No one will accuse me of arguing that Dulan junipers are a proxy for temperature. But twq has been to Dulan and says that they are. So let’s take twq’s assertion at face value and, a la O.J. Simpson, let’s reluctantly explore what happens if Dulan junipers were a temperature proxy as twq says. A post put up under duress, so to speak.
First here is a plot of the values of the Yang et al 2002 Dulan chronology. Again, I do not claim that this is a temperature reconstruction, but, you know, if it is, on twq’s say-so, I will reluctantly include it in reconstructions in place of (say) the Dunde data set. I noticed that there is a striking positive anomaly around the year 1000 – I wonder if anyone can come up with a good name for this period. Maybe Jonathan Overpeck can think of something.
Figure 1. Dulan chronology from Yang et al 2002 data (emailed by Dr Yang)
Just so no one thinks that Yang et al 2002 is the only study to observe this strange phenomenon, here’s an excerpt from Figure 3 of Zhang et al 2003. The resemblance of the two chronologies is hardly accidental, since Zhang et al 2003 use 55 cores from Kang et al 1997, which was relied upon in the Dulan chronology used in Yang et al 2002.
Excerpt from Figure 3 of Zhang et al 2003. Original caption: The 2326-year ring-width chronology of Sabina przewalskii Kom. and the number of samples for each year in Dulan area of eastern Qinghai-Tibetan Plateau. The smoothed curve superimposed on the chronology is derived after filtering to emphasize long-term fluctuations (the filter is a cubic spline passing 50% of the variance in a sine function with a wavelength of 64 years).
So, reluctant as I am to consider these ring width chronologies as temperature proxies, since we’re so easy to get along with here at climateaudit, we’ll give twq the benefit of the doubt – this time only.
Now twq thought that he’d try to compare the low-frequency variations of Dulan tree rings to the low-frequency Dunde variations. To assist him in this process, I’ve collated various Dunde versions below over the same period as shown for Dulan juniper chronologies above. The IPCC SPM told us that “additional data” from mulitple indicators showed “coherent behavior”; Lonnie Thompson’s ice core data is one of the most important such indicators, featured prominently in Inconvenient Truth. The Dunde ice core chronology (chronologies?) represents one of the triumphs of modern climate science. In a field fraught with dendroclimatological uncertainty, here is a data set where one can find firm ground under one’s feet so to speak.
Indeed, if one uses the Dunde data for statistical analysis, multiple linear regression takes on a whole new meaning, as here twq has the opportunity to perform multiple linear regressions on just one data set and, following dendro precedent, pick the version that he likes the best. Western dendroclimatologists call this sort of procedure “robust” and “rigorous”; I’m sure that there is a mot juste in Chinese.
Dunde Versions – legend as in Dunde: Will the Real Slim Shady Please Stand Up?
Climate Audit 
45 Comments
As I note in my Solar Science blog, the three major solar minima of the last thousand years were:
– The Spörer Minimum (1420 to 1570)
– The Maunder Minimum (1645 to 1710)
– The Dalton Minimum (1790 to 1810)
Now I’m not going to endorse the Dulan chronology as a temperature proxy either, but it does record solar activity pretty well.
It is still not proved that the ring widths in Qilianshan is a precipitation proxy as compared to temperture change. I have repeated several times that before you give me the in-site monitoring data or real phisiology modelling data I don’t think it as a precipitation proxy. You have turned around many times for a long time for this proxy. But every time it seems to not be much help for its explanation of growing mechanism. Can we be patient about it? Otherwise I am afraid of more mispresentation or misinformation that will occurs again, like the beutiful picture “a tree in the desert”. Also I noted recently that you posted so many Dunde chronlogies in your blog but never communicate with related scientists. I guess there is similar misinformation. Why don’t these high-respected scientists who make truely delve into this science stand up? One reason is that you have made too much irrelative remarks on the topic.
I have noted that the replication is very low during the period 9-12th centuries in zhang’s chronology.
Steve, I noted that you consider ring widths to be precipitation proxy based on these descriptions about correlation between precipitation and ring widths during the 50 years. I can not find any other evidence. Like many others indicated, correlation does not mean causation. My point is that both temperature and precipitation have definitely exerted thier effects on the ring withs, and you can not make final conclusion at present. We have to wait for more evidence including both phiological monitoring and modelling data acquired. I have spent much time for my point, but still you can not get it. Again, we try to suggest that we must be careful for the precipitation explanation of Qilianshan ring widths. What do you say? I wait for your reply to the point, not other irrelative posts.
From the NAS 1995 thread:
For AGwers who prefer to look at trends over levels, and argue that the current warming trend is “unprecedented”.
Re: #4
That, twq, is point that Steve M has been driving home and with which many posters here agree. That point seems to get lost when your rather obvious dissapproval of Steve M and perhaps anyone else making remarks and critiquing the work of dendroclimatologists becames the main theme of your posts. Maybe its a cultural thing but we here in the West are very much accustomed to this process. Many of us thank are lucky stars that a postal clerk some hundred years ago decided to look at the physicists’ problem of relativity. You would do yourself and the discussion a lot of good by making specific points and quit the lecturing of Steve M — we all have been aware of your message on that issue for some time now.
twq, why does sample depth hit a minimum in the 9-12th century, but then start to rise again prior to the 8th century? It looks like the low number of samples occurs during a time when temperatures appear to be at unprecedented levels. Is this a coincidence? Why would samples be disappearing when temperatures are unusually high?
I requested sample data from Thompson in 2003 long before the existence of this blog and he refused. I requested it from NSF in 2004 before this blog and they refused. I requested it from SCience in 2005 and they refused. I requested it from NAS in 2006 and they refused. Paul Dennis of the University of East Anglia, nonplussed by the inconsistent Guliya versions, requested sample information and, to my knowledge got nowhere. The refusal has nothing to do with me and everything to do with Thompson.
Your allegations of “misinformation” are very tiresome. All Dunde versions come from versions used in peer reviewed literature.
Actually the most interesting thing to come out of this little exercise is the awareness that we have two distinct millennial-length proxies so close: Dunde dO18 and Dulan junipers. Where else is there an ice core chronology cheek to jowl with a 1000-year tree ring chronology.
You’d think that climate scientists would be exceedingly anxious to reconcile the different perspectives?
Ken, a little quibble — Einstein was a patent clerk.
TQW in #4 says “My point is that both temperature and precipitation have definitely exerted thier effects on the ring widths”. For all I konw he may be correct. However, as a plant physiologist, I also know that increased CO2 will have also played a considerable role in upregulating photosynthesis of these high altitude trees over the last 50 years by virtue of relaxation of substrate limitation.
This will make any comparison of 20th Century chronologies and pre industrial ones meaningless. Unless, that is, you can find some means of deconvolutng the CO2 effect.
Don’t any plant physiologists work with dendroclimatolists? I think not.
Re #11 FWIW, I think it’s the plant physiologists who are under-funded, not the dendroclimatologists.
re #12 very true!
RE: why does sample depth hit a minimum in the 9-12th century, but then start to rise again prior to the 8th century? It looks like the low number of samples occurs during a time when temperatures appear to be at unprecedented levels. Is this a coincidence? Why would samples be disappearing when temperatures are unusually high?
Bender, I guess that there is no more tree cores available for cross-dating of this period. The dating for tree-ring chronology are based on the correlation of ring widths between overlaps of different periods. For one period, both dead and living trees are included in the cross-dating computation. Accordingly, there are various sampling depths in different spans of one chronology. If you check the samping depths in different periods the 2000-year long dulan chronology, you will note that the sampling depths vary with time. In sheppard et al. (2004) chronology, you will note that many tree cores are included in the first centuries of the first millennium.
RE: Your allegations of “misinformation” are very tiresome.
I guess L. Thompson and others should have the same feeling as you. Are you sure that all the allegations by you are 100% correct? I don’t think so.
And the question is: why? Chance?
From Chinese literature, there are only two or three (I am not sure now) cores from living trees which were used for crossdating. No other long living trees are found, and I guess the main reason is that almost all the old trees are cut and used for construction due to the special political and economical reasons during the the 1950s and 1960s.
That wins my special “Deepest Gloss put on a statement” Prize for 2007.
to which allegation do you refer? Provide an example so that we can judge for ourselves.
I am not of the impression that Steve believes every sign of smoke is proof of fire. The blog is like a discussion. Steve is sifting through material, highlight things which require further inspection. But by talking about them publicly, openly allows other people to proffer explanations for what is going on. Thus: all of the points brought up are not intended to be sure issues and few of the points under discussion should yet be regarded as allegations.
Speaking for myself, Mann and others have come-off brash. The breathless dismissal of criticism such as takes place at real climate and else where has worn my patience thin. So I feel sympathetic to the sometimes excessive but amusing ribbing with which Steve laces his comments. It gives me the sense that he’s frustrated too.
Re:#15
twq, I find your postings interesting and helpful to the extent you provide new facts/references/interpretations, and I thank you for that.
However, I don’t understand your post #15. The repeated refusals of Thompson to release data have been public knowledge for some time (years). Perhaps you might try requesting this data yourself. If the unavailability then turns out not to be “100% correct,” you could then supply a pointer to at least some of the data in question, which would certainly be a boon to the field in general.
RE: Jon,
Thank you for your information. In my opinion, since the beutiful picture “a tree in the desert” which is regarded as growing in the Dulan region is unbelievable (see the below), why the blog still put it there? I have been confused with it. I have suggested several times, but no response. Rather, I got some criticism, and it is unfair to me. Please see the following information. In addition, it is still not clear that what (temperature or precipitation) Dulan ring widths represent at all because its true mechanisms is not solved till now. However, I feel much criticism comes to me again.
“A CA reader emailed Dr Shao requesting information on the provenance of the picture, once its association with Qilianshan juniper was questioned. Dr Shao replied to the CA reader:
The last picture on my talk is not a Qilian Juniper. A friend of my put that picture for me and I don’t know where it was taken exactly.
I emailed Dr Shao requesting further details. Dr Shao replied that her friend couldn’t remember where the picture came from. In my experience with geological reports, geologists are careful to identify the locations of each photograph and this is part of scientific documentation. To the extent that dendroclimatologists use photographs from undocumented locations, this is a practice that, in my opinion, should be discouraged. “
The source code to produce the dunde graphic, cinlduing the collation of the various versions is here http://data.climateaudit.org/scripts/thompson/dunde.collation.txt
The script should function automatically donwnloading either public versions of data or, where I’ve received data by email, I’ve archived the data online so that the script downloads it.
The reason that no one disputes the existence of these multiple versions is that it’s true.
#22. twq, at your suggestion, I removed the picture of the “tree in the desert” from the Dulan juniper thread and created a new thread entitled “Tree in the Desert” with no implication that this particular came from the Dulan area as opposed to say the Dunhuang area. The phenomenon of the “tree in the desert” in Dunhuang and elsewhere is interesting and some of our readers have posted up other interesting pictures. I’m getting a little mystified at your continued complaint unless it’s just for the sake of complaining.
As to your point:
again, I fail to see your complaint. If the matter isn’t solved, then we must all be in agreement that the data should not be used as a temperature proxy in multiproxy reconstructions until the matter is resolved.
Steve, you need to review the R script that you referenced in #23
The line near the end…
…clearly won’t work for anyone other than yourself.
RE #24,
Thank you. Let’s see what the truth is at all.
I havea question regarding tree ring width temperature proxies for dead tree parts found in ancient construction or preserved over the ages in some way. Would somebody be so kind as to explain how the particular tree parts involved are known to have come from stressed trees growing on the correct side of the slope at the right height (and responders) so as to be that special sort of tree that gives a temperature signal in the tree ring widths? I apologize for not knowing this obviously elementary method. It strikes me as possible that with so few ancient dead tree parts available for analysis, the selectivity might be slightly reduced from that exercised upon living tree samples.
Re:#27
Well, it should be obvious, Joe – dendroclimatologists have the unique advantage of being able to throw away data that doesn’t fit their models (at least, so I understand from the public comments of one leader in the field). If it fits, it’s “good”, if it doesn’t, it gets thrown away and nobody even knows you looked at it.
Sorry if that’s a little condescending and/or snarky, but I’m still gobsmacked that the data mining techniques used in that particular field with, apparently, not one of the True Believers questioning this, err, questionable practice, are seemingly accepted as more “evidence” of AGW. Like most regulars here, I’m still waiting for the a priori selection methods for trees (and, indeed, any and all other biological proxies) that should show a temperature signal and, of course, how one seperates this “signal” from all other growth factors. Probably a good thing I’m not holding my breath…
Yes, so people like Follow the Money and KevinUK can post their conspiracy theories, speaking of tiresome.
Boris, are you suggesting there is no carbon lobby? Or nuclear lobby. You would be tiresome if you weren’t so comical.
Dear Steve
A most fascinating blog. Thanks very much
I am trying to glean an understanding of how temperatures are read from tree rings. Bear with an amateur for a moment.
Core samples are taken from a number of trees and the ring widths are measured and documented.
A set of temperatures from a local met station are collated for the last say 100 years.
It is assumed that wider rings are indicative of hotter years and thinner with colder years and all the sets of tree rings are compared to the met temperatures.
If a number of the ring widths match the temperatures, it is assumed that these trees are accurate temperature indicators and all the other samples are discarded.
We can then backcast the temperatures of yore.
I hope I have this all wrong, because bejasus I could backcast the GDP of Bangladesh with this methodology
Regards Dennis George
Slightly OT questions (links to relevant papers are appreciated) :
How long overlap is needed for reliable cross-dating?
How often the dating fails (no correlation found, even though there is true overlap / correlation found but no overlap) ?
re #17 twq says “I guess the main reason is that almost all the old trees are cut and used for construction due to the special political and economical reasons during the the 1950s and 1960s”. Yeah, I’m with this one. These guys must have lugged all their cutting gear up to 3500m, knowing that there were trees with lovely long, thick straight trunks….
I would be more inclined to believe this (and happy to be proved wrong) by a picture of said trees.
Another reason maybe be that the junipers live not so long generally in this special area, or they are destructed by people at historical times, or that they did not find those living long trees which is located far from the towns or villages. Compared young trees, it is of course these millennium long old trees are rare.
#31,
I’m afraid that you have accurately summarized the method of the dendroclimatologists.
As near as I can figure, twq is declaring that tree rings are temperature proxies, except when they aren’t?
Dennis, ah, but the tree rings don’t have to correlate with local temperature records for them to be declared a temperature proxy, apparently.
If they don’t correlate with local temperatures, you can try the gridcell average. If that doesn’t work, you can try an adjacent gridcell. If that doesn’t work, you can correlate them to the “global average” temperature. If that works you can explain the correlation by “teleconnection”, or so they say.
My worry is that if you try enough different ring width/thickness measurement series and temperature record combinations, you’re bound to find SOMETHING that correlates reasonably well with your ring width, even if it’s by chance. However, IANADC so don’t ask me whether this is reasonable.
RE: #37
If they don’t correlate with local temperatures, you can try the gridcell average. If that doesn’t work, you can try an adjacent gridcell. If that doesn’t work, you can correlate them to the “global average” temperature. If that works you can explain the correlation by “teleconnection”, or so they say.
Can you give us these examples? Then we can learn from them. Thank you!
Re: #10
Charles, I realized this after I hit the submit button, but not to worry because as of this AM I have promoted AE from postal to patent clerk. My point remains the same since as they say it is all relative.
re #34 “Another reason maybe be that the junipers live not so long generally in this special area, or they are destructed by people at historical times, or that they did not find those living long trees which is located far from the towns or villages. Compared young trees, it is of course these millennium long old trees are rare.”
Now I’m much more inclined to believe this, but tqw also states in #14 that “Bender, I guess that there is no more tree cores available for cross-dating of this period. The dating for tree-ring chronology are based on the correlation of ring widths between overlaps of different periods. For one period, both dead and living trees are included in the cross-dating computation. Accordingly, there are various sampling depths in different spans of one chronology. If you check the samping depths in different periods the 2000-year long dulan chronology, you will note that the sampling depths vary with time. In sheppard et al. (2004) chronology, you will note that many tree cores are included in the first centuries of the first millennium.”
So why is there still this deficit of chronologies in the 9-12th century compared with times both before and after?
twq : my statements were based on articles by Steve McIntyre that I read on this web site. Some of them are here and here. More related articles here and here. I can’t find the post any more which talked about correlating the proxies with temperature data from an adjacent gridcell, although I did find it earlier today. It’s here on the site somewhere.
bender,
I’m sure someone will accuse twq of fraud at any moment now. Comical indeed.
RE: #1 – I wonder what happens to the semi persistent semi tropical Central Asian High during minima? That’s not only what drives the Asian monsoon (e.g. by heating the interior) it also is what blocks the cold fronts. Only the area within a few hundred miles of the Pacific escapes its influence under “normal” / current conditions. But is that true when there is a minimum?
Re: #11
Don Keiller, as a plant physiologist could you guide us to any sources that would inform us of the effort that dendroclimatologists receive or request from plant physiologists. It is a question I should put to twq, but I fear he is too involved lawyering the “tree in the desert” case.
Twq and other dendrochronologists posting here have all made indirect references to the need for better understanding the growth models for trees, but I see no evidence that they are making progress or even trying. What bothers me is that they appear too much to me to have found the searching for regressions that fit their temperature proxy purposes (and without the real statisticians concerns for data snooping, cherry picking and model over fitting) to be more satisfying than first understanding the plant physiology — an understanding that could make their statistical cases valid through a prior selection of model variables and interactions.
Has the public/scientific attention to AGW and its potential for being a determining factor in what papers get published have anything to do with this seemingly short-cutted and backwards approach? I suspect it may, but I will withhold further judgment on this matter until such time as I hear a reply from the dendros or I feel that no reply is forthcoming.
#15. twq, I emailed Thompson in 2003 requesting data. In fall 2004, long before climateaudit was started, I sent him graphics showing inconsistencies between Dunde and Guliya versions and politely asking for an explanation, I got no response. I didn’t just start criticizing Thompson out of the blue. I’ve tried to get information politely and through channels and nothing has been successful. PAul Dennis of East Anglia also failed.