I think that Preisendorfer would roll over in his grave if he saw how Ammann, Schmidt and Mann were bastardizing his Rule N.
Here’s what Ammann and Schmidt said at realclimate::
For instance, the data can be normalized to have an average of zero over the whole record, or over a selected sub-interval. The variance of the data is associated with departures from the whatever mean was selected. …
MM05 claim that the reconstruction using only the first 2 PCs with their convention is significantly different to MBH98. Since PC 3,4 and 5 (at least) are also significant they are leaving out good data. It is mathematically wrong to retain the same number of PCs if the convention of standardization is changed. In this case, it causes a loss of information that is very easily demonstrated.
This follows pretty similar hyperventilation by Mann here:
Claims by MM to the contrary are based on their failure to apply standard ‘selection rules’ used to determine how many Principal Component (PC) series should be retained in the analysis. Application of the standard selection rule (Preisendorfer’s "Rule N’") used by MBH98, selects 2 PC series using the MBH98 centering convention, but a larger number (5 PC series) using the MM centering convention. Curiously undisclosed by MM in their criticism is the fact that precisely the same “Åhockey stick’ pattern that appears using the MBH98 convention (as PC series #1) also appears using the MM convention, albeit slightly lower down in rank (PC series #4) (Figure 1). If MM had applied standard selection procedures, they would have retained the first 5 PC series, which includes the important ‘hockey stick’ pattern.
I talked about the issue of centering "conventions" the other day. Preisendorfer, cited here as an authority, said quite explicitly that principal components are about the analysis of variance and are "by definition" centered on the time-average (not on a subset.) The MBH98 method is not a "convention" or even principal components analysis. Preisendorfer:
If Z’ in (2.56) is not rendered into t-centered form then the result is analogous to non-centered covariance matrices and is denoted by S’. The statistical, physical and geometric interpretations of S’ and S are quite distinct. PCA, by definition, works with variances i.e. squared anomalies about a mean. (p.27)
The cheekiness never ceases to amaze me. We said quite clearly that the bristlecones were demoted from the PC1 to the PC4. And yet, Mann here has the gall to say that "curiously undisclosed" by MM is this very demotion. I’ve posted on several occasions on whether MBH98 actually used the form of Preisendorfer’s Rule N, claimed in recent realclimate posts. It is impossible to replicate the actual selection in other networks using this rule. However, today, I wish to talk about Preisendorfer’s own justification for Rule N – which I’ve argued to be at most necessary for significance and not sufficient. This seems to be an obvious distinction, but obviously not understood by realclimate. I’ve illustrated this obvious point by showing what happens if you replace bristlecones with dot.com stocks.
Here are some quotes from Preisendorfer on Rule N and other selection rules:
The null hypothesis of a dominant variance selection rule [such as Rule N] says that Z is generated by a random process of some specified form, for example a random process that generates equal eigenvalues of the associated scatter [covariance] matrix S…
One may only view the rejection of a null hypothesis as an attention getter, a ringing bell, that says: you may have a non-random process generating your data set Z. The rejection is a signal to look deeper, to test further. One looks deeper, for example, by drawing on one’s knowledge and experience of how the map of e[i] looks under known real-life synoptic situations or through exhaustive case studies of e[i]’s appearance under carefully controlled artificial data set experiments. There is no royal road to the successful interpretation of selected eigenmaps e[i] or principal time series a[j] for physical meaning or for clues to the type of physical process underlying the data set Z. The learning process of interpreting [eigenvectors] e[i] and principal components a[j] is not unlike that of the intern doctor who eventually learns to diagnose a disease from the appearance of the vital signs of his patient. Rule N in this sense is, for example, analogous to the blood pressure reading in medicine. The doctor, observing a significantly high blood pressure, would be remiss if he stops his diagnosis at this point of his patient’s examination. ….Page 269.
Thus, even if the PC4 is "significant" under Preisendorfer’s Rule N, that doesn’t mean that the PC4 is a temperature proxy. It means that you need to look at it. That’s what we did. But we found out, as Mann had found out before us, that the hockey stick pattern came from bristlecones, which had been identified by specialists as being related to CO2 (or other) fertilization in the 20th century and not due to temperature. There was a "ringing bell" – it should have been paid attention to.
One of the specific criteria set out for significance by Preisendorfer is that the analyst be able to identify a plausible "dynamical origin" for the selection:
In all of this research, one basic question should be kept in mind: is there a dynamical basis for the candidate selection rule? (cf Preisendorfer, 1979a), For practiced meteorologists and oceanographers, such rules should always be checked for the reasonableness of the synoptic patterns in the first few retained eigenvector maps. Both theorists and synopticians should agree that the basis for a selection rule should as far as possible be the dynamical process underlying the data set. (p. 251)
Preisendorfer’s concept of a "dynamical origin" is not fulfilled by ragbag tree ring networks. Presiendorfer’s examples are all PC calculations for data sets consisting of time series of uniform fields (temperature, sea level pressure) over gridded geographical regions. The regularity of the geographical index is not explicitly stated, but is clearly assumed in Preisendorfer. The MBH collection has 16 sites in the gridcell containing Sheep Mountain and 1 site in other gridcells; it has lots of bristlecones. It’s not geographically regular. The need for geographical regularity can be seen in the discussion of "dynamical origins" which requires that the underlying process be representable by differential equations. Preisendorfer has a lengthy discussion of the "asympototic PCA property" for one-dimensional harmonic motion (a spring-linked mass model) and for two-dimensional wave motion on a rectangular domain of an ocean surface. Preisendorfer:
The PCA property (2.96) of Z turns out to have deep connections with various models of physical processes studied in meteorology and oceanography. As we shall see in the next chapter, various data sets generated by solutions of any of a large class of linear ordinary or of linear partial differential equations exhibit the PCA property in the limit of large sample sizes n. When this is the case, the eigenvectors of the data set resemble the theoretical orthogonal spatial eigenmodes of the solutions. In this way, empirical orthogonal functions arise with definite physical meaning… The eigenmotions U of the dynamical system manifest themselves as the eigenvectors of the PCA of the dynamical field’s output field Z. The formal connections of PCA with harmonic analysis exist on quite general levels (p. 87)
“3. Dynamical Origins of PCA: We will give examples of the asymptotic PCA property (2.125) for two systems. This will provide an objective basis for the selection rules stated in Chapter 5.”
Rule N is described by Preisendorfer as a "dominant variance rule" and the chapter formulating Rule N is entitled “5b. Dynamical Origins of the Dominant-Variance Selection Rules. "
MBH98 itself referred to the use of Preisendorfer’s Rule N only in the context of temperature PCs, which do meet Preisendorfer’s conditions. The basis of selection for tree ring PCs was not explicitly stated. The supposed use of Preisendorfer’s Rule N was not mentioned until there was a need to include the bristlecones from a lower PC series. The ragtag and motley collection of tree ring site chronologies obviously do not meet the conditions used by Preisendorfer to justify Rule N. But more importantly, as Preisendorfer clearly says and is obvious, passing a Rule N significance test is merely a "ringing bell". It is not a sufficient test of statistical significance. Roll over, Preisendorfer, and tell Pachauri the news.
Climate Audit 
86 Comments
The final sentence is a delightful allusion, but Tschaikowsky might feel
that he has been rolled due to his place in it being taken by Pachauri.
Anyone too young to be acquainted with the allusion, count your blessings. 🙂
Re #2 Welcome aboard TCO. You join many of a like mind that began by thinking it can’t be that hard to get the maths and science sorted and find out if anthropogenic global warming is the problem some say or the hoax that others say. If you continue to research it you will find it is unimaginably complex scientifically and politically. Like you I think it would help if there were a good common man’s guide to what’s wrong with AGW theory. There have been some and I think the late John Daly’s web site is still a good starting point and is full of useful guest papers. I think it is also worth reading that of one of his greatest critics and a regular contributor to this site to indicate the sort of bile that any one seriously questioning AGW must face. I assume you were referring to the journal “Nature” and to understand the difficulty of getting anything critical of the status quo published there you should read M&M’s account of their attempts. I can’t see what sceptics can do other than to try hold to account proponents of AGW. If you are sitting next to a nervous passenger on an plane that thinks every movement and sound is an indication of imminent peril it hard to think of a comprehensive rebuttal because you are dealing with a belief. You have to say “no, that noise is the flaps” “no the engines are fine the pilot just needs to slow down a bit” and address each irrational thought as it arises. Having said that, I am choosy about which carriers I will fly with and read the safety instructions.
Steve: Speaking of which, there was rather a spectacular plane crash in Toronto this week, with the Air France plane burning up about 50 yards from the 401 highway, which, at that point, is about 16 lanes wide and the busiest highway in Canada. Some of the passengers hitchhiked away from the plane crash. The emergency team was on the scene in 50 seconds and there were amazingly no fatalities. I hardly ever listen to CBC radio anymore with its culture of complaint and rural jam-making, but I inadvertently surfed through a phone-in on the crash where one of the callers expressed particular angst about the environmental impact of fire retardant and suppressant chemicals on a little creek. I presume that they were relieved by the person in charge’s statement (assuming that I heard right) that the rescue team was accompanied by an environmental consultant who was ensuring that the efforts to rescue the passengers did not have an adverse environmental impact. I hope that he was joshing the caller, rather than the airport is paying envinrmental consultants to stand by 24/7, but these days, you never know.
TCO, I would also like to welcome you to the circus. Regarding your “at least do the meta-analysis the “way it ought to be done”).” comment, we would all love to see it done right but there are a few things standing in the way.
First is that in order to do the analysis properly, you need access to all of the raw data and “they” are refusing to make it all available.
Second big problem is that the most contentious proxies (tree rings) may be totally useless as temperature proxies. Not enough data is available to determine whether they are or not. At least there is not enough data relevant to tree growth to allow for the isolation of the temperature component.
Third big problem is that if all the raw data ever gathered were readily available, it would still be too small a sample to make any conclusions about our climate. Think of all the climate data ever gathered as a single grain of sand and we are trying to use it to describe all the beaches of the world.
TCO re #2
As I am a non-scientist I don´t know much about “meta- analysis”. I have just heard that it
should be a sloppy way of using statistics, so now you make me curious. Could you please tell me then how to make a meta-analysis “the way it should be done” on a system of chaos!? You might be able to explain it clearly as you plainly consider it something that must be done “at least”!
Hans Kelp
From what I remember of Greek lessons when I was a kid, the word meta could be translated as about ten different prepositions in English. Always struck me as a singularly useless word.
TCO, re explaining the basics : be fair, its perfectly proper to tailor your material to your audience. Now that you’re here, you might like to check out some of the links in the sidebar, which include some more entry-level discussions of the issues.
1. aaargh: I already read the travails of the Nature paper (the reviewer comments and all that). Don’t make me read it again.
2. Why do you all want so much to reanalyze someone else’s work. At least get your heads straight that it might be possible to do an analysis yourself from first principles. Don’t tell me that you need M’s data. Just go get all the data you can from individual studies and analyze it as you think best. And regarding the tree rings being unreliable. Well are they or aren’t they? What’s your best guess. If you have all the data available and want to get the most telling analysis from it, what do you come up with. Surely you guys have wasted enough time looking at all this stuff, that you know how to get the subordinate references, have an opinion on how to choose which studies are trustworthy, have an idea how to do meta-analysis despite the non-perfections of individual sub-studies.
3. I mean I was on y’all’s side and still am some (the warping the studies based on 1920-1980 just seems wacked!!!) But the constant thinking only in reference to debunking, not to the actual attempt to find truth ITSELF gets tiresome. I would hope that if you all have developed the sophistication that you could redo all the work (and I don’t mean redoing his calcs or using only his data…get your mind out of that rut…I mean redoing the analysis based on whatever YOU can find…maybe even some experimental evidence that he missed) and find that the “hockey stick” still emerges, that you would publish it and that it would be a contribution to the literature. You could still school Mann (and in so doing instruct others) for methodology since his hockey stick would be based on errors cancelling and yours would not.
4. Whatever you do, don’t become like the Cold Fusion advocates who can’t discuss rationally the issues without seque into their publishing battles for respectability.
5. Just so this does not come across as too adversarial: I liked the reviewer comments who liked your work.
——————-
net, net though: You’ve spent all this time learning PC subtleties of methodology and the experimental literature of climate study (the source materials). Surely you have more to contribute than just pinning Mann’s wings to the page?
TCO, perhaps you haven’t noticed that a fundamental part of science is repeatability. And perhaps you haven’t noticed that many of the underlying studies are not themselves available. That the data involved are being closely held and the names come from the same incestuous circle.
Heh, sorry. I meant to try these posts :
http://www.climateaudit.org/index.php?p=166
http://www.climateaudit.org/index.php?p=63
and the documents they point to, which I found useful when I first came across this site.
Well, maybe so, maybe so. But given the amount of time already spent on this, it would be a useful exercise for MM to do a first principles analysis (as they see fit) of all the proxy studies that they can get. I would think that many of them are in the open literature. Also, that MM might find a few that MBH left out. Such an exercise might lead to NEW criticisms of MBH or new understanding of how to do work in this field. It might also lead to a useful result. But even purely as an exercise it would be useful to take the mind off of analysing the analysis for a second and to analyse the data.
Please put my post back. I apolgize if it was poorly written. I tend to think out loud. Please put it back.
Can’t we have a decent discussion? I’m not even trolling or anything. Heck, like I said, I’m 60% on your side. that doesn’t mean I won’t question the questioners (of the questioners) haha!
TCO,
You may not be aware that MBH98, as well as its offspring, has had a very
large influence on many people’s perceptions of recent climate history.
It has been a key part of what appears to be an effort to make the
“Medieval Warm Period”, and the “Little Ice Age”, disappear. It gave new
meaning to the phrase “climate change science”. It would seem to be well
worth some serious analysis.
But Steve is doing more than analyze MBH98. He has had a variety of
“learning experiences” which he has shared with us, and they go beyond
MBH98, and/or Michael Mann. His correspondence with the magazine
“Nature” may be quite edifying to many people, as well as, for example,
his correspondence with Crowley.
Also edifying may be the behavior of other practitioners of “climate
science” with respect to Steve and Ross’s papers. In a post on July 12
http://www.climateaudit.org/?p=262 Steve compared Bradley’s recent
comments with the March 3, 1999 press release announcing MBH99. As it
happens, on July 13 of this year I saved a copy of that press release
from: http://www.umass.edu/newsoffice/archive/1999/030399warming.html As
of a few minutes ago, it was gone. I do not know when it disappeared, or
whether it may reappear when someone realizes how tacky it was to delete
it, but in case you might guess that I may have misspelled the URL, you
can find it mentioned at the following AGU page
http://www.agu.org/sci_soc/prrl/prrl9906.html unless that page disappears
soon also.
What I am taking too long to say is that there is much more going on here than
analysis of one study.
Jerry,
I saw all that stuff. That’s fine and good. My question remains. Surely all the effort that MM have exerted makes them capable of doing original work in the field now, no? I practically feel I could after reading all this stuff here.
The phrase “can of worms” springs to mind.
Errrr, do we have another case of SpamKarma post-retro-whatever trashing a comment ? Lots of posts responding to #2 from TCO, without any post by TCO being visible.
TCO had a second post which also is missing at this time.
I figured out how to get TCO out of the penaltyy box. He was picked up by Spam Karma for multiple postings without a history. .
Re #13 , I may venture into some atttempts of my own. I’ve been reluctant so far for a couple of reasons. 1) Let’s suppose that you thought it was a good idea to sell Enron. Did you need to also have a buy recommendation? I didn’t want to mix up really well-based criticisms with speculation of my own. 2) I had statistical and math skills which were sufficient for the task at hand. At the start, I was much less knowledgeable about the various proxies than I am now. 3) Some of my ongoing interests are as much mathematical and statistical as to do with proxies. I’m interested in issues of autocorrelation and confidence intervals, which are wentirely different than proxy reconstructions. 4) From a practical point of view, if I’d presented my own view of the matter, why would anyone have paid any attention to it, even if it was meritorious? EVen now, I’d have a hard time attracting any interest. 5) I want to be sure that I udnerstand perfectly how other people do their work before trying my own. 6) I don’t like leaving half-eaten sandwiches: I finish what I start, especially if someone tries to make it difficult. 7) I’ve got some unfinished business on the other multiproxy studies.
TCO whose post is now back, I think.
What you seem to be assuming is that it’s POSSIBLE to find a legitimate proxy record from which to calculate past climates sufficiently accurately to compare with other sorts of past climate data. One of Steve’s findings is that without bristlecone pines the remaining proxies don’t have sufficient statistical merit.
Now obviously if Steve tried to claim that on the basis of lack of merit of Mann’s findings it’s impossible to determine past climates on the basis of proxy data he’d be over-reaching, but if he’s willing to serve as a ‘devils advocate’ for those who claim to have done so, he’s performing an important service for us all. Let those who want there to be proxy climate reconstruction try, and let Steve then try to debunk their claims.
You wouldn’t try claiming, I hope, that if a person debunked a spoon bender, it was sensible to demand that the debunker stop being a nay-sayer and get out there and start bending spoons himself, would you? So why demand it of proxy climate reconstruction?
A. You missed my first post, but we got the essential content here.
B. Is the method critique of centering really so simple as shown (substracting by the 1900s average prejudicing towards graphs that have extreme value in the 1900s?). I would think that this is pretty trivially wrong and not even debatable. How do they defend this. It’s like plotting on a log-log graph, getting a line and saying a relationship is first order! Although I would think it would select prejudicially for both blade up and down. Shouldn’t you get half of your random noise experiments giving blade down?
C. At the tree-hugger site, they were cackling about you messing up degrees and radians. What was the deal with that?
D. Little dissapointed by Crooked Timber and that general communitty that nooone engaged in looking into the PC4 and such issues. I’ve seen them do better on other stuff. I know that there are some shapr mathematicsl physicists in that community.
replies to replies:
1. a. I was an Enron skeptic. That’s a story I wish I could tell.
b. on an action/philosophy level, no, I think. IOW, your single criticism has value regardless. However, if you were Pekka Heital and I’d hired you to look over Goldman’s DCF-NPV and you’d found flaws (e.g the WACCincluded inflation, but the revenues and costs were in constant dollars, the tax effect of depreication was mishandled, yadayada) in it; I would also like you, dear Pekka, to have calculated the NPV as best you could yourself. For my kind use…and to push the thinking forward…and to make sure that your thinking is motivated by pursuit of best correct answer rather than advocacy of a viewpoint or flawfinding in opposing viewpoints in a tendetnious manner.
2. I feel like you’ve GOT to be knowledgeable about proxies by now. Heck any looking into something for a week or two (reading the literature) ought to leave you confident enough to take a swing at doing something on your own. After all, you do have a basic foundation of writing for publication in other fields.
3. So where will that take you? What sorts of problems, fields will you work on?
4. I think you underestimate the value of taking a swing at an answer in concentrating the mind. I could watch and criticizze another as he played the hand from the dummy seat, but I learned and did more from the declarer side. Also, I DO think you could get a competing meta-analysis published. Surely this happens in fields. And to the extent that you’ve been looking into this, I would find it surprising if you did not at a minimum have additional studies to include. Even the passage of time would give those.
5. It is good to want to understand very well how to do work properly before doing it (although even then, one may learn faster from jumping into the pool…it’s a judgement call), but studying a POOR METHODOLOGIST in detail? C’mon!
6. That’s great to see it to the end. I do think that you all seem a little too wrapped in the immediate kerfuffle and not thinking about the logical extensions. For instance, as simple as “what other fields, researchers, papers have similar errors”?
7. Dish.
in penalty box again…sheesh:
I haven’t seen MM make the argument that all proxy work is worthless. If that is their feeling, would be interesting to see it expressed. Even so, I think they could learn something even if they stay debunkers by walking a few steps in the other man’s moccasins.
TCO,
Don’t give up hope. It appears that Steve’s attempt to mollify Spam Karma was only temporarily successful.
I will venture the guess that John A may have more success tomorrow.
Re #19. The PC point is pretty simple (or else I wouldn’t have thought of it.) It was misrepresented in the original article, so no one would have thought about it unless they tried to calculate the principal components. It was a big data job even collecting the data. There were 232 series at WDCP. The formats are inconsistent. Also I wa specifically looking for a problem in the North American tree ring data.
I’ve spent a lot of time on basics as well: what is the statistical meaning of a tree ring chronology? I can replicate their methoids using linear mixed effects modeling, which can them impart some statistical meaning to the process. It’s pretty neat, but a big big job to write up.
How can you handle severely autocorrelated series properly? This is not as easy as it sounds. No one in paleoclimate even thinks about the problem. I keep poking at the papers trying to understand them better, but the math is hard. If you google Kiefer Vogelsang econometrics, you’ll see some of the papers that I’m trying to understand. Also many of Peter Phillips papers on spurious regression. They are on the internet. I’ve also been doing a lot of work on fractional Gaussian processes. I used them to model the series in the GRL article, which was a bit of artistry that’s been totally unappreciated. I hadn’t done any math for 35 years so these things are not a walk in the park for me. Long-run variance estimation is a topic that needs to be discussed in paleoclimate. These are not in Ross’ area of expertise. I need some help from some young math guy, whose more up-to-date than me. I’ve stumbled into a cornucopia of problems that no one’s handling, but it’s taking me a very long time to upgrade my skills. (The Mann stuff was trivial, but it took me time. I didn’t know an up-to-date computer language; I’d never written an academic paper etc.etc. I was still trying to make some money from time to time.)
I’m not convinced that many of the traditional proxies are any good. As matters stand, I probably need to deal critically with at least one other study and probably two. I’ve got a critical piece on Jones et al. 1998 about 80% done and I really should do one on Moberg.
I have some nice little pieces that I’ve been thinking about. About 2 years ago, it was one of the first things that I did, I did a distribution of ring widths for the Tasmanian series and found that it was bimodal; the narrow-ring lobe was all modern and the wide-ring lobe was all ancient. It was pretty neat.
Re #19C: I had nothing to do with the radian/degree thing, although they all pretend that. Michaels and McKitrick wrote a paper in which McKitrick inadvertently did calculations in Shazam assuming that the program read in degrees (it read in radians). Ross archived his computer script. Tim Lambert went through the script and fond out that Ross had goofed and played it up to the crowd. My take on this: anyone can make an error; that’s why it’s a good idea to archive your source code and data and allow people to check it. McKitrick acknowledged the error and corrected, but they still hold it against him. Since Michaels and McLKitrick were MM, they merge the papers together. Ironically Mann made a cos latitude error as well, by incorrectly calculating the weighting factor in his temperature PC calculations. I’ve been twitting Lambert on this, but Lambert is completely one-sided.
What is (was) your day job?
David H (#2): You said:
“There have been some and I think the late John Daly’s web site is still a good starting point and is full of useful guest papers. I think it is also worth reading that of one of his greatest critics and a regular contributor to this site to indicate the sort of bile that any one seriously questioning AGW must face.”
1. I assume the second sentence refers to me.
2. My Oxford Handy Dictionary defines “bile” as “peevishness”.
3. Please look at John Daly’s site (www.john-daly.com) and then at my critique of his site (www.trump.net.au/~greenhou/), and tell me why you think the former IS NOT bile, while the latter IS.
4. Please look at my postings on this site, and then at the numerous responses to those postings (and also, perhaps, to those of Peter Hearnden), and tell me why you think the former IS bile, while the latter IS NOT.
If AGW were blindingly obvious then esoteric statistical manipulations would not be necessary. That they and the arcana of specialists are resorted to back academic authority, suggests that it is pathological science.
John,
I am not saying that all the commentators critical of AGW are free of bile and peevishness. I have not looked lately at your site other than very briefly but we did have a private discussion in which I think you agreed that you had wrongly stated that John Daly had taken NASA-GISS graphs out of context and used them inappropriately. He had in fact used them almost exactly as they appeared and quoted verbatim. I did not see any change to your web site.
I believe that John Daly regardless of what faults he might have had or what errors he might have made will go down as one of the greatest “amateur” scientists of his day and will be remembered and referenced long after you and I are forgotten. I put amateur in quotes because he was not formally trained as a scientist but then so too were many of the greatest.
Steve: Karlen, a distinguished climate scientist and a coauthor of Moberg et al [2005], recenetly cited john-daly.com in an article in Ambio (about which I need to post a comment.)
I have just looked at John H’s website and I think the first few paragraphs justify my view of it. What is a bit odd is that so many of those passionately promoting AGW theory keep banging on about why do sceptics keep attacking the hockey stick. Yet here we have one of the more vocal (or perhaps verbal) AGW supporters with a web site exclusively devoted to attacking one private sceptic. I think John Daly would have seen it as a recognition that he was having an effect but I think its continuance now is just tacky. Perhaps, John, its time you started a “What’s wrong with ClimateAudit” site.
Re #28
If you think you’re hard enough …
If I had the time I’d do a “What’s eating John Hunter” site or better still a “John Hunter and the Isle of the Dead” site. But I don’t.
If only I could be an academic instead of having to do a real job. *sigh*
David H (#27 and #28): You may well “think” that I agreed that I “had wrongly stated that John Daly had taken NASA-GISS graphs out of context and used them inappropriately.” You also claim that “he had in fact used them almost exactly as they appeared and quoted verbatim”. Perhaps your recollection of our discussion differs somewhat from mine. You will see that my comment on John Daly’s article “Global Mean Temperature – Disputed Data” came under a section in which I was giving examples of “misrepresentation or misinterpretation” of data. Based on two plots derived from the GISS web site (Global and U.S. temperature records), John Daly concluded that they were “completely at variance” — a statement which was certainly not “quoted verbatim” from the GISS site. However, plotting the data on the same scales and also estimating their linear trends indicated that it was quite untrue to state that the records were “completely at variance”. In fact the trends differed by only 13%. So what is your problem with this?
You also said “I think John Daly would have seen it as a recognition that he was having an effect but I think its continuance now is just tacky.” You seem to forget that “Still Waiting for Greenhouse” is still active (although only a shadow of its former self). Does it therefore not seem reasonable to continue to address the distortions of that site?
TCO,
> Surely all the effort that MM have exerted makes them capable of doing
original work in the field now, no?
It would seem to me to be inappropriate, as in presumptuous, for me to
offer unsolicited advice to Steve and Ross on what to do with their time
beyond the occasional mention of some article which might be of interest.
Aside from that, it struck me as quite ironic that in one of your posts,
which is still among the missing, it seemed that your idea of original
work was for them to mimic the work of MBH.
Meanwhile, what they have been doing seems to be very original in the
sense that it seems that nobody else appears to have done it: not
“Nature”, not the IPCC, not AAAS, not NAS, not NSF, etc.
I hope that they have enjoyed what they have been doing, and that they
enjoy doing more of it.
Steve (#30): Can you please exert some control on John A’s gratuitous slurs and rudeness? I actually arrived home from work after midnight to read John A’s “if only I could be an academic instead of having to do a real job”. Now, Steve — you know as well as I do that, if I said such things, my posting would be snipped. As I’ve said before, this (coming from the man who runs your site) does your credibility no good at all. Any reasonable person would actually offer an apology.
JB: Actually what I want them to do would be MORE like MBH (in the sense of making new findings, attempting to construct an understanding), but LESS like MBH in the sense of “confining their work solely to finding mistakes in MBH”. And I guess I AM intemperate enough to ask them about/suggest where they should spend their time. This is the internet and is a chatty blog. I don’t see anything wrong in such chat (as well as thinking about extensions of work rather than the immediate debate.
JohnA, Hunter: *HANDS ON HIPS* this is my thread now. Move your little he-said/she-said to some other thread. The “shoddiness thread” is a good place–it couldn’t get ruined more.
Re: #33
Dang! There goes another perfectly good irony meter! They just don’t make ’em like they used to…
Re #6 etc., on why is ClimateAudit still focusing on MBH98. The simple answer is that it’s still highly influential. Two months ago I was interviewed at length by a reporter for a western Canadian farming magazine on the whole climate change/adaptation issue. We talked about the hockey stick issue among other things (mostly I tried to confine my comments to arguing that farmers shouldn’t invest heavily in one particular climate scenario but hedge themselves against a range of futures). Anyway I thought I had gotten through to the author, but when the article came out my comments were bracketed off with warnings about the mighty consensus of true experts, the marginal role of industry-funded skeptics etc. The voice of consensus was an IPCC lead author, whose rehash of the party line was reinforced with, of all things, a full-colour presentation of the Mann hockey stick. Along the same lines the Government of Canada not only still promotes it on the net but is shipping out packages to schools across the country showing it.
Core IPCC authors enjoy the privilege of having their words influence world leaders and public perceptions, but until they accept the corresponding responsibilty of admitting their mistakes, and until they honestly examine how they they made such a major a mistake in promoting the hockey stick so heavily even though it had never been subject to proper independent review, and until its promoters in Government and media begin making some public retractions; they can’t expect the people who laboured this hard to get their critique into play to simply drop the subject.
I’m not asking you to drop the ongoing argument. I do think it would:
a. strengthen your standing/position
b. teach you something
c. help you to criticize MBH even more astutely
d. be a good realization of your already invested time
e. (possibly) contribute to better understanding of what actually occurred
if you did some original work.
Ross (#35): I know this is an old and well-worn plea, but if you so disbelieve the existing temperature reconstructions, why don’t you convince Steve McIntyre to generate a new one? If he believes that proxy data is less reliable that workers have previously assumed, then he should simply estimate the uncertainties appropriately — it is not an excuse for not doing the reconstruction. A publication that showed that there is no useful statistical skill in any feasible proxy reconstruction (however well executed) would even be a useful advance.
I am one of those strange people who find much more value in knowing something about the world, with a realistic measure of uncertainty, than knowing a whole catalogue of claimed mistakes.
“A publication that showed that there is no useful statistical skill in any feasible proxy reconstruction (however well executed) would even be a useful advance.”
It would also be a miracle. You’re asking for proof of a negative. The word ‘feasible’ does leave a little wiggle room, but if it wasn’t for that I’d say it’s an impossible task. Further, Steve is only one person and is not backed by major grants, not even ones by the evil corporations.
I know it’s a broken record too, but why doesn’t Mann, et. al. come clean, correct any errors and provide substantive disproofs of anything M&M have wrong and then issue new proxy reconstructions which are produced in such a way that they can be quickly checked and thereby restore their somewhat shaky reputations?
Alternatively, you might try engaging in actual debate concerning Steve’s actual findings rather than worrying about trivialities.
John Hunter, I’ve actually spent a lot of time (really a lot of time) trying to properly understand the calculation of confidence intervals when autocorrelation is involved. Paleoclimatologists use stone age statistical methods. I’m not convinced that any of the standard reconstructions achieves confidence intervals lower than natural variability and their only “skill” results from an under-estimate of the standard deviation of natural variability which they masquerade as “skill”. The estimation of confidence intervals by Mann are a joke: he ignores autocorrelation. That’s bad enough. But when you have an R2 of ~0.0 in the verification period, you cannot obtain a confidence interval less than natural variability other than by incorrectly using the calibration period standard error. Recognizing that it’s a joke is one thing, but I’m just a one-eyed man. To actually estimate confidence intervals correctly in conditions of autocorrelation is a different thing.
Also, unless one has mastered all the multiproxy reconstructions and shown the problems with them, it’s premature to add to the cacophony of reconstructions. You have to show why each one of them is wrong. Also if you do that, you actually do learn something in the process. It takes far more time than it should because you’re fighting to figure out what each one of them actually did, where they actually got their data, etc. etc.
what I said…
Re #38 and #41: If Steve were to receive an NSF grant like this
http://www.nsf.gov/awardsearch/showAward.do?AwardNumber=0125670
and had a bunch of undergraduate students to do the leg work maybe he could consider your proposition.
Re #37,38: John, I take it as a compliment to Steve and I that people in the field would actually be interested in an “MM07” curve. However paleoclimatology is not my field, and we do not need to publish a reconstruction of our own to establish bona fides for our MBH98 work. I came to the job with economics training and that sufficed to do the particular job, but I won’t push my luck. Last year Steve and I spent a day in a tree ring lab where the scientists told about their craft and we told them about ours, and it was a lot of fun. But I’m never going to know enough botany to understand tree rings, or enough geology to understand borehole thermometry, etc. I grant that I am probably more pessimistic than the average paleoclimatologist about the usefulness of global reconstructions; this is in part from my understanding of the statistical difficulties. The challenge is not simply to generate a graph, but to also be able to say whether it is truly informative about the actual climatic history of the planet; or at least moreso than a list of random numbers. There’s more to it than “simply estimating the uncertainties appropriately”. Sometimes it’s not even clear what the nature of the uncertainty is, and how to quantify it, if it is even possible to do so.
1. Has he applied and been rejected for grants?
2. I agree that the MBH criticism stands (reasonably) on its own. Yes, it would carry even more weight if you were doing work in the field, and might come up with a few more critiques. But it stands on it’s own. That’s not my point. I’m not saying MM are full of guff and shouldn’t be trusted to critique MBH until they do work of their own. What I’m saying is that it might be worthwhile to do work of their own. Think about. I know it’s subtle.
3. Perhaps a general paper on the difficulties in doing paleoclimatology would be in order. you could even colaborate with a tree ring experimentalist or whatever. (This will help with (1) by the way…:))
I found post 38 by Hunter surprisingly good (you know…"smart for a Marine" hee hee). IN particular the third sentence. I’m not saying it quite as provocatively as Hunter, but I do think it makes sense.
Steve: Speaking of Marines, did you see Jacoby’s discussion about a Marine’s approach to data preservation and archiving here and here
Re#45,
I’d certainly love to see this “realistic measure of uncertainty” described and see the mathematical justification. For instance, let’s take Mann and Jones (2003) http://www.ncdc.noaa.gov/paleo/pubs/mann2003b/mann2003b.html . It would be great to see someone build from the uncertainties inherent in the proxy data and watch him/her extrapolate these uncertainties as the data are weighted, scaled, and used to represent areas up to thousands of square miles in size.
Something tells me “two-standard error uncertainties,” +/-0.3 deg C in this case (a total range of 0.6 deg C, or about 50% greater than the range of proxy data itself!) wouldn’t cover it.
Re#46 –
I should’ve said something like “reconstruction curve,” or “40-yr smoothed reconstruction curve,” rather than “proxy data.”
It’s a signal to noise problem. If you have enough data, then you can still extract a good signal, even given lots of noise. On the subject of proxies, I wonder if historical records are useful. Could we get written records from all over the world (that had writing) to see if the MWP was just a euro event or occurred at other temperate zones (e.g. China).
Steve and Ross: If there is an autocorrelation problem with the function that attempts to verify the tree ring/temperature relationship, and the process is non-stationary, then why don’t you demonstrate that by running an ARIMA program to show it? Or maybe you have. Also, it’s interesting that you spent a day in a tree ring lab. My question is, do tree ring experts use statistical methods to establish tree rings as a temp proxy, or do they use lab measurements of the temperature/growth relation that are then extrapolated to the field measurements?
Even better, does anybody know of a calibration study of treerings in the direct vicinity of a weather station?
-precipitation
-sunshine hours
-humidity
-temperature
RE #48 – we refer to some comment by Larson and Kelly in our E&E article. Our starting point were cedars, since Gaspe cedars were so integral to MBH and we wanted further info on the Gaspe hockey stick. Larson and Kelly had studied cedars in southern Ontario for about 15 years. There is a remarkable “forest” of cedars along the Niagara escarpment which is essentially a cliff that goes for 600 km. The “forest” is 600 km x about 100 meters. From the cliff above a 1000 year old tree, you can see the CN tower. They mentioned one “tree” which was 24 g and 150 uears old (if I remember right). Cedars apparently grow best under cool moist conditions and the biological growth curve is an upside-down U with temperature (which is typical of most conifers). In contrast, the multiproxy studies assume a linear relationship between temperature and ring width (without proving it) or between a weighted average of temperatures (weighted according to the EOF1) and ring width. The cedars had strip-bark like the bristlecones and LArson and Kelly had written a paper comparing the two.
For MBH, they used their MBH98 PC1 (bristlecones) as a temperature proxy without either – no biological growth curve and no statistical analysis to show that either the cedars or the bristlecones were a valid proxy (and in the face of explicit specialist literature to the contrary.) It is bizarre beyond belief. I think that slow acceptance of our investigations is simply that people find it too incredible that there should have been such a lack of ordinary statistical due diligence.
RE #51 – “people find it too incredible that there should have been such a lack of ordinary statistical due diligence.” I don’t think that the problem is just in the statistics, it is in the idea that temperature can be extracted from tree ring width. This has to be wrong.
The growth of a tree is determined by temperature, T, water, W, CO2, C, sunlight, S, and soil fertility, F. (I’m a physicist so please correct me if I’ve said something wrong biologically.) The tree ring width is related to these variables by some -> unknown
RE #51 – “people find it too incredible that there should have been such a lack of ordinary statistical due diligence.” I don’t think that the problem is just in the statistics, it is in the idea that temperature can be extracted from tree ring width. This has to be wrong.
The growth of a tree is determined by temperature, T, water, W, CO2, C, sunlight, S, and soil fertility, F. (I’m a physicist so please correct me if I’ve said something wrong biologically.) The tree ring width is related to these variables by some unknown function G which we need to invert to get T.
ring_width = G(T, W, C, S, F)
First problem: Suppose I know G. I would like to invert the equation above to find the temperature from the measured tree ring width for a given year. I can only do that if I know W, C, S, and F independently for that year. This is basic mathematics. I only have one equation so it can only be solved for one unknown. Based on this argument alone, if the only measurement made on the trees is ring width, it is impossible to use ring width as a proxy for temperature.
Second problem: What does G look like? It’s not linear. #51 points out that it’s shaped like an upside down U with temperature, T. Suppose we vary the amount of water, W. No water and the tree dies or too much (look at the dead full grown trees in a beaver pond) and it also dies. Another upside down U. The same sort of behavior is true for CO2, sunlight, and soil fertility. Ring_width goes to zero if any of the five variables is outside a restricted range. G is a complicated function and can’t be linear to behave in this way. Any claim that G is linear should be viewed with great suspicion.
Steve (#40): You say "paleoclimatologists use stone age statistical methods" and "the estimation of confidence intervals by Mann are a joke: he ignores autocorrelation". You may well be right. It is probably true to say that statistics are most scientists’ weak point – we are generally happy if we believe that the "uncertainty of our estimated uncertainty" is a bit less than "our estimated uncertainty". However, any useful critique of the the statistical methods of paleoclimatologists would generally concentrate on a broad range of paleoclimate papers and not just on those of the "hockey team". climateaudit would in fact be far more credible if it, say, considered also two other paleoclimate papers which have had a quite significant impact on the debate – those of Soon and Baliunas (2003, Climate Research, 23, 89-110) and Soon et al. (2003, Energy & Environment, 14, 2 & 3, 233-296). It would seem that these papers raise issues of both questionable statistical validity and neglect of the autocorrelation of the proxy records. Perhaps you could run a whole thread on these two papers.
Steve: Unless I’m mistaken, I haven’t seen any policy reliance on Soon and Baliunas. It also seems to me that there has been ample attention from other parties so, if you don’t mind, I think that my attention is better spent on Jones et al [1998], Crowley and Lowery [2000], Moberg et al [2005], Briffa et al. [2001], Esper et al [2002]. I’ve got a considerable inventory on other projects and don’t like to leave things unfinished. When I get through those, maybe I’ll take a look at Soon and Baliunas. Unfortunately I can’t clone myself. You could perhaps assist me by getting Briffa to identify the 387 sites and Esper to archive his data.
Steve (#54): You say that you “haven’t seen any policy reliance on Soon and Baliunas”.
Policy has many influences (e.g. your papers and climateaudit, via the Barton Letters, to the U.S. House of Representatives), so I think we would both agree that Soon et al. have had SOME influence on U.S. (at least) climate policy. However, you seem to be getting the analysis and data mixed up again. You brought up the subject of paleoclimatologists using “stone age statistical methods” to which I responded that you should perhaps extend that criticism beyond the narrow focus of the “hockey team”. So asking me to assist “by getting Briffa to identify the 387 sites and Esper to archive his data” seems a bit off target. I appreciate that you cannot clone yourself, but (a) you appear to have a number of qualified supporters on this site who could help you out (why don’t they?) and (b) you do appear to have the time to divert yourself to irrelevant issues such as gun statistics (i.e. the “Lambert, Lott and Mann” thread).
Once again, climateaudit would be more credible if you applied as much critical effort to the contrarian literature as you do to the mainstream literature. For another example, you might want to look at the paleo record described on page 1567-1568 of Khandekar et al., 2005 (Pure and Applied Geophysics, 162, 1557-1586).
“Policy has many influences”
And you have many ways of wasting Steve’s time. If you want him to do more, shut up on the trivial stuff and let him get his work done.
Anyway, the logic of your argument is sorely lacking. Just because there’s an indirect policy affect from Steve’s work doesn’t mean S&B has had such an effect. IOW you’re asserting what you were trying to prove. And even then, any effect Steve has is not from his own work per se, but from the fact that doubt has been cast on findings which were of major policy proposal significance. IOW, if Mann’s hockey stick wasn’t of significance to the public debate over AGW, then Steve’s showing it wasn’t statistically significant would have no policy significance.
However, let me backtrack a bit and admit that quite apart from the AGW debate, Steve’s findings may well have policy significance as it’s highlighted the need to make ease of replication a higher priority in research which can have policy significance. But this policy significance isn’t about any particular research area.
Commenting on # 50-53 and factors influencing growth.
The increased CO2 level (so far) has had little effect on plant growth in general. I draw my conclusion on the population variations of moths and butterflies in Finland. Using nearly 800 resident species it is possible to show that the mean of the population size reflect the summer mean temperatures within 0.1-0.01 centigrades resolution. The ziczac curve does not deviate (or drift away) from the temperature curve during the recent 70 years (from which I have a good record). Because these insects are competing on plants as a food resource, the shifts in resource are closely varying in accordance with the temperature and are not reflecting increased CO2 levels of recent decades. Accordingly I doubt that the tree rings would expose fertilizing effects of measurable levels during the 20th century.
Those who by chance are interested in historical malaria cases and mosquito populations in relation to temperatures in 1750-1850 in Finland should check:
Click to access 1475-2875-4-19.pdf
(Lena Huldén et al: an ‘indoor’ disease in northern Europe. Historical data analysed)
Precipitation variations are probably very important to consider in short term context. Fertilizing effects (because of soil changes, volcanoes, geology) may in addition have long term effects.
Larry Huldén
I agree that some more general criticism of the practices of paleoclimatology/proxies as a field would be useful (from research methods sense…I don’t care so much abot the warring parties point of Hunter).
A couple of points here. Paul, Dendrochronologists don’t use tree ring width directly, they generally use related indicators such as late ring density, which I take to mean the density of the wood at the end (outside, and autumn presumably) of each ring. It is claimed that this measure is better correlated with temperature, although I have not yet researched the literature on that point. If anyone has references on this point I would appreciate links or pointers.
Secondly, John Hunter, here is your chance, why don’t you allocate time to analysing Soon & Ballunias ? Even better, spend your time trying to get Realclimate to have some credibility by analysing MBH98 or 99. Their “Idiots Guide” to the Hockeystick could do with a thorough fisking, it as fine a piece of misinformation as I have seen outside of a Greenpeace website. Just for example, they manage to claim that the HS signal seen using PCs must be real (and by inference a real temperature signal) because you still see it when you don’t use PCs and just graph the entire set of proxies. This studiously ignores the fact that PCs were used because the number of BCP records exceeds the total of the other records and is extremely spatially imbalanced. Without the MBH PC workings, the BCP signal only shows up in the PC4, and while this can be included if you use “Rule N”, it would help if “Rule N” was used to select PCs in other intervals, which it isn’t. They also claim that the MBH 99 “correction” to the BCP records turns these into a valid temperature proxy, which only shows that none of them has ever looked at matter at all. All in all, quite precious. Now John, why not spend your efforts in getting them to recognise their misunderstandings and misapprehensions, THAT would increase the sum total of knowledge very effectively.
Dave Dardinger (#56):
I think you are being a little naive, so let me be a little blunt. I (and a few others) think that there are some people who believe that it is in their relatively short-term interests to act as if anthropogenic global warming is not happenning. Now, I don’t want to get into a discussion about whether such people do exist or (if they do) whether their belief is correct. Also I don’t want to get into pointing any fingers here — let’s just assume that there are some people like this. Now it would help such people if they were presented with information which appears to weaken the case for anthropogenic global warming — and, as in the case with any debate, the proponents do not actual have to believe this information. (The converse would also be true — it would NOT help such people if they were presented with information which appears to weaken the case AGAINST anthropogenic global warming.) So this is how the “logic” goes — someone says that the case for anthropogenic global warming relies heavily on the Mann et al. “Hockeystick” — someone points out some apparent errors in the Mann et al. “Hockeystick” — therefore the case for anthropogenic global warming fails. It’s not my “logic”, so don’t blame me!
So all I’m asking is that Steve and his able supporters do not concentrate solely on finding errors in the mainstream literature, but that they have a look at some of the contrarian literature as well. It’s simply a matter of balance (trivial to you maybe, but balance nevertheless).
Ed Snack (#59): You ask “John Hunter, here is your chance, why don’t you allocate time to analysing Soon & Ballunias?”.
I have — it’s rubbish, in the sense that it fails the test which I will call the “McIntyre Test” — if you feed their analysis with smoothed random noise (i.e. autocorrelated noise) instead of real proxy data, you will get rather similar results to the ones they got. I would just like Steve to verify this for me.
It is better to concentrate on errors of the main stream science rather than errors of science which in any case is ignored by the main stream science.
Interesting John Hunter, you are now clearly implying that the McIntyre test has validity, and hence you accept that MBH98 & 99 are, like S&B, rubbish. Glad that you can finally come around to accepting the compelling evidence Steve has accumulated on this score. Care to correct Tim Lambert and some of the other posters on Realclimate now ? Every bit of help in establishing a bit of clarity in the field would be welcome.
Concerning influences on tree ring growth, it is interesting that in the same year as MBH98 a shorter and simpler paper “A statistical study of the relationship between the solar cycle length and tree-ring index values.” was published by Zhou, Keqian and C. J. Butler. In their Abstract they say:
“We have determined the correlation coefficient between tree-ring index values and the sunspot cycle length for 69 tree-ring data sets from around the world of greater than 594 years duration. A matrix of correlation coefficients is formed with varying delay and smoothing parameters. Similar matrices, formed from the same data, but randomly scrambled, provide a control against which we can draw conclusions about the influence of the solar cycle length on climate with a reasonable degree of confidence. We find that the data confirm an association between the sunspot cycle length and climate with a negative maximum correlation coefficient for 80% of the data sets considered. This implies that wider tree-rings (i.e. more optimum growth conditions) are associated with shorter sunspot cycles. Secondly, we find that the climatic effect of the solar cycle length is smoothed and delayed by several decades and the degree of smoothing and the amount of the delay is dependent on the elevation and the geographical location of the trees employed.”
It does seem that if you look long enough and are selective of the data and methodology you can prove anything from tree rings. These authors do at least tell us that 20% of the series did not show the correlation.
Ed Snack (#63): You say “you are now clearly implying that the McIntyre test has validity, and hence you accept that MBH98 & 99 are, like S&B, rubbish”.
I think, in the trade, that’s called a non sequitur. The analyses are completely different. In Soon et al., the problem is blindingly obvious. The cases of Mann et al. are far more complicated and any “error” non-obvious.
Which makes me wonder why Steve didn’t attack the easiest problems (e.g. Soon et al.) first, before moving on to the hard stuff.
Re #61 and Hunter’s request that I "verify [a result] for him". While I have earned my living in humble occupations like business and Hunter undoubtedly sits at high table, I am not a tradesman to respond to his beck and call. I’m very busy on my own projects; my statistical skills are hardly unique; John H, get someone else to do it. Pay them if you have to. Please stop your repetitive hectoring on this.
Re #59- Ed, dendrochronologists emphatically do use ring widths. This is their primary measurement. All the bristlecone series are ring width. Schweingruber collected lots of latewood maximum density measurements, but is one of the few people to do that. A couple of Briffa et al [Jones, Schweingruber] ‘s widely used reconstructions (Tornetrask, Polar Urals) are based on Schweingruber’s desnity measurements. There seemed to be a hope about 15 years ago that this would be a magic bullet, since ring widths had so many issues, but I don’t get the sense that they still feel that this is a magic bullet. If you look at my posts on Tornetrask, Briffa “corrects” the downward 20th century density index by the ring width index.
John, I’m about 50% on your side. But I still think that even if McI (or Mann) are biased in the overall argument that their relevant contributions on specific issues either stands or falls on it’s own. IOW, I don’t think that GW advocates should try to build up a case for Mann (to defend themselves from the people trying to drive anti-GW) if the actual study by Mann is flawed. And vice versa.
RE#54
So they should get a statistician involved and give him/her some ackowledgement (possibly even co-authorship, depending on the amount of contribution). How hard is that?
RE#65
Seems to me that at least one Soon et al publication has been scoured over and essentially ripped to shreds already. Why would Steve (or anyone else) pile-on? But clearly no one had attempted to replicate MBH98 as Steve quickly found-out.
re #60
Well, your whole post amounts to saying “My political opponents want to diss Mann so I’m going to push for dissing them and therefore if Steve refuses to go along it’s proof all his activity is political.” Now the reverse point (which has been pointed out to you numerous times), is just as valid, if you refuse to diss Mann it’s proof all your actions are political.
And, of course, your red herring deftly avoids actually addressing my points. Or are you conceding your lack of logical rigor?
#57 & #59: The point is that tree growth is a multi-variable phenomenon and extracting one variable, the temperature, requires full knowledge of all the others. Whether one uses ring density instead of ring width makes no difference to this argument, just substitute one for the other in the equation. The strength of any fertilizer effect by CO2 changes the function G but it does not change the argument either. The conclusion is still the same, there have to be measurements of multiple independent variables for each year of tree growth to be able to extract the temperature. A single measurement, such as ring width or density, will not allow you to compute a temperature.
In fairness to the dendrochronologists, for ones interested in temperature, they will try to sample trees at temperature stressed sites e.g. high altitudes and high latitudes. They will then try to do linear regressions to monthly temperature at the nearest weather station. Unfortunately, this often seems to degenerate into data mining e.g. Jacoby. He starts out picking sites beleived to be temperature-stressed. But then he only archives the 10 most "temperature sensitive" sites i.e. the ones with upward 20th century chronologies so it becomes tautological.
The other huge problem is that the tree growth is even more autocorrelated than temperature. Some of the dendro people in the 1980s used to try to remove the autocorrelation and work with residuals, but then they didn’t have any long-term climate material. So in the 1990s, they’ve tended to leave in the autocorrelation which they describe as "conservative". So when you see the word "conservative" in tree ring literature, instantly think about huge autocorrelation. As a play on words, are political "conservatives" also "auto" correlated – in both senses?
The CO2 fertiliser effect may have been greatly overestimated. Recent results coming from FACE experiments seem to indicate that when you increase CO2 there is a rapid increase in growth, but this soon declines and more or less disappears.
Steve
If the dendrochronologists are looking for sites which are good temp proxies and the local temperature (thermometer)records show warming then by default they will be picking sites which show 20th C warming. Do they ignore sites where the local temperature has not increased or fallen? If they ignore temperature stressed sites where there is no correlation between temp and tree growth that seems reasonable.
re #0
Both senses? You mean they like both stock cars and Indy cars? Wal’ sure, there’s lots of correlation between the two groups.
re #71 Paul,
You’re missing the point. This process of ignoring non matching sites is precisely what data mining is about. To have a chance of showing legitimate analysis you have to include both the sites which agree with your premise and those which disagree. If there are sites which according to your apriori criteria are ‘stressed’ and they don’t show correlation to temperature, then that’s a good sign that there’s something else, possibly important, going on. The data, at the very least, should be archived for others to look at who may not have quite the same take on things as the authors.
This has been an extremely enlightening discussion for me. It is staggering to see the weakness of the models for these temperature proxies. Unlike what John Hunter wants to say the critics are concluding, it does not disprove AGW. But it literally screams that there is no there there. Yet. In reading M&M’s work, the point that there is little or no statistical validity to the proxies, the game is over. Tree ring width and density without dealing with moisture cannot be used as it currently is as a proxy for temperature. Period. A better model is needed. And my own paltry research has convinced me that the Bristlecone pine people need to account for the millions of tons of fertilizer that have been deposited on these trees since 1913 by the massive draining of the nearby Owens Lake.
But the MBH believers will never be convinced. Just as Keynesians from the 1960s never accepted that their theories were incomplete. John Kenneth Galbraith, at age 95 or whatever, is still spouting the same line he did 50 years ago. Those who are proved wrong just fade away and are not heard from as much. Crescat scientia!
Steve, thanks for the correction. I was thinking of the Polar Urals series where, as you point out, Briffa uses latewood density initially. Based on the comments of, I believe, Briffa about the superiority of the latewood density over simple width measurements, I did assume that new studies would tend to use that measure preferentially. Hence my interest in references showing just how and why it was supposed to be superior.
John, not a non-sequitur at all. You appear to accept the “test with noise data” criteria, that has been applied to MBH, with results every bit as conclusive as you claim for S&B, hence I am quite justified in assuming that you should believe MBH to be fatally flawed. You may wish to weasel away from that conclusion, but it is hard for you to sustain and remain a credible critic.
The flaws in MBH may or may not be more egregious than those in S&B (which, BTW, I am quite unfamiliar with), but there are almost certainly more errors in MBH, and those errors are ferociously defended by a large number of people who refuse to look to see them.
Dave (#69): You ask “are you conceding your lack of logical rigor?”.
Dave, one of the more tiresome aspects of the contrarian canon is the resort to “logical fallacies” and the like. We do not live in a binary world, where something is either true or not true. There are many shades of truth, clouded by uncertainty. You can play your erudite “logical” games, but leave me out of it.
“There are many shades of truth, clouded by uncertainty.”
So pi isn’t know to millions of digits, the sun doesn’t rise in the east in the mornings (usual caveats applying), and AGW is only one of many ‘truths’ concerning climate change. [My favorite ‘truth’ is that the world is warming slightly from the heat generated by butterflies beating their wings in frustration for being blamed for hurricanes and suchnot.]
One thing for sure, however. To many regulars here you appear to be wasting Steve’s time on trivial things. Therefore your distortions and lack of logic aren’t going to be overlooked. “And,” as Edith Ann would say, “that’s the truth!”
Re #77, John Hunter.
Oh, right. You’re an arts graduate.
fFreddy (#80): You say “Oh, right. You’re an arts graduate.”.
Is that just a gratuitous slur on arts graduates or an attempt at a serious comment? If the latter, then please explain (if Steve hasn’t sensibly censored your comment before then).
Re: #80 fFreddy, I would have said “post-modernist”
moberg, ambio daly?
The TCO cherry-popping (not picking) thread is still missing my initial comment…
Reading back over this is interesting. I think there is a tendancy for Steve to confound two changes when looking at an issue. For instance centering/standard deviation dividing. In this case, centering/preisendorfer n’ing. All are arguable, but they should be clearly layed out. For instance, here it would have been better to make the Preisendorfer n arguments in advance or just to show the effects in more of a full factorial BC manner.
It’s not even clear to me if Steve understood the decision on Preisendorfer “n”ing. If not, fine admit it. If so, why wasn’t this made more prominent in the paper?
Also, a while ago Ross said that the different PCs were not wieghted differently (PC1 more then PC2) within the reconstruction (other then the yes/no cutoff of course). One of our new commenters did talk about weighting. Which is most appropriate? Weighting or not?
One Trackback
[…] components who was cited in MBH98, provided the following advice in his text- advice quoted at CA here: The null hypothesis of a dominant variance selection rule [such as Rule N] says that Z is […]