritsch
    Posted Feb 28, 2007 at 5:50 PM | Permalink

    Two replies:


    Don’t you think it is the scientist him/hersefl who should determine if the paper was misprepresented? You are aware of emails about this issue.

    Dr. Knappenberger,

    Cheap spots are easy! I commented on an early version of the statistics used in hs paper. Not everyone has a sinister motivation although you might want to believe it to be the case if they disagree with you.

    I don’t know you biut I think you can do better than this.


    Posted by: Peter Webster at February 28, 2007 01:24 PM

    One can only hope that that post was by an imposter. Otherwise I would have to question what the heck PW is adding to the discussion other than misdirection.

  4. Gerald Machnee
    Posted Feb 28, 2007 at 7:04 PM | Permalink

    I find the following statement by Kossin a bit puzzling. It seems to go against what he is trying to say.
    **Still, the very real and dangerous increases in recent Atlantic hurricane activity will no doubt continue to provide a heightened sense of purpose to research addressing how hurricane behavior might change in our changing climate, and further efforts toward improvement of archival data quality are expected to continue in parallel with efforts to better reconcile the physical processes involved.**

  5. Steve McIntyre
    Posted Mar 1, 2007 at 12:28 AM | Permalink

    #3. That is almost certainly not an imposter.

  6. John A
    Posted Mar 1, 2007 at 2:10 AM | Permalink

    Re #4

    That’s the boilerplate standard weasel request for more funding. Translated from weaselese it reads:

    Although we found no evidence of anthropogenic influence in hurricane strength or numbers, nevertheless we need more funding to do some further torturing of the dataset just in case there’s a signal in there. Remember Katrina.

    Negative results like this are just as important scientifically as positive results. Nevertheless you don’t get further funding for purely negative results.

  7. Louis Hissink
    Posted Mar 1, 2007 at 3:08 AM | Permalink

    One of the more difficult problems in public announcements made by enthousiastic mining entrepreneurs to stock exchanges is that many of their statements are made on the basis of honestly held beliefs, no matter how unscientific they might be. They should be penalised for this? No.

    Belief if one hell of a hard human characteristic to counter. If some mining entrepreneur honestly believes he has the mother of all lodes on his property, and says so, then so be it. In this case it’s a case of caveat emptor. If you don’t accept his conclusion, then don’t invest in his prospectus.

    Academics are as human as mining entrepreneurs, and those whose science is dominated by the deductive method will make statements on what they earnestly believe to be right, forgetting that the basis of their deductive reasoning is intrisincally flawed, (what if the original assumption is wrong).

    Scientific empiricists, on the other hand, tend not to make blue-sky announcements.

    The problem is that taxpayer funded academics have a monopoly which we as buyers of “theory” have no alternative.

    Herein is the problem and the solution.

  8. trevor
    Posted Mar 1, 2007 at 5:10 AM | Permalink

    Re #7: Louis, Louis, Louis.

    Being part of the Australian environment, as I think you are (correct me if I am wrong), you know that an Australian mining entrepreneur simply cannot just say whatever it is that he believes. Any statements made by his company have to be signed of by a “Competetent Person” under the JORC Code. By definition, the Competetent Person must be a member of the professional organisation The Australian Institute of Mining and Metallurgy, and be able to demonstrate 5 years experience in the relevant field. The Competent Person is answerable to the Ethics Committee of the AusIMM, and there are numerous examples of people being held to account for breaching the rules. It is no trivial matter for the Competent Person to sign off on a public statement. He/she is required, in effect, to warrant that the statement is, to use Canadian terminology “True, Plain and Fair”.

    The Competent Person puts his/her reputation on the line by signing off on a public statement. He/she is saying that the statement is professionally acceptable.

    It seems to me that that is what is wrong with this whole climate science thing. There seems to be precious few checks and balances, or professionalism. “Peer Review” from a clique of mates doesn’t seem to cut it. It is even arguable that some of the statements made are deliberately false and misleading, and therefore open to legal action under the equivalent of the Australian Trade Practices Act.

  9. David Smith
    Posted Mar 1, 2007 at 6:05 AM | Permalink

    I see Kossin as a rather new PhD (2000), at a radical-left university, in a highly-polarized field, making a finding which is quite unwelcome by the ideologues of climate science. I’m not surprised that he put soft statements into his work and allowed the spinmeisters to spin.

    Sad, yes, but I think it’s about survival.

  10. Ken Fritsch
    Posted Mar 1, 2007 at 10:49 AM | Permalink

    I’m not surprised that he put soft statements into his work and allowed the spinmeisters to spin.

    Significant numbers of climate scientists, can evidently no longer simply publish cold hard facts (with uncertainties) about climate change, but need to point to the connection to AGW where applicable and on those occasions when the results seem to weaken at least part of the case for AGW, the authors appear obligated to make the deferential statements that their results should not be construed as evidence against any part of AGW.

    I think it is a compromise between being part of the consensus, for whatever that is worth, and being a disinterested scientist. If one simply ignores all the hype and adheres strictly to the facts of the publications (and from an ever vigilant skeptical point of view), I think one can obtain a better understanding of any conclusions that can be derived. Note the interest that PW evidently shows in having one look beyond those published facts.

  11. Steve Sadlov
    Posted Mar 1, 2007 at 10:52 AM | Permalink

    RE: #8 – To wit, please see the new post at Climate Science (Pielke Sr’s site) from yesterday, the first in series making the case for licensing of those producing software for climate modeling. It’s actually an entertaining piece, especially if you like toy sailboats (and if you ever did the toy catapult “Design of Experiments 101” excercise, you’ll probably appreciate it even more).

  12. Bill F
    Posted Mar 1, 2007 at 11:07 AM | Permalink

    The phrase that comes to mind is “don’t tell me about it, show it to me”. What I mean is that any person can write something that says anything they believe. However, the reader is the one who has to decide if what is written is believable. Some readers will accept the use of the peer review process as evidence of believability, others will take the qualifications of the writer into account in determining what to believe, and others will want to see the raw data and replicate the experiment for themselves before believing it. The problem with the way science is reported these days is that the careful words chosen by the author and agreed to by peer reviewers to describe their findings are often tossed out as too complicated, carefule, and boring when it comes to writing the press release announcing the publication. So somebody with no professional reputation to uphold, no stake in the review process and no direct ties to the work that was performed decides for themselves what the “take home message” of the research is and puts that out in the press release to the media. The media edits the press release to fit their column space and away we go! Meanwhile, the author is revelling in their fame and is understandably reluctant to go out and correct the misrepresentation for fear of appearing to backtrack on their published work. So the press release writer and the media end up framing the public’s view of the “take home message”, while the scientists are left trying to sort the wheat from the chaff to get to the real bottom line of the research. When the author or authors decide to limit access to their data or methodology, it just makes it that much harder for the more skeptical among us to replicate what they did so that we can decide if we agree with them.

    The only real solution in my mind is to revamp the way we report scientific findings such that the press release is included with the publication in the peer review process. That way, the author and the peer reviewers are forced to take ownership of not just the science that is published, but how it is spun when presented to the media. It isn’t a perfect solution, but at least it would force the authors to acknowledge how their research is being presented to the general public.

  13. Paul Linsay
    Posted Mar 1, 2007 at 12:44 PM | Permalink

    #12. The real solution is to get rid of the press releases, the science should stand on its own. The only reason for a press release is publicity to garner money and awards.

  14. Peter D. Tillman
    Posted Mar 1, 2007 at 1:55 PM | Permalink

    Re #13

    That’s not realistic — a lot of competent scientists can’t write standard English. That’s why both Science and Nature run a plain-English summary of the papers they judge most significant in each issue.

    The trick is to have the facts in the two versions straight. Which isn’t really that hard to do.

    Cheers — Pete Tillman

    “Fewer scientific problems are so often discussed yet so rarely decided
    by proofs, as whether climatic relations have changed over time.”
    — Joachim von Schouw, 1826.

  15. jae
    Posted Mar 1, 2007 at 3:10 PM | Permalink

    There’s a summary over at CO2science that supports Kossin’s conclusions.

  16. Steve McIntyre
    Posted Mar 1, 2007 at 4:02 PM | Permalink

    Here’s a presentation from the Ontario Securities Commission on disclosure standards for mineral exploration projects. http://www.osc.gov.on.ca/FastAnswers/fa_20050304_43-101_good-practices.pdf Reporters are obliged to state average or range of values ‘€” not just highest value in a drilling program. Compare that to Myles Allen (climateprediction.net) promotion of the highest value form their simulations. Rules apply to websites and press releases, not just prospectuses.

    In comparison, university and institution promotion of climate science through press releases are a Wild West show.

  17. Paul Linsay
    Posted Mar 1, 2007 at 4:14 PM | Permalink


    a lot of competent scientists can’t write standard English


    I repeat, the press releases are meant to garner money and awards, and tenure, which I forgot. Science by press release is a new phenomenon and a rather pernicious one at that. Press releases are a political tool, they have nothing to do with science. The people in your field who know and understand the science don’t need them to read your paper and figure out its import and validity.

  18. Roger Dueck
    Posted Mar 1, 2007 at 4:54 PM | Permalink

    #7 Louis
    See #8 and remember Busang (Bre-X).
    I’m not in mining, I generate prospects for oil and gas exploration (I must be the BIG OIL Dr. Su-fruity is talking about!). It’s fine to say caveat emptor but the entire point of the argument is that unqualified people are then asked to decide on the validity of a concept. The over-statement of a concept will only result in the perpetrator being discredited because you will eventually need to come through with the goods. Most times the biased expectation is optimistic, which is why people do this work in the first place. You may have heard of Enron and the fallout Bill 51-101 regarding disclosure and third-party assessment of assets. Geology is unforgiving. The public, on the other hand, is much more forgiving as the IPCC keeps admitting that it is lowering the size of the prize, we see more evidence of global nothing and yet the fear factor increases and more politicians jump on the bandwagon. I agree with trevor#8 in that the problem with the climate debate is the unwillingness of parties to subject their opinion to third-party disinterested assessment.

  19. Steve McIntyre
    Posted Mar 1, 2007 at 5:45 PM | Permalink

    I’ve discussed Bre-X in the early stages of the blog:

    Bre-X: Is de Guzman Alive?
    Some Thoughts on Disclosure and Due Diligence in Climate Science
    Bre-X #3: Core and Code
    Bre-X #2: the Skeptical Geologist
    Bre-X #1

  20. Gerald Machnee
    Posted Mar 1, 2007 at 7:01 PM | Permalink

    Here is one on the press release from the University of Miami on Dr. Soden et al 2005 on their paper “The Radiative Signature of Upper Tropospheric Moistening”. Check:
    It starts by:
    VIRGINIA KEY, FL (Oct. 6, 2005) ‘€” A new study published in this week’s issue of Science confirms evidence of global warming using satellite measurements of water vapor ‘€” a well-known greenhouse gas ‘€” in the troposphere.

    Using satellite measurements of water vapor from 1982 to 2004, Dr. Brian Soden from the University of Miami Rosenstiel School of Marine & Atmospheric Science, and colleagues were able to confirm the climate model simulations that have long indicated that moisture is increasing in the upper troposphere and this water vapor build up is exacerbating global warming.
    Notice that it “confirms evidence”.

  21. David Smith
    Posted Mar 1, 2007 at 7:55 PM | Permalink

    Re #20 Another press release is here . I’m wondering if the Hoyos et al results publicized in this release will be refigured, now that the uptrend in severe hurricanes looks more like a flatline.

    Hoyos showed three trendless factors (shear, deformation, humidity) and one uptrending factor (SST), and then linked the uptrending SST to uptrending hurricane severity. But, if storm severity is trendless, then it seems like that SST/storm connection is weakened.

    Perhaps the original reports on increases in severe storms are actually correlated with this instead of with rising SST.

    I suspect that the “out” is that Hoyos covers 30 years (including 10 of pre-modern satellite coverage) while Kossin covers but 20 years.

  22. Posted Mar 1, 2007 at 9:32 PM | Permalink

    Submitted to RealClimate on this post:

    Klotzbach 2006 uses the same dataset as Webster et al. 2005 (note typo in your post, not 2006) and finds no trends in numbers of Cat 4-5 or ACE other than in the Atlantic.

    So shouldn’t Kossin et al. actually reinforce Klotzbach 2006 and suggest that any global increase in storm intensity (or numbers of intense storms) necessarily occurred before 1986 (or 1983)?

    Whether that increase is due to data or climate is a separate issue, but isn’t there multiple lines of evidence now that there are no global trends since the early- to mid- 1980s, whereas in the Atlantic there has been?

    Klotzbach paper:

    Click to access klotzbach2006.pdf

  23. TAC
    Posted Mar 1, 2007 at 10:03 PM | Permalink

    Roger (#22),

    there are no global trends since the early- to mid- 1980s, whereas in the Atlantic there has been

    I would not concede the point too quickly. While the number of Atlantic hurricanes has increased in recent decades, it is not obvious that the increase is statistically significant. To determine significance one would have to employ a realistic null hypothesis — one that accounts for the time-series structure (e.g. regimes) of the natural process. I’m not aware of anyone who has done this.

  24. Posted Mar 1, 2007 at 10:33 PM | Permalink

    TAC- Thanks … but trends in the data are trends. They may result from starting at a low point in a cycle and ending at a high point, or they may result from some long-term secular change. But the fact that the Atlantic is more active 1995-present versus 1970-1994 is about as lock-solid a conclusion as you’ll get in science. Over longer time periods (pre-1970) this issue is certainly much more debatable. Thanks!

  25. Bob Koss
    Posted Mar 2, 2007 at 1:22 AM | Permalink

    For those that are interested, I just posted some world-wide data in the Elsner et. al 2006 thread. Includes a link to a Gif image of storms from 1983-2003 world-wide.

    I can’t seem to be able to get the link to another local post to work right. What’s the technique?

  26. Hans Erren
    Posted Mar 2, 2007 at 2:01 AM | Permalink

    re 18:
    The habit of taking high end values for reserve estimates resulted in a significant share price drop recently when values were adjusted to realistic values. In the end it doesn’t pay to exaggerate.

  27. John A
    Posted Mar 2, 2007 at 2:01 AM | Permalink

    Roger Pielke:

    Thanks … but trends in the data are trends. They may result from starting at a low point in a cycle and ending at a high point, or they may result from some long-term secular change. But the fact that the Atlantic is more active 1995-present versus 1970-1994 is about as lock-solid a conclusion as you’ll get in science.

    Trends in the data are trends? No they’re not. It maybe a statistical version of pareidolia – seeing a trend in a random process that has no trend.

    See Paul Linsay’s Poisson Fit post. The trend you refer to as “about as lock-solid a conclusion as you’ll get in science” turns out to be a spurious trend in a random dataset. The annual distribution of hurricanes in the Atlantic is random – there is no forcing and no trend.

    Perhaps you’d like to trump this by claiming a “scientific consensus”?

  28. Louis Hissink
    Posted Mar 2, 2007 at 2:31 AM | Permalink



    I was recently appointed to the AIG Complaints committee (area – diamond exploration) so I agree with you, but you seem to have unnecessarily emphasised certain aspects about “competent persons”.

    While the requirement to have a “competent” person sign off any ASX announcement, recent company announcements exaggerated the “on/in ground” assets. In those cases AIG wrapped a few knuckles and to date all is silent on the Western Front. Australia is fortunate that government has not officially entered into this area of human endeavour.

    However, it has in climate science.

    So my point, that caveat emptor, remains correct and the market, including the role of AIG, can police the reporting of specious mining properties.

    But as you correctly note, that is absent in climate science.

  29. Mark T.
    Posted Mar 2, 2007 at 2:33 AM | Permalink

    I definitely have to agree with John A on this one. Perhaps if it were worded as “spurious” trends, as John suggests, then yes.


  30. Louis Hissink
    Posted Mar 2, 2007 at 2:48 AM | Permalink

    Just as a quick point – mineral exploration is the best developed practice of the empirical method of science.

    Sciences which can’t immediately test their hypotheses, astronomy, geology, archaeology and economics, rely of the deductive method of science, hoping that the initial assumption is correct.

    That can’t happen in the empirically driven sciences but it does in the deductive sciences where artful debate is the principal method of convincing doubters.

    But how does the mug punter deal with the pronouncements of the deductive sciences? If they are correctly called sciences in the first place?

    History has shown that serious opposition to any deductively reasoned position is usually hazardous to one’s physical health. No one has died from questioning empirical derived facts.

  31. Posted Mar 2, 2007 at 3:53 AM | Permalink

    # 23 –

    I think the problem is the statistical definition of trend (statistically significant trend implies that we need statistics..)

    We have indexed family of random variables. We observe part of those, and apply a function f(Y_1,Y_2,Y_3..)=\beta . This function (statistic) outputs a scalar that somehow represents a trend. Usually it is just a function that performs least squares fit. To state ‘statistically significant trend’, we need something to compare. Significant wrt what? We need H0, ‘normal situation’. H0 is used to define a distribution for \beta . If observed realization of \beta is very unlikely under H0, we can start talking about significant trend.

    For example, Jones 90, in Table 1 finds that RAUS 1930-1988 linear trend is significant trend at 5 % level. Under his H0 P(\mid \beta \mid \geq 0.56) \leq 0.05 This implies that H0 can be used to find the distribution of \beta . Now, what is that H0? X are i.i.d Gaussian? (If I got it right, this is TACs point in #23, we have not seen H0)

    Of course, trend can be defined without statistics, but then one shouldn’t use term ‘statistically significant trend’.

  32. Louis Hissink
    Posted Mar 2, 2007 at 4:32 AM | Permalink


    I suspect that statistics come into play when a mathematical relationship isn’t obvious.

    And trends are essentially linear relationships between 2 variables, but if the coefficient of variation is low, say less than 50% of the data are explained by the trend, then that trend is spurious, or “not real”.

    Debate about trends and other mathematical or statistical results usually means that the theory and data have unlinked themselves from the physical objects.

    Take the present discussion about grid cell temps elsewhere here. Theoretically very interesting but physically and in relation to the real world? Nope. Average Grid cell temps mean nothing physically. The argument is no different to arguing over how many angels could be fitted onto a sewing needle pin head.

  33. TAC
    Posted Mar 2, 2007 at 5:47 AM | Permalink

    #24 Roger, I think I understand your point — “a trend is a trend” — but I want to be clear about mine. The word “trend” often is used to imply a nonstationarity in the stochastic process from which observed number of hurricane arrivals are a realization. Using such a definition, it is not at all clear that we are seeing anything other than natural variability in the number of hurricane arrivals.

    However, if you define trend (as I think you do) as an increase or decrease in the number of arrivals itself, then I agree with you. However, this definition is uninformative — really useless; essentially all interesting systems exhibit this kind of “trend.”

  34. Posted Mar 2, 2007 at 8:12 AM | Permalink

    TAC- (#33) Thanks, your formulation is far more precise, and I agree. The debate over whether hurricane statistics show signs of nonstationarity is (to me) far more interesting and relevant than the absence or presence of a trend.

    Part of the confusion arises from the fact that hurricane data from 1980 (say) to present might be interpreted as a sample from a larger population, or it can be treated as the entire population. In the case of the former one would wonder about what happened before 1980 and what will happen next. In the case of the latter all that matters is the data in hand.

    I think that a fair reading of Kossin et al. suggests that he uses the notion of “trend” in the latter sense. However, discussion of his paper, including in the UW/NSF press release, tends to conflate the two notions of trend.

    For this reason I think that discussion of “trends” (i.e., IPCC’s notion of “detection”) is apt to mislead if one is not precise in what is being discussed.


  35. Douglas Hoyt
    Posted Mar 2, 2007 at 9:04 AM | Permalink

    People also need to be reminded that there is no long-term trend in land falling tropical storms in the US, except perhaps a slight downward trend over the last century. I suspect this measure of tropical storm numbers is the most homogeneous time series available.

  36. Paul Linsay
    Posted Mar 2, 2007 at 9:15 AM | Permalink

    #30, Louis

    History has shown that serious opposition to any deductively reasoned position is usually hazardous to one’s physical health. No one has died from questioning empirical derived facts.

    Yes, science, following Galileo’s methods works much better than following Aristotle’s.

    #34, Roger,

    The debate over whether hurricane statistics show signs of nonstationarity is (to me) far more interesting and relevant than the absence or presence of a trend.

    As I showed on a couple of other threads here, hurricanes are generated by a stochastic process. I neglected to follow up with a discussion of the variations in the counts seen in the full NATL record back to 1851. They can easily be explained as a random walk. There is no need to invoke nonstationarity. The problem is that the behavior of random walks is highly unintuitive. To quote Feller, An Introduction to Probability Theory and Its Applications, 3rd edition, Wiley, 1968, p. 67,

    We shall encounter theoretical conclusions which not only are unexpected but actually come as a shock to intuition and common sense. They will reveal that commonly accepted notions concerning chance fluctuation are without foundation and that the implications of the law of large numbers are widely misconstrued.

    I’m in the process of writing up the hurricane notes. Contact me and I’ll send you a pdf when it’s done.

  37. Posted Mar 2, 2007 at 9:17 AM | Permalink


    And trends are essentially linear relationships between 2 variables, but if the coefficient of variation is low, say less than 50% of the data are explained by the trend, then that trend is spurious, or “not real”.

    ..and now we need to define spurious trend 😉 One example (a bit subjective): large realization of \beta , even though E(\beta)=0 (*). For example, highly autocorrelated stationary processes may look very ‘trendy’.

    (*) If the process Y is stationary, E(\beta) must be zero. (assuming f is a linear function that returns zero for constant input)

  38. Posted Mar 2, 2007 at 10:02 AM | Permalink

    Paul- Thanks, I look forward to seeing your analysis.

  39. Mark T.
    Posted Mar 2, 2007 at 10:50 AM | Permalink

    It seems you (Roger) fully understand the implications that John A, et. al. mention. John Brignell, of NumberWatch fame, has an interesting description of the problems (he uses a sine wave as his example) associated with throwing around the term “trend.”


  40. Gerald Machnee
    Posted Mar 2, 2007 at 11:17 AM | Permalink

    Re #21 Dave – Yes that press release sums up the spin line well. As long as you start in the 1970’s you can show hurricanes doing what you want to show as you leave out half a cycle. If a study would have been done in the 1970’s starting in the 1940’s you would have hurricanes decreasing as in the coming ice age at that time.

  41. Louis Hissink
    Posted Mar 3, 2007 at 2:00 AM | Permalink



    You seem to have misundersood me – The deductive position was the staus quo in Gallileo’s time – he questioned it using his empiricism from observation. When the deductive method takes precedence over the empirical, then such a science is actually a pseudoreligion.

    So science did indeed suffer but not from questioning an empiriclly derived fact.


  42. Paul Linsay
    Posted Mar 3, 2007 at 9:56 AM | Permalink

    #41, Louis,

    We agree 100%. I was pointing out a different aspect of the same thing. With AGW we have now reached the point where the heretics are being excommunicated, c.f., the CNN meterologist wanting to decertify others that don’t toe the line. When do they start burning the witches?

  43. DeWitt Payne
    Posted Mar 3, 2007 at 3:59 PM | Permalink

    #36 Paul,

    For the hurricane data up to 1995, I agree that there is insufficient reason to invoke non-stationarity. After 1995, however, I’m less certain. I’ve become somewhat more educated on poisson cusum charts and it still looks like S+ goes out of control and stays out starting in 2000.

  44. Paul Linsay
    Posted Mar 3, 2007 at 4:28 PM | Permalink

    #43, DeWitt,

    What are Poisson cusum and S+ and S-?

  45. Louis Hissink
    Posted Mar 4, 2007 at 5:02 AM | Permalink



    Phew – had me worried for a second. The CNN weather person is a qualified meteorologist? I get the impression TV weather-man types are more sceptical than we, so is that witch-burner a journalist rather than a qualified climate scientist?

  46. Louis Hissink
    Posted Mar 4, 2007 at 5:46 AM | Permalink



    What does your graph/plot show? What is a Poisson Cosum? All a bit fishy I think.

  47. bernie
    Posted Mar 4, 2007 at 11:08 AM | Permalink

    Speaking of trends can anyone comment on the parallels between the following:
    Figure 3 in Kossin: See http://www.ssec.wisc.edu/~kossin/articles/Kossin_2006GL028836.pdf

  48. Ken Fritsch
    Posted Mar 4, 2007 at 3:26 PM | Permalink

    Re: #46

    CuSum methods were originally developed for use in enhancement of QC monitoring of industrial processes in the 1950s and 1960s. The technique is well founded in basic statistics and in the case of a Poisson distribution would indicate when a trend indicates that the mean has changed. The link below gives a good explanation of its history, statistical grounding and use.

    I would be interested in hearing how sensitive the CuSum method and limits used by DeWitt in his CuSum chart are, i.e. how much does the mean have to change and over what time period to go over the S-/S+ limits. His CuSum charting is in line with some of the tests that RichardT did in calculating the sensitivity of a Chi square test in picking up a change in the mean of hurricane counts with the difference that CuSum picks up a trend in one direction (something along the lines I was looking) while I believe RichardT was combining Poisson distributions with lower and higher means that continued to give the same overall mean.

    Since hurricanes are a real life process and while their distribution can well be estimated over a relatively long period by a given mean — as in the case of the time period 1945 to 2006 — I have no doubts that that mean varies by small amounts over that time period. I would be curious as to how much of a change can be detected statistically, be that by CuSum charting or some other statistical technique.

    Link is here .

  49. Roger Bell
    Posted Mar 4, 2007 at 3:36 PM | Permalink

    Why do you think that hypotheses in Astronomy can’t be immediately tested?
    Roger Bell

  50. DeWitt Payne
    Posted Mar 6, 2007 at 1:37 AM | Permalink

    To produce the graph in #43 I used the Excel spreadsheet sspois.xls found here from Hawkins and Olwell of the University of Minnesota and based on their book “Cumulative Sum Charts and Charting for Quality Improvement” with lambda =5.5, lambda+ = 9 and lambda- = 2. That gives a K+ of 7.11 and a K- of 3.46. The control limits were calculated using the anygeth.exe program also found at the above link with an ARL of about 400 and the same lambda’s (poisson distribution, zero start and no winsorizing) . That gave the upper control limit h+ = 8.8 and the lower control limit h- = -5. I’m not a statistician and haven’t read the book so I may not be using the spreadsheet correctly. However, I think an ARL of 400 means about the same as +/- 3 sigma control limits. I tried a number of combinations of lambdas with pretty much the same result. I also tried using poisson distributed random numbers generated in Excel with about the same lambda and those stayed well inside the control limits.

  51. DeWitt Payne
    Posted Mar 6, 2007 at 1:58 AM | Permalink

    I tried to post details of how the chart in #43 was produced, but it was put in spam limbo.

  52. Lawrence Hickey
    Posted Mar 6, 2007 at 8:34 AM | Permalink

    We are in trouble if we don’t counter this p.r. when Gore says, offhandedly, that “we can measure the temperature too” as well as the CO2 concentration, and nobody answers. – when he ignores the vast uncertainty in the magnitude of the MWP, etc. We are in a lot of trouble from the AGW spinners like Gore. This sort of presentation, when unanswered in a public and effective forum, as opposed to a valuable technical one like this, will empower carbon credit elves and other mischief from people that want to regulate the world in the name of an eco-religion that will not tolerate any sort of apostasy. . Reasonable questions about the magnitude of recent warming and the cause of it, and a comparison with the past will be crushed by political power of a clique that has embraced AGW doomsday as an axiom. Can we really measure past temperature Al, to correlate it with the co2 record which we can measure? By slight of hand, Gore slides over that issue. Is CO2 a more powerful greenhouse gas than water vapor, or methane? What role to clouds play? And a really important issue- If the MWP really was warmer than today, and it was global, what does that say about your chart Al? What is the evidence for that? What of a review of the hockey stick and the sordid cherry picking brislecone cone pines over polar Ural disgrace? We are in a lot of trouble if we can’t somehow present this picture fairly before this theory ossifies into doctrine with so much capital, dollar and intellectual, behind it, we can’t stop it from being turned into policy.

  53. Steve Sadlov
    Posted Mar 6, 2007 at 8:56 AM | Permalink

    RE: #48 – What if, as of 1851, the “test strength” was lower, and more “failures” escaped, whereas, more recently (say, since 1980) the “test strength” increased and more “failures” are now being caught?

  54. DeWitt Payne
    Posted Mar 6, 2007 at 8:46 PM | Permalink

    Here’s a cusum plot of just 1944 to 2006 hurricane numbers. In this case lambda is 6, lambda+ is 8 and lambda- is 4. It’s still over the upper confidence limit in 2005 and 2006 and will stay there if there are 7 or more hurricanes in 2007.

  55. Ken Fritsch
    Posted Mar 7, 2007 at 11:07 AM | Permalink

    Re: #53

    RE: #48 – What if, as of 1851, the “test strength” was lower, and more “failures” escaped, whereas, more recently (say, since 1980) the “test strength” increased and more “failures” are now being caught?

    That is essentially what I have been proposing with the added variable of multi-decadal cyclical effects. The steady landfall hurricane counts with little or no trend also point in this direction. I would think that the message here would be to remove all the artifacts of man-made measurements before drawing any hard conclusion about how nature is actually operating here.

    Re: #50

    DeWitt, thanks for the information. I need to look into CuSum in more detail to determine actually how sensitive it is to a given change in mean. I know it ranks up there with MLE and MA statistical analysis so I would agree that it is signally a change in mean — I just do not know by how much. You can look at five year moving averages to get a feel for these changes also. It would be of interest to me if you did the same exercise using the period from 1945 to 2006 as I can show that by dividing the 1851 to 2006 period into 2 time periods that the fit for a Poisson distributions improves significantly and particularly from 1945 to 2006 period. The moving average graph shows that there might be a 50 to 60 year peak to peak cycle operating here. I do recall Willis E discussing removing a cyclical effect to give an improved fit to a Poisson distribution.

  56. Ken Fritsch
    Posted Mar 7, 2007 at 11:17 AM | Permalink

    This graph should be easier to read.

  57. DeWitt Payne
    Posted Mar 7, 2007 at 1:54 PM | Permalink

    Re: #55

    I really would appreciate some feedback from someone with a stronger background in statistics than myself. I think I have done the control limits (Decision Intervals) correctly, but I really don’t know. As you can see, I have posted the plot of hurricane numbers from 1944 to 2006. While it looks like a shift in lambda may have occurred in 1995, you really can’t reject the null hypothesis of a constant lambda until 2005.

    Here is the log from anygeth that I used to calculate the Decision Intervals for the 1944 to 2006 plot. It may mean more to you than it does to me. I think that an in control ARL of 474 means you can expect one out-of-control point every 474 data points and the OOC ARL of 13.2 means that a run of 14 points above the mean would be another reason to reject the null hypothesis. The Winsorizing constants used mean no Winsorizing was done, whatever that is.:

    Program to calculate cusum decision intervals
    Copyright 1997, D M Hawkins and D H Olwell
    Which distribution do you want?
    Enter the in-control and out-of-control means
    6.000 8.000
    The exact theoretical reference value is 6.952
    Enter the reference value you want to use :
    What are the Winsorizing constants?
    -999.00 999.00
    Want zero-start (say Z), steady state (say S) or FIR (say F)?

    Enter ARL
    h 10.0000 arls 132.3 119.6 128.2
    h 12.0000 arls 252.6 232.8 247.4
    h 13.0000 arls 346.5 326.7 340.9

    k 7.0000 h 13.0000 ARL 346.51
    h 14.0000 ARL 473.73

    DI 14.000, IC ARL 473.7, OOC ARL 13.2, FIR ARL 8.2, SS ARL 8.5
    Which distribution do you want?
    Enter the in-control and out-of-control means
    6.000 4.000
    The exact theoretical reference value is 4.933
    Enter the reference value you want to use :
    What are the Winsorizing constants?
    -999.00 999.00
    Want zero-start (say Z), steady state (say S) or FIR (say F)?

    Enter ARL
    h 10.0000 arls 243.1 222.1 237.1
    h 12.0000 arls 531.0 497.1 524.2
    h 11.0000 arls 360.1 339.1 353.4

    k 5.0000 h 11.0000 ARL 360.06
    h 12.0000 ARL 531.00

    DI 12.000, IC ARL 531.0, OOC ARL 11.7, FIR ARL 6.8, SS ARL 6.8

  58. bernie
    Posted Mar 7, 2007 at 5:10 PM | Permalink

    I was looking at these numbers too, and don’t you have to control for El Nino effects before looking for patterns.
    I was also thinking if there is an equivalent of El Nino in the South Atlantic? If not, why so much asymmetry?
    Are differences in shorelines and currents sufficient? If an equivalent does exist the
    interaction of the two should be interesting.

  59. Ken Fritsch
    Posted Mar 11, 2007 at 5:35 PM | Permalink

    Re: #57

    The Cusum plots, while showing a change in mean hurricane counts are not, in my mind, a particularly good analytical tool knowing what we do about the occurence and measurement of hurricanes. We suspect a trend due to counting efficiency (which could be confounded with a natural increase ‘€” but that is contradicted by the steady count of landfall hurricanes) and there is evidence of a reoccurring increase in hurricanes on a multi-decadal time scale.

    From Paul Linsay’s Poisson Fit thread, Willis E. detrended and removed a sine cycle from the 1851 to 2006 hurricane data and compared it to a Poisson distribution here:


    The Chi square goodness of fit test to a Poisson distribution for the untreated data yielded a p = 0.09, while the data worked by Willis E. to detrend and with cycle removal gave a Chi square goodness of fit to a Poisson distribution of p = 0.87. I would like to see more details on how Willis E. processed these data, but in my view it is line with changing count efficiencies and cyclical nature of hurricane occurrences.

  60. Willis Eschenbach
    Posted Mar 12, 2007 at 12:01 AM | Permalink

    Ken, you say:

    I would like to see more details on how Willis E. processed these data, but in my view it is line with changing count efficiencies and cyclical nature of hurricane occurrences.

    I first detrended the data in the normal way, removing the least squares linear trend.

    Then I used an iterative process (the “Solver” function in Excel) to determine the phase, amplitude, and frequency of the sinusoidal wave that removed the most variation from the hurricane data (minimizing the RMS error between the sine wave and the data).

    All the best,


  61. David Smith
    Posted Mar 12, 2007 at 6:10 AM | Permalink

    RE #58 bernie, the South Atlantic waters are too cool (below 26C) and the winds move in different directions at different altitudes (wind shear) which tears apart any seedling.

    On rare occasions a seedling will survive and form a storm.

  62. bernie
    Posted Mar 12, 2007 at 6:24 AM | Permalink

    My question is more about the mechanisms for the generation of El Nino conditions off the west coast of SA. What is it about that mechanism that means that El Nino type conditions are not generated in say the Bight of Benin? Is it to do with the Andes, since there is clearly no equivalent in West Africa. Basically what causes El Ninos?

  63. David Smith
    Posted Mar 12, 2007 at 8:48 AM | Permalink

    Sorry, bernie, I was under-caffinated when I read and then replied!

    There actually are something resembling El Ninos in the Atlantic ( see here ), driven by wind anomalies, but my understanding is that the Atlantic wind anomalies are small in magnitude and duration, and the amount of ocean potential energy involved is quite small compared to the Pacific.

    So, they happen, but they are babies (El Enano?) compared to the Pacific events.

  64. MarkW
    Posted Mar 12, 2007 at 9:24 AM | Permalink

    El ninito

  65. bernie
    Posted Mar 12, 2007 at 9:34 AM | Permalink

    Interesting. You indicate that the total energy involved is relatively small but is it strong enough to make the predictions of El Nino effects more problematic? Could it amplify or suppress El Nino effects? Could it lead to a shorter or longer El Nino?

%d bloggers like this: