Craig Loehle’s recent guest post on WUWT inspired me to download and compare various proxy reconstructions to see how the new Ljungqvist 2010 data compares to Loehle, Mann (2008), and Moberg. Most reconstruction data (save Loehle’s) is conveniently archived by the NCDC here. Loehle’s reconstruction (2008 corrected version) is available here. When we put them all together, we get a lovely spaghetti graph. I’ve added both the NH land and NH land/ocean temperature records at the end (as to which is a better comparison, I don’t know!). All the data is decadal averages.
(click to massively embiggen)
Updated. See the addendum at the end.
We can also add in the uncertainties for each:
Loehle’s reconstruction is the odd-man out runs highest for most of the period, and Mann et al ’08 CSP CPS series runs considerably lower than other NH reconstructions for 800-1100 AD. Only the Loehle et al reconstruction shows the medieval warm period close to as warm (e.g. within the uncertainty bounds) of the modern instrumental land/ocean temperature record. Bear in mind, of course, that this is only a Northern Hemisphere reconstruction, and thus only half the picture. Unfortunately, long-term Southern Hemispheric proxies are few and far between, though there has been a concerted push in the last decade to collect more.
As for vindicating anyone, the person whose result is most in line with Ljungqvist ’10 is Moberg, though Mann’s EIV series is close:
Update
Added in the Mann et al ’08 EIV series, which tracks much more closely to the other reconstructions.
Update #2
I may have been somewhat unfair to Loehle et al when I assumed it shared roughly the same baseline as the other reconstructions. When we re-baseline all records to zero during the period of instrumental record overlap (1880-1930 for Loehle et al, 1880-1970 or 1990 for others), it still runs a bit higher than other reconstructions but not as glaringly so. All the graphs have been updated accordingly.
Didn’t Loehle claim his was global? His coverage didn’t really justify the claim, but that’s what was claimed nonetheless.
How’d you line up Loehle with the instrumental period? If I recall, his was something like 50-year centered means, ending in the 1930s, or something like that. I care enough to comment, but not enough to check. I’m a jerk.
In any event, it’s a little unbecoming to just jump straight to the final answers, so to speak, without comparing the data sources and methods as well.
carrot eater,
Fair enough, digging into the methods is probably a more useful task. But the data is out there, and it is fun to compare to see whom Ljungqvist “vindicates”, so to speak.
I simply used the Loehle data as-is, which has the end overlapping with the instrumental record and other reconstructions. I could rebaseline it down a bit, but that would result in a very cold LIA and quite low recent temps, and would make comparing it to the instrumental temp record rather problematic.
Lucia —
How are you centering the series? This should be to zero over as long a common period as possible.
Note that my CI’s for Craig’s recon are also available on the file you link. See also http://econ.ohio-state.edu/jhm/AGW/Loehle/ .
An interesting question here is what was the Mann 08 particularly sensitive to. First guess would be volcanic eruptions, but that does not appear to be the case
Hu,
Its Zeke rather than Lucia authoring this post. I’m in the process of updating the figures to ensure a common baseline period.
Nice post Zeke.
Care to comment on the potential for loss of variance for these reconstruction methods? Jeff Id has posted many times on this subject, and he says there is a terrible potential for loss of variance when the S/N ratio in the proxy data is low.
(arrgh, I embiggened a graph in the midst of writing a comment… poof!)
Zeke, I think for Mann08 you mean EIV (error-in-variables) and CPS (composite plus scale).
On substance, my intuition … FWIW… is that it is the uncertainties rather than the anomalies that are most important. Two issues in particular.
1. How do the various reconstructors define “uncertainty,” and are their calculations sufficiently correct and complete so that the numerical values assigned to it are more-or-less correct?
2. What is the relationship between uncertainty-as-defined — UAD I guess 🙂 — and “uncertainty” as it is understood by citizen-scientists, policymakers, and the like? For instance, eyeballing the third graph (without embiggening this time!), Mann08 EIV at the year 500 is about -0.3C +/- 0.4C. As a citizen-scientist, I would interpret this claim to mean, “using their EIV Method, Mann08’s authors can be 68% sure that the Northern Hemisphere’s decadal average temperature around 500 AD was within 0.4C of being about 0.3C cooler than it was during the modern baseline period.” (“68%” would be “95%” if Mann08 and the others used a two standard deviation interval; I can’t recall.)
I suspect that the claims of all these reconstructionists are much, much, much weaker than that.
We might be curious about what Californians think about the November 2010 Senatorial race, and poll people: “Who will you vote for: Boxer, Fiorina, another candidate, or nobody”? In that case, we know that there are all sorts of uncertainties to contend with, some obvious and other subtle. People are now sophisticated enough to realize that a footnote like “the sampling error of this poll is +/- 4%” does not imply that the pollsters believe that they have fix on the election to “within 4%.” There are many other sources of of random error and systematic bias that are far beyond the scope of a sampling error calculation.
Paleoclimatologists don’t do nearly as well as pollsters at communicating the limitations of their methods and the tentative nature of their findings, in my opinion. It seems to me that many of them aren’t well informed about such issues.
Hu–
Yep. I have a bad cold and am busy trying to finally figure out what’s causing the huge memory surges at my blog. I may have to ask someone to help me out.
AMac,
Yep, I made a typo. Should be CPS.
It’s “Moberg”, not “Moburg”.
Mark,
Good catch. I fixed his name in the charts/text.
Still not entirely happy with my baseline for Loehle. The problem is that it overlaps so little of the instrumental record compared to the other series, so its tough to calibrate against that.
It’s all hockey sticks. Even Loehle indicates the current warming is unprecedented, and the blade is going up.
bugs,
Well, only with the instrumental record grafted on. Without that we’re about equal to the MWP. That said, without the instrumental record Loehle only goes to 1930, so it really doesn’t say anything about current warming.
To me the Hockey Stick was about two things.
1) A relatively stable temperature record prior to now, with no long term trends or rapid changes in temperature. We have that.
2) A rapid rise in temperature during this point in time. We have that as well.
That it should be a picture perfect replica of a hockey stick is a persistent but nonsensical straw man.
Zeke
That’s what I was getting at. If I recall correctly, Loehle never anchored his series to the modern record either, seeing as I don’t think he did a calibration/validation step.
Re: bugs (Sep 28 15:27),
> 2) A rapid rise in temperature during this point in time.
I agree that the instrumental record ~1890 – present shows a substantial rise in temperature, probably at least as steep as some prior rises in the past ~2000 years. Very possibly more rapid.
.
> 1) A relatively stable temperature record prior to now, with no long term trends or rapid changes in temperature.
I do not agree with this point. The failure of the “community” of paleoclimatologists to adhere to Feynman-like norms makes their work less trustworthy than it should be — especially when conclusions derived from their work support the policy views of those same people. Groupthink may or may not account for much of the AGW Consensus concerning paleotemperature reconstructions — who knows?
And as noted above, I think the uncertainty bounds on reconstructions are way too low. If the “correct” uncertainties range around +/- 2C instead of +/- ~0.4C, then the reconstructions are not very informative as to climate stability.
So,
Let’s see an evaluation of how Tamino compared these.
No comments about the people involved, just a comment on the
methods used to align these things.
Tamino: right or wrong
Mosh,
Tamino did the same thing I did initially, namely download the Loehle series and use it as-is. In retrospect I would argue that this is not the most appropriate comparison, though if you do calibrate it against the brief period of overlap with the instrumental record you get something close (but a bit lower) than Tamino’s version.
Re: steven mosher #52977,
In an in-line reply to a comment at his post, Tamino explains that his comparison method is the acceptable one:
Pingback #52971 by Jeff Id (this thread) links to his post Honesty in Blogging, challenging that choice by Tamino.
It’s a shame that Ljungqvist study still shows that pesky divergence problem.
Bugs, this one’s for you:
Is this a hockeystick which I see before me,
The handle toward my hand? Come, let me clutch thee.
I have thee not, and yet I see thee still.
Art thou not, fatal vision, sensible
To feeling as to sight? or art thou but
A stick of the mind, a false creation,
Proceeding from the heat-oppressed brain?
I see thee yet, in form as palpable
As this which now I draw.
Oh heck, I should have done the literature search first:
http://althouse.blogspot.com/2010/02/oh-climategate-professor-is-sad.html?showComment=1265593465482#c3031281537052505319
Maybe one should just restrict the question to “was there a Little Ice Age cold period”.
And the answer to that is “yes, there was one” – and then this points to the second conclusion that “yes there are cycles in the climate” and etc.
Now to the major problem – not all years in the LIA were colder than average because the high resolution proxy estimates show there were some very cold years and there were also some very warm years in the period.
I’d like to see the 50 year smoothing algorithms pulled out of all the series because the climate does not operate on a 50 year smooth – maybe 1 or 2 (or 3 on the outside) months at a time and that is all. The variability is hidden in these smoothing routines (and the warmth of the MWP and the cold and warmth of the LIA is hidden as well).
The emails suggest the authors of certain series were aware of these issues.
Zeke, os there any way to get an average of all the graphs, so that we can see only one piece of sphaghetti?
DeNihilist,
Surely you aren’t suggesting that the multi-reconstruction mean is inherently more skillful than any individual reconstruction? 😛
Zeke — it looks to me that if you follow the 0 deg line across it is included in the error range for all years other than a couple in the LIA — so do I conclude there has been no significant change through out the reconstructed period for any of the reconstructions?
lucia (Comment#52962) September 28th, 2010 at 2:02 pm
Hu–
Yep. I have a bad cold and am busy trying to finally figure out what’s causing the huge memory surges at my blog. I may have to ask someone to help me out.
——-
I hope you feel better soon.
The thing is, does Loehle even have a calibration period? Unless I’m really missing something, I don’t think it did. So oddly, Tamino might be assuming that Loehle was more competent than he actually was. I don’t see any calibration to the instrument record, and I don’t see any validation against it. There probably isn’t enough overlap to do both, anyway. From my vantage, this looks like a case where Loehle isn’t right, and he isn’t even wrong. His work is just half-baked, and not worth much attention in the current form.
Two things that stand out on the series is 1) there is a real LIA (thanks guys for telling us the obvious) and 2) the MWP was a world-wide event. Mann is wrong low about this (more so than Loehle is wrong high) in his strange jihad to rid the world of the MWP.
But truthfully I couldn’t care, I already knew both of these to be true from historical sources, if you can’t get these right, your method (like Mann’s long series of the same wrong paper) is just “half-baked, and not worth much attention in the current form”. (/poke)
Whether the max MWP is equal to e.g. 1980, that is important for trivia buffs, but irrelevant to me: We still have CO2-driven radiative heating in either case.
I would be delighted in the usual suspects neutrality, when they admit that Mann is not confirmed by the other series (leave off Loehle as unworthy if it makes you feel better).
Woo hoo! Finally we’ve cast off the terrible burden of the southern hemisphere.
Zeke, nope, just can’t find my glasses right now, so can’t make out all the different strands of sphaghetti.
🙂
Umm Carrick.. if you’re going to snark, get your head out of your rear first. Ljungqvist is apparently NH only, and at that, I guess only extratropical (I say apparently because I haven’t read it yet, and probably never will). Loehle might have put the word global in his paper, but a glance at the proxies tells you how global that study was. And yet you’re using the above to make global conclusions to match whatever a priori thoughts you had. Wonderful. Anyway, whatever you don’t like about other published paleo works, Loehle is on a very different level of being inadequate. Not even bothering with calibration/validation is just the start.
pough, try this out for the SH –
Kellerhals T et al. JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES Volume: 115 Article Number: D16123
a bit of the abstract:”That given, the reconstruction reveals that Medieval Warm Period- and Little Ice Age-type episodes are distinguishable in tropical South America, a region for which until now only very limited temperature proxy data have been available. For the time period from about 1050 to 1300 AD, our reconstruction shows relatively warm conditions that are followed by cooler conditions from the 15th to the 18th century, when temperatures dropped by up to 0.6 degrees C below the 1961-1990 average. The last decades of the past millennium are characterized again by warm temperatures that seem to be unprecedented in the context of the last similar to 1600 years.”
Max_OK–
The cold is mild. I’m more concerned about the memory issue. Hunting around, it seems that WP may have gotten pretty bloated for self hosting. I may end up hosting at WP for a while.
Mr. Eater of Carrotts, Dr. Loehle does use some South African proxies, and admits that they may make his anomolies “hotter”.
Gee Zeke, Ya blew this one worse than the Nepal one
So,
Let’s see an evaluation of how Loehle compared these on WUWT.
No comments about the people involved, just a comment on the
methods used to align these things.
Loehle: right or wrong
One of the ways of comparing two sets of reconstructions is to reference them to the geomagnetic field strength, the independent variable following same trend during the last two millennia.
http://www.vukcevic.talktalk.net/LL.htm
Bill Illis (Comment#52985) September 28th, 2010 at 6:33 pm
Maybe one should just restrict the question to “was there a Little Ice Age cold periodâ€.
And the answer to that is “yes, there was one†– and then this points to the second conclusion that “yes there are cycles in the climate†and etc.
= = = = = = = = = = =
Surely this merely points to variability within the climate system and not that that variability is necessarily cyclic.
Eli:An interesting question here is what was the Mann 08 particularly sensitive to. First guess would be volcanic eruptions, but that does not appear to be the case
.
I guess the obvious thing to do would be to identify which proxies cause most of the difference. One could do a lot of “leave_one_out” reconstructions, but that would require a full replication of Mann 08 algorithms.
.
A possible poor man’s alternative is to scrap through the Mann 08 proxies, and find the proxies whose difference from the final Mann 08 reconstruction correlates best with the difference between Mann 08 and “consensus” reconstructions.
Presumably these would be the proxies that push Mann 08 away from the consensus (?). But even that would require obtaining the appropriate scaling parameters for each proxies – I don’t know if they’re available.
.
Bonus trolling points if the culprits turn out to be Finnish varve of American tree series :).
A comment on the methods used to align these things:
The principles involved are the same as with other issues in climatology, and throughout science. In particular, to do a statistically-meaningful comparison with everyday methodologies, the criteria for the comparison must be determined beforehand, rather than after looking at the data (in this case, the reconstructions).
If you look at the data and then decide how to align, you are free to seek the offset that maximizes similarities between curves (Loehle), or choose one that does not do so (Tamino).
Either of these choices — or any other — is fine, as long as the statistical method then chosen to calculate a P value accounts for the alignment as well as for the similiarities. The most familiar such method is the Buonferroni correction, though it likely isn’t the one applicable here. The basic point applies: if the best alignment can be picked from more than one choices, the P value has to account for this choice. (The greater the number of lottery tickets purchased, the likelier it is to hold a winning ticket.)
On the other hand, if the point of the exercise is to look for similarities — e.g. “cool” Little Ice Ages and a “warm” Medieval Warm Periods that are shared among reconstructions — then one would search for the optimum offset that minimizes differences between patterns. That would allow the observer to use informal visual-recognition skills and say, “Hey, look at that!” But at the price of disallowing statistical comparisons (at least, those performed without Buonferroni-like post hoc adjustments).
So if Loehle did (or plans to) present a formal estimate of significance with a simple statistical tool, he has fallen into an error that is common in climatology — and Tamino is right.
If Loehle intended to informally point out similarities in patterns, then his approach is valid — and Tamino is choosing an approach that minimizes visual congruences among reconstructions.
Then what earthly use was his reconstruction?
Suppose that Loehle had presented an reconstruction that the pro-AGW Consensus advocates liked.
To go back to a classic Blackboard dispute, suppose that looking at a recent 10-year average of global 10-year temperature anomalies favored the AGW position, while a 30-year average favored the skeptical stance.
Would scientists and bloggers hold to the same principals and modify their views on AGW accordingly?
Would they suddenly realize that their previous stance on alignment/averaging/whatever needed to be amended?
Or, more cynically, would they declaim that their views have always been what they now are (“Oceania has always been at war with Eastasia”)?
Dear Lucia,
At “Klimazwiebel”, Mr Von Storchs climate blog, he made a survvey among sceptics and asked some questions about climate blogs. I often read your blog and I trust this blog MUCH more than many other climate blogs.
But I have one question to you :
Reading all these comments about climate reconstructions and obviously understanding what they mean, what do you think about the end points of the graphs? Does it mean, that the MWP might as well have been as warm as the modern warm period. Or do you personally think that Micheal Mann is right when he claims that the last decades were much warmer than any period of the last 1200 years?
Or do you still have scientific doubts about that special point in a way that you could say, well we don’t really know at all (50/50)?
Or would you prefer to say that the last decades were very probably much warmer than the MWP?
Best regards
Yeph
AMac,
I am not sure you are correct that you have a choice how you align. The two reconstructions should be consistent, i.e use the same baseline. Loehle’s approach does this (both are centered over the mean of the full reconstruction period). Tamino’s does not. Loehle is right. Tamino is wrong.
Bugs and CE,
Loehle’s paper says that he used “data for long series that had been previously calibrated and converted to temperature by their respective authors”. Seems maybe a little lazy but I don’t think an unacceptable approach.
Re: brid (Sep 29 07:57),
But here we are, having a post hoc discussion on the subject! This is the sort of dispute that other fields have developed procedures to handle. Paleotemperature reconstructions, not so much.
AMac,
Good point – there really should be generally accepted ex ante tests for basic paleo questions like this. However, it think is indisputable that Tamino’s approach is flawed.
Well, AMac, all other reconstructions use the calibration period with instrumental temperature to align. The defacto baseline seems to be 1961-1990.
Stop lying Bugs. The Loehle graph line shows the MWP warmer than it is today. What is your problem? You don’t have to lie to people to get them to believe you. I think global warming is a hoax in so far that I do not believe co2 can have a significant impact on temperature. Also, I find it very underhanded to hold a congressional hearing and intentionally make the room hot. But I would say around 2,000ppm of co2 in the atmosphere, could cause the temperature to rise a little. On the other hand, you keep trying to tell people that the earth has never been warmer which is a complete lie. Please stop the revisionist history or I will hose you down with a can of ultra strength raid.
> Stop lying
Bad netiquette. C’mon. It’s not just you (and Tamino) who get so irritated and offended that rational argument is supplemented with snark, sarcasm, insult. One thing that paleo has taught is the corrosive effect of this form of discourse.
.
Zeke — re: baseline, I agree. But (1) that doesn’t invalidate my earlier points, and (2) IIRC Loehle’s reconstruction ends in the 1930s, making a 1961-1990 alignment more difficult.
Yeph-
I’m not sure exactly what you mean when you ask, “what do you think about the end points of the graphs?”
Are you asking if I think someone has to rebaseline using the overlap period in the surface temperature? This is a question discussed by Jeff Id. I think they don;t have to be rebaselined that way– that is I agree with JeffId. Does this mean there is ambiguity in the result? Yep.
I don’t know enough about reconstructions to know how to interpret the computed uncertainty. Likely, the uncertainty in the reconstruction, along with any uncertainty in pinning baselines is such that we can’t be sure the MWP was not warmer than it currently is. But I say that without knowing much about creating reconstructions.
@Lucia
Thank you very much. I think this answer is ok for me.
Who knows enough about these reconstructions? 😉
At least you understand the statistics behind the reconstructions. I’m happy to have asked your opinion. I’m a little bit afraid about the extreme “violence” of the “climate wars” and being a lay man who wants to know what he is talking about, I’m happy to get real personal views of the bloggers who are important for me.
Thanx
Yeph
Dr. Bart, it depends on what Dr. Loehle was trying to portray. According to one of his comments, he was showing the correlation of the rises and drops only. If this is true, then his way is correct.
If Tamino’s interpertation is correct, that he was trying to show that his anomolies were quite matchable, then his way was wrong.
What this tempest in a teacup to me signifies is the great divide between “us and them”. Did Tamino, or you or any of the consensus team take 15 seconds and ask Dr. Loehle, just what was he trying to compare with his example? No. Automatically it was ass-u-me-d that he was trying to pull the wool over WUWT readers eyes.
For your information, I thought right away that his representation was about the undulations and not the temps. But then I am not a climate scientist, so maybe I am just ignorant.
🙂
AMac @ September 29, 2010 at 5:14 am
“If you look at the data and then decide how to align, you are free to seek the offset that maximizes similarities between curves (Loehle), or choose one that does not do so (Tamino). ”
Tamino didn’t say that was his intent, but I suppose one can assume it was.
Here’s what I don’t understand – if you want to show how the reconstructions relate to the modern temps, wouldn’t you want to calibrate each reconstruction to the modern instrument record? Otherwise if you adjust to the mean of each reconstruction, how can you say much of anything of how they relate to modern temps?
That is my understanding of why Tamino took the approach he did – to show the relationships to modern temps. So to say his approach was not to maximize the relationship between the curves is a bit misleading, IMO, by omitting the reason for the approach he took.
That said, I can understand why Loehle took the approach he did – to show the relationships between the paleo reconstructions.
So it seems to me that trying to compare Loehle’s approach with Tamino’s is like apples and oranges
“According to one of his comments, he was showing the correlation of the rises and drops only. If this is true, then his way is correct. ”
Yeah, but that’s not really what people complained about when they complained about his methods. So his use of the example as “Vindication” is rather bizarre.
Carrot Eater, you make a valid point that the different proxy reconstructions often use very different geographical weightings (e.g., NH, NH extratropical, land only, etc). But I think you make too much of that.
If you think about it, the main effects one would expect on long term climate are firstly a change in the scaling, the implications of which I think are contained in this plot
If one does land only, one should expect a greater scaling than ocean only or ocean plus land. If one does NH only, you’d expect a larger scaling because of the increase in temperature sensitivity as you approach the arctic… and so forth.
The second change is that one might expect a baseline shift as the geographical weighting changes (I can expand on why I think that would happen, if it isn’t obvious).
The third change is that the relative amplitude of regional scale climate fluctuations increase as you make the size of the region of the Earth you are averaging over smaller (see e.e. Hansen’s works), however most of this is expected to be “short period” (less than 30 years max), so the predictable effects of this are distortion in the overall scaling (miscalibration) and a shift in the baseline.
I think you can compared Loehle to the other reconstructions if you first rescale and shift its origin, assuming of course there is any validity to Loehle at all, which given it is based on real proxies, there probably is.
Also, here is Mann’s 2008 paper.. I hadn’t seen it linked. The supplementary information is interesting too.
Your comparison is still flawed. Loehle has thirty year averages not “decadal” and Loehle is not a “northern hemisphere” reconstruction, just look at the map of proxy locations:
http://www.worldclimatereport.com/wp-images/loehle_fig1.JPG
Clearly locations South of the Equator.
Carrick wrote:
I don’t the the paper you linked to offers much support for the idea of a “jihad to rid the world of the MWP.”
In other words, Mann is putting his chips down on the EIV reconstruction as regards the MWP, a reconstruction that is throughly, almost boringly, in the middle of the pack. Of course you could argue that he ought to throw out the CPS reconstruction, rather than just laying out the reasons he thinks EIV is likely, but not certainly, superior. But I don’t think you’ve made that argument, and it doesn’t seem the stuff of “jihad” in any case.
Robert:
Robert, I was just picking on Carrot a bit…for old times sake. I liked Mann’s EIV reconstruction better as well… and wish Zeke would have used that instead of the CPS one to start with.
The CPS reconstruction looks very poor even compared to Mann’s other reconstructions, and it is probably worse that Loehle (especially a baseline shifted, rescaled Loehle).
I’ll let others debate whether Mann is an anti-MWP jihadist, as you know he has gained a reputation for that in some circles.
Andrew_FL, do you remember how Loehle did his average? Was it straight unweighted?
DeNihilist if you look at Loehle vs Ljungqvist the correlation is not very good.
In the comment of 9/29/10 5:18pm, Robert contests Carrick’s characterization of the Mann team’s stance on the Medieval Warm Period, and approvingly notes the features of the Mann08 EIV reconstruction.
Robert’s quote of Mann08 finishes with
Building on that, Robert notes
Steve McIntyre’s 8/1/10 post The No-Dendro Illusion addresses recent findings about the confidence that should be placed in Mann08’s EIV reconstructions. Mann09 (Science not PNAS) revealed that Mann08’s EIV and CPS reconstructions require their tree-ring proxies to demonstrate “skill” prior to 1500. MWP ~ 950-1250, says Wikipedia, probably correct despite Wiki’s unreliabilty on matters touching AGW.
The tree-ring proxies include the extremely-contentious bristlecone pines and Siberian larches, about which so much ink has been spilled.
Mann09 and Gavin Schmidt say that Mann08’s non-dendro CPS and EIV reconstructions are skillfull for many centuries prior to 1500 AD…
… but only if the Tiljander varve series are included in the proxy set.
Wow, AMac, Steve has discovered the amazing fact that if you remove a shitload of data from a reconstruction, the reconstruction loses skill. The world’s best science blog at work.
Also, lol at Carrick and his “jihadist” comment. Stay classy.
Carrick (Comment#53071)-I believe it was a simple average. Really not the best way to combine series, I think, but certainly a lot simpler.
I should note that two of the series were actually composites over China and North America, so these actually represent sets of several series effectively given lower weights to each individual proxy.
I’m no Loehle apologist, in fact I have yet to see anyone, including Loehle, do what I would probably do to create a proxy for this. Unfortunately for me I have neither the time nor the access to data necessary to really do what I’d like to. But as far as I’m concerned, no one has created a satisfactorily certain reconstruction of the climate of the last 1000-2000 years. So my opinion is that there is no actual way to really know, at this point, what the whole “MWP-CWP” current dispute amounts to, and nobody has definitively shown what the global situation looks like in this regard to date. We don’t know absolutely nothing, but we know very little.
McIntyre, the one trick pony. If he had anything of substance to say about AGW, he would have moved on from the paleoclimatology by now.
I am in total agreement with Carrick. How can anyone not believe that he was trying to discount the MWP? And this a big part of the problem with the theory. If you are convinced your theory is correct, you do not need to make a fake graph.
Boris wrote in Comment#53094 —
> Steve has discovered the amazing fact that if you remove a [boat]load of data from a reconstruction, the reconstruction loses skill.
No, that is not the point that the ClimateAudit post The No-Dendro Illusion addresses.
1. One of the prominent claims of Mann08 — arguably the central claim — was that they were producing “skillful” reconstructions going back well over 1,000 years, without having to rely on tree-ring proxies.
2. If one subscribes to the general premises of the reconstructors, that would be an important advance. And AGW Consensus scientists and boosters ardently defend these premises. See for example IPCC AR4 WG I Chapter 6.
3. What “The No-Dendro Illusion” points out is that statements in the Online Supplemental Material of a later publication (Mann09 in Science) undermine this central point of Mann08 (PNAS).
4. The Mann08 CPS and EIV reconstructions shown by Zeke in this post’s figures depend on tree-ring proxies. Most of the tree-ring signal comes from data sets that were collected from a few sites. In some cases, they come from only a few trees. McIntyre and others have raised a series of questions about these data sets–questions that pro-Consensus scientists and boosters have failed to adequately address, in my opinion.
5. Mann08 also presented CPS and EIV reconstructions that were built exclusively from non-tree-ring data series. The similarities of these no-dendro traces to the dendro traces were offered as strong evidence that the reconstructions are meaningfully representing the temperature-anomaly signal, over many centuries of the Earth’s recent past.
6. However, Gavin Schmidt recently pointed out that these no-dendro CPS and EIV reconstructions fail to achieve the threshold of “skillfulness” prior to 1500 AD if the Tiljander proxies are not employed.
7. And the Tiljander data series as used in Mann08 are not proxies for temperature. They are not calibratable to the instrumental temperature record due to extensive land-use contamination in the 19th and 20th centuries. Two were accidentally used upside-down with respect to the sole published authority’s interpretation. The fourth of the four data series used by Mann08 (“thickness”) doesn’t even contain independent information–it’s just the sum of #2 (“lightsum”) and #3 (“darksum”). Citation: my blog post The Tiljander Data Series: Data and Graphs.
8. So the pre-1500 validity of the no-dendro CPS and EIV reconstructions rests on their use of four–actually three–invalid proxies. That’s awkward.
Boris, I again offer you a guest post at my blog to contest or discuss any of these statements. I have also gone back and numbered the preceding paragraphs, to help you take aim at those statements that you disagree with.
bugs: “If he had anything of substance to say about AGW, he would have moved on from the paleoclimatology by now.”
OMG, you have me ROFL on that one!!
Without the paleo there is no AGW. Just a rather pleasant, slight temperature increase from the Impending-Ice Age-Seventies, that has proven to be quite beneficial to the human condition (not to mention how great it has been for polar bears).
Wow, AMac, Steve has discovered the amazing fact that if you remove a shitload of data from a reconstruction, the reconstruction loses skill. The world’s best science blog at work.
.
yeah. anti-science at its best.
.
nobody is keeping “sceptics” from collecting long proxy data up till now on multiple continents. nobody stops them from covering every square mile of the planet with a thermometer.
.
but their focus ALWAYS is on removing datapoints, instead of adding some.
.
and the reason for this is simple: additional data points would simply confirm, what the existing ones show. the only way to show the opposite of reality, is by removing data.
Without the paleo there is no AGW. Just a rather pleasant, slight temperature increase from the Impending-Ice Age-Seventies, that has proven to be quite beneficial to the human condition (not to mention how great it has been for polar bears).
.
yeah. if we ignore physics.
.
if we don t, we see much more pleasant temperature increase coming up.
.
—————
.
most sceptics would profit a lot, from taking a closer look at the Loehle graph and paper. it shows that it is warmer today, than at any time in that 1935 years period.
.
but the average “sceptic” sees every graph with a minor peak between 500 and 1300 as a confirmation of the failure of AGW theory.
One detail missed in the discussion of why Loehle’s graph is an outlier is the same point I make at tAV every few weeks. Regression based methods on a window of data cause variance loss in the rest.
Loehle avoided some of that problem by using pre-calibrated proxy studies and using a technique little known in paleocliamte.
averaging
His plot has less of the mathematically insured pre-calibration variance loss of the other studies and therefore greater peaks and valleys. The result is certainly more representative of the data than a regression method would be but it is still proxy stuff.
One detail missed in the discussion of why Loehle’s graph is an outlier is the same point I make at tAV every few weeks. Regression based methods on a window of data cause variance loss in the rest.
Loehle avoided some of that problem by using pre-calibrated proxy studies and using a technique little known in paleocliamte.
averaging
His plot has less of the mathematically insured pre-calibration variance loss of the other studies and therefore greater peaks and valleys. The result is certainly more representative of the data than a regression method would be but it is still proxy stuff.
.
one detail often missed by “sceptics” is the fact that he throws away very good data. by removing all dentro proxies, he loses data that can be accurately dated.
.
using only pre-calibrated proxy studies leaves him with an arbitrary and limited choice of time periods spanned (that is why he ends in 1935) and places covered.
.
the attempt to transform this into an advantage is absurd.
Re: Tamara (Sep 30 08:04),
> Without the paleo there is no AGW.
I think it’s almost the opposite — It’s the physics — the radiative transfer equations (RTEs), as Steven Mosher has diligently noted — that connects rising CO2 and other trace gases (e.g. CH4) to rising average global temperature.
Solving those equations yields a direct temperature rise of about 1.2C for a doubling of CO2 concentration. Google Jeff Id’s posts at the Air Vent for details.
The contentious questions among the scientifically literate revolve around the issue of indirect rises to temperature that will be caused by rising GHG concentrations — the feedback issue.
Paleotemperature reconstructions of known accuracy and precision would be helpful in placing AGW concerns in context, and in “hindcasting” exercises for GCM models with important empirical parametric components (i.e. all GCMs, AFAIK). Conversely, reconstructions with exaggerated accuracy or precision might mislead modelers as they tune their models (my suspicion is that this is likely an ongoing problem).
But AGW’s validity does not depend on paleotemperature reconstructions.
.
Re: sod (Sep 30 08:33),
> yeah. anti-science at its best.
That’s just silly. I numbered the paragraphs in Comment#53106 to help you avoid vague sweeping statements and focus your arguments. Which paragraph do you think is “anti-science”?
sod:
One thing you miss is that dendro proxies aren’t “good data”. It’s a fools game to try and use them, at least as they are being used now, since in general they aren’t simple proxies of temperature.
Fundamentally trees are strong proxies for precipitation, weak proxies for temperature. Picking trees that reside near tree lines is a method to try and find “temperature limited” growth, but because moisture is still very important even there, shifts in climate equals shifts in precipitation to temperature patterns equals confounding factor. That doesn’t even address shifts in tree lines from climate change (e.g., MWP ← LIA ← current warming period).
The argument that “these are most of the available data” is also a very poor one. The sum of 10 wrong answers doesn’t make one right one.
Eli:
What is your definition of “not very good”?
The cross correlation coefficient is 0.59 for 193 points… even adjusting for autocorrelation, p < 0.001.
bugs:
I thought this quote was going after Michael Mann. LOL.
AMac,
I agree that the non-dendro reconstruction is pretty worthless. Too bad we don’t have enough information to get some good results for the MWP. Maybe more proxies will be developed in the near future.
But that doesn’t make the dendro reconstruction worthless. The issue with strip bark bristlecones is in the 20th century, not the MWP. And, in any case, the latest research” shows that strip bark bristlecones do not show anomalous growth in the twentieth century as compared to whole bark specimens.
Unfortunately dendro reconstructions are the only way to reliably reconstruct past temperatures. Some people have attacked the entire field because they don’t particularly like the results. I don’t think this is very interesting or notable.
Andrew_FL, thanks for the comments. Layman Lurker, Jeff and I have been talking a bit about the “right way” to compare these series.
It’s a little bit disturbing that he’s combined “blended” proxies together with individual ones. Ideally, you’d like to maintain control of how the proxies get combined yourself.
Boris:
I missed this.
Where did Amac say exclusively non-dendro reconstructions are “worthless”?
This is what I get from Amac:
So why do you think it’s worthless? You’re telling us we can throw away e.g. the Volstok deuterium ice core data now?
I meant they’re worthless for comparing the MWP to now.
And that is specifically the one from Mann ’08, not all non-dendro reconstructions.
Amac:
You’ve made a good summary. I just wanted to raise one other issue, which is the water-vapor amplification of CO2 radiative forcing isn’t particularly contentious, at least among people who understand the arguments fully.
There is some argument regarding its magnitude (typically AOGCMs using it make the assumption of constant relative humidity), but that feedback is certainly real and very tangible.
What I would say the disagreement lies in is what other feedbacks are associated with the increase in water vapor (cloud formation is the low-hanging fruit there).
Gotcha Boris. You’re referring to this from Amac:
Makes sense.
AMac,
You are looking at the issue in a much different manner than what I meant. I’m not arguing that GHGs have an effect on our climate and temperature. (I would argue about their prominence and the sign of potential feedbacks.) You are looking at the theory from its most basic interpretation – Greenhouse Theory + man’s use of fossil fuel. Simple, easy. Only if we know nothing about our past climate history, since the greenhouse theory can’t answer all of the unknowns that generated the major and minor climate swings we’ve seen in the past.
Much of the argument over Mann’s reconstruction centers around the claim that the current rise in temperature is “unprecedented.” That cannot be demonstrated by the simple physics of CO2 concentration. Is it possible to accurately predict/hindcast the temperature of the planet just from atmospheric concentration and solar intensity?
JeffId speaks alot about loss of variance. We know the past 30 years’ temperature very well, the last 150 years pretty well, and before that we don’t have a method of measurement that has a comparable precision.
Climate sensitivity is key to the projections of temperature increase. We could calculate sensitivity if we knew all of the factors, feedbacks, forces, etc. at work, but we don’t. So we guess at sensitivity based on what we know, and use hindcasts to support the guess.
So, our interest/concern about AGW has to be based on our understanding of the historical (paleo) climate.
Put another way, if there were no climate scientists, our perception of climate change would be based on the memories of the oldest living members of society. We might not be concerned for decades into the future because we have written histories, archeological evidence and fossils that demonstrate warmer periods in the past.
It is sort of a chicken/egg question. Which came first, the heat or the hysteria?
Re: Boris (Sep 30 10:52),
Dendro paleoreconstructions have a series of very large challenges to overcome, some of which are mentioned by Carrick upthread. It would be interesting and notable if a research team could come up with a validated temperature history of the past millenium or two, accompanied by robust uncertainty estimates. Even if those uncertainties were very large, it would be useful, e.g. in helping GCM model hindcasters understand the limits of their quest.
I agree that a number of people ardently attack some work (and fervently defend other work), largely because they don’t (or do) particularly like the results. Feynman would disapprove of the skeptics who do this. He would disapprove of the Consensus boosters who do this. It isn’t the practice of science (the practice of non-postmodern science, if you want).
Sod: “nobody is keeping “sceptics†from collecting long proxy data up till now on multiple continents. nobody stops them from covering every square mile of the planet with a thermometer.
.
but their focus ALWAYS is on removing datapoints, instead of adding some.
.
and the reason for this is simple: additional data points would simply confirm, what the existing ones show. the only way to show the opposite of reality, is by removing data.”
That’s funny, it is my impression from reading McIntyres blog that he is strongly in favour of adding data ponts. The most recent data points of many of these proxies are getting old and he encourages bringing these proxy series up to date. The old man even did at least one field trip himself to help adding those recent data points. The reason for the apparant disinterest of many climate scientists in bringing these proxy series up to date is simple – the di…
No, I don’t believe the explanation is necessarily that simple. Do you, sod?
Niels, I agree with you. I can’t think of many examples where “skeptics” were in favor of removing data. Quite the opposite, they usually demand more data than is really necessary. (See all the harping that Al T. has done on this blog on this topic recently for example.)
If anything, it’s the True Believer’s who don’t want more data points added, probably because they are fearful of it rocking their deeply held belief.
Amazing. Do you really think there is no research going on?
It is pretty amazing. I linked to a study with updated proxies in this very thread.
Boris,
“The issue with strip bark bristlecones is in the 20th century, not the MWP.”
Do you realize how foolish this comment is? Have you read any of the literature on the divergence problem?
Hmm, Santer et al 08 omitted data after 1999. Did the dog eat it?
“Carrick (Comment#53047) September 29th, 2010 at 1:12 pm
Also, here is Mann’s 2008 paper.”
Glad you pointed this paper out. His first conclusion:”we find that the hemispheric-scale warmth for the past decade for the NH is likely anomalous in the context of not just the past thousand years…, but longer..for at least the past 1300 years…this can be extended back to at least past 1700 years…”
This is a very curious conclusion when the graphs clearly show as many as 5-8 decadel periods with similar abrupt changes in temperature, both positive and negative of ~.5-.6 degreeC. This obvious feature of the data is completely ignored. The interpretation would be that sudden decadel changes of this size are well-represented in the record and the last decade is not anomalous.
The question then becomes whether or not the current temperature is anomalously high. The article contains plenty of caveats that the proxy data, particularly further back in history, is not reliable enough to justify a claim that the most recent decade is anomalously high, even though that claim is made.
Carrick, compare it to the cross correlation between Ljungqvist and Moberg or Mann EIV
bugs (Comment#53141)-“Amazing. Do you really think there is no research going on?”
The complaint dates to 2005:
http://climateaudit.org/2005/02/20/bring-the-proxies-up-to-date/
In 2007 we were told it couldn’t be done:
http://climateaudit.org/2007/10/26/eli-rabett-explains-why-realclimate-scientists-cant-update-the-proxies/
(hehe, ass end of nowhere)
Steve has actually made some posts about proxies that finally are up to date. Look through:
http://climateaudit.org/category/proxies/post-1980-proxies/
So yep, there is some effort to update the proxies, but it is not nearly good enough. We had reconstructions throughout the 2000’s being made with proxies all ending in 1980 or earlier. Now a few proxies are updated to two of the last three decades. Great. But would it really have been asking too much to have this more completely done, five years ago?
Eli:
Why not Mann CPS?
Oh I know. Rabbits love picking cherries! 😉
Not that much different actually (using NH when available).
Mann CPS: 0.58
Mann EIV: 0.76
Moberg: 0.57
Your Leporidae eyes are getting fooled because Loehle has a higher variance than the other series (of course cross correlations take care of this, nor do they care about the offsets).
Eli, in terms of spectral performance, all of these other proxy reconstructions stink in comparison to Loehle. I know for a follower of the Mann in the Red Cape that isn’t news you wanted to hear.
Loehle-vs-Wild.
They look like low-passed noise (Mann ran his series through a 10-year low-pass Butterworth filter, but sampled them at 1-year intervals.)
Mann’09 does a bit better. Sorry too lazy to plot on the same curve as Loehle, maybe I need more vitamin A.
tell me one of you believes the CIs in the last figure around the year zero.
go ahead. and then explain how a proxy estimate of year 0 can have less certainty than CRU 1850 estimates.
Osborn pointing this out to Mann about one of manns recons.
So, I’m channelling Tim, tonight
Stephen Mosher:
Or in general why the uncertainties don’t grow as you go further back in time. Seems to me they have to, if you’ve done the analysis properly.
Re: steven mosher (Oct 1 03:47),
> tell me one of you believes the CIs in the last figure around the year zero.
I estimate that Ljungquist’s reconstruction estimate for the year 0 is 0.0C +/1 0.1C. I certainly believe that — “believe” in this case meaning that I accept that Ljungquist did their uncertainty arithmetic correctly, and that that number turns out to be a tenth of a degree.
The important question — of course — is what “uncertainty of 0.1C” means. In other fields, people offer numbers and graphs like this to signify something akin to “Our findings allow us to estimate that the probability that the NH temperature anomaly was within 0.1C of 0C is 68%” (or 95% for a two-sigma convention).
I believe that such an interpretation of the calculated “uncertainty” values of Lungquist, Mann, Loehle, et al. is absurd.
It would sure help if these folks would state in plain language what they think their uncertainty values actually mean. I think that exercise would cause some raised eyebrows among citizen-scientists, lay people, and policymakers. (In fairness, alone among these names AFAIK, Loehle has been frank about the limits to the real-world meaning of his reconstructions.)
bugs said (Sep 30 15:35)
and Boris responded (Sep 30 18:11)
That study is Salzer et al, “Recent unprecedented tree-ring growth in bristlecone pine at the highest elevations and possible causes,” in PNAS, 2009 (full text online, yay).
1. Salzer took new bristlecone pine tree-ring series in the Intermountain West, and found that trees at the treeline (at the elevation above which they could not grow) laid down tree rings with a good correlation to inferred regional temperature. As expected, the correlation wasn’t present for trees growing at elevations well below the treeline.
2. The important and novel finding was that this at-treeline correlation continued beyond 1960. In fact, the mean width of the tree rings for 1950-2000 was unprecedented, greater than for any other half-century, stretching back to 1400 (Sheep Mtn, Fig 3), or back millenia (all records, Fig. 2). This implies unprecedentedly high temperatures at the treeline trees’ habitat, 1950-2000. Preliminary data extend the findings to 2005.
Boris, bugs — Agree or disagree with this summary?
3. To me, what’s interesting is the contrast between Salzer et al (2009) and the data that led to dendrochronologists’ recognition of the so-called Divergence Problem. This was floated by Briffa, Schweingruber, et al in Nature in 1999, “Reduced sensitivity of recent tree-growth to temperature at high northern latitudes” (Abstract). Esper et al see the DP as a methods problem in “Trends and uncertainties in Siberian indicators of 20th century warming”, Global Change Biology (2010; full text PDF). A recent review is Esper and Frank, “Divergence pitfalls in tree-ring research,” Climactic Change (2009; full text PDF).
4. The DP is what led many climatologists to decide to truncate the calibration periods for many paleoreconstruction studies near the 1960 mark. By doing so, they could avoid spoiling the goodness-of-fit between treerings and instrumental temperature records, and produce temperature reconstructions with tighter “uncertainties.” Some of these discussions are captured in the leaked Climategate emails.
Boris, bugs — Agree or disagree with the narrative in #3 and #4?
5. The DP has been discussed in some lengthy threads at The Blackboard, including these two. Pro-AGW Consensus advocates haven’t addressed the concerns that Steve Mosher, Carrick, myself, and others have raised about the theoretical and statistical problems with the way the DP has been handled by Consensus scientists, in my opinion. And graphical depictions of paleotemperature reconstructions have obscured rather than explained the uncertainties that an airing of the DP adds to our understanding of past climate — a significant communication deficit.
Boris, bugs — Three opportunities for comment.
A. Salzer et al basically say that in their properly-selected and properly-analyzed data set, the DP doesn’t exist.
B. Esper & coauthors in the 2009 and 2010 papers basically say that the DP can be seen in properly-selected data sets, but that sophisticated methods of analysis can account for most of it.
C. Scientifically-literate lukewarmer commenters at The Blackboard basically say that the DP was incorrectly handled by Consensus scientists, 1998-2008. Don’t obscure conceptual problems in your field — Post-hoc statistical analyses are invalid unless acknowledged and corrected for — Consensus scientists’ communication to lay audiences by graphical means falsely exaggerated the strength of the methods and the data that were used to build paleotemperature reconstructions.
Your thoughts?
Mosh, Carrick, a good check on uncertainty (error) through time would be to standardize and measure different reconstructions (or proxies) against themselves. If uncertainty truly is a function of time, then the differences between reconstructions should increase as you move further into the past. In the Loehle/Ljungqvist case, this comparison can be done by rescaling and offsetting one series WRT the other. After doing this, here is the graph of the differences:
http://picturepush.com/public/4275449
Layman Lurker, another good way to look at it is to look at the time series of the cross-correlation coefficient windowed, e.g., over 500-year periods. I’ll try and do that later today and show the results here (and on Jeff’s thread).
AMac, Not sure what you are looking for here.
The divergence problem was always considered to be most likely a twentieth century phenomenon and not necessarily a phenomenon due to the decoupling of ring growth and temperature in warmer periods. There was lots of evidence to this effect: stable year to year changes in growth, the relatively small set of trees affected by the DP and known tree growth characteristics that serve as the fundamentals of dendrology. However, the possibility that tree growth might reach some upper barrier in periods about as warm as the late 20th Century could not be eliminated from consideration, so studies of the DP mentioned it. But the expert opinion of the guys in the field who had done this stuff for decades was against it. As a result, in some cases (though not all) this expert judgment led to dropping the late twentieth century lack of growth because it was very likely spurious.
Now we find the literature catching up to expert opinion and we have both more qualitative and quantitative analyses which show that the DP is not a phenomenon that invalidates temperature reconstructions. I still don’t think causes are 100% settled, but we are much closer.
So in drafting the IPCC reports the experts used expert judgment, which is what we are asking them to do. I don’t see any fault with that approach, nor do I see a problem with other experts disagreeing. I think the TAR and AR4 give plenty of caveats about tree ring reconstructions related to this and other problems, which is absolutely sufficient.
Layman Lurker, here’s the correlation coefficients over time.
As you can see, the correlation does tend to get worse as you go backwards in time. These are all against Ljundqvist.
Proxy Correlations Versus Time
I think sod makes a valid point, inasmuch as “skeptics” are mostly focused on criticizing and reinterpreting existing data sets that support the theory of AGW. NAN’s citation of a rumor that — perhaps — McIntyre had once been on a “field trip,” maybe even more than one, contrasts humorously with the many hundreds of thousands of hours logged in the field by the “True Believer’s.”
Of course, the disparity may not be there because of philosophical differences; it may simply reflect the fact that “skeptics” are extremely unlikely to possess the training or experience in climate science that would lead someone to trust them with the budget to do field research. So you rarely see the “skeptics” add data points for the same reasons you rarely see clinical trials on vaccine efficacy from Jenny McCartney et al.
Still, you’d think the “skeptics'” friends in the fossil fuel industry, who have been so generous in support the “skeptics'” outreach to the public, could fund some primary research to crack the whole “AGW” scam right open.
Interesting Carrick. Thanks for that.
Re: Boris (Oct 1 13:40),
> AMac, Not sure what you are looking for here.
Boris, you linked Salzer et al., saying “And, in any case, the latest research shows that strip bark bristlecones do not show anomalous growth in the twentieth century as compared to whole bark specimens.”
You and bugs were then amazed that this citation didn’t effect a conversion of commenters to the AGW Consensus position.*
The reason is that Salzer highlights the problems with the Divergence Problem as much as it provides a solution.
Many climatologists give themselves tremendous leeway in how they frame and attempt to solve the problems in their subject area.
Researchers working in other fields in the physical sciences do not (can not) do so. To my knowledge.
In particular, the extent to which treering growth of a member of a species growing at the current upper-elevation limit of its habitat has been correlated with temperature throughout the past 1,000 or 2,000 years is not settled science. To the extent hypotheses on the subject can be formed and tested, they should be.
But to start with,
— “Let’s use all the data in constructing a calibration curve,”
then change it to,
— “Post-1960, the data diverges from expectations, so let’s end the calibration period at 1960,”
then segue to,
— “Different data-manipulation methods make post-1960 data look like we think it should, so let’s use them with these methods,”
ending with,
— “Hey, new tree samplings don’t show a DP, so DP isn’t a problem,”
is unsatisfactory.
Calibration curves tailored from the periods of best correlation will, naturally, lead to reconstructions that have narrower calculated uncertainties. Taken at face value, the reconstructions will seem more robust than would otherwise be the case.
Somehow, climate scientists fail to see the problems with these sorts of procedures. It’s a small step from there to producing illustrations of the reconstructions that “Hide the Decline.”
——–
* that’s hyperbole on my part. But the meme that “people who don’t fully agree with the AGW Consensus must be anti-science” has become an article of faith among some in the advocacy community.
AMac,
You misunderstood. I wasn’t referencing the divergence problem, I was referencing the problem with bristlecones specifically.
The issue was this: Strip bark Bristlecones show a lot more growth than whole growth forms. In 2006, the NAS panel recommended that Bristlecones not be used in climate reconstructions. A lot of dendro people disagreed and set out to show that there was useful temperature information in Bristlecone series.
Enter Salzer 2009 which updated the Bristlecone proxies and showed that the difference between whole bark and strip bark was an artifact of the processing used. They used the raw data to show that ring growth in whole bark and strip bark was the same in the late twentieth century.
To be clear, Salzer is only discussing the divergence between the two types of Bristlecones and not the big picture divergence problem.
I don’t even know what this means. It seems like generalized smearing to me.
I have no idea what you mean by “settled science.” But maybe you have some citations you were thinking of?
You guys are always hung up on procedures and you forget there are actual physical trees out there. It’s not like dendros shop around for a procedure that gives them the result they want. They notice a problem and attempt to find a solution. But then, the divergence problem is not as widespread as “lukewarmers” would have people believe. How many of Mann’s series show a post 1960 divergence? Do you even know?
Tamino is flat wrong. Craig’s series gives anomalies relative to the mean of the whole period, while Lundqvist’s is relative to the 1961-90 reference period of CRUTEM3+HadSST2. Any comparison has to be made relative to a common period. (In fact, Lundqvist matched the mean of his series over 1850-1989 to that of the CRU+HAD series over the same period, which could be slightly different, but close enough for this discussion.)
Zeke’s revison compares Loehle centered on zero 1880-1930 to the others centered on zero over 1880-1970 or 1990. This is a big improvement over Tamino, but still a) any “short” period of only a century or so contains a lot of noise that will affect the whole comparison. Using the longest available common period would average most of this noise out, and made the comparison most valid. And b) Zeke’s periods still aren’t the same for all series, and so still aren’t strictly comparable, even aside from the noise issue. Craig’s original comparison on WUWT is the most valid.
No — it just shows that the MWP was warmer than the tridecade centered on 1935 (1921-1949), at least in terms of point estimates. Since 3 proxies dropped out by 1950, leaving less than half of the full 18, the reconstruction ends there. It says nothing about temperatures today.
Although the CIs I contributed show that Craig’s MWP was significantly warmer than the bimillenial average and even moreso than the LIA, the difference between the MWP and 1935 isn’t going to be significant at the 5% level. See http://www.econ.ohio-state.edu/jhm/AGW/Loehle/.
Re: Boris (Oct 2 08:20),
Thanks for the careful read.
> You misunderstand.
OK, thanks for the clarification. Sometimes it helps to be specific about why you’re bringing up something, when you first raise it. In any case, Salzer et al 2009 was an interesting paper.
My interpretation was that Salzer sampled new trees, not that they updated the records of already-sampled trees. Is that right?
As far as the Divergence Problem, my understanding was that bristlecone pines are major players in that story. The DP is explicitly discussed in Salzer’s text.
Boris, you accuse me of engaging in “generalized smearing” of climate scientists. That’s a new one for me. I’ve been very specific as to my concerns–in this thread, in conversations with you (e.g. about Tiljander) and in (nearly) all of my comments at the Blackboard and elsewhere (e.g. C-a-s). If you review what I’ve written on this thread, comments have been on unlikely uncertainty calculations (#52961), conflating science and advocacy (#52976), performing post-hoc statistics without corrections (#53013), preferring preferred results more than robust methodologies (#53018), Mann08’s EIV and CPS no-dendro reconstructions requiring contaminated uncalibratable Tiljander to have “skill” prior to 1500 (#53082, #53106); confirming the C02-temperature connection (#53116); unlikely uncertainties again (#53137, #53179), and finally Salzer and the DP (#53193, #53217).
If that’s generalized smearing: do your comments live up to your standard?
For the phrase “settled science”, you can get context by Googling AGW science is settled. Citations for “treering growth at current upper-elevation limit of its habitat” — linked in #53193. Briffa 1999, Esper 2010, Esper 2009.
Yes, No (forget about actual trees–huh?!), and It sometimes looks that way (e.g. Mann08).
No, I don’t know; it’s an interesting question. What’s the answer?
However, recall that Mann08’s calibration period extended to 1995, though he had to extrapolate some data series to reach that date. It thus seems to me that his calibration period deals with whatever the DP served up, post-1960. So that’s one instance where Mann08 used the better approach.
Hu:
Agreed. Very sloppy analysis in this case, and the only way he gets away with it is running an echo chamber with no dissenters allowed.
As I showed above, Craig is correct that his series generally agrees with Ljungqvist, Tamino also doesn’t understand the issues well enough to recognize why the Loehle series might legitimately have a larger variance.