Juckes Reply #3

Juckes also replied to CA reader Mark Rostron and there were a couple of interesting aspects to the response.

1. In a Millenial reconstruction, it would be helpfull to know how many data points were
available for each measured time period over the thousand years. Fig 2 indicates that
the number in the early years may be quite small.

(1) Figure 2 refers to data used in the Mann et al. (1999) paper. In our reconstruction
we use only data which are available throughout the study period, 13 time series in the
revised manuscript.

2. How much recent data is there? Does the sampling end in 1980, 26 years ago?

(2) Many of the proxy series end in the 1980’s. For calibration purposes it is important
that proxies are from sites which have been undisturbed. This is now rather harder to
ensure because of the spread of agriculture and nitrogen pollution.

3 Although Climate Models are mentioned, there are no references to any climate
model that has accurately predicted temperature, are any available?

(3) There is a discussion in the IPCC report.

4 Appendix A2 says that “This method starts out from the hypothesis that different
proxies represent different parts of the globe”, but there are no correlations shown to
local temperature see Table 1 and 2. Has any test been done to see if the Proxies
correlate well with local temperatures. Is any local temperature data and correlation

(4) Local correlations are only meaningful if the signal to noise ratio is greater than
unity. This is not the case here.

5 What is the justification for including low correlation proxies listed in Table 1?

(5) As above.

6 Although Table 2 shows a cross correlation between different data bases, there is no
cross correlation between individual proxies. Do the various Proxies correlate well with
each other in the non instrumental temperature period?

(6) Yes: Jones et al (1998) comment on this. We have looked at this, but have not
found any new results.

Here are a couple of points that caught my eye here. The original Juckes et al submission had 18 proxies; now they have 13 proxies. I wonder which proxies have been subtracted. I presume that the duplicate versions of Tornetrask have gone to one version, but who knows. There’s an amusing aspect to this. Juckes’ statistical model for CVM stated:

This method starts out from the hypothesis that different proxies represent different parts of the globe

This can obviously be said about anything. So Juckes et al used two Tornetrask versions – did these “represent different parts of the globe”? Maybe one Tornetrask series “represented” Sweden and the other Tornetrask series “represented” Antarctica or South Africa. Anything is possible in their model as stated. Alternatively, if one wanted to test the hypothesis that the Juckes proxies represented different parts of the globe, one way of testing the hypothesis would be to plot the locations. Now Juckes had multiple different geographical locations for Tornetrask, including the middle of the Baltic Sea, but I think that he’s prepared to grudgingly concede that all the data comes from northern Sweden. So I guess he was just kidding us when he said that the proxies in the first submission “represented” different parts of tghe globe.

There are still some puzzles in the next step of this beauty contest. The submission included two series from one Quelccaya ice core – one for dO18 and one for accumulation. Did these two series from the same ice core “represent different parts of the globe”? Maybe one of the series “represented” northern Sweden? Or Australia? Or Crawford, Texas? Or Minneapolis? He had 4 different bristlecone and foxtail series, including series only a few tens of miles apart. Did these “represent” different parts of the world through Mannian teleconnection? Maybe the Upper Wright Lakes foxtails represented Poland, Boreal foxtails Afghanistan, the Methuselah Walk bristlecones Thailand and the Indian Garden bristlecone the South Atlantic? IT sounds silly expressed like this, but the assumptions of the “model” do not preclude this.

Here’s something else that’s fun. We’ve talked before about Mann’s explanation for the failure to use up-to-date proxies: difficulties in deploying “heavy equipment” to remote parts of the world. Bringing the bristlecones up to date for example would require the coring of bristlecones in Sheep Mountain CA, which is at least an hour’s drive from the nearest airport in Bishop CA. Updating these records is clearly impossible without a commitment equivalent to the space program. Or is it? Hasn’t Hughes already updated these records? It’s just that he hasn’t reported this in the 5 years since re-sampling in 2002. I’ve speculated that, if bristlecone ring widths were off the charts as Mannian methodology would predict, we would have heard about it by now, just as we would have heard about Thompson’s drilling at Bona -Churchill if there had been an increase in dO18.

Instead of blaming the update failure on the insuperable logistics of going one hour from Bishop CA< Juckes points to potential site contamination by "nitrogen" pollution. Excuse me, wasn't potential nitrogen fertilization one of the possible problems with bristlecones (e..g. NAS Panel)? If fertilization and such are problems in updating the proxies, isn't it possible that fertilization was a problem before 1980?



  1. James Lane
    Posted Mar 9, 2007 at 1:22 AM | Permalink

    (4) Local correlations are only meaningful if the signal to noise ratio is greater than
    unity. This is not the case here.

    We get this also in the reply to McIntyre. What sort of nonsense is this? I think an eight year old would understand that if tree ring size was not correlated to local temperature then it is useless as a temperature proxy. And as a poster on another thread observed, if the S/N ratio is less than unity, the proxy should be discarded, rather than promoted.

  2. Jean S
    Posted Mar 9, 2007 at 2:52 AM | Permalink

    I think an eight year old would understand that if tree ring size was not correlated to local temperature then it is useless as a temperature proxy.

    Well, the teleconnections, you know. Trees is the magical world of the Team are responding to the average hemispheric temperatures, not to the local ones.

    And as a poster on another thread observed, if the S/N ratio is less than unity, the proxy should be discarded, rather than promoted.

    I don’t think discarding data is a good approach. One could instead try to get rid of the noise, e.g., by filtering the proxies. But still, if it really is that SNR of the proxy is below the noise level, I think there should be a serious discussion if these really are good proxies for the temperature. Also, I’d like to know how this SNR was calculated.

  3. bernie
    Posted Mar 9, 2007 at 6:45 AM | Permalink

    Jucke’s statement that “Local correlations are only meaningful if the signal to noise ratio is greater than
    unity. This is not the case here.” makes me wonder about the entire logical basis for the various tree rings
    and ice cores as bases for estimating temperature. How well settled is it that tree rings are reasonable
    proxies? If not, then isn’t the entire HS exercise fatally flawed. Why not use price indices for wheat and
    other more moisture dependent crops as proxies for temperature? Different agricultural products might allow
    for a factoring out of precipitation impact on pricing. It seems to me that these series would be both more granular
    though perhaps less geographically dispersed.

  4. James Lane
    Posted Mar 9, 2007 at 7:33 AM | Permalink

    Leaving aside the SNR, I don’t understand how these people can contend that a series that doesn’t correlate with local temperature is a valid proxy for global temperature. Teleconnect me not, I want a physical explanation. I understand that increased temperature over there might have increased precipitation here etc but surely this is all part of the “noise”?

  5. BKC
    Posted Mar 9, 2007 at 9:26 AM | Permalink

    Re. #2

    Even if you filter or average to reduce the noise, I think the bigger problem is determining what part of the signal(s) you see is temperature, precipitation, C02 fertilization or some other factor.

  6. James Erlandson
    Posted Mar 9, 2007 at 9:28 AM | Permalink

    A single tree will show growth based on temperature but it will not be possible to separate that from growth based on other factors ‘€” CO2, sunlight, fertilizer, moisture. For this study, temperature is the (small) signal while the others are (big) noise. The hypothesis is that temperature is the only signal that will affect simultaneously all trees everywhere. So by averaging tree ring data from many trees from all over the world, the noise (CO2, sunlight, moisture, fertilizer and others) will average out to zero leaving only the temperature signal.

  7. Allan J
    Posted Mar 9, 2007 at 9:34 AM | Permalink

    Re: #2

    I am sorry if this is a foolish input. It has been decades since I did any S/N analysis. But, if my memory is correct, in the olden days to work with SNR less than unity you needed pretty good knowledge of the nature of the signal and of the noise. If this is still true shouldn’t examination of the SNR calculations include assessment of whether or not both the signal and the noise data were good enough to support the calculations?

  8. jae
    Posted Mar 9, 2007 at 10:33 AM | Permalink

    #3, 6. Trees are good proxies for moisture in temperate zones. They are very poor proxies for temperature. Moisture almost always limits growth in the areas where these trees grow. As noted by Veizer during photosynthesis a plant has to exhale almost 1,000 molecules of water for each molecule of CO2 absorbed (the water use efficiency). Only in tropical areas and rain forests is there enough water.

    I really get a kick out of Juckes reasoning for not including samples beyond the 80s. What a cop out!

  9. Mark T.
    Posted Mar 9, 2007 at 11:53 AM | Permalink

    But, if my memory is correct, in the olden days to work with SNR less than unity you needed pretty good knowledge of the nature of the signal and of the noise.

    Depends upon what you mean by “pretty good knowledge.” Most noise sources are assumed Gaussian (which is not always true to the detriment of the algorithm). Most signals, in the context of this discussion, are typically narrowband w.r.t. the sample rate. Furthermore, multiple observations allow processing to remove noise. All is not lost with less than unity SNR. Having more information about the signal is definitely a plus, however.


  10. John Hekman
    Posted Mar 9, 2007 at 11:58 AM | Permalink

    Temperatures in the 1990s and following have been unprecedented. This is shown by tree rings, although not by tree rings during the time when temperatures were unprecedented.

    Tree rings cannot be shown to respond in a statisitcally significant way to temperatures, at least the temperatures that trees actually experience. But taken in the aggregate, tree rings are indeed a proxy measure for temperatures.

    Tree rings tell us that the recent increase in temperatures is unprecedented in at least the last thousand years. If we actually took tree ring measurements for these recent unprecendented years, the tree rings would bear us out on that assertion.

    But it is too much trouble to measure these thirteen tree ring series to find out.

    ….I have tried not to attribute any motivations or feelings to why the assertion of attribution to human activity has been made. Am I making a mistake in stating the facts as we now know them?

  11. jae
    Posted Mar 9, 2007 at 12:08 PM | Permalink

    #10: Congratulations, you have got it!

  12. John Hekman
    Posted Mar 9, 2007 at 12:21 PM | Permalink

    May I just add, it is a damn shame that some of the billions of dollars that are being spent on global warming research cannot be allocated to measuring the recent growth of these thirteen tree ring series.

  13. DeWitt Payne
    Posted Mar 9, 2007 at 2:33 PM | Permalink

    If one were really interested in determining/proving a linkage between temperature and tree ring width/density with an individual record SNR less than one, wouldn’t one do some sort of meta-analysis of the different tree ring records against the weather history at each site rather than against a hemispheric average? Not to mention that the confounding variables like CO2 fertilization are strongly correlated with the temperature record. I guess if one is convinced that high CO2 = high temperature, maybe it isn’t really a confounding variable.

  14. bernie
    Posted Mar 9, 2007 at 5:20 PM | Permalink

    #11, #12, #13

    I will only trust the measurements of the tree rings in thelast few years if the trees are cut and measured by
    those without a dog is this fight. I am still baffled at the complexity of measuring a tree ring on a tree
    slice to begin with.

    With respect to controlling for all those confounding variables, do they have tree slices from multiple
    trees at a single site and first average those or do they rely on single trees from each of a number of widely
    dispersed geographic sites. My BS detector is beginning to sound an alarm, but I would truly like to hear
    from a dendrologist on this subject – preferably one, as I said, without a dog in this fight.

  15. nanny_govt_sucks
    Posted Mar 9, 2007 at 6:19 PM | Permalink

    And what is the point of taking cores from multiple trees at a site and averaging if not to find the LOCAL “signal” from the “noise” of the many individual trees?

    Why average individual trees to come up with a single proxy? What is the justification for this attempt at local signal detection if local signals are not represented in the final multi-proxy result?

    Can an average of a group of tree rings at a site act as a better teleconnector than an individual tree ring?

  16. jae
    Posted Mar 9, 2007 at 6:50 PM | Permalink

    Bernie: slices are rarely used, since the tree must be cut down to obtain a slice. (one could get into megatrouble by cutting down a 4,000 year old bristlecone pine).. Instead, they use a hollow “increment borer” that extracts cores from the trees, and they often have to take multiple cores just to locate the pith (so the rings are perpendicular to the axis of the core). Several “good” cores should be taken from each tree, and several (>10, generally) trees per site should be sampled, if possible (if there’s only one very old tree, you may be stuck with that.) It gets tricky getting multiple “good” cores from trees with irregularly shaped boles, which is usually the case for very old trees.

    See here for info on tree rings and databases.

  17. bernie
    Posted Mar 9, 2007 at 8:18 PM | Permalink

    Many thanks. I knew about the cores but I had no idea that living bristlecone pines can be 4000 years old.
    That is amazing and a little humbling and I afraid my ignorance is showing. My comments about slices came from
    a misreading of a picture of Mann on RealClimate.org where he is holding what I took to be a slice.
    Seems that the picture is somewhat misleading.
    I take it from your last comment that you see the analysis of the cores fundamentally difficult.
    Thanks for the reference site.

  18. Andrey Levin
    Posted Mar 10, 2007 at 2:29 AM | Permalink

    Elevated CO2 atmospheric concentration is extremely important factor for tree growth. See, for example, CO2 science:

    “…flooding of the air with CO2 has likely already resulted in mean yield increases of 70% for C3 cereals, 28% for C4 cereals, 33% for fruits and melons, 62% for legumes, 67% for root and tuber crops, and 51% for vegetables, all due to the aerial fertilization effect of atmospheric CO2 enrichment and its concurrent anti-transpiration effect, which together greatly enhance plant water use efficiency”


    Is it possible that “unprecedented” rise of temperature ‘€” according to tree ring proxies ‘€” during last two centuries is in fact result of just CO2 fertilization? CO2 increase in atmosphere is global phenomenon, not a local noise.

    Could it be that tree ring temperature proxies are the darling of AGW crowd (and not ocean sediment, stalagmite, etc. proxies) for exactly this reason? Or tree ring proxies could be more easily manipulated?

    They definitely prefer weather station data instead of satellite and weather balloon measurements even for last 25 years, for quite obvious reasons.

  19. fFreddy
    Posted Mar 10, 2007 at 4:49 AM | Permalink

    Re #14, bernie

    My BS detector is beginning to sound an alarm, but I would truly like to hear
    from a dendrologist on this subject

    I think my favourite quote from a dendrologist was the one about “The great thing about dendroclimatology is that you can select which tree rings to use to enhance your signal” or words to that effect.

    Can anyone remember the exact source and/or provide a link for this ? Let’s see if we can cause bernie’s BS detector to blow a fuse …

  20. Bernie
    Posted Mar 10, 2007 at 8:45 AM | Permalink

    From http://sonic.net/bristlecone/growth.html
    “Spring comes to the bristlecone pines in early May with the melting of snow and higher temperatures. Each year the tree increases in girth only 1/100th of an inch, often less, and new cones and twigs are formed. In this subalpine zone there are only three warm summer months, often only 6 weeks, to produce growth and reserves for overwintering. All of this must be accomplished on a mere 10″ (25.4cm) precipitation. During cool summers, the trees live on energy produced that summer and grow on the reserves stored from the previous year. To live so long under such conditions, the bristlecone has established several strategies.”

    This description of the apparent way bristlecone rings develop raise even more questions in my mind. However, I am sure that there are experts visiting this site who can clarify how the measurement process actually works.
    I also assume that we have accurate local climate measuresments that allow the modeling of the relationship between tree ring width, temperature, precipitation, cloud cover, etc. Or do the above comments by JH and DP summarize the lamentable lack of foundation for the use of these proxies.

  21. fFreddy
    Posted Mar 10, 2007 at 9:05 AM | Permalink

    Re #20, Bernie

    I also assume that we have accurate local climate measuresments that allow the modeling of the relationship between tree ring width, temperature, precipitation, cloud cover, etc.

    In the case of the bristlecones, whose increase in ring width in the 20th century is responsible for the hockey stick shape of MBH, then yes, there is a climate station reasonably nearby.
    Unfortunately, it shows pretty much flat average temperatures over the 20th century.

  22. bernie
    Posted Mar 10, 2007 at 12:18 PM | Permalink

    Clearly I need to do more reading. I have read through Wegman et al and the subsequent questions. I am off to read the NAS report. Are there any other authorative discussions on BristleCone proxies? A solid defense of their use might also be informative.

  23. bernie
    Posted Mar 10, 2007 at 12:23 PM | Permalink

    That is odd. I am off to read the NAS report, but are there any other authoritative articles on the use of Bristlecones as proxies? Any reasonabe defenses of their use as temperature proxies?

  24. Steve McIntyre
    Posted Mar 10, 2007 at 2:32 PM | Permalink

    #23. Look at the citations in our EE 2005 article. I tried to review relevant literature on bristlecones and there were multiple problems. Also if you go to Bristlecones in the Categories frame here, you’ll see a vaiety of posts and discussion about bristlecones.

    At the end of the day, it’s amazing how intransigent the Team is about keeping bristlecones and foxtails in their calculations – I wonder why.

  25. bruce
    Posted Mar 10, 2007 at 3:23 PM | Permalink

    Re #24:

    At the end of the day, it’s amazing how intransigent the Team is about keeping bristlecones and foxtails in their calculations – I wonder why.

    What a superb example of a rhetorical question! Wikipedia (reliable on most matters other than William ‘Stoat’ Connolly’s version of climate change) says:

    A rhetorical question is a figure of speech in the form of a question posed to make a statement or initiate introspection rather than for the purpose of getting an answer. (“How many times do I have to tell you to stop walking into the house with mud on your shoes?”).

    A rhetorical question seeks to encourage reflection within the listener as to what the answer to the question (at least, the answer implied by the questioner) must be. When a speaker declaims, “How much longer must our people endure this injustice?” or “Will our company grow or shrink?”, no formal answer is expected.

    However, just to see if I have got it right. Could it be that their corpus falls apart if the bristlecones and foxtails are taken out of their calculations?

  26. bernie
    Posted Mar 10, 2007 at 4:36 PM | Permalink

    You are spoiling the fun!!

  27. DocMartyn
    Posted Mar 11, 2007 at 6:32 AM | Permalink

    Anyone want to produce a plot of tree ring width vs. area of land under irrigation, in the state of California over the last 100 years? Even from the air you can see the areas of California that are irrigated, huge circles of green against the background of brown. Irrigation must raise the water vapor content of the air, and raise the dew point. Is the irrigation water making its way to the Bristlecones?

  28. BradH
    Posted Mar 13, 2007 at 6:26 AM | Permalink

    One thing I will say about Juckes is that, at least he replies.

    I recall when he first appeared here, he was asked whether he really (seriously) believed that his study was defensible. His response was something to the effect that, “Well, I certainly hope so.”

    To give him his due, whilst we may not be satisfied with the responses he is giving, at least he engages somewhat, rather than the standard Hockey Team response of, “Run away!” This leads me to believe that he might actually believe in what he’s supporting.

    I remain intrigued to see at what point he stops defending his study and either claims victory, or claims that continued correspondence is a waste of time. [An end with a concession that the study has too many flaws is too much to expect.]

    I can respect someone who is at least prepared to try and defend their position.

    I cannot stand cowards such as Mann and Jones, who toss cake from their windows at Versailles, then return to the fawning adoration of their courtiers.

  29. Willis Eschenbach
    Posted Mar 13, 2007 at 7:04 AM | Permalink

    Heck, BradH, I think all of the team seriously believe that their studies are defensible. And I don’t think that there is some kind of conspiracy.

    I also am not impressed with Juckes “replies”. Responding to pages and pages of clear problems in his paper with a few content free sentences that basically say “No, no, you’re wrong and I’m right” is not replying in my book. In some ways, I prefer Mann’s method … at least he’s not pretending to reply.


  30. bruce
    Posted Mar 14, 2007 at 12:15 AM | Permalink

    Re #29: Willis, I agree with your comment.

    In effect, the questions and comments put up in response to the Juckes et al paper are those of generally informed reviewers, and as I understand it, it is the intent of the COPD modus operandi to expose new papers to such review.

    It can be understood that an author might want to dismiss trivial or obviously uninformed questions. However, to this observer, there were precious few, if any of those. Sound scientific practice surely requires that an author respond constructively to reviewer’s questions. That Martin Juckes has chosen not to do so is highly visible to interested observers, and also to the COPD people.

    Scientists will just have to accept that the development of the internet and blogs brings transparency to their work these days, and they must expect their work to become subject to detailed scrutiny.

    Confident scientists welcome this scrutiny, and will engage with those who question their work. Scientists less confident about their work may engage in less than co-operative practices that we have seen from a few working in the area in recent times, now apparently joined by Martin Juckes et al.

    It would be very interesting to know whether the ‘et al’ co-authors endorse Martin Juckes approach.

  31. Steve Sadlov
    Posted Mar 14, 2007 at 10:25 AM | Permalink

    RE: #27 – You have an excellent point. In particular, I have to wonder what impacts all the mid to late 20th century new irrigation in Western Nevada and the Colorado Valley has made on precipitatable water of moonsonal convective storms, which serve as the primary source of warm season moisture beyond snow melt remnent moisture. This is particularly true during the August – October time frame, when the snow is at maximum melt back and there is no way that any mid latitude cyclonic storms can make it past the combination of the Pacific High and the Sierra Nevada.

  32. bruce
    Posted Apr 19, 2007 at 1:41 AM | Permalink

    At CoPD on 18 April 2007, one Hugues Goosse who describes himself as Editor has posted a response to the Reviewer 1 Comment on the Juckes paper.

  33. Steve McIntyre
    Posted Apr 19, 2007 at 8:05 AM | Permalink

    A link to the Goosse response is here . It’s hard to figure out why the Editor is responding under the comments of a specific reviewer rather than as a separate thread as in the Burger and Cubasch paper. They mention a “revised” version although no such revised version has appeared online. It looks to me like that the revised version made only cosmetic changes, that the editor is not satisfied with the situation and has required a substantial re-write.

    Editor Goosse has coauthored recently with Mann. Our criticisms of Juckes were not even mentioned by Goosse.
    Unlike the Burger and Cubasch paper last summer, where online re-submission was required for a major re-write, he seems to have created a situation where the authors can circumvent online scrutiny for their re-write. We shall see.

  34. Posted Apr 19, 2007 at 12:39 PM | Permalink

    Our criticisms of Juckes were not even mentioned by Goosse.

    damn those denialists 😉 No link to CA at RC, etc.

    http://www.webster.com/dictionary/denial :

    a psychological defense mechanism in which confrontation with a personal problem or with reality is avoided by denying the existence of the problem or reality

  35. Stan Palmer
    Posted Apr 19, 2007 at 1:52 PM | Permalink

    So if this paper is published by editorial fiat with the peer reviews being ignored, how could it ever be usedthe IPCC which relies on peer-reviewed research. Publication by editorial fiat would render this paper useless.

  36. Armand MacMurray
    Posted Apr 19, 2007 at 2:38 PM | Permalink

    Steve’s link is to the Ref#1 review; the link to Goosse’s comment is: http://www.cosis.net/copernicus/EGU/cpd/2/S964/cpd-2-S964.pdf

  37. Posted Apr 23, 2007 at 2:49 PM | Permalink

    I think we’ll see very interesting peer-reviewed papers this year:

    Juckes et al will be accepted to CP
    Mann’s Climate over Climate Over the Past Two Millennia (*) will be published soon..

    .. and it relies heavily on Robustness of Proxy-based Climate Field Reconstruction Methods, which we will see soon as well.

    And then we’ll get Wahl and Ammann. Peer review works.

    (*) Storch assumes unrealistically large inflation of variance, even larger than the one we saw in MBH99 🙂 !

  38. hadenough
    Posted Apr 23, 2007 at 3:18 PM | Permalink

    These guys are being paid by ‘We The People’! Right??

%d bloggers like this: