We sent in the following two supplementary comments to the NAS Panel, one commenting on answers provided by Mann subsequent to our presentation and noting up points related to the revisions of Wahl and Ammann [2006] and rejection of the Ammann and Wahl submission to GRL here and the other responding to a question from one panellist on our suggestions on how to do a reconstruction better here.
-
Tip Jar
-
Pages
-
Categories
-
Articles
-
Blogroll
- Accuweather Blogs
- Andrew Revkin
- Anthony Watts
- Bishop Hill
- Bob Tisdale
- Dan Hughes
- David Stockwell
- Icecap
- Idsos
- James Annan
- Jeff Id
- Josh Halpern
- Judith Curry
- Keith Kloor
- Klimazweibel
- Lubos Motl
- Lucia's Blackboard
- Matt Briggs
- NASA GISS
- Nature Blogs
- RealClimate
- Roger Pielke Jr
- Roger Pielke Sr
- Roman M
- Science of Doom
- Tamino
- Warwick Hughes
- Watts Up With That
- William Connolley
- WordPress.com
- World Climate Report
-
Favorite posts
-
Links
-
Weblogs and resources
-
Archives
42 Comments
Site note: I got a "404 Not Found" error for the second link: "The requested URL /pdf/NAS.recommendations.pdf was not found on this server."
Steve: Fixed
re: #1 Me, too neither.
I don’t think you left many stones unturned in your submissions. I can’t imagine that anyone with even a half-objective view can buy into these reconstructions. Great comments to NAS!
I just received a pdf reprint copy of Howard Falcon-Lang (2005) “Global climate analysis of growth rings in woods, and its implications for deep-time paleoclimate studies” Paleobiology 31, 434-444.
The conclusions would make a worthwhile submission to the NAS panel, prior to their conclusion. Here it is:
“1. Quantitative climate analysis of growth ring characteristics (mean ring width, mean sensitivity, percentage latewood) of modern trees was undertaken at the global scale. Results help quantify, for the first time, the enormous variability in tree growth response to present-day climatic forcing.
“2. Modern data allow the quality of growth ring data in pre-Quaternary fossil wood to be assessed. In light of such analyses fossil data are found to be inadequate in distinguishing paleoclimate signals from the background noise of variability.
“3. Quantitative growth ring analysis of fossil woods may be used only in well-constrained paleoecological studies where taxonomic and climatic sources of variability can be controlled, and additionally, of course, as a qualitative tool in paleoclimatic and paleoecological analyses.”
In other words, no conclusions regarding specific cliimate indicators are possible from growth ring studies, though general and qualitative paleo-ecological assessments are possible. Other than hand-waving ecology-based dismissals of other growth influences, I don’t see how anyone can possibly claim that confounding “taxonomic and climatic sources of variability” can be controlled for growth rings stretching across centuries and even millennia. To get the knowledge permitting climate-specific conclusions, clearly the greenhouse grunt-work studies of species-level (at least) tree growth under systematic variations of conditions will be necessary. Such basic spade-work in dendroclimatology is very non-sexy but clearly very necessary.
So far as I’m concerned, and presuming Falcon-Lang’s conclusions continue to be supported by others’work, paid has been put to the validity of any paleoclimatology from tree rings. I don’t see how anyone could currently find otherwise. Dendro-paleotemperatures appear to be a dead issue; and that’s not just dead trees.
Pat: Great, FINALLY a real expert assessment of the use of growth rings. It confirms my own intuitive notions. Isn’t it odd that nobody did this before dozens of scientists spent 10-15 years doing “reconstructions” using inappropriate data. How on earth could they avoid considering the very fundamental assumptions inherent in their work. I wonder who started the idea that trees are good temperature proxies.
Re #3,
FWIW, the wording of conclusions 2, and 3, seems to be limited to fossil wood.
3:
Dendro-paleotemperatures appear to be a dead issue; and that’s not just dead trees.
1. Agree with JerryB in 5.
I hope this doesn’t tar his name with the commenters here.
2. I believe I have set up an adequate way for you to test the courage of your convictions with folks who do this sort of work for a living.
If you are confident in the courage of your convictions, Pat, would you like to assert them on a dendro list serve and see how you do? Tell folks whose life work is in tree rings that it is as you say.
Let us know Pat whether you’d like to assert your convictions to professionals. [won’t be back home until some time tomorrow, so reply will be delayed]
Best,
D
#5, Unless you’re willing to suppose that the variability of the growth patterns of ancient trees was inherently different than the inherent variability of the growth patterns of more modern trees, then I don’t see how your qualifying reference of fossil vs. modern has any bearing on the issue.
#6, Dano, I don’t have to defend my claim anywhere. Falcon-Lang’s paper is published, and from the text, he has published similar preliminary findings previously. Let the professional dendro community deal with that. I can add, though, that reading the sort of qualitative assessments of climate that are made in order to assign temperature as the prime driver of growth rings in certain instances, or moisture for that matter, I find it more and more incredible that anyone can suppose that tree rings can be used to reconstruct a past temperature to within +/- a degree C. Or even +/- 2 degrees C.
The statistical tests of goodness-of-fit only tell us about the internal consistency of the data-set and the numerical stability of the fit itself. Statistical tests tell us nothing about the physical accuracy of the data. The tests are sufficient to exclude data from scientific validity, but they are only necessary for the inclusion of data. They are not sufficient for the latter. For sufficiency of inclusion into science, the observed variable must be shown by valid theory and reproducible observation to have a causal relation to the parameter of interest. I don’t see that criterion met in dendroclimatology papers. Do you? And if so, where?
Re #7,
Pat,
The comparison may not be ancient vs modern. Some species of trees live for thousands of years. I simply called attention to the wording that limited conclusions 2, and 3, to fossil, i.e. dead, wood.
I do not suggest that those conslusions may not also apply to living wood, I simple note that the wording seems limited to fossil wood.
They are very well-written, even-tempered communications. Kudos. Really.
One tiny editorial comment: I would define the acronym HS and MWP, the first time used.
#8, Jerry, sorry I was unclear. F-L looked at the how the growth of modern trees varied with several measures of modern climate, such as mean annual T, mean annual T-range, mean cold month T, mean warm month T, and so forth. The variability he observed was very large for any given climate parameter, showing that the natural biological variability of the trees tended to swamp the impact of climate. This modern variability was then taken to be a reason to be cautious about ascribing tree ring measurements taken on ancient trees to be measures of ancient climate.
But you’re right that some trees live for more than 1000 years. I’d suppose the response variability F-L found for living trees obviates their use to reconstruct past millennial climates as well.
Agree with TCO. 2 very well written documents. They make their points extremely well and are full of substantive points and recommendations that have to be taken seriously.
Love it!
As an aside note, recent trees growing near airports produce radiocarbon dates older than the cities around the airport, as a result of the “old carbon” in aircraft fuel.
I also pointed out elsewhere some time ago that changes in electric fields also affect plant growth and thus ring width. That comment was in response to Ido et al noting the Bristlecones having growth spurts unrelated to temperature Steve has previously pointed out.
So I have to stand next to Pat Frank on this issue if a division is called for.
I think its most telling that McIntyre and McKitrick have to point out statements made by Mann which are materially false this late in the game:
This is one of the failings of the NAS Panel. If Mann had been before a Congressional Committee, swore an oath to tell the truth and made statements like that, he wouldn’t have been allowed out of the building until he explained himself.
The NAS Panel (and I’m going to be specific and personal about this: Doug Nychka) should have immediately challenged these claims. The Panel (especially Doug Nychka) cannot possibly have been unaware of the revelation of what the source code contained and yet no-one challenged Mann on it. The NAS Panel (espcially Doug Nychka) should have immediately challenged Mann’s assertion about the R2 statistic especially coming from a self-acknowledged non-statistician.
So do we send him a bill for the carbon credits, or does his nose get longer, or what?
============================================
re: #13
John
I have re-read the revised terms of reference for the NAS panel.
the panel have no remit to even ask the questions you are suggesting. They get to comment (a very vague word) on the overall accuracy and precision of individual studies, and they get to assess the methods used.
While you may disagree with what the NAS panels terms of reference should be, there is nothing in it which says that Mann, or the author of any other reconstruction, should go on trial for their methods. Besides which, the NAS panel takes evidence; it does not cross-examine, nor does it badger, witnesses; and if the panel really does know something so much better than the witness, there is little point in having called the witness 🙂
yours
per
re: #6, Dano, I’m still waiting for one of you tree ring experts to come to this blog and defend the basic science behind the really weird idea that tree rings can be used as temperature proxies. I will not believe it makes sense, until someone demonstrates otherwise. It is critical to a whole bunch of studies purportedly demonstrating AGW, and I am simply flabbergasted that no clear evidence is forthcoming on this subject. If no other important variables but temperature changed in the thousand year life of a tree that just happened to be growing in a temperature sensitive location, then a temperature signal might be obtained (although probably not a linear one). But that would be a rare occurrence, for sure.
I know a better way: have one of those folks come over here and educate us.
Re #16 you’ve clearly missed Rob Wilson’s posts here?
I will be replying to Rob’s posts, which are far from being the last word on tghe topic.
Re: #15
Not quite true. The Panel have given themselves no remit to ask those questions. How can they assess that which they do not give themselves the right to examine?
Quite correct. But since somewhere along the line they chose to ignore clear specific questions by Congressman Boehlert on Mann’s research that his Committee wanted answered, and who is footing the bill for this show, I’d say someone was taking a big political risk telling the Committee a lot less than they want to know.
Why did the Panel not commission a small team of statisticians and auditors to go through Mann’s research and then with that report in hand, question Mann, Hughes and everyone else? Because it wasn’t possible or because they couldn’t be bothered?
Because either way, the NAS Panel is on a hiding to nothing. If it finds for Mann, everyone will question its competence to do so given its behavior. If it finds substantively against Mann, then there’ll be howls of protest about how poorly the Panel conducted itself, and how much evidence it could present etc.
This is where the lessons of Jan Henrik Schoen and Hwang woo Suk should have been learned, and haven’t. Neither Bell Labs nor South Korea’s authorities threw their hands up and convened a nice little Panel to discuss wide issues of experimental design and results in materials science and stem cell biology. They focussed on the narrow issues and no-one could possibly fault their findings.
#17: No, Peter, I did not miss Rob’s posts. His work looks better than the usual stuff, but I have issues, such as:
This cherry-picking thing is never allowed by the scientific method, unless there is a clear valid reason for doing it. Maybe picking locally “temperature sensitive” chronologies is valid in some way, but I don’t see how.
Also, Rob states:
Now, exchange the words “temperature” and “CO2” in the last two sentences. Temperature is probably qualitatively no different than C02 as a variable affecting tree growth. If you can’t find a CO2 signal, then why is it that you can find a temperature signal in this variable soup. It seems to me that there are just too many critical variables (not just minor ones) to allow one to show a long term trend caused by just one of them (i.e., temperature). And I think this is exactly why cherry-picking is necessary. And I do not believe that CO2 is any less of a limiting factor to growth than temperature. There are hundreds of studies that demonstrate that plant growth is very sensitive to CO2
Finally, a very interesting project being undertaken by Swiss researcher can be read here:
Funny, just last weekend I went to an open house in a greenhouse, and guess what, they are actively pumping co2 into the greenhouse to enhance productivity. Why would they make such an investment if it didn’t matter?
I see it’s the caveat “some species” isnt it. Could it be that perhaps bristlecone pines are a member of this “some species” group?
I collected a number of articles from the Swiss group some time ago. Hattenschwiler et al New Phytologist 2004 includes Walter Ammann as one of the coauthors – I wonder if he’s related to Caspar and what he thinks of how Caspar handles bristlecones.
My question is: how does Rob Wilson know which trees are “temperature sensitive” without comparing them to… temperature? And, btw, since he is picking “temperature sensitive” proxies, then by definition, he has contradicted his other statement “HOWEVER, we did not cherry pick for a HS signal” which is the temperature signal in question.
No matter how he slices it, the cart is before the horse.
Mark
Re: #19
I see a distinction between the NAS, and the NAS panel. It strikes me that we are agreeing that they have precious little remit to try MBH.
As regards whether the NAS is disregarding clear questions by Boehlert, that is an interesting political issue. I have no idea of how the remit of the NAS panel came to be thrashed out; who knows, maybe different senate committees disagreed about what questions should be asked ?
yours
per
per,
I think it’s becoming evident that the Mannite concern is with having his work scrutinized/with being grilled. We heard a lot of concern that CONGRESS shouldn’t be holding science hearings. So fine. They pass it to NAS. But if NAS, then wants to do a thoothless circle jerk, it shows that the issue is not Congress versus scientists, it’s with doing a tough investigation.
Mark, I asked a similar question. I think what Rob means when he says “temperature sensative” is really “agrees with a previous reconstruction”. It’s not a correlation to the local instrumentally observed temp or even “global climate field” (as shown by instrument).
I think it’s likely that he’s a nice guy and is interested in trying to be intellectually honest (thus the engagement with Steve). But I don’t get the impression of a penetrating intellect.
#17 – Peter, I searched Climateaudit, and Rob Wilson discussed his proxy methods under several subject headings:
http://www.climateaudit.org/?p=541
http://www.climateaudit.org/?p=602
http://www.climateaudit.org/?p=610; especially post 25, item 6.
That last is the only one that specifically addresses temperature and the divergence problem, and implies, but does not say outright, that tree ring series that correlate with modern temperatures are assumed to also correlate with past temperatures.
In the absence of greenhouse-controlled studies on the response to temperature, moisture, soil composition, CO2, etc., of tree rings within the field-selected tree species, or at least on a closely related species, I don’t see how the assumption of linearity of response across time can be justified. That especially in light of the very large natural variation reported by Falcon-Lang.
Speaking of soil composition, I wonder if it would be possible to take ground cores from any given region and reconstruct the paleosoils. One could then determine past fertilization effects, of at least the mineral variety, and attempt to justifiably normalize them out. I haven’t investigated this, but it may even be possible to reconstruct paleo-moisture.
You use the lower elevation trees for moisture and the upper elevation ones for temperature. And then you do some sort of regression thingie.
#29. that’s their theory. But they base this on a 1974 (!) study. But then why do the upper border sites have bigger ring widths than the lower border sites. I’ve seen mention of inversions in the mountains whereby the valleys are colder than higher up in the spring and this affects the ability of seedlings to grow- maybe the lower border is also a temperature border in some cases. lots of issues.
#28. There has been considerable recent work on the impact of CO2 on high altitude conifers in Switzerland. I’m planning to post on these but if you google Kàƒner, you’ll get a variety of articles on the topic reaching up to 2005. One of the interesting issues is that one school of thought appears to hold that the upper forest border in low latitudes is due to carbon limitation – thus carbon dioxide increases would have a direct impact. Exactly what causes a low-latitude altitudinal border (in direct physiological terms) seems to be unresolved.
I have a sneaking suspicion that the “divergence problem” is simply the result of selecting from random(~ish)data sets those which best fit the required trend over a calibration period. Once beyond the calibration period over which the records have been selected, the data will tend to revert to the mean. This is the same as the apparent “little ice age” seen in David Stockwell’s lovely hockey stick from random data reconstruction.
It would be interesting to “Cherry Pick” TWO sets of tree ring chronologies from the same source to match temperatures over, say, the last 50 years and the 50 years before that. Then compare them. Are they the same populations? How well do they predict temperatures out of their calibration intervals??
Time to re-emphasize the old “Bring the Proxies Up to Date” thread?
I’m confident in the courage of my convictions regarding tarot card readings, but if I assert them on a list serve full of folks whose life work is in tarot card readings, I don’t imagine I’d have any impact other than to be assaulted.
#31. John D, I’ve been doing a few more comparisons of nearby site chronologies by diffreent authors, and sometimes at the same site. The results are interesting and I’ll be posting them up.
#30: A. But do they actually use both series? How do they combine them? I don’t think this is trivial, since there will be some offset, but still would expect rainfall(high) or temp (high) to correlate strongly with rf(low), temp(low). Do they actually try to do what I said or do they just use the top trees and publish a temp reconstruction (and occasionally the bottom and publish a precip recon)? If you have high trees, which vary 80% with temp and 20% with precip and bottom trees which are 40%temp, 60% rainfall; you should be able to use that data together to get insights.
B. Being from 1974 is no big deal, Steve. You’re waxing Mannian. Next you’ll tell me that MBH is old news. I’ve used lots of good papers from the 70s. A better critique would be that the initial study was thin, spotty, limited in species and that a whole new house of cards (studies in 90s, 00s) is being built on it…so that we should do a re-examination, restudy of this foundational work. But don’t diss the 70s. A good foiundational X-ray spectrum for Cu from the beginning of the century is still completely relevant to x-ray crstallography of new compounds using a Cu(alpha) source.
31: I thought MBH claimed to have done exactly that. Now you can still cherry-pick even more to see which series follow what you talked about. And it would be interesting to see the foundational work of a nature that you mentioned (without even any emphasis on getting a recon from it…just to assess tree ring series in general). But I think MBH knew the concept….
#30 (newA or C if you continue from earlier): The upper sites could have wider rings for a variety of reasons. Perhaps moisture is more limiting than temp? Or competition? (Or your alpine inversions). Your question actually opens a broader issue (interestingly related to your interest in treeline tracking): What happens at the treeline? Should you get some narrowing to zero of rings? Is there some other mechanism of “not being able to grow” that stops trees from growing (establishes a treeline), but is not reflected in the rings?
Shameful! Those are the folks we need to blame for events like Katrina, melting glaciers, etc. What a waste of GHG emissions. What a bunch of polluters 🙂
People are throwing money at Kyoto, and it doesn’t (significantly) matter.
#37. Why treelines exist is very much speculated on in the botanical literature. The ability of trees to germinate seems to be distinct from their ability to exist once germinated. For the bristlecones, the fact that they compete with big sagebrush means that aridity is obviously an issue. You’re right that a study from 1974 is not necessarily outdated – I like to read old studies to get different perspectives. In fact, my beef is not really with the date of the study but with the lack of proof in the original study.
There’s also a curious discrepancy in the citation history of the early Lamrache articles: on the Lamarche et al 1984 article on CO2 fertilization, the botanical literature in the 1990s (which I’ll review soon) accepts that Graumlich 1991 showed that the upper border sites are confounded temperature-precipitation signals – which seems logical given the sagebrush competition. Graumlich’s concept of interaction seems very plausible. I don’t think that Graunlich’s analysis refutes the CO2 fertilization as the statistical analyses in her and similar articles seem not to be of the best quality, but it would take a fair bit of work to disentangle.
But if you rely on Graunlich 1991 against CO2 fertilization, you also have to apply it against simplistic upper border-lower border divisions of Lamarche and Stockton 1974.
Superb letters Steve.
Clear, factual, and relentless.
TCO, I wonder if you are missing the point a bit regarding the use of 1970’s data. There is, I believe, nothing against old (in this regard) data per se, but with a phenomena that is supposed to be particularly evident in the instrumental data from the late 1970’s on, it seems a bit odd that there has been so little apparent effort to collect more modern data. It would seem a given that if dendro data is a good temperature proxy, then the signal should be particularly obvious in the period 1980-2005. That there are problems with the proxies is apparently the reason, but real scientists would be out collecting data to elucidate this point, not continuing to rely on old data and extrapolating it forward.
I agree that it is odd that there is not more current work to solidify the foundational studies. But net/net, the real kvetch and area of examination should be the adequacy of the foundational studies, not their antiquity.