Q.e.d.

One point that Wahl and Ammann and ourselves agree on, but which Juckes appears to contest, is that principal components methodologies applied to AD1400 MBH98 networks result in upweighting or downweighting of bristlecones. Their Scenario 6 shows reconstruction results without bristlecones for covariance, correlation and Mannian PCs. I have no dispute with these calculations although I would characterize the situation differently.

The Figure below shows results for the key cases for WA Scenario 6 (no bristlecones) as compared to their Scenario 5. Understanding this graphic goes a long way towards understanding the oceans of words on this topic. Pink shows results with 2 covariance PCs; red with Mannian PCs; blue – 2 correlation PCs and 5 covariance PCs; orange – Mann’s unreplicated archived results, which are lower in the 15th century than any replication.

The left graphic is identical to what was either illustrated or discussed in MM05(EE); I have no major disagreement with the calculations in WA Scenario 5 – only with their failure to acknowledge that their results were equivalent to the MM05 (EE) results. Wahl and Ammann Scenario 6 – without bristlecones – yields results that are essentially equivalent to results with 2 covariance PCs in the left panel (what people call the “MM” results, though they are really just MBH with lower bristlecone weight.) If there are no bristlecones in the network (as in the right panel), then the different standardization procedures don’t “matter”. But the methodology makes a difference to the final reconstruction if bristlecones are in the system.

This diagram also illustrates rather clearly the lack of substance to Juckes’ complaint that using 2 covariance PCs “effectively eliminates” much of the data. This is not correct. The results using 2 covariance PCs are essentially identical to results without bristlecones. It is simply false that “much” of the data is “effectively eliminated” using 2 covariance PCs. The only data whose downweighting is material are the bristlecones.

Figure 1. MBH98-style NH Temperature Reconstructions. Left – WA Scenario 5 as previously described. Right – WA Scenario 6 with bristlecone series excluded. Orange – MBH98 for reference. Red – with two Mannian PCs; magenta – with 2 covariance PCs; blue – one graph with 2 correlation PCs; one graph with 5 covariance PCs. All smoothed with 21-year gaussian filter. wahl.c11.gif
This direct connection between bristlecones and PC methodology (mentioned in MM05 articles) was acknowledged by Wahl and Ammann as follows:

Restricting the PCs in MM05a/b to only the first two (5d) indirectly omits the information carried by the bristlecone/foxtail pine records and thereby leads to a non-meaningful reconstruction. (33)

Now Ross and I categorically agree with Wahl and Ammann that an MBH-style reconstruction without bristlecones is non-meaningful. Our point of difference is that we assert that an MBH98-style reconstruction with bristlecones is also non-meaningful. This is a different issue than whether the 15th century values of the reconstruction are higher than 20th century values.

If the reconstruction isn’t meaningful, it doesn’t matter whether the 15th century portion is higher or lower than the 20th century. (However, if it is higher using Mannian methods, then that is a rather neat refutation of the claim that 20th century uniqueness has been established with Mannian data and methods – regardless of whether the reconstruction is meaningful or not. This logic is a nuance of our approach that many controversialists lose sight of.)
In my recent European presentations (including at KNMi to Mitrie coauthor Nanne Weber), I summarized our position post-NAS panel and post-Wahl and Ammann as follows:

1) Wahl and Ammann and ourselves agree that an MBH98-type reconstruction without bristlecones is non-meaningful.

2) the NAS panel agreed that strip-bark sites (which include all the relevant bristlecone, foxtail and even a couple of limber pine sites) should be avoided in temperature reconstructions for a variety of reasons.

Q.e.d.

I see no remaining wiggle room for MBH supporters. However rather than squarely facing up to the bristlecone problem, Juckes simply avoided any discussion of the impact of bristlecones, even though this was squarely in the middle of the agenda, both as a result of our work and even of Wahl and Ammann. Instead, as we’ve seen here, Juckes reverted to realclimate code words, “effective elimination of much of the data”, rather than careful analysis distinguishing between bristlecones and everything else.

1. PHE
Posted Nov 30, 2006 at 3:35 AM | Permalink

My understanding from the ‘hockey stick’ debate was that without the bristlecone pines, the tree-ring proxies do not show a significant 20th century rise. From this post, it seems like I misunderstood this point. What is the explanatoin?

2. Posted Nov 30, 2006 at 5:05 AM | Permalink

As you know, McIntyre, my statement “effective elimination of much of the data” was not referring to Bristlecones. What are you trying to imply with your idiotic reference to “code words”?

As far as I can tell above, you are showing than using the Mann et al. regression technique without the Mann et al. validation technique can produce a meaningless result? Our paper also concludes that inverse regression is not suitable for the AD1000 to present reconstruction with the data we had (I add this caveat because the choice of statistical algorithm depends on the data being used).

McIntyre puts a lot of weight on one sentence in the NAS panel report, who devote a whole paragraph to this issue. I prefer to base my decision on reading the literature on which the report was based. The most detailed study on strip-bark trees, by Bunn et al. (2003), confirms anomalous growth in strip-bark samples, but does not in any meaningful sense correlate with CO2 (substantial differences observed in 1850s). A recent review by Kostiainen et al (2004) states “We conclude that elevated [CO2] had only minor effects on wood properties while fertilization had more marked effects and thus may affect ecosystem processes and suitability of wood for different end-use purposes.”

Re #1: The 20th century rise in the above figures comes from the measured temperatures which the proxies are regressed against (using inverse regression). You should also note that the technique used by Mann et al. uses different amounts of data for different periods. There is greater uncertainty about the 15th century because of the relatively small amount of proxy data that extends back that far. The later parts of the reconstructions shown above use far more data and thus have less sensitivity to the various permutations being discussed.

3. Jean S
Posted Nov 30, 2006 at 5:18 AM | Permalink

PHE, the explenation is shortly that there are other hockey stick shaped series, and with the Mannian reconstruction (after the PC calculations), those get weighted. Mannian methodology per se is like every other work of the Team: it (over)fits a set of proxies (or any series, I can do it even with some correlated noise) to the instrumental record.

Let me try the explain the point in the above graphs as it is IMHO somewhat hidden in Steve’s post:

1) Notice that this is AD1400 step. This means that only 1400-1499 (this is the period W&A used, in real MBH it would be 1400-1449, but this is not essential here) part of the graphs is meaningful.
2) Now look this AD1400 part of the orange curve (in both figures). This is the “real MBH98”. The red one in the left graph is essentially the “W&A emulation” of that. The small difference shows that W&A can pretty much replicate MBH98, although
Mann still “beats” them.
3) Compare that (i.e., orange or red curve on the left) to the rest of the curves:
a) The left figure: pink curve (covariance PCs used): the construction goes up around 0.3C! Blue curve
(correlation PCs), still the graph goes up about 0.2C. So in summary: in order to have the MBH AD1400 part
where it is, it is essential to use Mannian PCs.
b) The right figure (bristlecones exluded): Now all scenarios (even Mannian PC methodology) go up to where only covariance PC scenario is in the left figure. Summary: bristlecones are needed in order to have MBH98 AD1400 part where it is.

Conclusion: It is essential to use BOTH Mannian PC calculations AND bristlecones to have MBH98 AD1400 step as low as it is. Doing something else would raise the “tail” of MBH98 at least 0.2C, more likely 0.3C.

[Steve: I think magnenta is missing from the right!]

4. TAC
Posted Nov 30, 2006 at 5:40 AM | Permalink

#1 Over the past couple of years, SteveM has meticulously dissected the MBH methodology and found flaws at nearly every step. For multiple reasons, the MBH findings are meaningless.

Perhaps a more careful study could extract some useful climate-relevant signal from existing proxies, but so far no one, including SteveM, has done so (I am not convinced that it would even be possible, given all of the issues that SteveM and Ross have identified). So the most we can conclude is that the existing proxy studies contribute little or nothing to our understanding of the millennial history of climate.

Obviously, a lot of work remains to be done. Some of the flaws in existing reconstructions could be corrected. Perhaps more and better proxy data can be found. Better statistical methods can be developed. Until these things are done, however, there isn’t a lot to say. The important thing is to recognize the situation for what it is.

The role of the statistician is to quantify ignorance, not expand knowledge.

5. mikep
Posted Nov 30, 2006 at 9:03 AM | Permalink

re 4
And knowledge of ignorance is the beginning of wisdom.

6. bender
Posted Nov 30, 2006 at 9:28 AM | Permalink

PHE, you didn’t “misunderstand the point”. You just have to pay very careful attention to what these graphs illustrate. It takes some effort, but follow the instructions in #3 and you’ll see why #2 is flaming angry, to the point of accusing Steve M of idiocy.

7. bender
Posted Nov 30, 2006 at 9:39 AM | Permalink

Dr Juckes, ignore Steve M’s use of the phrase “code words” and focus on his substantive point: you guys don’t understand the data you’re working with. i.e. You do not understand how these data interact with various codes to produce the full range of flavors of reconstruction that can be produced. Is this not true? There are only two conclusions available to me: (1) you are ignorant of these interactions; (2) you are cognizant of them, but don’t care to reveal the dynamic at play. Which is it?

You are complaining about Steve M’s insulting choice of words, and not attacking his argument. Why not? I suspect it is because his analysis is correct.

Posted Nov 30, 2006 at 9:57 AM | Permalink

A bit of context switching from the world of signal integrity and stability of large highly complex multinode electronic circuits (as in computer systems). With BCPs the waveform is a square wave or even one time step function, whereas without BCPs it is more akin the a sawtooth wave, possibly even a quasi periodic one. Given the various interacting, harmonic rich oscillations of the oceans and atmosphere, this would intuitively make sense, whereas, the step function would not. Some in the climate science community have had very little if any exposure to this way of looking at things, very few of them have taken analogue electrical engineering courses or other analogous ones (maths wise) in other disciplines, so I innately excuse them for their ignorance. However, Mann has a physics background and assuming he passed undergrad upper division E&M courses ought to know better, which makes his behavior therefore all the more disturbing.

9. bender
Posted Nov 30, 2006 at 10:30 AM | Permalink

To be fair, SS, not all the indiviudal bcp trees show the square-wave/step-function you describe. There is the full range of slopes from gradual to abrupt. As it is the individual tree that responds to its environment, not the collective population, this inter-tree variability should not be overlooked.

10. John Hekman
Posted Nov 30, 2006 at 11:32 AM | Permalink

In #2, Martin Juckes makes several statements that can be addressed by Steve or others here.

First, he acknowledges the NAS report’s request to avoid the use of Bristlecones, but says that he pays more attention to the literature to which it refers. So, does that literature allow him to ignore the conclusion that Bristlecones should be avoided?

Second, he discusses the Bristlecone growth anomaly and states that it has been found not correlated with CO2 trends. So, is he arguing for the use of Bristlecones? Does he concede that the growth anomaly is not related to temperature changes, since it has been reported, if I remember correctly, that temps in the White Mountains did not increase significantly over the relevant period?

Reading Juckes’ comment here, I am left saying, yeah, but what about…?

11. EP
Posted Nov 30, 2006 at 11:45 AM | Permalink

Conclusion: It is essential to use BOTH Mannian PC calculations AND bristlecones to have MBH98 AD1400 step as low as it is. Doing something else would raise the “tail” of MBH98 at least 0.2C, more likely 0.3C.

In short: bristlecone data is an outlier?

12. Ross McKitrick
Posted Nov 30, 2006 at 11:47 AM | Permalink

Martin, the full paragraph from the NAS Report (p. 50) is:

The possibility that increasing tree ring widths in modern times might be driven by
increasing atmospheric carbon dioxide (CO2) concentrations, rather than increasing
temperatures, was first proposed by LaMarche et al. (1984) for bristlecone pines (Pinus
longaeva) in the White Mountains of California. In old age, these trees can assume a “stripbark”
form, characterized by a band of trunk that remains alive and continues to grow after the
rest of the stem has died. Such trees are sensitive to higher atmospheric CO2 concentrations
(Graybill and Idso 1993), possibly because of greater water-use efficiency (Knapp et al. 2001,
Bunn et al. 2003) or different carbon partitioning among tree parts (Tang et al. 1999). Support
for a direct CO2 influence on tree ring records extracted from “full-bark” trees is less conclusive.
Increasing mean ring width was reported for Pinus cembra from the central Alps growing well
below treeline (Nicolussi et al. 1995). Free-Air CO2 Enrichment (FACE) data for conifer
plantations in the Duke Forest (Hamilton et al. 2002) and at the alpine treeline (HàÆà⣴tenschwiler et
al. 2002) also showed increased tree growth after exposure to atmospheric CO2 concentrations
about 50 percent greater than present. On the other hand, no convincing evidence for such effect
was found in conifer tree ring records from the Sierra Nevada in California (Graumlich 1991) or
the Rocky Mountains in Colorado (Kienast and Luxmoore 1988). Further evidence comes from
a recent review of data for mature trees in four climatic zones, which concluded that pine growth
at treeline is limited by factors other than carbon (KàÆà⵲ner 2003). While “strip-bark’ samples
should be avoided for temperature reconstructions, attention should also be paid to the
confounding effects of anthropogenic nitrogen deposition (Vitousek et al. 1997), since the
nutrient conditions of the soil determine wood growth response to increased atmospheric CO2
(Kostiainen et al. 2004). However, in forest areas below treeline where modern nitrogen input
could be expected to influence dendroclimatic records, such as Scotland (Hughes et al. 1984) and
Maine (Conkey 1986), the relationship between temperature and tree ring parameters was stable
over time.

How does the full paragraph change the conclusion that the bcp’s should not be used? Pointing to literature that posits other forms of contamination than CO2 does not alter the point that the growth spurt is not a temperature signal.
When you say

As far as I can tell above, you are showing than using the Mann et al. regression technique without the Mann et al. validation technique can produce a meaningless result?

all I can say is go back and re-read it. Using the Mann et al. regression technique and applying the Mann et al. validation technique still leaves invalid results.

Take out a small group of proxies that have been red-flagged numerous times as invalid for temperature measurement, and both the hockey stick shape and any possibility of verification significance goes away. Agreed? Now add them in again. First, whatever results depend on use of faulty data are themselves faulty. Second, the ‘validation technique’ involves verification-period test scores, such as r2, CE and RE. On this point, once again referring to the NAS Report, p. 107:

Regarding metrics used in the validation step in the reconstruction exercise, two issues
have been raised (McIntyre and McKitrick 2003, 2005a,b). One is that the choice of
“significance level” for the reduction of error (RE) validation statistic is not appropriate. The
other is that different statistics, specifically the coefficient of efficiency (CE) and the squared
correlation (r2), should have been used (the various validation statistics are discussed in Chapter
9). Some of these criticisms are more relevant than others, but taken together, they are an
important aspect of a more general finding of this committee, which is that uncertainties of the
published reconstructions have been underestimated.

The verification scores themselves — even with the bcp’s included — show that the model has no reconstructive skill prior to the late 1700s. Referring to the negative CE scores over this interval the NAS (p. 91) said

Reconstructions that have poor validation statistics (i.e., low CE) will have
correspondingly wide uncertainty bounds, and so can be seen to be unreliable in an objective
way. Moreover, a CE statistic close to zero or negative suggests that the reconstruction is no
better than the mean, and so its skill for time averages shorter than the validation period will be
low. Some recent results reported in Table 1S of Wahl and Ammann (in press) indicate that their
reconstruction, which uses the same procedure and full set of proxies used by Mann et al. (1999),
gives CE values ranging from 0.103 to –0.215, depending on how far back in time the
reconstruction is carried.

Without the bcp’s there is no hockey stick one way or the other and verification scores are clearly insignificant. With the bcp’s the graph shape depends on faulty data and the verification scores are still insignificant. The results can be seen to be unreliable in an objective way regardless of which way the analysis is done. QED.

13. bender
Posted Nov 30, 2006 at 12:01 PM | Permalink

Re #12 Thanks for the summary. Well done, I thought.

14. jae
Posted Nov 30, 2006 at 12:28 PM | Permalink

In short: bristlecone data is an outlier?

Yes, and not only that, but the data should not be used because:
1.) ring widths are magically correlated to “global temperatures,” but not to local temperatures.
2.) many scientists have argued that there are non-climatic reasons for the growth characteristics of these trees.

15. Posted Nov 30, 2006 at 1:10 PM | Permalink

Martin:

Reading your response #2, I was immediately reminded of the following Arabian proverb (see p. 47 in Chapter 3 Completeness and Complexity in Fagin, R. et al. 1996, Reasoning about Knowledge, MIT Press.)

There are four sorts of men:
He who knows not and knows not he knows not: he is a fool – shun him;
He who knows not and knows he knows not: he is simple – teach him;
He who knows and knows not he knows: he is asleep – wake him;
He who knows and knows he knows: he is wise – follow him.

16. bender
Posted Nov 30, 2006 at 1:38 PM | Permalink

Re #14 (2)
Even if the response is “climatic” that doesn’t mean it’s a linear proxy for temperature. (1) The response could be nonlinear and synergistic between precip and temp. (2) The response could be attenuated the further you go back in time.

Most dendroclimatologists know this. Follow them. The Teams do not act on this knowledge. Shun them.

17. Jean S
Posted Nov 30, 2006 at 2:09 PM | Permalink

Martin:

A recent review by Kostiainen et al (2004) states “We conclude that elevated [CO2] had only minor effects on wood properties while fertilization had more marked effects and thus may affect ecosystem processes and suitability of wood for different end-use purposes.”

Here’s the abstract of “the review paper”

Kostiainen, Katri, Kaakinen, Seija, SaranpàÆà⥃ Æà⢬ Pekka, Sigurdsson, Bjarni D., Linder, Sune & Vapaavuori, Elina: Effect of elevated [CO2] on stem wood properties of mature Norway spruce grown at different soil nutrient availability. Global Change Biology 10 (9), 1526-1538.
in full:

The objective of the present study was to investigate the interactive effects of elevated [CO2] and soil nutrient availability on secondary xylem structure and chemical composition of 41-year-old Norway spruce (Picea abies (L.) Karst.) trees. The nonfertilized and irrigated-fertilized trees were, for 3 years, continuously exposed to elevated [CO2] in whole-tree chambers. Elevated [CO2] decreased concentrations of soluble sugars, acid-soluble lignin and nitrogen in stem wood, but the effects were not consistent between sampling height and/or fertilization. The effect of 2*ambient [CO2] on wood structure depended on the exposure year and/or fertilization. Radial lumen diameter decreased and annual ring width increased in the second year of exposure (1999) in elevated [CO2]. In the latter, the CO2 effect was significant only in the nonfertilized trees. Stem wood chemistry and structure were significantly affected by fertilization. Fertilization increased the concentrations of nitrogen and gravimetric lignin, annual ring width, and radial lumen diameter. Fertilization decreased C/N ratio, mean ring density, earlywood density, latewood density, cell wall thickness, cell wall index, and latewood percentage. We conclude that elevated [CO2] had only minor effects on wood properties while fertilization had more marked effects and thus may affect ecosystem processes and suitability of wood for different end-use purposes.

I don’t see how the above supports your claim that the elevated CO2 would have no/little effect on bristlecone pines. I sure can understand why it was cited by NAS (as Ross quoted in #12). But I’m not a biologist, so I may be completely wrong here.

18. Brooks Hurd
Posted Nov 30, 2006 at 6:12 PM | Permalink

Martin Juckes,
This morning, I was perusing the posts when I woke up at 5:40 AM. I saw your Post #162 in the Pea Under the Thimble thread which is missing for some reason this afternoon. Believe me, I have been called worse things in the past, however what you are doing is nothing more than a simple logical fallacy known as an ad hominem.

When I said that the “Hockey Stick” is almost the cornerstone of the IPCC TAR, it is because the Summary for Policymakers is what most people in the press, and government would read. The TAR itself is quite large and it is highly unlikely that busy reporters or elected officials would get very far beyond the Summary for Policymakers. I do not think that the title of this document can be any clearer.

There are two graphs on this document: Jones’ instrument data, and the hockey stick from MBH98. In the text of the summary it says:

It is also likely7 that, in the Northern Hemisphere, the 1990s was the warmest decade and 1998 the warmest year (Figure 1b).

This comes right out of MBH98. Figure 1b is the hockey stick.

MBH99 states in the conclusions:

…the 1990s are likely the warmest decade, and 1998 the warmest year in at least a millennium.

On page 3 of the summary, under the topic There is new and stronger evidence that most of the warming observed over the last 50 years is attributable to human activities, the hockey stick is referenced again:

Reconstructions of climate data for the past 1,000 years (Figure 1b) also indicate that this warming was unusual and is unlikely7 to be entirely natural in origin.

If any policymakers would go beyond their summary, they might read the Technical Summary, the Scientific Basis. On page 1 of the Technical Summary under the topic of Surface temperatures during the pre-instrumental period from the proxy record we see the hockey stick once again labeled as Figure 5. The second sentence of this section states:

The 1990s are likely to have been the warmest decade of the millennium in the Northern Hemisphere, and 1998 is likely to have been the warmest year.

In the body of the TAR, the hockey stick is labeled Figure 2.20. Figure 2.21 on the same page is is a spaghetti graph with the hockey stick and other reconstructions. From 2.3.2.2 Multi-proxy synthesis of recent temperature change

Mann et al. (1999) concluded that the 1990s were likely to have been the warmest decade, and 1998 the warmest year, of the past millennium for at least the Northern Hemisphere.

The next page of the TAR is 2.3.3 Was there a “Little Ice Age” and a “Medieval Warm Period”? The conclusions of this page as posited by the section title can be found in paragraph 5:

As with the “Little Ice Age”, the posited “Medieval Warm Period” appears to have been less distinct, more moderate in amplitude, and somewhat different in timing at the hemispheric scale than is typically inferred for the conventionally-defined European epoch.

This conveniently supports the temperature reconstructions in MBH98 and MBH99.

Although I could continue, I think that my point that the hockey stick is the essentially the cornerstone of the TAR, at least in the area of pre-instrumental temperatures. Without MBH98 and MBH99, what would the temperature history in the TAR have been based upon? Phil Jones’ instrument record? Two centuries would hardly have been as persuasive as the ten century graph and quotes from MBH98 and MBH99.

19. bender
Posted Nov 30, 2006 at 6:58 PM | Permalink

But I’m not a biologist, so I may be completely wrong here.

I am a biologist, and you’re not wrong, Jean S.
Low-elevation Norway spruce and high-elevation bristlecone pine can be expected to be limited by very different factors. Furthermore the CO2 effect in strip-bark bcps is posited to be a result of enhanced water-use efficiency. If you were to grow Norway spruce on dry, rocky slopes to the point where they are half-dead they would probbably respond to CO2 the same way the bcps do.

20. Posted Dec 1, 2006 at 5:39 AM | Permalink

Re 18: Yes, with the caveat “in the area of pre-instrumental temperatures” that makes a lot more sense. A agree that pre-industrial temperature is an important issue, but there is a lot more in the IPCC TAR. Without MBH the temperature history would probably have been based on Jones et al. 1998. The MBH work didn’t change the central estimate much, but it did put error bars on it which led to a concise statement about temperatures of the late twentieth century likely exceding those of the past millenium. Note that IPCC did not echo MBH’s conclusion of 95% certainty.

Re 11: In the context of our study the Bristlecones are not outliers.

Re 17: True, but orange trees (the study used by Graybill and Idso) are also not bristlecone pines. In the absence of any studies on bristlecones, we need to make do with what is available. This work is also cited by the NAS report in their discussion of Bristlecones. Kostiainen et al, however, clearly think it is a minor effect, at least in the trees they studied. What is missing is evidence of a substantial effect in bristlecones.

Re 12: As I said, I checked the references referred to in the NAS report and I don’t find any justification for omitting the Bristlecones. It doesn’t matter how many times you wave the red flag if you have no basis for waving it. There is a bit more to science than flag waving. I re-read the post: Wahl and Amman say that the perturbed graphs on the left fail the MBH validation tests. Are you disputing this?

Re 7: You need to read beyond the first paragraph of my post.

21. Louis Hissink
Posted Dec 1, 2006 at 6:01 AM | Permalink

As a comment from far left field, but PC Analysis of non-linear tree ring widths is, simply, mathematical incompetence.

22. welikerocks
Posted Dec 1, 2006 at 7:19 AM | Permalink

I checked the references referred to in the NAS report and I don’t find any justification for omitting the Bristlecones. It doesn’t matter how many times you wave the red flag if you have no basis for waving it.

Did ya miss the part that said Bristlecones were not good proxy for temperature Dr. Jukes?

Science is facts; just as houses are made of stones, so is science made of facts; but a pile of stones is not a house and a collection of facts is not necessarily science. ~Henri Poincaré

You’ve been presenting a pile of stones and calling it a house, its got no structure and it’s crumbling down Dr. Juckes.

23. EP
Posted Dec 1, 2006 at 11:47 AM | Permalink

Presumably there are a variety of less “fickle” proxies (i.e not the suspect bristlecones) for which a huge data set can be used and still be enough to generate a plot? I don’t see what the problem is if you can replicate your findings with another subset.

Why do you *have* to include bristlecones?

24. Jean S
Posted Dec 1, 2006 at 1:30 PM | Permalink

Martin, you are just amazing. In #2 you say:

I prefer to base my decision on reading the literature on which the report was based.

So, as I’m rather curious about the issues, I checked the literature referred by the NAS report. Let’s go through the relevant NAS report paragraph and referred studies.

The possibility that increasing tree ring widths in modern times might be driven by increasing atmospheric carbon dioxide (CO2) concentrations, rather than increasing temperatures, was first proposed by LaMarche et al. (1984) for bristlecone pines (Pinus longaeva) in the White Mountains of California. In old age, these trees can assume a “stripbark” form, characterized by a band of trunk that remains alive and continues to grow after the rest of the stem has died. Such trees are sensitive to higher atmospheric CO2 concentrations (Graybill and Idso 1993), possibly because of greater water-use efficiency (Knapp et al. 2001, Bunn et al. 2003) or different carbon partitioning among tree parts (Tang et al. 1999).

Ok, lets see.

LaMarche et al: Increasing Atmospheric Carbon Dioxide: Tree Ring Evidence for Growth Enhancement in Natural Vegetation, Science, Volume 225, Issue 4666, pp. 1019-1021.:

A response of plant growth to increased atmospheric carbon dioxide, which has been anticipated from laboratory data, may now have been detected in the annual rings of subalpine conifers growing in the western United States. Experimental evidence shows that carbon dioxide can be an important limiting factor in the growth of plants in this high-altitude environment. The greatly increased tree growth rates observed since the mid-19th century exceed those expected from climatic trends but are consistent in magnitude with global trends in carbon dioxide, especially in recent decades. If correctly interpreted, these findings have important implications for climate studies involving tree ring observations and for models of the global carbon dioxide budget.

From Graybill and Idso (1993) one should read carefully the part starting from “One perplexing problem…” in the end of p. 82 until mid of 84.

Ok, next. Knapp et al: Detecting potential regional effects of increased atmospheric CO2 on growth rates of western juniper, Global Change Biology, Volume 7 Issue 8 Page 903 – December 2001.

Evidence of an atmospheric CO2 fertilization effect on radial growth rates was uncovered by examining climate–growth relationships for seven western juniper tree-ring chronologies in central Oregon using multiple regression models. Consistent upward trends of the residuals from dendroclimatic models indicated a decreased ability for climate parameters to predict growth with time. Additionally, an assessment was made of whether enhanced growth was detectable under drought conditions, because a major benefit of elevated atmospheric CO2 is the reduction of water stress. Mean ring indices were compared between ecologically comparable drought years, when atmospheric CO2 was lower (1896–1949), and more recent drought years that occurred under higher atmospheric CO2 concentrations (1950–96/98). The results presented herein show that: (i) residuals from climate/growth models had a significant positive trend at six of seven sites, suggesting the presence of a nonclimatic factor causing increased growth during recent decades; (ii) overall growth was 23% greater in the latter half of the 20th century; (iii) growth indices during matched drought and matched wet years were 63% and 30% greater, respectively, in the later 20th century than the earlier 20th century; and (iv) harsher sites had greater responses during drought periods between early and late periods. While it is not possible to rule out other factors, these results are consistent with expectations for CO2 fertilization effects.

Western juniper isn’t a bcp but still… next:

Andrew G. Bunn, Rick L. Lawrence, Gabriel J. Bellante, Lindsey A. Waggoner, and Lisa J. Graumlich: Spatial Variation in Distribution and Growth Patterns of Old Growth Strip-Bark Pines, Arctic, Antarctic, and Alpine Research, 35:323-330, 2003.

Postindustrial rises in CO2 have the potential to confound the interpretation of climatically sensitive tree-ring chronologies. Increased growth rates observed during the 20th century in strip-bark trees have been attributed to CO2 fertilization. Absent in the debate of CO2 effects on tree growth are spatially explicit analyses that examine the proximate mechanisms that lead to changes in rates of tree growth. Twenty-seven pairs of strip-bark and companion entire-bark trees were analyzed in a spatially explicit framework for abiotic environmental correlates. The strip-bark tree locations were not random but correlated to an abiotic proxy for soil moisture. The strip-bark trees showed a characteristic increase in growth rates after about 1875. Furthermore, the difference in growth rates between the strip-bark trees and entire-bark companions increased with increasing soil moisture. A possible mechanism for these findings is that CO2 is affecting water-use efficiency, which in turn affects tree-ring growth. These results point to the importance of accounting for microsite variability in analyzing the potential role of CO2 in governing growth responses.

They didn’t find CO2 directly responsible for strip-bark pines, but they suggest an indirect method, don’t they? Next:

Tang K.; Feng X.; Funkhouser G: The delta13C of tree rings in full-bark and strip-bark bristlecone pine trees in the White Mountains of California. Global Change Biology, Volume 5, Number 1, January 1999, pp. 33-40(8).

Dendrochronological work at Sheep Mountain in the White Mountains, CA has demonstrated that bristlecone pine trees in two forms, full-bark and strip-bark, have experienced different cambial growth rates over the past century or longer. The strip-bark trees showed a greater growth increase than the full-bark ones. A calculation of the plant water-use efficiency in response to anthropogenic CO2 released into the atmosphere shows that W of trees in both forms has increased for the past 200 years. However, there is no significant difference between the two tree forms in the rate of increase in W. This implies at least two possibilities with respect to the CO2 fertilization effect. First, the biomass in both tree forms might have increased, but carbon distribution among different parts of a tree was different. Second, the biomass may increase without causing any corresponding change in the plant water-use efficiency.

Oh, for a layman the above is saying that bcps of different age responded differently during the last past century or longer. Wonder if that causes any problems to bcp tree ring indecies.

The NAS report continues:

Support for a direct CO2 influence on tree ring records extracted from “full-bark” trees is less conclusive. Increasing mean ring width was reported for Pinus cembra from the central Alps growing well below treeline (Nicolussi et al. 1995). Free-Air CO2 Enrichment (FACE) data for conifer plantations in the Duke Forest (Hamilton et al. 2002) and at the alpine treeline (HàÆà⣴tenschwiler et al. 2002) also showed increased tree growth after exposure to atmospheric CO2 concentrations about 50 percent greater than present.

No need to quote those papers. Let’s continue:

On the other hand, no convincing evidence for such effect was found in conifer tree ring records from the Sierra Nevada in California (Graumlich 1991) or the Rocky Mountains in Colorado (Kienast and Luxmoore 1988). Further evidence comes from a recent review of data for mature trees in four climatic zones, which concluded that pine growth at treeline is limited by factors other than carbon (KàÆà⵲ner 2003).

Lisa J. Graumlich: Subalpine Tree Growth, Climate, and Increasing CO_2: An Assessment of Recent Growth Trends, Ecology, Vol. 72, No. 1. (Feb., 1991), pp. 1-11.

LaMarche et al. (1984) hypothesized that recent trends of increasing ring widths in subalpine conifers may be due to the fertilizing effects of increased atmospheric CO”2. Five tree-ring series from foxtail pine (Pinus balfouriana), lodgepole pine (P. murayana), and western juniper (Juniperus occidentalis) collected in the Sierra Nevada, California, were analyzed to determine if the temporal and spatial patterns of recent growth were consistent with the hypothesized CO”2-induced growth enhancement. Specifically, I address the following questions: (1) Can growth trends be explained solely in terms of climatic variation? (2) Are recent growth trends unusual with respect to long-term growth records? For three of the five sites, 20th-century growth variation can be adequately modeled as a function of climatic variation. For the remaining two sites, trends in the residuals from the growth/climate models indicate systematic underestimation of growth during the past decade that could be interpreted as either CO”2 fertilization or as a response to extreme climatic events during the mid 1970s. At all five sites, current growth levels have been equalled or exceeded during some preindustrial periods. Taken together, these results do not indicate that CO”2-induced growth enhancement is occurring among subalpine conifers in the Sierra Nevada. While the results presented here offer no support for the hypothesized CO”2 fertilization effect, they do provide insights into the response of subalpine conifers to climatic variation. Response surfaces demonstrate that precipitation during previous winter and temperature during the current summer interact in controlling growth and that the response can be nonlinear. Although maximum growth rates occur under conditions of high winter precipitation and warm summers for all three species, substantial species-to-species variation occurs in the response to these two variables.

Those “adequately modeled as a function of climatic variation” were all foxtail pines. The two “not adequately modeled” were a lodgepole pine and a western juniper. Oh, I find also this very interesting (from the “Discussion”):

The foxtail pine response surface analysis presented in this study suggests a third type of temperature-precipitation interaction in which drought stress limits growth in years of low winter precipitation and cool temperatures limit growth in years of high winter precipitation. While the contribution of precipitation in governing of subalpine trees in the southern Sierra Nevada and the nearby White Mountains has been recognized (LaMarche 1974, Scuderi 1987), the important distinction that the effects of temperature and precipition on growth are nonlinear and multiplicative rather than linear and additive has not been fully appreciated.

Felix Kienas and Robert J. Luxmoore: Tree-ring analysis and conifer growth responses to increased atmospheric CO2 levels, Oecologia, Volume 76, Number 4 / September, 1988.

Tree-ring data of naturally grown connifers were analyzed to evaluate the possibility of enhanced tree growth due to increased atmospheric CO2. Tree cores were obtained from 34 sites in four different climatic regions in the northern hemisphere. In each of the four regions, the sampling sites were located along ecological gradients between the subalpine treeline and low elevations and, sometimes, the arid forest border. Growth trends after 1950, when the atmospheric CO2 concentration increased by more than 30 mgrl·l-1 indicate an increase in ring-widths at eight of the 34 sites. These chronologies were from sites which moderate temperature or water stress. In four cases the growth increase in the post-1950 period coincided with favorable climatic conditions. In the remaining four cases, the growth increase exceeded the upper bound response expected from CO2 enrichment experiments with seedling conifer species. Therefore, increased growth in any of the tree-ring chronologies examined could not be solely attributed to higher atmospheric CO2 concentrations.

I was not able to download that, so I could not get exact details of trees studied. So there might be some postive evidence for the use of bristlecone pines. Don’t know.

Christian KàÆà⵲ner: Carbon limitation in trees, Journal of Ecology, Volume 91 Issue 1 Page 4 – February 2003.

The ongoing enrichment of the atmosphere with CO2 raises the question of whether growth of forest trees, which represent close to 90% of the global biomass carbon, is still carbon limited at current concentrations of close to 370 p.p.m. As photosynthesis of C3 plants is not CO2-saturated at such concentrations, enhanced ‘source activity’ of leaves could stimulate ‘sink activity’ (i.e. growth) of plants, provided other resources and developmental controls permit. I explore current levels of non-structural carbon in trees in natural forests in order to estimate the potential for a carbon-driven stimulation of growth.

Thus the KàÆà⵲ner (2003) study concers the present (and future), so no relevance for the past series. The NAS report continues:

While “strip-bark’ samples should be avoided for temperature reconstructions, attention should also be paid to the confounding effects of anthropogenic nitrogen deposition (Vitousek et al. 1997), since the nutrient conditions of the soil determine wood growth response to increased atmospheric CO2 (Kostiainen et al. 2004).

Vitousek, P. M., J. D. Aber, R. W. Howarth, G. E. Likens, P. A. Matson, D. W. Schindler, W. H. Schlesinger, and D. G. Tilman. 1997. Human alteration of the global nitrogen cycle: sources and consequences. Ecological Applications 7: 737-750.

This paper has very little discussion about tree growth. Anyhow, even if nirtogen was somehow linked to those bcp series, that would be a human caused non-climatic effect, which would only support the fact that bcps should be avoided.

Then we come to Martin’s “review” paper I already quoted.

Kostiainen, K., Kaakinen, S., SaranpàÆà⥃ Æà⢬ P., Sigurdsson, B. D., Linder, S. & Vapaavuori, E.: Effect of elevated [CO2] on stem wood properties of mature Norway spruce grown at different soil nutrient availability. Global Change Biology 10 (9), 1526-1538.

The objective of the present study was to investigate the interactive effects of elevated [CO2] and soil nutrient availability on secondary xylem structure and chemical composition of 41-year-old Norway spruce (Picea abies (L.) Karst.) trees. The nonfertilized and irrigated-fertilized trees were, for 3 years, continuously exposed to elevated [CO2] in whole-tree chambers. Elevated [CO2] decreased concentrations of soluble sugars, acid-soluble lignin and nitrogen in stem wood, but the effects were not consistent between sampling height and/or fertilization. The effect of 2*ambient [CO2] on wood structure depended on the exposure year and/or fertilization. Radial lumen diameter decreased and annual ring width increased in the second year of exposure (1999) in elevated [CO2]. In the latter, the CO2 effect was significant only in the nonfertilized trees. Stem wood chemistry and structure were significantly affected by fertilization. Fertilization increased the concentrations of nitrogen and gravimetric lignin, annual ring width, and radial lumen diameter. Fertilization decreased C/N ratio, mean ring density, earlywood density, latewood density, cell wall thickness, cell wall index, and latewood percentage. We conclude that elevated [CO2] had only minor effects on wood properties while fertilization had more marked effects and thus may affect ecosystem processes and suitability of wood for different end-use purposes.

Finally, we have:

However, in forest areas below treeline where modern nitrogen input could be expected to influence dendroclimatic records, such as Scotland (Hughes et al. 1984) and Maine (Conkey 1986), the relationship between temperature and tree ring parameters was stable over time.

Those are low altitude moist regions which have little to do with the series in question.

For me, these papers referred by the NAS report give only evidence that suggests strongly to avoid the bristlecone pine series. In fact, as a layman in biology and a professional in statistics, I came to a conclusion that one should be very cautious in applying linear methods to any tree ring chronologies.

On the other hand, after reading the same literature Martin comes to this conclusion (#20):

As I said, I checked the references referred to in the NAS report and I don’t find any justification for omitting the Bristlecones.

How can our conclusions be so different?

25. KevinUK
Posted Dec 1, 2006 at 1:53 PM | Permalink

#24 Jean S

Well here are two reasons why your conclusions might differ.

1. You haven’t stood as a prospective Councillor for teh Cowley ward in Oxford, England for the Green Party.

2. You’re not funded by the UK NERC.

Now Jean, given your statistics expertise, can I ask if the Euro HT had asked you to be their stats guru, would you have included bcps in your reconstruction? No? I didn’t think so. Neither would I, but then again you and I have actually read the NAS Panel and Wegman reports.

KevinUK

26. Dan Hughes
Posted Dec 1, 2006 at 2:12 PM | Permalink

The issues under discussion are far removed from my areas of expertise, but maybe you will kindly let me make a comment.

Pelletier, and others, have shown that the results of power spectral density analyses of atmospheric temperature follow well-defined patterns. Maybe this could be a zeroth-order filter test to apply to proxies. If the raw data from the proxies do not follow the expected response of actual measured atmospheric temperatures, then they cannot be proxies for temperature.

Maybe the other stats that are being used here do the same thing?

My lay-person impression is that if tree-ring width is a proxy for temperature it must be an average value of the temperature over the time period during which the trees produce growth rings. I understand that the temperature values used in calculations for the ‘Global Average Temperature’ are averages over the entire year of daily values of (Tmax + Tmin)/2. And that the ‘Global Average Temperature’ is some kind of area-weighted value of specific locations on the planet. These latter temperature values are not the same as a time-weighted average over the tree-ring-growth period.

It then seems to me that comparing temperature values extracted from tree-ring widths to values of (Tmax + Tmin)/2 might be comparing apples to oranges.

27. bender
Posted Dec 1, 2006 at 2:24 PM | Permalink

This, to me, indicates that Dr Juckes is going to stick to his guns and try to defend the use of bcps. This despite what NAS et al. recommend.

This upping of the ante could mean that you will need to fund a study to actually experimentally quantify bcp responses to T and P and C. If I am right, you will see that the responses are nonlinear and synergistic. And, once this is published, no reviewer will ever allow bcps to be used in a temperature reconstruction again.

The people who published the CO2/water-use-efficiency paper are the ones for the job.

28. KevinUK
Posted Dec 1, 2006 at 2:28 PM | Permalink

#25

Looks like I’m not the only one who is questioning Martyn’s motives.

KevinUK

Posted Dec 1, 2006 at 2:36 PM | Permalink

RE: #25 – RE: “You haven’t stood as a prospective Councillor for the Cowley ward in Oxford, England for the Green Party.”

Fascinating. Another “scientist” with a possible sociopolitical bias likely driving his or her work. Ever so Stalinesque. The politisation of scientists has been a real loss for society as a whole. What a waste of talent.

30. EP
Posted Dec 1, 2006 at 2:38 PM | Permalink

Whoever that M Juckes is on the other blog has some peculiar ideas about the merits of CAP

Politics I think should be left out of the discussion here, although it certainly motivates many on either side and can set the context. I’m not sure we can dismiss someone’s work because they are motivated politically. What needs to be dismissed is the lack of scientific rigour, when it occurs.

Posted Dec 1, 2006 at 2:40 PM | Permalink

RE: #26 – A bit down in the first post, I noticed the poster even managed to nick a sentence off of me …. LOL!

32. KevinUK
Posted Dec 1, 2006 at 2:53 PM | Permalink

#30 EP

I disagree. Read some of the other threads on this blog and you’ll see that Isaac Held and Judith Curry professed to not be interested in the politics of global warming only the science. Do some searching on Google and you see that this is untrue. Martyn is just another example of a politicised scientists, in this case a politicised scientist funded by the UK taxpayer i.e. ME!

Now another question? Other than squash, what do Steve M and Martyn have in common? The answer is contained here. Doa a search for Juckes on this page.

KevinUK

33. anonymous
Posted Dec 1, 2006 at 3:23 PM | Permalink

http://geosci.uchicago.edu/~rtp1/BushWatch/index.html

34. jae
Posted Dec 1, 2006 at 3:35 PM | Permalink

27: Bender:

This upping of the ante could mean that you will need to fund a study to actually experimentally quantify bcp responses to T and P and C. If I am right, you will see that the responses are nonlinear and synergistic. And, once this is published, no reviewer will ever allow bcps to be used in a temperature reconstruction again.

35. EP
Posted Dec 1, 2006 at 3:37 PM | Permalink

Martyn is just another example of a politicised scientists, in this case a politicised scientist funded by the UK taxpayer i.e. ME!

And I agree that politicised scientists are a problem when their work is flawed and influences national (and supra national)policies.

Ultimately any clouding of scientific judgement caused by political opinion should be weened out at the review stage. Any clouding will manifest itself in that step.

36. EP
Posted Dec 1, 2006 at 3:44 PM | Permalink

I meant to say “weeded” not “weened”😉

37. bender
Posted Dec 1, 2006 at 3:50 PM | Permalink

Re #34
I have commented on that previously, but I will repeat it here: altitudinal and latitudinal treeline confiers as a class may exhibit the same problematic sort of nonlinear-synergistic pseudoclimatic response. I can’t see why cedars or junipers, for example, would necessarily be any different from bcps, given that they all exhibit strip-bark forms. I would recommend all these species be subject to a proper experimental calibration test.

38. jae
Posted Dec 1, 2006 at 4:17 PM | Permalink

37: OK, I agree. Now, aren’t these problems also associated with full-bark trees? IOW, how is it possible to use tree growth data to reconstruct temperature, given the lack of understanding of these interactions and the lack of data to address the interactions, even if they were understood. I still question whether tree growth data should even be considered as suitable proxies for temperature.

39. bender
Posted Dec 1, 2006 at 10:25 PM | Permalink

aren’t these problems also associated with full-bark trees

Re #38 Be careful about your degree of presumptuousness over what is/is not a proven “problem”. However, yes, that is part of the hypothesis. By outlining my concern about species that have strip-bark forms, I was not restricting my concern to just the strip-bark samples. Quite the opposite. The responses in full-bark samples concern me too. My hypothesis would be that the srtip-bark forms have a stronger CO2/water-use efficiency response coefficient, but that the nonlinear P*T interactions in full-bark forms are strong enough to be problematic for a linear univariate reconstruction model.

Note, jae, that these alleged nonlinear synergies are not a proven fact, just a compelling, IMHO, hypothesis.

40. jae
Posted Dec 2, 2006 at 10:23 AM | Permalink

Note, jae, that these alleged nonlinear synergies are not a proven fact, just a compelling, IMHO, hypothesis.

Surely this has been investigated by biologists? I’ll try to find some info.

41. Steve McIntyre
Posted Dec 2, 2006 at 11:59 AM | Permalink

Juckes stated:

McIntyre puts a lot of weight on one sentence in the NAS panel report, who devote a whole paragraph to this issue. I prefer to base my decision on reading the literature on which the report was based.

As Jean S stated above, the NAS panel;s recommendation not to use bristlecone/foxtails is not a gotcha or isolated from the literature. Here is another clear statement from Biondi et al (including MBH coauthor Hughes):

The average of those sites [a network of high-elevation temperature-sensitive tree-ring sites in the Great Basin and Sierra Nevada of Hughes and Funkhouser, unpublished], plotted in Figure 5, is based on many ring-width series, each one being 500 years or longer, without individual growth surges or suppressions and from “strip-bark” five-needle upper forest border pines of great age. Such record is not a reliable temperature proxy for the last 150 years as it shows an increasing trend in about 1850 that has been attributed to atmospheric CO2 fertilization [Graybill and Idso, 1993]

The network is presumably the network in Hughes and Funkhouser 2003 where the high-elevation sites are bristlecones and foxtails that are the mainstay of Mann’s PC1: Sheep Mountain ca534, Campito Mountain ca533, Spring Mtn Upper, Pearl Peak Upper nv512, Timber Gap Upper ca529.   The salient point is whether it is a “reliable temperature proxy”; the attribution of why it is unreliable would be nice to have, but is not necessary to determine that it is not a “reliable temperature proxy”.

42. Chris O'Neill
Posted Dec 4, 2006 at 4:39 AM | Permalink

Steve McIntyre wrote:

2) the NAS panel agreed that strip-bark sites (which include all the relevant bristlecone, foxtail and even a couple of limber pine sites) should be avoided in temperature reconstructions for a variety of reasons.

which was in the chapter on Treerings (apart from the “for a variety of reasons” phrase). The conclusions to this chapter (shown at the beginning on page 45) include the statement:

Surface temperature reconstructions based on tree rings require attention to confounding factors; guidelines exist to identify and account for these factors.

So even though the NAS panel said that “strip-bark” samples should be avoided, they concluded that guidelines exist to account for the confounding factors that such samples introduce to reconstructions.

43. Spence_UK
Posted Dec 4, 2006 at 5:10 AM | Permalink

So even though the NAS panel said that “strip-bark” samples should be avoided, they concluded that guidelines exist to account for the confounding factors that such samples introduce to reconstructions.

… which appears to me to be referring to treerings OTHER than the strip bark samples, which should be avoided, as noted in the first statement you quote. These two statements can only be self-consistent in this way.

44. Jean S
Posted Dec 4, 2006 at 5:32 AM | Permalink

#42 Chris:

So even though the NAS panel said that “strip-bark” samples should be avoided, they concluded that guidelines exist to account for the confounding factors that such samples introduce to reconstructions.

How is the HT applying these guidelines? This the relevant paragraph from the NAS report (the one before CO2-paragraph quoted #12/#24, and one after discussing the “divergence issue”):

An especially suitable strategy to minimize confounding effects is to sample sites along ecological gradients, such as elevation or latitude (Fritts and Swetnam 1989, Bugmann 1996). For example, (Naurzbaev et al. 2004) selected sites along latitudinal (from 55 to 72°N) and elevational (from 1120 to 2350 m above sea level) transects, and used the parameters of the Regional Curve Standardization to infer climatic influences and past temperature variability. Other strategies are available to improve tree ring reconstructions of surface temperature. Some of these strategies involve using maximum temperature instead of mean temperature (Luckman and Wilson 2005), combining multiple tree ring parameters related to temperature (Helle and Schleser 2004), sampling species with opposing responses to temperature (Biondi et al. 1999), and applying mechanistic models to tree ring records (Anchukaitis et al. 2006).

Which one of the cited works applies to “strip-barks” and possible past CO2 fertilization? Finally, the NAS tree ring chapter (Chapter 4) ends (Chris’ quotation comes from a summary bullet):

In conclusion, tree ring science provides useful insights into past temperature variability. Promising areas of current and future research can be summarized as:
1. updating site chronologies that were collected 20-30 years ago,
2. increasing the number and geographic coverage of temperature-sensitive tree ring chronologies longer than 1000 years,
3. quantifying the precision and accuracy of low-frequency temperature signals,
4. performing experimental studies on biophysical relationships between temperature and tree ring parameters,
5. refining mechanistic models of temperature effects on tree ring parameters at multiple spatial and temporal scales.

Pretty strong words. I’m tempted to characterized that conclusion as “treering based temperature reconstruction field is still very immature”.

45. Steve McIntyre
Posted Dec 4, 2006 at 7:07 AM | Permalink

#42. OK, Chris, I’ll bite. What are these “guidelines” – ther than avoid using potentially contaminated chronologies?

46. Posted Dec 4, 2006 at 8:46 AM | Permalink

Re 23: All the proxies are influenced by a variety of factors other than temperature.

Re 24: How about leaving out the ones which say it is a possibility that CO2 has some effect, as we have no disagreement on that. Lots of people saying there is a possibility doesn’t make a certainty.

Post 23 points at a fact you appear to ignore: the question is not whether the bristlecones are affected by something other than temperature, the question is whether they are affected to a substantially greater extent than other proxies.

Re 41: Biondi et al only cite Graybill and Idso, they don’t produce any new evidence on this. Graybill and Idso cite controlled experiments on orange trees.

47. Jean S
Posted Dec 4, 2006 at 8:52 AM | Permalink

Re 24: How about leaving out the ones which say it is a possibility that CO2 has some effect, as we have no disagreement on that. Lots of people saying there is a possibility doesn’t make a certainty.

Post 23 points at a fact you appear to ignore: the question is not whether the bristlecones are affected by something other than temperature, the question is whether they are affected to a substantially greater extent than other proxies.

So if that is at least possibility, shouldn’t you then be cautious and avoid them as NAS suggested?

48. EP
Posted Dec 4, 2006 at 9:34 AM | Permalink

Re 23: All the proxies are influenced by a variety of factors other than temperature.

Post 23 points at a fact you appear to ignore: the question is not whether the bristlecones are affected by something other than temperature, the question is whether they are affected to a substantially greater extent than other proxies.

Then the issue is to what extent the bristlecone attributes are dependent on external influences and whether these influences can be factored out as affecting certain attributes. Any subset of data of sufficient size should not give significantly different results from any other subset. If this conflict occurs then there is something wrong with relying on that attribute to estimate temperature.

49. Brooks Hurd
Posted Dec 4, 2006 at 11:26 AM | Permalink

Martin Juckes,

Thank you for your comments on the effects of various factors on proxies. You cited:

Graybill and Idso cite controlled experiments on orange trees.

Are you aware of any other controlled experiments on the effect of temperature, rainfall, fertilization, etc on the proxies?

Mathematically, I would expect a description of the proxy parameter to be described by an equation such as:

$aT^(bx)+cR^(dy)+eF^(fz)+…………$

Where T=temperature, R=rainfall, F=fertilization, and a, b, c, d, e, and f are constants.

50. Brooks Hurd
Posted Dec 4, 2006 at 11:27 AM | Permalink

Unfortunately my attempt at using LaTex failed to produce the desired equation.

51. Posted Dec 4, 2006 at 11:32 AM | Permalink

Dollar signs won’t do here, see http://www.climateaudit.org/?p=660

$aT^(bx)+cR^(dy)+eF^(fz)+$

52. Posted Dec 4, 2006 at 11:34 AM | Permalink

oops, should try with WinEdt first (with dollars)😉 , you need to use { and }

$aT^{bx}+cR^{dy}+eF^{fz}+$

53. Posted Dec 4, 2006 at 12:01 PM | Permalink

Martin Juckes:

Lots of people saying there is a possibility doesn’t make a certainty.

If you get a sudden midnight knock on your door, that’ll be the Irony Police. I suggest you confess and go quietly.

54. jae
Posted Dec 4, 2006 at 12:11 PM | Permalink

44:

Pretty strong words. I’m tempted to characterized that conclusion as “treering based temperature reconstruction field is still very immature”.

ME TOO!

55. Hans Erren
Posted Dec 4, 2006 at 3:35 PM | Permalink

re 53: Nobody can escape the spanish inquisition.

Dr. Juckes, would your publication pass an MSc mathematics exam comittee?

56. Earle Williams
Posted Dec 4, 2006 at 3:43 PM | Permalink

Re #55

Our three main weapons are: GCMs, tree rings, and ruthless deconstruction of denialist arguments.

.

.

.

And a fanatical devotion to Michael Mann!

(OK, maybe it’s time to eat some lunch. Cheers!)

57. John Baltutis
Posted Dec 4, 2006 at 3:44 PM | Permalink

Re #46

Glad to see you’re still visiting. Any responses for the unanswered questions, post #28, at http://www.climateaudit.org/?p=897 or have you just moved on?

58. Brooks Hurd
Posted Dec 4, 2006 at 3:56 PM | Permalink

UC

Thanks very much for the equation help.

Even this simplified equation indicates a more complex relationship between proxy parameters and temperature than I have seen in temperature proxy reconstruction papers.

Based on what I have read, I am confident that a, b, and t and not equal to 1 and the other constants are not equal to zero. In other words, it is not a linear equation. In the case of some tree rings the literature indicates that the relationship is more parabolic than linear.

Lacking specific studies showing linearity of one parameter to another, I have always been suspicious of assuming linear realtionships. The ideal gas law assumes linearity and is a reasonably good approximation for high temperature / low pressure systems. However, low temperature / high pressure gas systems deviate from linearity.

59. Dave B
Posted Dec 4, 2006 at 4:21 PM | Permalink

pardon me, dr. juckes…you said:

“the question is not whether the bristlecones are affected by something other than temperature, the question is whether they are affected to a substantially greater extent than other proxies.”

shouldn’t the question be:

is there any evidence for bristlecones being a proxy for temperature?

60. Hans Erren
Posted Dec 4, 2006 at 5:00 PM | Permalink

and why don’t you reject all proxies that do not correlate with local temperature, or in laymans terms: with proof are not temperature proxies?

61. jae
Posted Dec 4, 2006 at 5:05 PM | Permalink

59, Hans: Yes, that fact by itself invalidates many proxies, IMO. One really does not have to consider anything else. These series call into question the very basis for using tree rings as proxies. How can anything be clearer than that? Has anyone on the Team even addressed this problem?

62. Posted Dec 4, 2006 at 5:09 PM | Permalink

[ tex]aT^{bx}+cR^{dy}+eF^{fz}+ \ldots[/ tex] produces $aT^{bx}+cR^{dy}+eF^{fz}+ \ldots$

63. Paul Linsay
Posted Dec 4, 2006 at 5:20 PM | Permalink

#49 and #52. The function has to be a product of the factors and not a sum. This is because too little or much of any growth variable will kill the plant. For example, too little(much) water will dry out(drown) the tree even if temperture, sunlight, etc., are in the perfect range. If the function is only a sum, just “zeroing” the water won’t do anything to the other terms. This makes the growth function even more non-linear and complicated.

64. Posted Dec 4, 2006 at 5:31 PM | Permalink

Re #62

Actually each term should be an inverse quadratic partial differential, making the resultant an n-dimensional landscape.

65. John Baltutis
Posted Dec 4, 2006 at 5:59 PM | Permalink

For John A.

Links to Road Map bring up blank page. All others I tried work without any problem.

66. Posted Dec 4, 2006 at 6:01 PM | Permalink

Re #64

I know. I’m waiting for a new server to be delivered by webserve.ca so I can start a migration as soon as possible.

67. jae
Posted Dec 4, 2006 at 6:06 PM | Permalink

It’s beginning to look like modeling tree ring growth is about as simple as GCMs.

68. Posted Dec 4, 2006 at 7:12 PM | Permalink

Re: #62

If tree ring width is given by

$W = aT^{bx}cR^{dy}eF^{fz}$

then taking logs of both sides gives

$\log(W) = K + \alpha\log T + \beta\log R + \gamma\log F$

where $K\equiv\log(a)+\log(c)+\log(e)$, $\alpha\equiv bx$ etc because you cannot identify b and x separately. The coefficient of log temperature then tells us the temperature elasticity of tree ring widths. That is, the percentage response in tree ring width to a one percent change in temperature (keeping everything else constant).

One would intuitively presume more non-linearity than is captured with the logarithmic model but I do not presume to know anything about that. I just wanted to note that a multiplicative model is not hard to transform into a linear model. The word linear in linear regression refers to linearity in coefficients, not variables.

To apply OLS to the model above, one would normally presume a multiplicative error term that is log-normal.

PS: I am trying the LaTeX notation John A. showed in #61 for the first time. Hope it works.

69. bender
Posted Dec 4, 2006 at 7:36 PM | Permalink

log-linearization does not work if you have bot additive and multiplicative effects

70. John S
Posted Dec 4, 2006 at 8:28 PM | Permalink

Re #68

Then do a first order Taylor expansion and hope you don’t stray too far from the point you linearize around.

71. Posted Dec 4, 2006 at 8:31 PM | Permalink

Re: #68

Agreed. I am not claiming that the model ought to be one way or another, just pointing out that a purely multiplicative model is not hard to deal with. Since these are just curve fitting exercises, why not try alternative specs?

72. Jean S
Posted Dec 5, 2006 at 9:24 AM | Permalink

re #68/#69/#70/#71: I had the same idea a while back, and tested it a bit. Based on the PCA-like analysis on NOAMER set, I would say that the multiplicative approach ought to work better than the linear. However, I think there is a problem with RCS standardisation of the chronologies, which essentially assumes linear model. So here is something someone ought to test: start with individual tree ring measurements, and build a chronology with RCS but in log domain. Then using these type of chronologies, build a temperature estimate with correct regression models again working in log domain. I would do the regression in a grid cell level with two climate indicators: average temperature (in growing season) and (preceding annual) precipitation.

BTW, it would be interesting to see the r values of the logged chronologies against the grid cell temperatures (compared to non-logged correlations) . Since correlation coefficient measures linear correlation, we should see a clear increase in logged correlations if the multiplicative model have any truth to it.

73. Chris O'Neill
Posted Dec 6, 2006 at 10:37 AM | Permalink

Jean S wrote:

“This the relevant paragraph from the NAS report…”

That wasn’t the only paragraph referring to confounding effects. On page 7 it says:

“Like other proxies, tree rings are influenced by biological and environmental factors other than climate. Site selection and quality control procedures have been developed to account for these confounding factors.”

NAS were clearly being general in referring to “confounding factors”.

Steve McIntyre wrote:

“What are these “guidelines” – ther than avoid using potentially contaminated chronologies?”

That’s beside the point of this thread which is whether NAS’s statement (which doesn’t occur in any of their conclusions that I could find) that strip-bark sites “should” be avoided amounts to proof that these proxies are vastly inferior to other types of proxies. I think you need a much stronger word than “should”.

74. Steve McIntyre
Posted Dec 6, 2006 at 11:12 AM | Permalink

#73.

“Like other proxies, tree rings are influenced by biological and environmental factors other than climate. Site selection and quality control procedures have been developed to account for these confounding factors.”

One of the “site-selection and quality control procedures” that has been developed (e.g. Biondi et al 1999) is not to use bristlecone/strip-bark sites. The NAS panel re-iterated this “site selection and quality control procedure”. MBH failed to observe this QC procedure.

Since when is there any need to “prove” that bcp’s are “vastly inferior” to other types of “proxies”. If these studies are worth anything, they should survive not using bristlecones and foxtails. It is ridiculous that Juckes’ Union reconstruction uses 4 bristlecone and foxtail sites (out of only 18 proxies); Hegerl uses 2 out of 12; etc. It’s demented.

75. jae
Posted Dec 6, 2006 at 12:26 PM | Permalink

#73: Chris, doesn’t it bother you that there is little or no correlation between local temperatures and growth rates for these trees? Do you also believe in telecommunications?

76. Steve McIntyre
Posted Dec 6, 2006 at 4:27 PM | Permalink

Mew Matlab script from Jean S: http://data.climateaudit.org/scripts/mitrie/brownmannpca.m

77. Jean S
Posted Dec 6, 2006 at 4:32 PM | Permalink

Oh dear! I played again a bit with that idea of multiplicative model. This time I plotted the autocorrelation function of some these three ring chronologies (from NOAMER set). I admit I should have done that long time ago. Anyhow, the shape looked familiar … hmmm .. how about lognormal sample? Bingo! So I played further with those series with a nice Matlab time series toolbox available here:
http://hci.tugraz.at/schloegl/matlab/tsa/
It sure seems that taking log of those tree ring series really reduces the order (and magnitude) of autocorrelation. Some of the series I tested are pretty close to a lognormal sample…

Anyhow, I also tested Mannian PCA on geometric Brownian motion… I get a nice Hockey Stick about 95% of time even with lognormal!!!! Steve, please insert code here. http://data.climateaudit.org/scripts/mitrie/brownmannpca.m

78. Steve McIntyre
Posted Dec 6, 2006 at 4:49 PM | Permalink

These results are reminiscent of MM05a on the RE statistics and, even more, in our Reply to Huybers. The Team would argue that all 18 proxies are not random walks. IF all 18 proxies are random walks, you generate a mean correlation of 0.65. However, you only need one random walk to generate a lot of spurious results. With one random walk and 17 white noise processes, you get this quantile under Juckesian CVM:
# 90% 95% 97.5% 99% 99.98%
#0.4331643 0.4570457 0.4739879 0.4946343 0.5359809
If you have 2 random walks and 16 white noise processes, the yield is further increased:
# 90% 95% 97.5% 99% 99.98%
#0.4887333 0.5181645 0.5293909 0.5497349 0.5946711
We did something very similar in our simulations using our simulated PC1s from MM05a as the ginger corresponding to a random walk, but the concept was similar. The statistical point is cleaner with a random walk though. The interesting statistical issue is the impact of mixing the two types of noise.

79. Jean S
Posted Dec 6, 2006 at 4:55 PM | Permalink

Steve, the key in #77 is that it is geometric. The lognormal sample (change in my code AR1=0) is i.i.d., and equivalent of a white noise process in a multiplicative system!

80. Steve McIntyre
Posted Dec 6, 2006 at 5:14 PM | Permalink

#79. My #78 didn’t refer to the log-normal situation but the CVM on random walks. The distribution of ring width chronologies are quite interesting – especially when both MXD and RW are available. I posted up some distributions last year in connection with Esper – the joint distributions require some kind of copula.

81. Jean S
Posted Dec 7, 2006 at 4:48 AM | Permalink

UC, please, could you again produce a figure or two with #77 to your page for people to see🙂

82. Posted Dec 7, 2006 at 5:11 AM | Permalink

#81

Sure.

AR 0.2

AR 0.0

AR 0.7

Didn’t cherry pick, first 3 runs, I promise😉

83. Chris O'Neill
Posted Dec 7, 2006 at 5:47 AM | Permalink

Steve McIntyre wrote:

“One of the “site-selection and quality control procedures” that has been developed (e.g. Biondi et al 1999) is not to use bristlecone/strip-bark sites.”

That doesn’t mean it’s the only one.

“The NAS panel re-iterated this “site selection and quality control procedure”.”

They didn’t. They just used the word “should”.

“Since when is there any need to “prove” that bcp’s are “vastly inferior” to other types of “proxies”. If these studies are worth anything, they should survive not using bristlecones and foxtails.”

That’s another issue beside the point you were trying to establish in this thread. If you want to retract the assertion that bristlecones and foxtails cannot be used in reconstructinos based just on a non-defintive statement by the NAS panel then do so and we can move on.

84. Nicholas
Posted Dec 7, 2006 at 6:35 AM | Permalink

“One of the “site-selection and quality control procedures” that has been developed (e.g. Biondi et al 1999) is not to use bristlecone/strip-bark sites.”

That doesn’t mean it’s the only one.

That’s a specious line of reasoning. Just because it’s not the only one, does not mean it’s not one! They could have 2357813 different site selection and quality control procedures, but as long as one of them says to avoid bristlecone/strip-bark sites, Mr. McIntyre is correct.

“The NAS panel re-iterated this “site selection and quality control procedure”.”

They didn’t. They just used the word “should”.

re·it·er·ate
tr.v. To say or do again or repeatedly.

The NAS panel referred to the guide-lines and mentioned they “should” be followed. How is that not re-iterating them? They mentioned them again. That is one of the definitions of “re-iterate”, to say again.

85. Cliff Huston
Posted Dec 7, 2006 at 6:51 AM | Permalink

#83
Chris O’Neill,

Instead of badgering SteveM with your hair-spliting, you might try looking up the definition of the word ‘should’. You will find:

should: verb, past tense of shall – Used to express duty, obligation, probability, expectation, or contingency. (Webster’s New Riverside Dictionary)

If you are still confused by the NAS report usage of ‘should’ and ‘should not’, I suggest you contact the report authors and ask them what they intended.

86. Chris O'Neill
Posted Dec 7, 2006 at 8:40 AM | Permalink

jae wrote:

“doesn’t it bother you that there is little or no correlation between local
temperatures and growth rates for these trees?”

Are you referring to 5 degree X 5 degree grid cell “local” temperature or actual site
local temperature?

87. Jean S
Posted Dec 7, 2006 at 8:57 AM | Permalink

Are you referring to 5 degree X 5 degree grid cell “local” temperature or actual site
local temperature?

Are you implying that there is not a high correlation between 5×5 grid cell temperatures and (some) local conditions in a monthly/yearly level? What are such locations? Any proxy series from such locations?

Chris O, why do I have a feeling that you are trying hard to be another CA troll?

88. Steve McIntyre
Posted Dec 7, 2006 at 9:19 AM | Permalink

#86. In this case, it’s true of both. There’s little correlation with gridcell temperature. High-altitude Sierra Nevada temperatures have been recently collated and there’s less correlation with them.

89. Chris O'Neill
Posted Dec 7, 2006 at 9:25 AM | Permalink

“Just because it’s not the only one, does not mean it’s not one!”

Didn’t say it wasn’t. Just means there are other methods.

“re-iterate
tr.v. To say or do again or repeatedly.”

The report said “”strip-bark” samples should be avoided for temperature reconstructions” just once.

90. Mark T
Posted Dec 7, 2006 at 9:38 AM | Permalink

So, what… for the conclusion that strip bark series should not be used is invalid because it is only stated once?
Does that mean it would need to be stated two, maybe three or four times before being correct?

Mark

91. Chris O'Neill
Posted Dec 7, 2006 at 10:08 AM | Permalink

Jean S wrote:

“why do I have a feeling that you are trying hard to be another CA troll?”

I responded to someone who was effectively trolling me and couldn’t resist because that is quite an interesting subject. I don’t know the details involved but know that generally, high mountain weather is more related to weather further away than is low altitude weather. But that’s not the issue of this thread.

92. Steve McIntyre
Posted Dec 7, 2006 at 10:19 AM | Permalink

#89. By “re-iterate”, I meant that NAS said it once again – NAS wasn’t the first to observe the problem – not that it said it repeatedly. Jeez, this is silly hairsplitting.

93. Chris O'Neill
Posted Dec 7, 2006 at 11:46 AM | Permalink

“While ‘strip-bark’ samples should be avoided for temperature reconstructions,…”

A few people are not getting the point here which is that the NAS panel did not say that strip-bark samples must be be avoided for temperature reconstructions. This means you can use them if you know how to deal with the issues involved.

I think it’s also worth pointing out the quote from Biondi et al that has some bearing on this issue:

and from ‘strip-bark’ five-needle upper forest border pines of great age. Such record is not a reliable temperature proxy for the last 150 years as it shows an increasing trend in about 1850 that has been attributed to atmospheric CO2 fertilization [Graybill and Idso, 1993]

An obvious implication of this statement is that they’re not expressing any concerns about using it a temperature proxy before about 1850. So as long as you can work out the right coefficient for it, it can give you a good temperature proxy before about 1850. Obviously, there’s not much point trying to use it for a reconstruction after about 1850 because there are so many less problematic proxies available.

94. Jean S
Posted Dec 7, 2006 at 12:04 PM | Permalink

Chris O:

So as long as you can work out the right coefficient for it, it can give you a good temperature proxy before about 1850. Obviously, there’s not much point trying to use it for a reconstruction after about 1850 because there are so many less problematic proxies available.

Sure, how would you calibrate them? How many instrumental temperature series before 1850 exists in these regions?

95. Mark T
Posted Dec 7, 2006 at 12:20 PM | Permalink

Rendering them impossible to use prior to 1850, and hence, should not be used altogether.

Mark

96. Ken Fritsch
Posted Dec 7, 2006 at 12:26 PM | Permalink

Sure, how would you calibrate them? How many instrumental temperature series before 1850 exists in these regions?

I thought I had a shot at defending the Hockey Team as goal tender with my rationalizations for teleconnections in proxy responses, but I see Chris O has bigger pads and gloves than I do.

97. Steve McIntyre
Posted Dec 7, 2006 at 12:29 PM | Permalink

The big problem with bristlecones is ultimately that the ring width chronologies make no sense in terms of the ecological information. The bristlecone chronologies show essential ice age cold in from 1000-1250, in the period when Miller’s ecological analysis shows that tree lines were 300-500 meters higher and temperatures were 3.5-3 degrees warmer. Perhaps the proxy gets screwed up because of narrower medieval ring widths due to drought (a Graumlich type explanation); perhaps the comparison is screwed up because of wider modern widths. But the main point is that the “proxy” is about as crappy a proxy as you can imagine for representing long-term change in California. There’s no evidence of Ice Age level temperatures in medieval California. The demand to use this flawed data is simply ideological.

98. Dave Dardinger
Posted Dec 7, 2006 at 12:29 PM | Permalink

re: #93

An obvious implication of this statement is that they’re not expressing any concerns about using it a temperature proxy before about 1850.

You mean “it’s an obvious logical fallacy is to claim that because the NAS report says the last 150 years’ records are suspect, it implies that before 150 years they’re ok.” E.g. Saying, “Millions of people have died of the effects of hunger in the past century” implies that nobody died of hunger before 100 years ago. It doesn’t follow. It may be that it’s true that the panel didn’t mention problems in earlier records, but you have no way of knowing if this is because they didn’t think there were any (in which case they should have said so) or that they couldn’t decide if there were any or something else.

99. Mark T
Posted Dec 7, 2006 at 12:52 PM | Permalink

Logic: the bane of modern hype.

Mark

100. Chris O'Neill
Posted Dec 8, 2006 at 5:16 AM | Permalink

Steve McIntyre wrote:

“The big problem with bristlecones is ultimately that the ring width chronologies make no sense in terms of the ecological information. The bristlecone chronologies show essential ice age cold in from 1000-1250, in the period when Miller’s ecological analysis shows that tree lines were 300-500 meters higher and temperatures were 3.5-3 degrees warmer.”

So you don’t disagree that Biondi et al professed no concern for using strip-bark proxies for reconstructions before about 1850 (the NAS panel’s statement was based on Biondi et al’s concern about using them for reconstructions after about 1850) and instead are making your own argument for such concern. That’s OK but it means that your original argument has lost its basis and you’re moving on to a different argument.

101. Willis Eschenbach
Posted Dec 8, 2006 at 5:50 AM | Permalink

Chris, thanks for your post above. A couple of points:

1) Steve M.’s original argument is shown in point 2) of the head post, viz:

2) the NAS panel agreed that strip-bark sites (which include all the relevant bristlecone, foxtail and even a couple of limber pine sites) should be avoided in temperature reconstructions for a variety of reasons.

Steve is not “moving on”, he is merely explaining some of the “variety of reasons” that bristlecones don’t make good proxies.

2) We have no instrumental data in the area older than 150 years. Biondi says not to use bristlecone proxies post-1850. Perhaps you can explain how we can use the earlier (pre-1850) portions of the bristlecone record as temperature proxies when we have no instrumental data pre-1850 to either calibrate or validate them against?

w.

102. Posted Dec 8, 2006 at 6:37 AM | Permalink

Perhaps you can explain how we can use the earlier (pre-1850) portions of the bristlecone record as temperature proxies when we have no instrumental data pre-1850 to either calibrate or validate them against?

First make an overfitted reconstruction using other proxies than bristlecones, and then overfit (pre-1850) bristlecone data with that reconstruction. Remember to CVM at each step.

103. Mark T.
Posted Dec 8, 2006 at 12:25 PM | Permalink

Willis, Chris apparently teleconnected the post 1850 information to draw conclusions about
pre 1850. Very slick. Now teleconnection has a time travel machine as well.

Mark

104. Ken Fritsch
Posted Dec 8, 2006 at 1:20 PM | Permalink

So you don’t disagree that Biondi et al professed no concern for using strip-bark proxies for reconstructions before about 1850 (the NAS panel’s statement was based on Biondi et al’s concern about using them for reconstructions after about 1850) and instead are making your own argument for such concern. That’s OK but it means that your original argument has lost its basis and you’re moving on to a different argument.

Chris O, I thought it was better equipment that was going to lead you to defending goal for the Hockey Team, but I think now it has more due to your dedication to the Team and a hard headed ability to keep the puck in play while never admitting defeat — very desirable Team attributes.

105. Steve McIntyre
Posted Nov 6, 2007 at 10:51 PM | Permalink

#100. Chris O’Neill has raised this thread over at Pharyngula. His final comment above said:

So you dont disagree that Biondi et al professed no concern for using strip-bark proxies for reconstructions before about 1850 (the NAS panels statement was based on Biondi et als concern about using them for reconstructions after about 1850) and instead are making your own argument for such concern. Thats OK but it means that your original argument has lost its basis and youre moving on to a different argument.

In 101, Willis responded (and I agree with his response):

) 1) Steve M.s original argument is shown in point 2) of the head post, viz:

2) the NAS panel agreed that strip-bark sites (which include all the relevant bristlecone, foxtail and even a couple of limber pine sites) should be avoided in temperature reconstructions for a variety of reasons.

Steve is not moving on, he is merely explaining some of the variety of reasons that bristlecones dont make good proxies.

2) We have no instrumental data in the area older than 150 years. Biondi says not to use bristlecone proxies post-1850. Perhaps you can explain how we can use the earlier (pre-1850) portions of the bristlecone record as temperature proxies when we have no instrumental data pre-1850 to either calibrate or validate them against?

You didn’t respond to this so it’s pretty unreasonable for you to argue over at Pharyngula that I had ceased to cite the NAS panel reasoning and had moved on to alternative reasons.

Reviewing the post itself: obviously there has been much controversy over the impact of Mann’s erroneous principal components method. Mann’s associates have argued that the error doesn’t “matter” because they can still “get” a hockey stick a different way. In his Reply to Questions, Wegman provides a convincing and devastating rejection of this tactic, observing that statisticians cannot ex post modify their methodology to “get” a preferred answer as follows:

A cardinal rule of statistical inference is that the method of analysis must be decided before looking at the data. The rules and strategy of analysis cannot be changed in order to obtain the desired result. Such a strategy carries no statistical integrity and cannot be used as a basis for drawing sound inferential conclusions.

In the post in question, I observed that the NAS panel report provided an alternative argument for establishing the meaninglessness of the Mann reconstruction without principal components being invoked, as follows:

1) Wahl and Ammann and ourselves agree that an MBH98-type reconstruction without bristlecones is non-meaningful.

2) the NAS panel agreed that strip-bark sites (which include all the relevant bristlecone, foxtail and even a couple of limber pine sites) should be avoided in temperature reconstructions for a variety of reasons.

Q.e.d.

This is obviously not the only defect in MBH, but was a short and sweet argument after the NAS panel. In your comment above, you seek to rebut this argument on the basis that:

you dont disagree that Biondi et al professed no concern for using strip-bark proxies for reconstructions before about 1850 (the NAS panels statement was based on Biondi et als concern about using them for reconstructions after about 1850

While Biondi et al do not profess concern about strip bark proxies prior to 1850, there is no support in the NAS Panel report for your your assertion that “the NAS panels statement was based on Biondi et als concern about using them for reconstructions after about 1850″. In the section (STR Preprint, 56) in which the NAS Panel says that strip bark trees should be “avoided” in temperature reconstructions, Biondi et al 1999 is not cited (but a number of other publications are cited, including Graybill and Idso 1993, Vitousek et al. 1997, Knapp et al 2001, Tang et al 1999 – some of which we drew to their attention in our presentation. While Biondi et al 1999 may have contributed to their conclusion, there is no evidence that their statement was “based” on Biondi et al 1999 as you allege and there is evidence to the contrary.

You go on to allege:

instead are making your own argument for such concern. Thats OK but it means that your original argument has lost its basis and youre moving on to a different argument

In comments to the post, I agree that I observed that, in addition to issues raised by the NAS Panel, the results from Miller et al 2006 (which was cited favorably by the NAS Panel) were inconsistent with interpreting [Graybill’s] bristlecone ring widths as MWP temperature proxies. This was simply additional support for the NAS panel position that strip bark trees should be “avoided” in temperature reconstructions – a position that seemed sensible to me at the time and still seems sensible.

106. Steve McIntyre
Posted Nov 7, 2007 at 12:06 AM | Permalink

At a certain point, one does get a little impatient with arguments like the ones presented by Chris O’Neill here, for which he’s proclaiming vindication at Pharyngula. Look at what he said in #93. He quotes the NAS panel:

“While ‘strip-bark’ samples should be avoided for temperature reconstructions,…”

which he characterizes as follows:

A few people are not getting the point here which is that the NAS panel did not say that strip-bark samples must be avoided for temperature reconstructions. This means you can use them if you know how to deal with the issues involved.

Huh?? He’s just finished quoting the NAS panel as saying that they should be avoided. HE then says that they did not say that they must be avoided, which he transforms in a dizzy flight of reasoning to a claim that you “can” use them.

Has Chris nothing better to do than cavil over the meaning of the word “should”? If he’s in need of guidance, the meaning of the word “should” is reviewed in #85.

But hey, isn’t this an illustration of climate science at its finest? The NAS panel says that strip bark should be avoided; and the Team concludes that the NAS panel said that they can be used.

This is a lame parlor trick even by Mannian standards.

107. bender
Posted Nov 7, 2007 at 4:58 AM | Permalink

Another rhetoritician, master word parser. Trouble is – this is about data, data modelling, and data interpretation.

bcps ought to be avoided. And the divergence problem ought to be resolved. (NAS forgot to mention that last bit.)

108. MrPete
Posted Nov 7, 2007 at 7:28 AM | Permalink

This can be apparently summarized in four easy steps:
1) The Team has a desired goal to be demonstrated.
2) They choose data series and “model” methods that fulfill their goal.
Unfortunately, they do <i>not</i> “know how to deal with the issues involved”…
3) …in data processing, because they are not statisticians.
4) …in interpreting tree rings from strip bark trees, because they know nothing about the physical reality involved.
And Chris doesn’t understand the NAS report for the same reason. And doesn’t recognize that his Team has these weaknesses.

109. Paul Dennis
Posted Nov 7, 2007 at 7:37 AM | Permalink

Pete….I know it’s off topic but did you email me re: precipitation isotope analysis? We discussed it briefly but I can’t find any email messages.

110. Spence_UK
Posted Nov 7, 2007 at 7:42 AM | Permalink

Chris O seems to cling to the notion that MBH98 is a great paper and has been caught distorting facts in the past – he was the poster on d**toid who created a complete fabrication to justify the CENSORED directory (full story here)

In the example of the NAS panel report, Chris is exploiting the fact that scientists are naturally wary of drawing conclusions which go beyond their remit. Scientists will typically avoid saying you must not use the bristlecones – never say never. But since all of the literature points to serious problems with the bristlecones, you are going to need extraordinary evidence that you can tease a temperature signal out of it.

The use of BCPs in MBH99 with the correction used to inflate RE (noted by Steve recently) is a great example. Where is Mann’s extraordinary evidence for using a flawed proxy, that this adjustment helps tease out a temperature signal? Well, someone had the temerity to ask this question over at RealClimate (thanks as ever to Armand and nanny!). I’m surprised they posted it at all, as their censorship policy was quite strict at the time, but I think Mann likes the opportunity to be dismissive. His extraordinary evidence was a bunch of unquantified hand waving, followed by “google scholar” when pressed further. I’m unaware of any better attempt to justify the correction. The only extraordinary thing here is the cheek of the Mann.

111. Sam Urbinto
Posted Nov 7, 2007 at 10:24 AM | Permalink

I don’t think Chris understands panel-speak. That’s pretty strong language and it translates into English as “Don’t use them.” They are making the point that while strip-bark is good for some purposes, temperature reconstructions isn’t one of them.

They should not be used.
{they get used}
What did you do that for, we told you they shouldn’t be used.

Even forgetting that, “should not” doesn’t equal “can”, it totally ignores what Bloomfeild said “We had much the same misgivings about his work that was documented at much greater length by Dr Wegman.” Such report saying: “[Mann’s] decentered methodology is simply incorrect mathematics.” and “The papers of Mann et al. in themselves are written in a confusing manner…” and

We found MBH98 and MBH99 to be somewhat obscure and incomplete and the criticisms of MM03/05a/05b to be valid and compelling.
Overall, our committee believes that Manns assessments that the decade of the 1990s was the hottest decade of the millennium and that 1998 was the hottest year of the millennium cannot be supported by his analysis.

Dr. Wegman’s report also says such things as A) The results with MBH98 can’t be reproduced with the material available and B) The graphic is based on incorrect use of PCA C) The papers re-using the proxies cannont claim to be independent verifications D) The paleoclimate community has a self-reinforcing feedback mechanism E) There is too much reliance on peer-review

So if you throw out all this tainted “verification” work, what’s left? Even Dr. North, when asked by Barton “…if you want to ask your statistican expert…[the] methodology cannot be documented and connont be verified by independent review.”, all Dr. North says is “Do you mind if he speaks?” and had of course earlier said about Dr. Wegman’s report “We don’t disagree with their criticism” of the conclusions or methodology of the report.

But of course, Dr. Wegman’s report isn’t peer-reviewed, so it must be meaningless…

112. MrPete
Posted Nov 7, 2007 at 11:07 AM | Permalink

(Paul – sorry, buried in Real Work right now… not forgotten! And, we need to get in touch with our friends who run the coop precip collector)

113. James Erlandson
Posted Nov 7, 2007 at 1:31 PM | Permalink

re: Should

You should not stick your hand into the lion’s cage.
You should not stick your finger into the fan.
You should not pour gasoline on the fire.
You should not leave a loaded gun where little children can get at it.
You should not jump out of an airplane without a parachute.
You should not make jokes about bombs while in line at airport security.
AND
You should not use “strip bark” samples for temperature reconstructions.

114. Paul Dennis
Posted Nov 7, 2007 at 1:36 PM | Permalink

Pete – don’t worry. I was just concerned that the spam filter at work might have been overzealous!

115. Sam Urbinto
Posted Nov 7, 2007 at 6:40 PM | Permalink

Oh, wow, while reading the transcript, actually Dr. Wegman lists 6 people that peer reveiewded his report.

What a surprise, none publish with him. Or at least not activly collborate on research papers.

116. Michael Jankowski
Posted Nov 7, 2007 at 7:44 PM | Permalink

[Manns] decentered methodology is simply incorrect mathematics.

[sarcasm on]But just because it is “simply incorrect” doesn’t mean you can’t do it that way, right? [\sarcasm off]

117. steven mosher
Posted Nov 7, 2007 at 8:37 PM | Permalink

You shouldnt do incorrect math. I’m not saying you must not do it.

118. Geoff Sherrington
Posted Nov 7, 2007 at 9:13 PM | Permalink

If I may, a brief reiteration of a point on another thread that had got old.

One of the plant nutrients between the major N, P, K tio and the traces like Mo and B and Se are the intermediate nutrients. One of these is sulphur, taken up in fair quantity. We have the inverted U probability again. No sulphur and the tree dies. Too much sulphur (commonly derived from atmosphereic SO2) and you get acid rain and growth reduction. The sweet spot interacts in complex ways with other nutrients and growing conditions. I have little doubt that S deficiency or excess would affect growth ring dendrothermometry.

Common sources of SO2 have risen in the last decades, from sources such as combustion of dirty pyritic coal and smelting of sulphide mine ores. I cannot see the logic of assuming that CO2 and methane are the main culprits when SO2 has a similar ability to disturb. It might be the case that SO2 has been studied, but in casual reading I gave not seen it mentioned apart from the context of aerosols, Richard Lindzen style.

119. Clayton B.
Posted Nov 7, 2007 at 10:17 PM | Permalink

118 Geoff,

Common sources of SO2 have risen in the last decades, from sources such as combustion of dirty pyritic coal and smelting of sulphide mine ores.

By sources rising, do you mean SO2 emissions have increased? I’m not very familiar with coal industries but many sulphide smelting operations are currently undergoing SO2 abatement projects (especially in Canada). Also, with the implementation of more efficient acid plants, I would expect SO2 emissions to have decreased in last decades. I’m not sure we’re on topic here…

120. JDN
Posted Nov 7, 2007 at 11:25 PM | Permalink

Clayton …

China is a leading emitter of SO2, with heavy use of low grade, high-sulfur content coal.

121. Geoff Sherrington
Posted Nov 8, 2007 at 4:18 AM | Permalink

Simply pointing out that C fertilisation (nutritional) is not the only potential gaseous atmospheric effect on dendrothermometry; that CO2 is only one choice; and that anomalous SO2 possibly has a longer record of influence on dendro than CO2 as it can come from volcanos in large quantities. Was not just talking about the last few decades, though I did not make this clear. Have visited a number of Chinese mines and would think that SO2 in the air has not reduced much, even though amelioration has been mandated in many countries. Main point relates to dendrothermometry. No matter how good a recent calibration period might be, it will be upset by a SO2 factor from smelters and fossil burning. Besides, who knows what went on in the centuries prior, especially volcanics?