Divergence: the Young Dendros Rebel

I’ll try to do reports on various interesting aspects of the AGU conference over the next few weeks. Today I’m going to post on the session on the Divergence Problem, initiated and chaired by Rob Wilson, and which, for the most part, consisted of young dendros probing critically at the issue of the failure of ring widths to record recent warmth. This issue was touched on by both the NAS Panel and IPCC AR4, both of which arm-waved through the problem, relying on a rather lame explanation by Cook (who does a lot of solid work, but this wasn’t one of them).

There were 8 oral presentations in the 4pm Friday time slot, the worst possible time slot at AGU. All but two of the presenters were fairly young. Cook was in attendance, but neither Hughes nor Mann bothered coming. Perhaps the most remarkable aspect of the session was what wasn’t mentioned: none of the presenters mentioned the NAS panel or the IPCC analysis of the Divergence Problem.

There was an ancillary poster session in the morning, to which I was consigned. Had I been giving an oral presentation, I would have analysed the Divergence Problem as set out in my abstrat, but, since it was a poster presentation, I presented Almagre results in a divergence context notwithstanding my abstract. Rob Wilson teased me about it a little but was OK with it. I got quite a bit of traffic – I think that I talked to all the presenting dendros at the poster, something that would not have been possible at the oral session, where the presentations run in military sequence.

The “Divergence Problem” has been an issue that’s been covered since the start of CA. I’ve made a new keyword category Divergence to collect many of the posts and, for anyone interested in the topic, these posts make a pretty good review. I’ve certainly done what I can to keep this issue in the public eye, but the interest of Rob and the various presenters obviously pre-dates the interest expressed at CA.

The Divergence Problem and Asking for MBH Data
Actually the Divergence Problem is one of the things that prompted my initial inquiry to Michael Mann for his data. In April 2003, my attention had been drawn to the following quote by Malcolm Hughes, originally made in 1999 in connection with the publication of Vaganov, Hughes et al 1999:

The recent marked weakening in the correlation between tree growth and temperature means that past climate reconstructions are even more reliable than previously thought, but forces scientists to rethink the role of the vast northern forests in the global carbon cycle, said Malcolm Hughes. Hughes, professor and director of the Laboratory of Tree Ring Research at the University of Arizona in Tucson, is co-author of the Nature paper. Currently he is doing field research in the northern Rockies, but will return to the Tucson campus on July 20.

“The recent weaker correlation between tree growth and temperature clearly affects the reliability of our reconstructions of the past. Actually, it means past climate reconstructions (before the 1960s) are better than we thought they were. And, as a result of this, it means that we underestimated the differences between the present century and past centuries,” Hughes said.

The idea that worse correlations showed that their reconstructions were “better” made no sense to me. A few days later, I noticed a statement from Simon Brown also observing problems with tree ring reconstructions as follows:

Dr Simon Brown, the climate extremes research manager at the Meteorological Office at Bracknell, said that the present consensus among scientists on the IPCC was that the Medieval Warm Period could not be used to judge the significance of existing warming. Dr Brown said: “The conclusion that 20th century warming is not unusual relies on the assertion that the Medieval Warm Period was a global phenomenon. This is not the conclusion of IPCC.” He added that there were also doubts about the reliability of temperature proxies such as tree rings: “They are not able to capture the recent warming of the last 50 years,” he said.

For some reason, within a day or two of reading these news releases, I emailed Mann asking for the location of MBH98 data (something that I’d never done before). My inquiry was very much related to the divergence problem as I wanted to see if any of his proxies were breaking new ground in the 1980s and 1990s – something that seemed inconsistent with the divergence issue.

I mention this background merely to show that the divergence issue has been very influential in my approach to this problem.

Briffa et al on Divergence
I’ve done a number of posts on Briffa and divergence, including 529, 536, 570, 586, 996. The Divergence Problem was originally raised in Briffa et al 1998, where there is an important figure from Briffa et al. [Proc Roy Soc 1998] showing the relative decline of MXD and RW relative to temperatures since 1960. While there have been a number of subsequent articles on MD divergence, this is a very rare figure on ring width divergence.

Briffa et al. 1998 Original Caption. Figure 6. Twenty-year smoothed plots of averaged ring-width (dashed) and tree-ring density (thin solid line), averaged across all sites in Figure 1, and shown as standardized anomalies from a common base (1881-1940), and compared with equivalent-area averages of mean April-September temperature anomalies (thick line). [SM – it looks to me like the labels in the caption are reversed between density and temperature]

One of the Briffa “cargo cult explanations” for divergence came from Briffa et al 2002, which Ross and I quoted in our NAS presentation:

Briffa et al. (1998b) discuss various causes for this decline in tree growth parameters, and Vaganov et al. (1999) suggest a role for increasing winter snowfall.… In the absence of a substantiated explanation for the decline, we make the assumption that it is likely to be a response to some kind of recent anthropogenic forcing. On the basis of this assumption, the pre-twentieth century part of the reconstructions can be considered to be free from similar events and thus accurately represent past temperature variability. [Briffa et al. 2002]

The Divergence Problem was an issue that grew legs during the NAS presentation after Cuffey noticed the difference between the D’Arrigo recon and temperature – to which Rosanne D’Arrigo answered: “That’s the Divergence Problem” referring him to the Briffa cargo cult explanations. See posts 570, 584,714 and 715 for the NAS panel on this. On presentation day, D’Arrigo and others were totally unable to give a coherent explanation of the matter, so this was an issue that I was very interested in seeing how the NAS panel treated it. (See 714 for my preview). They ended up relying (unsatisfactorily IMHO) on Cook et al as follows:

(111) The observed discrepancy between some tree ring variables that are thought to be sensitive to temperature and the temperature changes observed in the late 20th century (Jacoby and D’Arrigo 1995, Briffa et al. 1998) reduces confidence that the correlation between these proxies and temperature has been consistent over time. Future work is needed to understand the cause of this “divergence,” which for now is considered unique to the 20th century and to areas north of 55°N (Cook et al. 2004)… also that the difference between northern and southern sites found after about 1950 is unprecedented since at least A.D. 900.

In my review comments on IPCC AR4, I was extremely critical of their truncation of the “inconvenient” bits of the Briffa et al 2001 recon where it went down after 1960. IPCC decided that it was “inappropriate” for the hoi polloi to be shown a downtrending recon, but did insert a few sentences on the divergence problem, following a similar line as follows:

All of the large-scale temperature reconstructions discussed in this section, with the exception of the borehole and glacier interpretations, include tree ring data among their predictors so it is pertinent to note several issues associated with them….. Several analyses of ring width and ring density chronologies, with otherwise well-established sensitivity to temperature, have shown that they do not emulate the general warming trend evident in instrumental temperature records over recent decades, although they do track the warming that occurred during the early part of the 20th century and they continue to maintain a good correlation with observed temperatures over the full instrumental period at the interannual time scale (Briffa et al., 2004; D’Arrigo, 2006). This “divergence” is apparently restricted to some northern, highlatitude regions, but it is certainly not ubiquitous even there. In their large-scale reconstructions based on tree ring density data, Briffa et al. (2001) specifically excluded the post-1960 data in their calibration against instrumental records, to avoid biasing the estimation of the earlier reconstructions (hence they are not shown in Figure 6.10), implicitly assuming that the “divergence” was a uniquely recent phenomenon, as has also been argued by Cook et al. (2004a). Others, however, argue for a breakdown in the assumed linear tree growth response to continued warming, invoking a possible threshold exceedance beyond which moisture stress now limits further growth (D’Arrigo et al., 2004). If true, this would imply a similar limit on the potential to reconstruct possible warm periods in earlier times at such sites. At this time there is no consensus on these issues (for further references see NRC, 2006) and the possibility of investigating them further is restricted by the lack of recent tree ring data at most of the sites from which tree ring data discussed in this chapter were acquired.

This lack of recent data was elsewhere explained by Michael Mann and apologist Eli Rabett as due to the inordinate “expense” and time required to update these sites.

The Young Dendros
The “young dendros” presenting at the session, aside from Rob Wilson as chair, included many who’ve been discussed here – nearly always in favorable terms BTW: Andrew Bunn (220), Martin Wilmking (76,527,1319), Trevor Porter (an associate of Mike Pisaric 1278,1283,1287), Greg Wiles, plus some not as young dendros Glenn Juday (76,527,1319), Brian Luckman (many mentions here). Ze’ev Gedalof of Guelph, who’s been very helpful with the Almagre analysis, had an interesting poster on CO2 fertilization and it was disappointing that he didn’t get face time at the oral session, although it would be hard to say who should have been bumped.

Hughes had presented his bristlecone results at an earlier session on High Mountain Environments. It’s not that he couldn’t justify presenting at that session, but it would have been more stimulating to see what the young dendros worried about divergence would have made of Hughes’ results. However, Ed Cook was there. (I actually have a lot of time for Cook’s articles which I find to be typically very substantive – although I think that his article on the Divergence Problem is one of his weakest efforts.)

Bunn and Lloyd
The “keynote” presentation was by Andy Bunn, together with Andrea Lloyd. (In passing, Bunn mentioned to me that he’s placed a number of his dendro functions in an R package.) Bunn’s presentation was one that most CA readers would have endorsed.

He noted that the Divergence Problem had arisen out of a concern over paleoclimate reconstructions, but observed that there were other reasons for concern over the problem, including forest management. In a poster, he and Andrea Lloyd showed that there were large areas of Arctic “browning” as well as “greening”.

As to possible causes of Divergence, he noted that trees were notoriously very plastic both in phenotype and genotype; that linearity was all too often assumed when the full range of variability had not been established; and that, until the mechanism for divergence had been established, one could not assume that the present divergence was unique. (I note once more that this latter point is a direct contradiction of a key NAS panel conclusions, one that I sharply criticized at the time.)

He noted all the different factors that could affect tree growth: temperatures, soil moisture, dimming, ozone, nitrogen, CO2 fertilization…

He then presented a simulation very similar to one that David Stockwell worked up a couple of years ago in the context of CA discussions. He showed a simple simulation of CA type in which one factor (T) had an upside-down U response and a second factor (P) had a sigmoid-response and then showed a “reconstruction” based on plausible values of the two inputs, showing that a divergence problem easily arose in the past (“medieval”) period, in which the reconstruction failed to recover high values of Factor 1. In his words, if all you have is a value of 6, you can’t tell whether you got it through 5+1 or 4+2.

Bunn’s emphasis on the need to provide an actual explanation of divergence was very healthy. He didn’t rebuke the cargo cult theories as I would have done, but the message was still there.

Martin Wilmking
Martin Wilmking reported the difference between “positive” and “negative” responders – trees that might be only 10 meters apart. CA has covered Wilmking’s work in the past and, on one of the occasions that CA was being criticized by the one of the old generation of dendros, Wilmking observed that our coverage of his work had been very fair.

A point that he made in passing, but which deserves emphasis IMO – was that the existence of positive and negative responders in close juxtapostion argued against the Divergence Problem being the result of ozone or dimming or some factor that would apply to all trees – a powerful argument against the cargo cult explanations of a still unknown anthropogenic factor. He said that pos and neg responders had responded similarly until recently when substantial differences arose. He argued that a chronology based on positive responders matched temperature well. The weakness in his argument – and this criticism was made by Andy Bunn in slightly different terms – is that any after-the-fact division of autocorrelated series into “positive” and “negative” responders might well generate something similar.

Laxton (of Greg Wiles’ group) reported on three sites in the Gulf of Alaska: Mt Eyak, Glacier Bay and Columbia Glacier – in each case showing informative maps linking the tree sites to the nearby glaciers. She observed that the retreat of the Columbia Glacier had been accompanied by a derease in ring widths at the nearby Great Nunatak site, She hypothesized that the glaier might have generated a small high-pressure system, whih affected precipitation. IT was a pretty dramatic divergence example.

Juday presented results from productive white spruce sites. He presented temperature data from several sites, including a site Talkeetna, which he vouched for as having no UHI contamination. He observed that there had been dramatic warming in the 20th century across a broad range of temperature measurements – in particular noting almost a doubling of frost free days from 60 to 120 at one location – so it was quite clear that these were not adjustment artifacts, but real increases in temperature.

He said that his model worked well. He showed a fit through to 1996 and that the results through 2006 were consistent. (I’m not sure whether this was an actual out-of-sample demonstration (in the sense that the fit was done originally in 1996 and worked for new data or whether it was a calibration-verification split). In either case, it will be interesting to follow up.

Brian Luckman (whose Alberta tree rings have been discussed previously here) presented results from a very large collection of Yukon sites – none of which have been archived. Luckman has an abysmal record for not archiving data. Luckman came by my poster and I had a nice chat with him, but made a point of criticizing him to his face for not archiving his measurement data. I don’t get the reluctance. Maybe authors who archive data should stop citing papers by authors who refuse.

In his presentation, he said that climate-tree relationships in the Yukon sites were weak. He referred to the Jacoby-D’Arrigo TTHH site in passing – a site mentioned in several peer reviewed articles. Luckman said that he couldn’t figure out where the TTHH site and the location shown on the map was simply a guess. So it’s not just me that’s frustrated by dendro secrecy. (One dendro mentioned to me that Jacoby and Cook were probably not being obstinate in refusing to identify the Gaspe location – it was more likely that they failed to make a map and no longer knew where the site was.)

Porter (Pisaric)
Trevor Porter of Carleton presented some results on isotopes in the Mackenzie Delta (Mean Annual temp -9.6 deg C; mean summer 11.8 deg C; mean ann precip 254 mm). He had sampled 3 trees for O18 and C13 (Max Berkelhammer had previously reported O18 for 2 bristlecones in another session discussed ….) Values of dO!8 were typically between 19 and 22 with a dwonspike to about 17 around 1895.

Mary Gagen reported on stable C isotopes from 5 European sites: LArnita, Finland (1640-2002); Turku, FIN (1895-194); Altai, RUS (1901-2000); Sandringham UK (1895-1994) and Glens, Scotland 1425-2005. dC13 levels at Larnita on her graphic were in the -23 to -25 range. She observed that dC13 values had remained very stable at most sites in the 20th century despite dramatic increases in atmospheric dC13. According to botanical understanding, this reflected a very sharp increase in Water Use Efficiency (a topic previously discussed at CA last year in connection with bristlecones here) This increase in Water Use Efficiency meant, in her words, that trees were behaving as though it were “wetter” than it really was.


The final presenter was Kevin Anchukaitis, working with Hughes, Vaganov and others, who is operating a “Forward” model of ring widths based on non-linear (trapezoid) responses to Temperature and Water. He reported on 3 northern sites: Twisted Tree, Coppermine and Hornsby Cabin (three Jacoby-D’Arrigo sites, for which only old versions have been archived.) HE said that no divergence arose relative to the model for Coppermine, but that divergences arose for Hornsby Cabin/Thelon and TTHH.

Focusing the session on the single topic of Divergence made for a good session and Rob Wilson deserves commendation for promoting the idea. In many sessions, it seems that authors tend to simply recapitulate their recent articles without necessarily making an effort to link their presentation to the topic of the session. But in this session, the authors all seemed to pay attention to the topic.

It’s not as though any answers were forthcoming, but the young dendros said that they viewed the situation as an opportunity, which is surely the right attitude. While many readers are totally dismissive of the data in the dendro collections, I’m not nearly so dismissive as it has the important advantage of being well-dated, the sites typically contain a “signal” even if we don’t know exactly what the “signal” is. Plus there’s a reasonable case that some of the other proxies (e.g. ice cores) are getting a bit of a free ride in terms of rigorous analysis, since they are so sporadic in provenance that statistical comparisons are much harder to do.

But again, my take-home impression was the dog that didn’t bark: the complete, and I mean witness protection complete, omission of any mention of the IPCC AR4 (and NAS Panel) excuses for divergence. The young dendros didn’t even waste their time arguing with them. They didn’t have an answer, but at least they faced the question.


  1. Larry
    Posted Dec 15, 2007 at 4:53 PM | Permalink

    The recent marked weakening in the correlation between tree growth and temperature means that past climate reconstructions are even more reliable than previously thought,

    Say what??

  2. Steve McIntyre
    Posted Dec 15, 2007 at 4:56 PM | Permalink

    Hey, that’s just Malcolm Hughes – so who cares other than the IPCC, Al Gore and world governments. But none of the young dendros say things like this.

  3. Steve Moore
    Posted Dec 15, 2007 at 5:01 PM | Permalink

    The recent marked weakening in the correlation between tree growth and temperature means that past climate reconstructions are even more reliable than previously thought,

    Let’s see…

    Last year, my checks cleared the bank.
    This year, they’re bouncing.
    That means that last year, my math was good.

    I can write more checks now.

  4. Larry
    Posted Dec 15, 2007 at 5:06 PM | Permalink

    Oh, I get it. Hughes uses the Microsoft defense. It’s not a bug, it’s a feature.

  5. Jeff A
    Posted Dec 15, 2007 at 5:36 PM | Permalink

    Hughes uses the “I hope you’re an idiot and not actually listening to what I’m saying” defense.

    That is a profoundly idiotic statement.

  6. Dave Dardinger
    Posted Dec 15, 2007 at 5:39 PM | Permalink

    I’m not sure what Hughes meant, but maybe it’s along the line that the nascent divergence was throwing a bug into the analysis even before 1980 so that if corrections were made for it, there’d be a closer fit of proxies to temperature in past times. I’d like to see it spelled out, however. It doesn’t seem very likely. More likely it’s just something he said off the top of his head one time and then it caught on.

  7. Steve McIntyre
    Posted Dec 15, 2007 at 6:00 PM | Permalink

    Enough about Hughes.

  8. T J Olson
    Posted Dec 15, 2007 at 6:21 PM | Permalink

    Worse…means better.

    That’s one’s worth a Homer Simpson ‘D’oh’!

    But seriously, the absence of critical and contextual appreciation of an apparent problem among young grad students doesn’t surprise me at all. I’ve seen this occur repeatedly in scholarly fields (ie, most of the humanities and much of the social sciences) over the past two decades, after they became politicized. Once inquiry is ruled by politically correct received opinion, productive, critical thinking only occurs at the boundaries of a topic, while the obvious must go unnamed, unidentified, unconfronted.

    Thus, as happened in the 1980s, the best work in these fields was often the product of Washington, DC think tanks instead of academe. So, Steve, sometimes history does repeat itself. Still, it is grand that Andy gets it.

  9. jae
    Posted Dec 15, 2007 at 6:22 PM | Permalink

    Thanks for the summaries, Steve. Those “young dendros” seem to have a very open mind. This may allow them to find out what the signal is. I hope it’s not a signal of a bacteria or virus problem, similar to chestnut blight (which completely distroyed this marvelous tree), dutch elm disease, pine blister rust (which decimated Western White Pine, and the pathogens that are after the oaks in California. This type of problem could explain why trees within a given stand are showing different signals.

  10. steven mosher
    Posted Dec 15, 2007 at 6:33 PM | Permalink

    I would hazard that you will see more of the “young dendros” in the future, for purely sociological
    reasons. Guys like JEG ( not a dendro I know) will feel this need to break out of the mainstream.
    to make a name. to be independent. When “climate science” was not yet recognized the pressure to
    conform is great. Now, that it seems on a secure path, watch for the pioneers, for the fellow heading
    off the reservation ( bad mixed metaphor there tonto) Anyway you get the point.

  11. Steve Moore
    Posted Dec 15, 2007 at 6:51 PM | Permalink

    RE #6:

    …along the line that the nascent divergence was throwing a bug into the analysis even before 1980 so that if corrections were made for it, there’d be a closer fit of proxies to temperature in past times.

    Dave D,
    I’d be more willing to concede that as a possibility if there wasn’t such an “apparent” close fit before 1960

    I think we’re seeing the effect of land use changes miles away – humidity, CO2 fertilization, airborne pollutants, dust, etc (oops, that sounds like teleconnection).

    Maybe “The Young Dendros” can puzzle this one out.

  12. John Andrews
    Posted Dec 15, 2007 at 7:02 PM | Permalink

    A quick look at the graph and my first thought was that the divergence is due to the effect of air conditioning pollution of the temperature data. I note that most of the reports only look at one side of the data and only one even considers the errors in the temperature data. As an interested observer of this global warming discussion, I have no professional expertise (radiological engineering doesn’t count) but it may be something to look at.

  13. david abrams
    Posted Dec 15, 2007 at 7:15 PM | Permalink

    “The weakness in his argument – and this criticism was made by Andy Bunn in slightly different terms – is that any after-the-fact division of autocorrelated series into ‘positive’ and ‘negative’ responders might well generate something similar.”

    Couldn’t you test this by taking the 10 best trees after 1960, and seeing how they stack up against the temperature record from 1880-1960?

    Anyway, I’m not a dendrochronologist but it makes sense to me that temperature changes would affect trees differently. Depending on the location of the tree and how its neighbors are affected.

    In general, it seems to me that if weather becomes more favorable for a tree, the same weather will help the tree’s neighbors, which may in turn hurt the tree.

  14. jeez
    Posted Dec 15, 2007 at 8:14 PM | Permalink

    Crap, let’s try again. The Young Dendros and their assault on the Mann.

    I’ll need to practice links under this software.

  15. John Norris
    Posted Dec 15, 2007 at 8:38 PM | Permalink

    from Steve’s Martin Wilmking paragraph:

    …was that the existence of positive and negative responders in close juxtapostion argued against the Divergence Problem being the result of ozone or dimming or some factor that would apply to all trees …

    Okay, this brings me back to my hypothesis about hiking tree huggers and a little shot from their water bottle. I’ll bet all the positive responders are either right next to the trail, or look like they could use a little extra help from a concerned environmentalist.

    See my comment42here

    : No. THese sites are in remote Alaska.

  16. Dan White
    Posted Dec 15, 2007 at 10:34 PM | Permalink

    I need a little help understanding something. I’ve surfed around this site but haven’t been able to find a discussion that answers my question. I understand the problems with the BCP data and how Mann used it. However, I realize that I don’t have a good feel for how various tree ring samples from different studies correlate with local (and “global”?) temperatures. The fact that there is a “divergence problem” implies that there is good historical correlation between ring width and temp. On the one hand, I see discussion of how tree growth has more to do with soil moisture than temperature (I think Mr. Pete’s expert horticultural group even laughed at the notion to the contrary), yet these studies appear to show good correlation to temperature. Are the supposed temp-RW correlations simply caused by data manipulation?

    Can someone give me a quick recap on this, or point me to the right discussion?


  17. Geoff Sherrington
    Posted Dec 15, 2007 at 11:22 PM | Permalink

    Re # 12 John Andrews

    I agree with your comment on airconditioning effects affecting dendro. If there was less in offices, scientists would spend more time in the field. (Written only partly cynically. We were not allowed to have airconditioners is our geological vehicles because the place to spend time was with the rocks, not with the cooler).

  18. John Norris
    Posted Dec 15, 2007 at 11:24 PM | Permalink

    re #15, Steve’s comment

    Steve: No. THese sites are in remote Alaska.

    in my best Maxwell Smart, agent 86, voice:

    … ahhhh would you believe Iditarod routing?

  19. Tim
    Posted Dec 16, 2007 at 12:01 AM | Permalink

    Re: #12 – I was thinking the same thing. Along with the numerous other factors described above, might part of the divergence also be due to the temperature record (such as Ross M. recently explored and is painfully apparent at surface stations.org). I wonder where the “Apr-Sep” temp. data in the chart comes from.

  20. MrPete
    Posted Dec 16, 2007 at 6:35 AM | Permalink

    #16 Dan, here’s my layman’s take on the issue:
    The problem is, we don’t have solid historical data, because sensors hadn’t yet been invented 1000 years ago. So, how does one obtain high-confidence historical correlations? It’s all a big guessing game, fitting curves together that ALL are subject to a great deal of “tuning.”
    This is why extending the record is so important.
    From a common-sense perspective, it’s easy to say “ridiculous — precipitation is the primary driver.”
    Yet when you look at the various data sets and squint a little, you DO see similar curves. So people are tempted to go from correlation to causation, and the tuning begins. And they make a science of it, so to speak. (Not sure that phrase communicates well to non-English speakers, sorry!)
    Is it simply data manipulation? I don’t think so; others here have said as much as well. It’s pretty complex. Often, there’s a certain amount of correlation between drought and warmth. But not always. And not necessarily in the way climate scientists are assuming. (They are wanting to measure overall “average” change. Yet, for example, if days are warmer and nights are colder, the average might remain steady even while midday heat causes further evaporation/sublimation.
    Bottom line: people have made claims about temp-growth linkage, but there’s not much non-calibration data to go on. So we get new data to see if they’ve really come up with something or just fine-tuned to fit the elephant and giraffe into the same box. The discussion is just beginning; unfortunately, many are claiming the game’s over already. Science-by-proclamation. I love it.

  21. Ron Cram
    Posted Dec 16, 2007 at 9:07 AM | Permalink

    Steve, I know you don’t want more about Hughes but I can tell you what Hughes meant. I don’t buy it, but I understand him. I think understanding is important.

    Hughes believes the Divergence Problem is a result of some unknown anthropogenic forcing (this is the part I’m not convinced of). Hughes therefore believes the older reconstructions done prior to 1960 (when the anthropogenic forcings were much less) are actually more reliable than they thought in the early 1990s because those reconstructions were unaffected by this mysterious anthropogenic forcing (this part does follow logically, if the first were true).

    Keep up the great work, Steve. Your work is extrememly important.

  22. jae
    Posted Dec 16, 2007 at 9:09 AM | Permalink


    Often, there’s a certain amount of correlation between drought and warmth. But not always. And not necessarily in the way climate scientists are assuming. (They are wanting to measure overall “average” change. Yet, for example, if days are warmer and nights are colder, the average might remain steady even while midday heat causes further evaporation/sublimation.

    I suspect they will ultimately find that the interaction of moisture and heat is the critical variable. Similar to the drought severity index. That could explain the divergence, especially if it really is getting hotter.

  23. Dan White
    Posted Dec 16, 2007 at 9:20 AM | Permalink

    Thanks MrPete. Let me follow up a bit. You say:

    Bottom line: people have made claims about temp-growth linkage, but there’s not much non-calibration data to go on. So we get new data to see if they’ve really come up with something or just fine-tuned to fit the elephant and giraffe into the same box.

    Let’s say we have a 1000 year record of tree rings, and a 100 year record of local temperature measurements. I assume step 1 is to compare the two (the calibration period?). I know detrending and smoothing is performed on the RW data, but are other adjustments typically made to create a better fit between the two datasets?

    Giraffe and elephant? That reminds me of a joke:

    Question: How do you fit a giraffe into a refrigerator?
    Answer: You open the refrigerator door, put the giraffe in, and close the door.

    Question: How do you fit an elephant into a refrigerator?
    Answer: You open the door, take the giraffe out, put the elephant in, and close the door.

    OK, enough of my sophisticated Sunday morning humor…

  24. welikerocks
    Posted Dec 16, 2007 at 9:29 AM | Permalink

    do they take soil samples in these tree ring studies? Most gardening books mention that trees need good soil..water, and nutrients. Fungi, bacteria algea etc are needed in the soil (which break down the dead leaves/planets/animals, along with insects and burrowing animals to release nutrients and CO2 ) Seems to me that this would matter to tree ring growth too or lack there of.

  25. Murray Duffin
    Posted Dec 16, 2007 at 2:08 PM | Permalink

    For nearby trees there is also an issue of shady or sunny side of the hill. Murray

  26. MrPete
    Posted Dec 16, 2007 at 3:50 PM | Permalink

    There are quite a few adjustments available to help make things fit better. Remember, the measured temperature data typically receives half a dozen or so adjustments (which is why the “measured” temperatures of the past keep changing on us.) There are plenty of other adjustments available for trees.

    I don’t know which is most significant, but I suspect this one is up there: according to certain scientists, a unique advantage of dendroclimatology is the privilege of picking the data that best shows the “desired” signal. Obviously, if you can pick and choose which tree cores to use from the several dozen available, you can generate exactly the kind of fit needed.

    Then, if you statistically reprocess the data using Mannian methodologies, you can generate your desired signal (hockey sticks, etc) even from pure noise sources.

    Some of this stuff reminds me of the time I did a snow job on one of my high school teachers: I had access to some hifalutin mainframe statistical analysis tools. Generated 200 pages of results, wrote up a half page summary, and turned it in. I got an A just for impressing the guy; I don’t think he had a clue what all the analysis meant — I sure didn’t 😉

    RockPhilia 😉 — we took soil samples, and perhaps will see the results sometime. But my sense is such a sample is of minimal value. As has been discussed before, the BCP’s are not growing in a nice plowed orchard. It’s a mixture of inhospitable rocks, sand, bits of humus, etc. Mapping out root->nutrient sources, and mapping root->shoot connections seems a rather difficult problem.

  27. welikerocks
    Posted Dec 16, 2007 at 4:30 PM | Permalink

    #27 Mr.Pete, RockPhilia? I like it. LOL
    Yes, I was talking to my husband about my question. Wonder how BCP’s evolved to survive like that and what do they drink in from the ground and the bits, or even if their roots do anything like regular roots do. Plus the soil gets eroded and churned by the bugs, weather and such so fast. Your results would be nice to know and would be interesting if it contained anything unique. Thanks for replying!

  28. trevor
    Posted Dec 16, 2007 at 5:28 PM | Permalink

    Just an observation. Here in New South Wales this season we have had good rainfall at the right times (for a change!). Driving through the country some two hours SW of Sydney yesterday, the good health and vigour of the trees is palpable. There is abundant and strong fresh new growth everywhere, and the trees in the landscape are just thriving. I could just see a thick tree ring being added to each tree for this season.

    Yet. It has been unseasonably cool through this growing season. Probably to do with El Nina this year. To a lay person though, this seems to indicate that the tree rings are likely to tell us more about precipitation, and the timeliness of precipitation, than it does about temperature.

  29. Roger Dueck
    Posted Dec 16, 2007 at 7:43 PM | Permalink

    Has there been any work done on the rate of erosion of the substrate that the BCPs grow on? I can see a number of factors coming into play. It would seem to be an important part of the growth equation as the exposure of the root systems would compromise the vigour of the growth. Over a period of thousands of years the rate of erosion could be very significant. It seems to me that so many factors are “corrected” for and the “right responders” are then chosen that the resulting correlation is at best an approximation to local weather conditions and at worst totally random with the cherry-picking necessary to ascribe any relationship to temperature.

  30. Tom C
    Posted Dec 16, 2007 at 8:56 PM | Permalink

    Has anyone ever provided justification based on physics and botany for the assumption that the difference between temperature in a given year and average temperature over some time frame should correlate linearly with ring width difference vs. the average? The whole notion seems preposterous to me. If it is not linear then how can any correlation be constructed given some unknown non-linear response? Seems that it is just playing around with curve fitting and spurious correlations.

  31. MrPete
    Posted Dec 16, 2007 at 9:32 PM | Permalink

    Roger, Tom — I’ve yet to see evidence that anyone has looked at any of the more significant elements affecting tree growth. Sure, they avoid reaction wood. Yet — seems that there should be plenty of studies of erosion, precipitation, nutrient topography, micro-geology and more, all compared to dendro growth. I’m no expert — to be sure — yet I’ve not heard of any discussion of the complexities. It all seems to get swept into a general sense that the curves more-or-less match, so let’s run with it.

    Reminds me again of Pat Frank’s favorite bottom line: how can it be science if there’s no physical theory or at least hypothesis to back it up?

  32. Doug
    Posted Dec 16, 2007 at 10:11 PM | Permalink

    I did a study of Norway Spruce in the Swiss Alps for a college term paper in 1972. It quickly became apparent to a freshman student that growing season limited the ring width at high altitudes, but competion from deciduous trees limited them at low altitudes. Other controls surely exist in soil type, water, insects, fungi, aspect etc. How many factors can exist before the temperature signal is unresolvable? Tree rings are wonderful things to date, but hard to extract much else from.

  33. bender
    Posted Dec 17, 2007 at 7:28 AM | Permalink

    It’s nice that the younger cohort are not dodging the question. What I would like to know is whether any of these bright young people is taking an experimental approach, growing trees under a wide range of set conditions, to determine the precise shape of the nonlinearity and the degree of synergy between growth limiting factors. The survey approach – where you take cores and correlate them with natural variations – is clearly defunct if you want the level of precision required to reconstruct gloabal summer temperatures to within a fraction of a degree.

  34. welikerocks
    Posted Dec 17, 2007 at 9:38 AM | Permalink

    I found this page with some information: link

    Bristlecones generally grow above 9,000 feet (essentially above tree line) in alpine outcroppings of dolomite — a magnesia-rich sedimentary rock that resembles limestone. The dolomite reflects more sunlight and contains greater moisture than surrounding rocks, which allows cooler root zones.

    While nearby sandstone often hosts other hearty species such as limber pine and sagebrush, pure stands of bristlecone pines can only be found in dolomite soils.

    And interestingly, the bristlecones that reside in moister soil with greater nutrients grow fat and tall but have a life expectancy of only 1,000 years or so. These trees (such as the Patriarch mentioned earlier) can reach a height of 60 feet with a girth in excess of 30 feet.

    The oldest bristlecones live in the least nutritious soils and reside in the most exposed locations on mountain peaks, and are deceptively small. The first bristlecone that was found to be more than 4,000 years old (called “Pine Alpha”) has a girth of only 4 feet.

    It seems soil with a high dolomite content produces a denser, resinous wood that is resistant to bacteria, fungus and insects. The wood of these bristlecones is also incredibly resistant to rot. Trunks can remain standing for more than 1,000 years after the tree has died. And when they do finally fall, it’s usually due to the erosion of the soil around the roots rather than from decay.

    The dry climate and high altitude in the White Mountains and in other Southwest alpine areas make for a very short growing season for the bristlecone — usually about 45 days in midsummer. During this window the tree forms new needles and cones and stores reserves for the winter. A typical bristlecone will add only a hundredth of an inch to its girth every year. And bristlecone pine needles live somewhere between 20 and 30 years.

  35. ALee UK
    Posted Dec 18, 2007 at 10:14 AM | Permalink

    Have the wider community taken any notice of this session – has it been discussed anywhere else in the blogosphere, or are there links to presentations?

  36. Scott Lurndal
    Posted Dec 18, 2007 at 4:11 PM | Permalink

    As an amateur woodworker and furniture maker, I’ve always wondered at the assumption that the width of a given tree ring is constant over time. We prize old-growth timbers because of the tightness of the growth-rings (and the corresondingly more esthetically pleasing grain figure); Is the tightness (close-spaced nature) of the growth-rings purely a function of the temperature (or precipitation), or do they compress as the tree builds sapwood around the heartwood?

    In other words, is the fact that older tree rings are narrower (and thus are used to show that the temperature was less), simply due to the age of the tree?

  37. jae
    Posted Dec 18, 2007 at 5:40 PM | Permalink


    Is the tightness (close-spaced nature) of the growth-rings purely a function of the temperature (or precipitation), or do they compress as the tree builds sapwood around the heartwood?

    There’s no compression going on. In general, tree rings tend to become thinner as the tree gets older, because the surface area of foliage exposed to sunlight (for photosynthesis) does not increase in proportion to the diameter x height of the tree (especially in dense stands, where there is very little sunlight below the canopy). In other words, the tree can make only so much biomass, and it has to be spread out over a larger and larger circumference and a greater and greater height. This makes the growth rings smaller and smaller. In closely spaced trees, water deficits can also cause this effect, as there is more and more competition for the available water. When dense stands are thinned (by fire or otherwise), you will usually see wider rings (a growth spurt), due mainly to more moisture availability. These effects are not nearly as pronounced in trees grown in the open.

  38. Peter D. Tillman
    Posted Jan 9, 2008 at 1:52 PM | Permalink

    Posted here since this thread has recent activity, and it’s definitely a novel proxy, not discussed here before per Google:

    Grape ripening as a past climate indicator
    Nature, Vol. 432, 18 November 2004.

    http://www.junkscience.com/MSU_Temps/Burgundy.html (proxy jpg)

    Abstract: French records of grape-harvest dates in Burgundy were used to reconstruct spring-summer temperatures from 1370 to 2003 using a process-based phenology model developed for the Pinot Noir grape. Our results reveal that temperatures as high as those reached in the 1990s have occurred several times in Burgundy since 1370. However, the summer of 2003 appears to have been extraordinary, with temperatures that were probably higher than in any other year since 1370.

    Complete paper, with original graph: ftp://ftp.ncdc.noaa.gov/pub/data/paleo/historical/france/burgundy2004.txt

    How very French! And looks reasonable: a 635(!) year historic record, albeit not very geographically diverse. Probably why RC hasn’t commented, right? {evil grin}

    Cheers — Pete Tillman
    (Poche’ on my Mom’s side, LA French, ex-Acadia way-back-when)

  39. Peter D. Tillman
    Posted Jan 9, 2008 at 2:06 PM | Permalink

    Re 38, grape-ripnig proxy

    Sorry, wrong link for pdf of paper:

    Click to access 2004_Nature_432_289_290.pdf

    — which has, as a bonus, David Parker’s “Large-scale warming is not urban” pseudo-answer. FWIW.


  40. bender
    Posted Jan 9, 2008 at 2:51 PM | Permalink

    Kennan on Chuine

  41. Peter D. Tillman
    Posted Jan 9, 2008 at 10:25 PM | Permalink

    Thanks, Bender. For some reason, Google searches of CA often don’t work — must be the page structure of the blog?? I’ve searched for an exact phrase in my own posts and had no joy. Try it yourself, prepare for dismay. i really count on Google to find things.

    Is there a better search engine for searching CA, does anyone know?

    TIA, PT

    Steve: the search function in Editor mode is excellent. Try contacting WordPress and see if they can make it available to readers.

  42. steven mosher
    Posted Feb 22, 2008 at 11:55 AM | Permalink

    A different Divergence problem cited on RC

    The authors use a couple of coral records and a marine sediment core from the Caribbean to reconstruct first wind shear and then major hurricane activity in the tropical Atlantic. First they find a good correlation of their proxy records to wind shear measurements in the mean hurricane development region (MDR). There are two interesting features here: the coral proxies show a positive correlation to wind shear over the MDR, but a negative correlation north of it. Thus, in relation to hurricane activity there is an opposite effect of wind shear in different regions (Fig. 2 of Nyberg et al.). We’ll come back to that later. The sediment proxy shows a positive correlation to wind shear over the MDR and no correlation north of it. Thus, since the two proxy records are correlated to wind shear in the opposite direction, one would expect that the proxies show opposite patterns and trends. This is more or less true for the period 1950-1990 used to calibrate (or ‘train’) the statistical model. However, for the preceding 230 years, the long-term trend is the same for both proxies. This clearly indicates, that at least one of the proxies has a wrong long-term trend, because they show an opposite long-term trend of wind shear. Thus the long-term trend of a reconstruction using both proxies does not seem very reliable. Nyberg et al. did not comment on this basic problem in their reply (Nyberg et al. 2008).

  43. Rearden
    Posted Jul 24, 2008 at 6:04 PM | Permalink

    Steve, I am interested to get your take on a recent dendro paper by Wilmking and Singh in the journal Climate of the Past Discussions:

    Eliminating the “divergence problem” at Alaska’s northern treeline
    M. Wilmking and J. Singh
    Clim. Past Discuss., 4, 741-759, 2008

    Recently, an increasing off-set between tree-ring based temperature reconstructions and measured temperatures at high latitudes has been reported, the so called “divergence problem” (here “divergence effect”). This “divergence effect” seriously questions the validity of tree-ring based climate reconstructions, since it seems to violate the assumption of a stable response of trees to changing climate over time. In this study we eliminated the “divergence effect” in northern Alaska by careful selection of individual trees with consistently significant positive relationships with climate (17% of sample) and successfully attempted a divergence-free climate reconstruction using this sub-set. However, the majority of trees (83%) did not adhere to the uniformitarian principle as usually applied in dendroclimatology. Our results thus support the notion, that factors acting on an individual tree basis are the primary causes for the “divergence effect” (at least in northern Alaska). Neither different detrending methods nor factors acting on larger scales such as global dimming or an increase in UV-B radiation could explain our results. Our results also highlight the necessity to adapt the methods of paleoreconstruction using tree rings to account for non-stable climate growth relationships as these are found in the vast majority of sampled trees and seem to be the norm rather than the exception.

  44. jeez
    Posted Jul 25, 2008 at 4:22 AM | Permalink

    I asked ten people to each flip a coin and each record their individual results

    I end up with 6 people with the result heads and 4 with the result tails.

    “Heads” does not correlate well with our external data, so we toss those results and select the four coin tosses that came up “tails” for use in my study.

    Result: A tossed coin will always come up tails.

  45. bender
    Posted Jul 25, 2008 at 7:12 AM | Permalink

    we eliminated the “divergence effect” in northern Alaska by careful selection of individual trees with consistently significant positive relationships with climate

    1. The divergence effect does not need “eliminating”, it needs epxlaining. Once you’ve explained it, then you can remove it.
    2. How is this selection process “careful”? Seems to me it demands very little care.

    Our results thus support the notion, that factors acting on an individual tree basis are the primary causes for the “divergence effect”

    Ambiguous statement. If these “factors” differ from tree to tree then this conclusion contradicts the Briffa interpretation. If it’s just one factor that is differentially affecting some trees, not others, this would be consistent with the Briffa interpretation.

    This paper doesn’t get us very far. But at least the authors recognize that:

    This “divergence effect” seriously questions the validity of tree-ring based climate reconstructions

    That would be true.

  46. Rearden
    Posted Jul 26, 2008 at 2:07 PM | Permalink

    They say they used a

    “…careful selection of individual trees with consistently significant positive relationships with climate (17% of sample) and successfully attempted a divergence-free climate reconstruction using this sub-set.”

    Their period of instrument record is from 1901-2000. What justification do the authors have to think that the trees that they used (17% of sample) historically (pre-1901) responded consistently to the same climate factor?

  47. bernie
    Posted Jul 26, 2008 at 7:31 PM | Permalink

    Who reviewed this article?

  48. rafa
    Posted Jul 16, 2009 at 4:22 AM | Permalink

    dear Steve, I thought this recently published article might be interesting for you, it surprises me was submitted long time ago.

    A forced response to twentieth century climate conditions of two Spanish forests inferred from widths and stable isotopes of tree rings

    It might explain part of the divergence problem, apparently the researchers found out from the rings and isotopes that trees react very quickly to changes in temps. and precipitation extending the “growth” period ( spring-summer in NH) if needed, and they spend reserves if needed too (so temps. and precipitation conditions for previous periods have to be considered). In introduces new doubts on rings vs. instrumental from the calibration point of view.

    BTW, it uses PCA, doesn’t describe the stopping rules they did follow 🙂


  49. Skiphil
    Posted Mar 28, 2013 at 3:22 PM | Permalink

    More than 5 years on since this thread, and now Jim Bouldin is challenging the scientific basis of tree-mometry on a more fundamental basis than “divergence” problems:


    Yet, so far he is meeting total silence from the field….

One Trackback

  1. […] a 2007 AGU session discussed at CA here, I observed that the “young dendros” were not content with attempts to simply pretend […]

%d bloggers like this: