Julien Emile-Geay has made many forceful criticisms of the Loehle reconstruction. For example, he says:
Relationship of each proxy to *local* temperature is not even discussed. We are just shoved a list of references (hey Craig , have you heard of tables ? They are a great means that scientists use to convey information clearly). How is it that tree-rings are seen as the penultimate antichrist but that, to take but one example, d18O of speleothem calcite (e.g. Holmgren et al., 1999) is a flawless paleothermometer ? Shouldn’t one discuss one by one, and with great care, the pros and cons of each proxy as a temperature recorder ? d180 in forams is a notoriously flawed temperature proxy, is being replaced by Mg/Ca where possible, and one can only surmise why the Keigwin[1996] series is here.. Is it just a form of screening for proxies with a low “hockey-stick index” (in McIntyre parlance) ?
Emile-Geay notably did not compare Loehle’s proxy selection with those in the canonical studies. Here I’d like to observe the overlap between Moberg’s low-frequency network and Loehle’s network. Because of this overlap, such criticisms apply a fortiori to Moberg et al, about which Emile-Geay has, to my knowledge, has been entirely silent.
Moberg used 11 proxies in his low-frequency network (retaining tree rings only for high-frequency information.) The 11 low-frequency proxies are listed here. Here is a plot of the 11 Moberg series.
Attentive readers will see that no fewer than 8 of the 11 series are used in the Loehle network. The only exceptions are the Arabian Sea G Bulloides (discussed a lot here) which is uncalibrated to local temperature (actually having an inverse relation!), the Agassiz melt percentage also uncalibrated to local temperature and the Lauritzen speleothem, digital information on which was not made available at the time.
Through a Materials Complaint to Nature, I am now in possession of a digital version of the Lauritzen data (without restriction) and have posted it up at www.climateaudit.org/data/moberg/
17 of the 18 series used by Loehle used temperature calibration by the original authors. There was only one exception: the Holmgren speleothem and this one exception was seized on by Emile-Geay and criticized. I think that there are definite advantages for a Loehle-type reconstruction in maintaining a policy of using temperature reconstructions rather than uncalibrated proxies. Even single-proxy temperature reconstructions have their own problems, but at least the attempt at constructing a temperature reconstruction by the original authors avoids some defects arising in uncalibrated proxies. I’ve objected to the use of uncalibrated series in the past (e.g. the Moberg G Bulloides series) and believe that the same objection applies against the use of the Holmgren speleothem. Obviously there are occasions in which one has to broaden the inclusion criterion, but that doesn’t apply here.
If Emile-Geay believes that multiproxy authors should not use uncalibrated proxy series such as the Homgren speleothem to which he objected, then he must surely also criticize the use of the even more objectionable Arabian Sea G Bulloides series by Moberg and by Juckes. If he thinks that the consideration of the Loehle series has been inadequate, then he must also conclude that the consideration by Moberg was also inadeuqate. However, to my knowledge, he has been thus far silent in his criticisms of Moberg.
And of course there’s an even more powerful example of the use of a proxy uncalibrated to local temperature: Mann’s PC1. This was never calibrated against local temperature. If it were, the implied MBH98-99 calibration would entail an implausible ice age in California, a point that I’ve often made. The MBH PC1 is able to have such a large impact on reconstructions precisely because it is uncalibrated and has such a negative value in its early stages.
So I welcome Emile-Geay’s suggestion that authors should examine “one by one, and with great care, the pros and cons of each proxy as a temperature recorder”. Indeed, much of the volume of ClimateAudit has been devoted precisely to such examination – an examination which has been notoriously absent in the canonical multiproxy studies. I hope that Emile-Geay is consistent in his policies and will be as unsparing in his criticism of Moberg et al, Mann and Jones and MBH, and perhaps even come to appreciate the texture of the post history at Climate Audit in which we have tried to carefully examine the pros and cons of each temperature proxy – not a small job.
Climate Audit 
45 Comments
Some remarks to one of the used series. The Mangini d18O Speleothem record (with a nice representation of the MWP).
I think this record has very serious problems and the “calibration” of the original authors hardly merit this name.
1) The calibration of this European (Austrian) high-altitude cave record is done using 3 points: Today, 1900, Little Ice Age. For the LIA the authors take an assumed mean European value of what the LIA is supposed to be averaged over the entire continent. This is no calibration in the usual sense of the word.
2) The 18O cabonate record is INVERSELY related to temperature. What does this mean? After subtracting the direct fractionation effects of calcite formation there is a negative relationship between the 18O of dropwater and Temperature. Warmer temperature, more fractionation ie more negative drop water. The drop water is of course mainly controlled by meteorological water and therefore this relations ship is the inverse of what you find everywhere in Europe or of what is used interpreting high latitude ice cores. It is simply an unphysical relationship. Mangini et al argue that infact the drop water is not at all controlled by the isotopic composition of the water falling in the catchment of the cave but is controlled by unknown and unmeasured soil processes. No verification, no explanation whatsoever.
Sorry for the question, but I must have missed something. Steve McIntyre stated eariler that the Sargasso proxy needed to be “put out to pasture” and JEG stated “..and one can only surmise why the Keigwin[1996] series is here.. Is it just a form of screening for proxies with a low hockey-stick index (in McIntyre parlance) ?”
Is this proxy mutually agreed to be flawed?
Steve: It’s not that it’s “flawed”. It’s just that it’s low resolution – look at the posts http://www.climateaudit.org/?p=2306 and
http://www.climateaudit.org/?p=2300 about the Richey results in the Gulf of Mexico or David Black’s results in Cariaco – all done in more detail. Keigwin’s series needs to be re-visited too achieve equivalent resolution to the Richey results and, if the core lacks modern ersolution, drill a new box core to get modern resolution. This is not a criticism of Keigwin who’s excellent – just that the demands being placed on this data, in part because he did it in the first place – are now greater than in 1996.
The holmgren data are shown as temperature in the publication I cite and the author sent me temperature calibrated data. The problem is the lack of archiving of the calibration and the data. I have emailed the author.
If “tree-rings are seen as the penultimate antichrist”, then what, may I ask, is the ultimate?
In Steve’s and my E&E-05 paper, Section 5.3, we respond to a criticism from Mann over our concern that his proxies had no correlation to local temperatures (on average, a negative correlation, as we later showed in our GRL reply to von Storch and Zorita). Mann’s complaint was that we OBVIOUSLY don’t understand his marvelous methodology, which is too sophisticated to be concerned with such trifles.
We doubt the authors really believe the idea of a temperature proxy exhibiting no relationship to local temperature makes much sense.
The ‘instrumental training patterns’ are temperature PCs. On this view, as long as a proxy correlates in any degree with any pattern in any temperature PC anywhere in the world, it is in. Except for the networks that have no correlation even to the PCs; they get used anyway.
Newbies to CA who feel inspired to mount a rhetorical attack on authors who use proxy indicators with no correlation to local temperature have their work cut out for them: every user of Mann’s PC1 awaits your attention. Why fuss about a new paper in E&E? Go for the big fish: see the listing at http://www.climateaudit.org/?p=2245.
5
LOL! Nice turn of phrase….
Ross M
We doubt the authors really believe the idea of a temperature proxy exhibiting no relationship to local temperature makes much sense
To be pedantic, and b/c I don’t know enough about the relation between the variables (or the biological/physial) process involved, why the limitation to a linear response?
Is it possible to determine (after hypothesizing based on some a priori) a non-linear relationships b/n proxy and temperature?
An obvious candidate would be tree-ring width. The area of the growth might a linear relationship, implying thata the width the is proportional to inverse of the square of temperature (or temp difference from some constant)
Steve,
The link you used 2nd para below the graphs doesn’t work.
http://data.climateaudit.org/data/moberg/
is correct but your hyperlink puts out
http://www.climateaudit.org/www.climateaudit.org/data/
Keep up the good work!
Jimmy, the indentation might have obscured what was being quoted from what. The assumption of a linear response to temperature is in the original MBH98 method, and is critical to all the tree-ring calibration work. If the response is nonlinear, e.g. quadratic, then there may be non-unique temperatures associated with ring width/density values. The discussion of the Divergence Problem has raised the possibility of nonlinearity, which needs to be pursued further.
What we were arguing was that the proxy records from various locations showed no pattern of correlation to the local temperatures. Their response was that it doesn’t have to: it could be correlated with weighted averages of temperatures on the other side of the world and still be a valid proxy (or words to that effect). My view, admittedly that of a non-expert, is that relatively few tree species watch the Weather Channel, so a tree is typically not going to know, much less care, what the temperature is on the far side of the world, unless the faraway temperature change happens to teleconnect to its local temperature conditions. But in that case the correlation will be local as well as distant. Distant correlation + local non-correlation = opportunistic data mining.
Reference Steve’s comment on #2: I am sorry to hear that. I thought his approach was very responsible and his explanation to why thinner samples were not tested quite reasonable. Perhaps in the not too distant future his pragmatic approach may be adapted people trying to glean too much from too little.
As i recall, in Moberg’s non-tree-ring proxy, they developed a graph with mixed temp and non-temp indicators which had a fairly high amplitude, and then SCALED it to have unit variance which reduced the amplitude quite a bit. I asked Moberg about this but he never replied.
With regard to “teleconnection”, is it conceivable that tree rings could be a proxy for “global” rather than local temperature if both tree rings and global temps are responding to an independent stimulus? For example, if both tree rings and global temps respond in the same way to insolation, it could be that local temps are irrelevant; they could be due to regional fluctuations(ocean currents etc). But local tree rings could be a global proxy because they respond to insolation not local temperature.
Having said that, it would be very hard to prove, and the burden of proof would be on those proposing such a system.
Braddles, you hit the nail on the head; “With regard to teleconnection, is it conceivable that tree rings could be a proxy for global rather than local temperature if both tree rings and global temps are responding to an independent stimulus? For example, if both tree rings and global temps respond in the same way to insolation, it could be that local temps are irrelevant; they could be due to regional fluctuations(ocean currents etc). But local tree rings could be a global proxy because they respond to insolation not local temperature.
Having said that, it would be very hard to prove, and the burden of proof would be on those proposing such a system.”
Indeed it would.
Why don’t grapes, or palm trees, for that matter grow in Iceland? Is it because there aren’t the relevant “teleconnections”? Or maybe, just maybe (silly thought really- and I’m sure some expert dendroclimatologist could tell you otherwise) it is the local climate that is the determining factor.
Ref 14: Every morning the palms that were stunted by Hurricane Wilma ask me why they can’t play the proxy game. I just don’t know what to tell them.
I am perhaps the epitome of Mr “general public” and have been lurking here for about 6 monthes, and so normally have nothing to add as I am learning. I cannot resist commenting from the 70s T.V. show with Mr. T in it “I love it when a plan comes together” Pre planneded on not, to do a study that mimics some of the same flaws as Mann,yet shows contrary results, IE a MWP and LIA,invite critics, fix the problems, IE establish true best estimate tempeture calabrations to the proxies, reject the questionable proxies, do the r2 stats, reveal the code, etc. In short do everything in a matter of a few weeks or monthes that the hockey team has refused to do for years, and then show the results, which I speculate will still show a MWP and LIA, is brilliant. And guess what, there is nothing the proponets of drastic measures to AGW can do about this. Their best response would have been none at all, but to late for that. It is said that only truth can swim upstream, and I enjoy watching it happen. So beware proponents of AGW, “anything you say can and will be used against you”
Mann, Climate Over the Past Two Millennia (2007)
Re #17, “CFR methods do not require that a proxy indicator used in the reconstruction exhibit any local correlation with the climate field of interest, but instead make use of both local and nonlocal information by relating predictors (i.e., the long-term proxy climate data) to the temporal variations in the large-scale patterns of the spatial field.”
Anyone who believes in Mann’s climate reconstructions should read and re-read the above statement until the obvious finally sinks in.
That is “what a load of male bovine excrement”.
braddles,
Hey, that’s a plausible idea.
1) Trees respond to solar – we can use global temperature directly, as global temperature responds to solar as well. But then, we would have a divergence problem after 1980 or so.
2) Trees respond to CO2 – well-mixed gas, we can use global temperature directly, as global temperature responds to CO2 as well. But if we state a priori that A-CO2 effect is much greater than any natural factor, we don’t need hockey stick studies, because we know the answer.
3) Something else that Mann means in #17 quote, which I don’t understand. But I don’t have to, as I’m not climate pro. However, I’d like to learn it.
Fishing and proxies. I just noticed that Chesapeake bay is included as a temperature proxie. Did you know that the entire ecosystem of Chesapeake bay has been distorted by European/Americans since the 1740’s? Originially the water was so clear you could see through to the bottom, then all the large clams, which filtered the water, were removed and eaten. Then the bottom was dredged by fishing boats.
I can not think of a body of water anywhere, except the North sea and its esturies, that has been more impacted by human being over the last 150 years, than Chesapeake bay.
As I understand it, the Arabian sea has also been over fished in the last 25 years and greatly changed.
UC/braddles,
I don’t follow the argument here. Tree growth is affected my many factors, some not directly related to climate at all. For instance I have/had 9 mature trees in my back garden, all planted to my knowledge at the same time about 90years ago. The differences between them are more striking than the similarities. Disease has claimed 3 of them, there are wide variations in ‘genetic health’ of the stock, the largest tree being the weakest. Branch drop, changes in local water, wind, shade, fungus, bird and insect populations due to other species arriving or dying.
To suggest that something that we haven’t monitored on at least hourly basis since it’s planting can somehow be an accurate indicator is hard to swallow.
Consider :
Trees are a passive sensor, not always ‘on’ and there are no exact quantified metrics for sensor variations (read growth) under all climatic, georaphic, biological and ‘local’ (see above comments) conditions.
For example engineering metrics for certain species under certain conditions exists but the margins of error are huge.
Man (meterologists etc.) are active sensors, using calibrated instruments, non-local measurements, historical precident and real ‘teleconnections’, yet as Anthony Watts and crew finds regularly that they are not that accurate localy never mind globally.
JEG has posted a reply to Loehle at his blog.
===========================
Interesting.
Link for JEG’s reply to Loehle:
http://thatstrangeweather.blogspot.com/2007/11/open-letter-to-craig-loehle.html
I’m not sure how JEG can nearly completely dismiss Loehle 2007 and yet still honestly argue at that MBH98 is a “valid contribution to paleoclimatology”. Thanks to MM, MBH98 has been shown to be deeply flawed on numerous levels — so badly in fact that results are pre-ordained regardless of input.
That said, as a lurker I appreciate JEG taking the time to engage with Loehle and CA’s core smart guys. The discussions are much more illuminating than the complete and utter silence from the RC folks regarding the important statistical issues raised by Steve.
Dev, it’s cuz he’s a newbie. Give him time.
===========================
Dang! I hate blogger. I’m not going to sign into blogger to leave comments at JEG’s blog! 🙂
I think a collaboration between the Lohle and JEG is something to be strongly hoped for. Part of the reason is the issue JEG broached of whether or not an approach or scientific judgment is being made in good faith. There are so many accusations of bad faith everywhere. In reality, there is probably less bad faith on ‘the other side’ than people sometimes wish to suspect. (There is probably the ordinary amount of human frailty & blindness all around.)
With luck, the two of them will achieve some sort of understanding that permits them to work together– it would probably work out well.
Re #24, Dev
The answer lies in Jeg’s post here, where he is rude about Loehle’s simple averaging approach, then says :
In short, JE hasn’t a clue what MBH is doing, but is willing to assume it must be frightfully clever.
RE 26.
Same here. I was gunna leave a thankyou note to JEG…
RE 28.
http://www.blogger.com/comment.g?blogID=2630818366822533683&postID=2246097806869987561&isPopup=true
Comments on this blog are restricted to team members.
🙂 🙂 🙂
I see that JEG still clings to the his pseudoscience claim against Loehle, saying that his definition is “any text or speech that wears the attributes of science while lacking the necessary rigor in the essential fulcrums of its reasoning.”
Wikipedia has a slightly different take…one that is more in line with what I think of when I hear the term:
and
and finally,
The examples they cite are astrology, superstition and the like. I think that JEG’s claim is a little over the top.
RE 30. JEG occupies the rhetorical ground between competence and camp.
he needs to tone it down or amp it up.
@steve mosher– Both competence or camp at a blog require open comments. Camp requires prompter responses.
Nothing wrong with JEG’s definition. It’s his unwillingness to apply his public judgements evenly across the board that is problematic.
It’s not what most people usually use the term pseudo-science for.
This is a large part of the issue; what exactly is being said. If I tell you
“A watt is equal to one joule of energy per second.” or “If 1 volt of potential difference is applied to a resistive load, and a current of 1 ampere flows, then 1 watt of power is dissipated.” or “1 kW-h is when something uses 3.6 MegaJoules.” or “One joule is the work done, or energy expended, by a force of one newton moving an object one meter along the direction of the force.” or “One joule is the work required to move an electric charge of one coulomb through an electrical potential difference of one volt; or one coulomb volt, with the symbol C·V.” or “1 joule = 1 newton-metre = 1 watt-second” or “1 joule = 2.7778 ×10−7 kilowatt-hour” How can you argue with any? It’s all various ways to say the same basic thing from various points of reference for various reasons. Why pick one over the other? Who cares.
But this isn’t a case of that; how exactly are we supposed to understand something if you use “pseudo-science” in a way that’s non-standard?
How exactly do you expect somebody to parse “This paper is BS, garbage. It’s illogical, pseudo-science babbling and it proves you’re an idiot and don’t know what you’re doing.”? What exactly is somebody using that type of language (or implying it either accidently or purposely) trying to accomplish?
So for me, this is what it boils down to. Either you’re trying to elicit a response based upon the misapplication of terms, or you don’t know how to write using the correct terminology so others understand you. If you don’t want things to be infered a certain way, be clear.
“I ate a piano.” Then after a hundred posts, it turns out “ate” means “played” and “piano” means “guitar”. Blah.
OK, here’s Merriam-Webster:
JEG’s answer to his own question:
is yes.
I grant that the article needs to have a more academic tone, and be fleshed out in various details (and hopefully this will happen), but to use the word “psuedoscience” is simply an attack.
Attacks such as this are useful to the Greek Choir that AGW has.
Here’s a quote on Scholarly Activism by Henry Farrell of GWU on the blog Crooked Timber
(Italics in original) So everyone spends time/bandwidth arguing about words and responding to attacks, and not talking about scientific issues.
Hate to break it to you optimists, but here are the money quotes:
and (emphasis mine):
Why is Dr. Loehle (and Climate Audit, for that matter) so obsessed with this twenty-something who is close-minded about AGW, and not nearly as competent in statistics as half a dozen or so CA principals or regulars? If you think you’re going to “win him over,” you’re just being naive.
Seems to be a good fit to some of Mann’s work.
For a moment, I thought the object of this discussion was sod …
Mike B,
Your motive hunting is leading you astray. JEG appears by all counts to be a free-thinking person. No one is “obsessed” with JEG. His perspective has value. Calling him a 20-something the way you did is an ad hominem attack. Are you going to dismiss Kristen Bynres because of her age? His “objective” (which you seem to mistrust) appears to be robust estimation of confidence intervals. This is a serious flaw in the Loehle paper, and I agree with his judgement on that count. His comment that Loehle (2007) is “pseudoscience” is about as unfair as labelling MBH9x “pseudoscience” (Although the papers have very different problems, they do share a common problem: confidence interval estimation.) Using that term is a rookie mistake that he may regret, but I doubt it. Loehle certainly isn’t all broken up about it. Finally, in what position are you to judge his statistical competence? Tell us some things that he’s doing wrong and how he could fix them.
Furthermore, JEG appears to be a man of action. Rather than whine about Loehle he is looking into Monte Carlo bootstrap methods as a way of calculating robust confidence intervals. If he’s so agenda-driven, why would he waste his time doing such a laborious, yet honest calculation? Skeptics should welcome his efforts to clarify the amount of uncertainty on the MWP. Skeptics should welcome him bringing statistical rigor to the Mann lab. Moreover he appears to be accepting of the principles of disclosure and due diligence.
RE 26 Lucia:
Think the irony is intentional?
JEG wants answers. But to get answers, he needs data. And to get data, he needs collaborations. Power to the man looking for answers.
@pp — 40
It appears some snark about JEG comments only being open to the “team” and my grousing about blogger comments may be a bit misplaced.
It appears JEG has posts open for comments generally, and a few are restricted. That’s a bit unusual for blogs– but it’s also fine. I happened to want to make what I thought an encouraging comment on the open letter, but had no particular comment on the other articles.
In cases were comments are truly open, it appears JEG also chose the option that permits people to log in without logging into their blogger account– so JEG is really and truly open for comments most of the time. (That said: I still don’t like blogger’s interface, but that’s hardly JEG’s fault. Blogger is free. So, even though I would love JEG to spend the $50 or so a year of his money to buy a domain name run a WordPress blog for my convenience, using blogger is hardly nefarious.)
I would second Bender’s comment regarding JEG. It appears JEG’s work is on the right track: clearly state a parameterized error model for the proxies then use Monte Carlo to estimate confidence intervals. Then people can shift their efforts to more productive arguments about the sources of error, the appropriateness of particular error models, and their parameterizations. That smells like progress to me.
re32. yes. I lean toward camp, but my slow response here wouldnt show that
Translation: Those proxies that show what I’m looking for count more than those that don’t.
One Trackback
[…] http://www.climateaudit.org/?p=2389 […]