A CA reader has provided a link to an extremely interesting presentation by dendro Brian Luckman of U of Western Ontario (Rob Wilson’s thesis supervisor) at the 2008 Canadian Society of Petroleum Geologists. Reader Erasmus de Frigid draws attention to the inhomogeneity in the tree ring record created when the tree was scarred by a glacier, evidenced by a terrifically interesting cross-section picture of the results of glacier scarring on ring widths. It sure looks to me like the net result of glacier scarring resulted in strip bark – something that’s obviously an important issue.
 |
 |
The presentation is available in both pdf and webcast ( I had to run the webcast in IE rather than Firefox).
The graphic below (slide 3) is remarkable in several aspects. While Luckman doesn’t use the term “strip bark”, “strip bark” trees are scarred trees. “Classic” strip bark bristlecones in the dry White Mountains end up with only a “strip” of surviving bark and are predominantly “scar”, but some of the Almagre strip bark trees have what must be a similar appearance to this Engelman spruce – half or more of the trunk covered with bark and half scar.
Note, as reader Erasmus de F observed, the tremendous growth pulse in the surviving part of the trunk immediately following the glacier scar.
If you drilled a core in the center of the surviving “strip” bark in the scarred spruce, you would get a huge growth pulse in the late 19th century; if you drilled a core at the edge of the surviving strip bark, you would get correlated but narrow widths. This is exactly the situation that we hypothesized at Almagre strip bark (our Tree 31 discussed here.) Here’s a ring width plot from the prior post. The glacier-scarred tree would yield a graphic like this:
Black – drilled from S in center of strip; red – from SE from edge of strip
At the time, I did the following pseudo-contour to illustrate the effect – surely this is almost uncannily comparable to the Luckman cross-section.
There is a final item that links the Almagre strip bark to the Luckman graphic really nicely. Luckman dated the scar precisely dated to an advance of the Dome Glacier in 1846. Now look at the date where the growth pulse at Almagre started – the late 1840s! (A photo gallery of the Almagre trees are online at http://picasaweb.google.com/Almagre.Bristlecones.2007).
Later in his presentation (slide 13), Luckman provides a couple of equally interesting graphics showing the sorts of mechanical inhomogeneities that can occur in tree ring cores and how these extremely difficult statistical problems are “handled” by dendros. Luckman provides the following equation listing factors affecting growth. A_t denotes an age effect; C_t a climate effect.
The first delta-D effect is defined as “the occurrence of disturbance factors within the forest stand (for example, a blow down or tilting of trees)”. This sort of “disturbance factor” is illustrated in the following graphic. In his oral commentary, Luckman said that the dendrochronologist needs to use trees that are “normally” grown, that the tree illustrated below had been tilted several times and that you would “exclude that sort of record because there’s been some disturbance”/
The second delta-D effect is defined as a site-wide disturbance i.e. “the occurrence of disturbance factors from outside the forest stand (for example, fire or an insect outbreak that defoliates the trees, causing growth reduction)”. Luckman also mentioned logging in his oral commentary. An example is shown below. Luckman said orally that this sort of disturbance is “much more difficult to remove”.
Luckman defined the term E_t as “random (error) processes not accounted for by these other processes.” He then stated that, if you “take care” of these four factors (age, within-site disturbance, sitewide disturbance and random error), you then get a disturbance-free chronology that is related to common climate, using the diagram:
At slide 11, Luckman described standardization as a “black art”. Another description of the statistical procedures involved in cancelling out strip bark and similar effects is provided in the diagram below.
Climate Audit 
209 Comments
Very, very interesting. I had been thinking about fire scars all along, but always thinking they were not relevant. Of course, scarring could occur by any accumuluation of ice and snow – not just glacier-sized.
Well done, by the way, Erasmus.
Interestingly the presentation admits trees are better rain gauges (60% of the variance) than thermometers (30-50% of variance) in the pdf (pp. 33 and 38).
Re: Gary (#3),
Doh
Steve:
Is this a new version of a slightly earlier post or am I misremembering?
Re: bernie (#5),
I accidentally placed a version online while I was still writing it. I just finalized the post.
So the hockey stick is a combined response to water: the response to the liquid flux is fast and the response to the solid accumulation is slow. How parsimonious.
Re: bender (#6), who’d have thought that?!?! /sarc
I have difficulty seeing how they can “remove” these impacts from fossilised trees when they have little knowledge of events in those times.
Nice to see a few people helping Steve out in this ongoing endeavour.
hmm, my comment disappeared…
Steve: try again. As I noted above, I had inadvertently placed a working version online while I was writing. Maybe something happened to your comment in the process.
Folks, I don’t want to discuss precipitation or issues that have been discussed 100 times before. Focus on what’s new here.
Luckman seems to be suggesting culling your sample to remove trees that have been damaged or distorted (which sounds like a great idea)–pretty clear you need to cut them down to do that. Quite a bit of subjectivity still in this process, however.
Hmmm… perhaps the trees cut down for firewood at the Almagre “DeadCar” site could be useful here. Fully exposed rings! Another expedition in the spring….
Following up on Craig’s point, are there always visible signs of such disturbances so that one can avoid taking a core from a “disturbed” tree? Can one tell that a tree has experienced a tree specific disturbance from a single core? Luckman’s figures are cross sections so you can easily see where any damage has occurred.
Re: bernie (#13),
One obvious visible sign is the presence of scarring. But “disturbance” and strip bark are opposite sides of the same coin. And very old trees tend to get strip barked at some point. The phenomenon has been most discussed in connection with bristlecones which can be VERY old obviously, but the effect exists in younger trees.
Thus, Graybill sought out strip bark trees. These are the hallmark of many multiproxy studies.
Luckman’s Jasper-Alberta-Athabaska-Icefields chronology occurs under several aliases in many multiproxy studies as well. It would be nice to know precisely how he (and Rob Wilson) dealt with trees like the scarred tree illustrated here.
But, needless to say, the measurement data isn’t archived. Nor, to my recollection, is the handling of insite or sitewide disturbances discussed in the articles on the Jasper chronology.
Bernie
The problem is that the people who take the cores and report the metadata are not the people who upload the measurements into their models, sprinkle pixie dust and announce that “temperatures in the late 20th Century are the warmest in the last X hundred years”. I doubt that the Hockey Team have shown any interest in the non-climatic aspects of tree growth any more than they have shown due diligence in other areas.
Also cores rather than cross-sections make it nigh on impossible to spot those non-climatic disturbances. But cross-sections are destructive and no-one can be bothered taking multiple cores of the same tree (apart from the Starbucks Team of course)
The “miracle” is that after a very cursory guess at the non-climatic effects on tree growth, they imagine that they can remove those influences through some pseudo-statistical procedure and get a meaningful result accurate to tenths or hundredths of a degree.
Steve: John A, the trouble with going a bridge too far is that people use such overstatements to discredit the valid points. It’s not true that “no one” can be bothered taking multiple cores – at least 2 cores is supposed to be standard practice and Schweingruber, for example, was pretty consistent about taking 2 cores. Unfortunately the most popular multiproxy chronologies tend to be from one-core chronologies: Yamal, Taimyr, Graybill. Odd, isn’t it.
Re: John A (#15),
That suggests a good project for a paper – take the most popular existing chronologies, leave tree rings in but systematically replace any one-core chronologies with the nearest available multi-core chronology. Does the MWP/modern relationship change?
Re: John A (#15),
question. according to some trees are supposed to be selected according to their corellation with instrumented temps.
Assume that’s true. would that argument extend logically to cores. you core a tree from two directions. one direction corellates well, the other doesn’t. why would you average them. you are just adding noise. Again, i’m just speaking about the logical implications of the argument that one can select trees based on a correlation with instrumented temps.
Re: steven mosher (#44),
There must be trees that correlate very well with the Dow Jones Industrial Average – and for the same reason.
Re: John A (#45), im just suggesting a way to illustrate the problem with the argument
Intresting about Yamal conditions of increased reindeer populations and other well observed disturbances that might cause a growt pulse in certain selected trees.
Reindeer herding and petroleum development on Poluostrov Yamal
“The regional scale of habitat destruction in northwest Siberia, including the Yamal Peninsula, was recently summarized by Vilchek and Bykova (1992) and Vilchek (1997). These authors observed that plant cover is already completely destroyed over 450 km²within gas and oil fields and 1800 km²along the main pipelines. They estimate the total area of destroyed vegetation to be about 2500 km². Based on the Tyumen Oblast’s present plans, they assert that the area of explored gas and oil fields will increase to 16200 km²and the portion with completely destroyed vegetation will increase to 5500 km². These figures can be misleading because they do not include the further degradation that is expected to occur due to overgrazing by reindeer, nor cumulative impacts such altered hydrology and blowing sand/dust from roads, quarries and abandoned drilling sites (e.g., Forbes 1995). The three most widespread types of disturbance are off-road vehicle traffic, exploratory drilling, and sand excavation (Vilchek and Bykova 1992; Khitun 1997). Assisted revegetation programs designed to control erosion on affected areas have met with limited success due to their immense expanse and the prevalence of nutrient poor, well-drained and highly erodable sands, in conjunction with the cold, dry climate (Martens 1995; Forbes and Jefferies 1999).”
Clearly not new, but new to me. the Equation (Principle of Aggregate tree Growth) is Additive (Aggregate) not multiplicative, as in the RCS I have been looking at. Now that is a big difference and surely this is testable, or is it?. When building the RC (averaging over equal age) does the SD(age) not give us a steer. Obviously you need a lot of trees with a good distribution of start dates spanning periods long enough to reduce the climatic effect to just noise with an identifiable statistical distribution. Is it the last bit (statistical distribution) that is the problem?
Chopping up a series into bits and aligning by the start of segment may produce a noisy looking mess but how does it affected autocorrelation? I think I will go and play with sinusoids for a bit. That will keep me quiet.
For the Yamal cores, superficially at least, the smooth of the SD(age) tracks the smooth of the RC(age) but Variance(of the SD(age)-smooth(SD(age)) looks like it increases with age. I am not at all sure I understand what this means but it could suggest that after the division by RC(age) stage the result needs some age related scaling to produce a consistent signal. An obvious choice would be to take a power such that the strength of the signal is scaled down with the age of the tree.
Or is this just an artefact due to only having a couple of hundred trees that are not necessarily well distributed for start dates or that the noise is not simple. If only we had a magnitude more core samples.
Anyway I am surprised by the Aggregate Equation, not what I was expecting at all.
Alex
Re: Alexander Harvey (#17), wrote, “Equation (Principle of Aggregate tree Growth) is Additive (Aggregate) not multiplicative…” or not some exponential term.
The equation is no more than a qualitative heuristic for how to think about elements entering into tree ring widths or densities. It’s not even known to be a complete list; hence the need to add an unspecifiable Et term which covers everything else unaccounted in the first four terms. And the presumption that everything else amounts to random is, well, presumptive.
On the other hand, if Luckman really restricted himself to “climate change” rather than temperature proxy record, then he’s on safer ground. “Climate” includes temperature and its ramps and timing, elevation, precipitation, and overcast vs. sunlit days. It doesn’t include other factors such as soil drainage, soil micro-mineral content (soluble zinc, copper, and iron are particularly important for photosynthesis), and other nutrient access such as phosphate and nitrate; not to mention commensal mycorrhizal fungi and rhizobacteria. Tree rings respond to all of that, and if all of that is included in the analysis, one can decide whether a tree was growing in a beneficial climate, or not, with the proper biological qualifiers on any conclusion.
snip – piling on
Steve:
That the most iconic tree series are the ones where the least amount of non-climatic metadata are returned? Not odd, just consistently sloppy.
I’m also curious, after the deletion of the non-climatic terms in the equation, how the expansion of the term
goes.
All the exceptionally well-qualified amateur dendrochronologists who regularly visit this site will no doubt have heard that a bunch of old trees have been given a shot of climatic growth hormone. Turns out, they’ve grown faster in the past 50 years than they have in 3.7 millenia. Hear that? Millenia!
“This is a cautionary tale,” says Michael Mann, who uses tree rings to gain insights into past climates at Pennsylvania State University’s Earth System Science Center, most famously to create the “hockey stick” graph showing an increase in temperature. “Only the human impact of increasing greenhouse gas concentrations can explain that warming.”
Re: windansea (#20),
Tree growth is not this simple, but the article quotes “Tree rings in the last half a century have been 58 mm wide on average, compared to 38 mm wide in the 4,750 years preceding the period, the study found.” Simple multiplication of 38 mm by 4,750 years by 2 (both sides) gives a diameter of 360 metres. Some butt! Now derive the correct units.
Re: Geoff Sherrington (#23),
5.8 mm seems more reasonable than 58 mm.
Re: bender (#25),
0.58 mm reasonable still, especially if we are talking about bristlecone pines.
Re: bender (#26),
0.58mm
Does anyone realise how small that is?
Steve: Yes.
Rising temperatures are causing some of the oldest trees on Earth to grow faster, new research suggests. But scientists are divided over whether or not the change will benefit the climate, as it may simply cause the trees to die more quickly.
Previous research (pdf) suggested that Great Basin bristlecone pines located in the mountains of western US are growing more rapidly. But the reason for the growth spurt – and whether or not it is unusual – was unclear.
A new analysis of tree rings suggests that the growth spurt is indeed unprecedented: bristlecone pines have grown faster in the past 50 years than they have in 3.7 millenia.
“This well-designed study suggests that climate change is the factor causing the growth spurt,” says Greg Wiles, who reconstructs past climates using tree rings at Wooster College in Ohio.
http://www.newscientist.com/article/dn18161-climate-change-gives-ancient-trees-growth-spurt.html
No glaciers in Australia, mate. Not for 10,000 years or so. Tallest mountain is 7,300 feet. Natural “strip bark” is not a term I have heard used here except for aborigines making canoes. Thermometer & rainfall records commonly go back to the 1860s.
So, some of the deltas are absent from the master calibration equation quoted above, or easier to handle.
So why does Australia fail to be treemometry heaven?
Our problem seems to be twofold.
First, dendros seem to have adopted a maxim that stressed trees are best trees for thermometry. I’m not sure what defines a stressed tree on this continent. No tree psychiatrists are listed in the online phone pages.
Second, an elimination process has selected the Huon Pine Lagarostrobos franklinii as the most-used candidate, though its range is mainly remote west Tasmania. The elimination process seems to proceed with feeback which says “This genus looks suitable because our preliminary work shows it compliant with our preconception of temperature in history”. Unfortunately, there are no really reliable temperature stations within cooee of the tree sites.
As a general statement, the art of dendrothermometry described above is subject to so many uncontrollable past variables (and a few present ones like divergence) that one can never have confidence that a residual mathematical signal is due to past temperature. Stationary is not to be assumed because too many exceptions are documented.
It’s a bit like the Monty python sketch where the knight loses successive limbs in a sword fight (“It’s only a flesh wound”). One wonders academically if a thermometer in his mouth would have shown a temperature response as the damage proceeded.
Surely the whole method is now shown to be without credit. Even the author, Brian Luckman, concedes that trees “can” provide “up to” 30-40% explanation of total variance as due to temperature “on a regional scale”. That’s a woeful conclusion based on a showing of what might be the best data to save the sinking ship.
Aerial CO2 fertilization is the reason for the increase in plant growth, crop yield, and presumably tree growth for the recent past, where some trees grow at faster rates than others by increasing airborne carbon dioxide concentrations.
Increasing near-ground air temperatures, no matter how intuitive “it-may-feel”, has not been shown to cause increased plant growth with real statistical dynamism. However, the technical literature field is rife with these kinds of circumstantial evidentiary type studies proving again that careful auditing is required before slipshod science is transformed into poor policy.
Can I request something for the 500th time: please do not keep complaining about every dendro issue on every thread. It’s not that these things aren’t relevant – but we’re talking about strip bark and scarring and/or statistical methods for allowing for disturbances.
Please take anything else to Unthreaded or a topical thread.
Re: Steve McIntyre (#28),
I wonder what the chances are of getting Drs. Wilson or Luckman to address the group. This is pretty interesting stuff. For example, what is the very best study done to date attempting to reconstruct these glacial scarrings?
Steve: Next to zero. Luckman got mad a year or so ago when I tried to get data from him. Rob has always been under a lot of pressure and the Briffa thing will make the pressure on him unbearable. The only morale that the Team will draw from Yamal is that they should have stonewalled even more and that Wilson should have talked to us less.
Re: Steve McIntyre (#28), Sorry Steve — lost my head (again). 🙂 I’ve always loved that cartoon, by the way.
Re: Steve McIntyre (#28),
I hope you don’t consider my #6 OT? If you do, perhaps I should point out that the slow response to scarring by solid ice is directly connected to the question of snowpack depth and the fast response to summer soil moisture availability. They two questions are not, in fact, independent. And note they have little to do with the “p” word, unmentionable in this thread in its summer liquid form. To my knowledge, this has never been mentioned at CA.
.
Next, I would offer the following hypothesis: glacial scarring should be most intense when (1) ice is built up high, and (2) flowing fast downhill. And the condition that leads to (1) are high snowfall, whereas the condition that leads to (2) is heavy melt. From this analysis, one may expect a single disturbance (or narrowly defined window of disturbances) as one moves out of the glacial-interglacial transition. That transition should give you a HS blade.
Re: bender (#33),
theories of stripbark formation are fine on this thread. Pete Holzmann observed incipient strip bark formation in an Almagre tree where a branch was in the process of ripping off and taking bark with it – something that he attributed to heavy snowfall.
Brunstein’s article on Almagre (AAR 1996 by recollection) reported that trapper records said that the 1840s had huge snowfall – he mentioned that these extremely heavy winters might have been the cause of so many bison being killed. (I’ll try to remember to upload the Brunstein article).
Under PEte;s idea, you could get strip bark being formed from heavy snow without needing glaciers.
In these smallish data sets, you don’t need a whole lot of 6 sigma growth pulses to affect the average.
Re: Steve McIntyre (#34),
My other crazy theory, based on observing the trees, is that lightning can have a tremendous impact on tree morphology. (Duh)…
My question: is there a known relationship between climate or (???) and prevalence of lightning in a given season?
Just a general question:
Is it even possible to account for disturbance factors using cores? It seems like you would almost have to have a cross-section to fully study the effects of disturbance factors for a species at a given site.
Steve: Well, you’re not going to cut down ancient bristlecones to check for disturbances. People in the mineral exploration business work with cores as well and try to get as much as possible from them. My contour diagram was a mineral-influenced method. I got a favorable email from a dendro on my contour diagram- saying that it looked interesting and that he’d never seen anyone do that before. You can work back and forth from cores to cross-sections. The Luckman crosssection was really interesting because it sort of proved my interpretation of the Tree 31 contour diagram.
Re: Mike B (#29),
SMc
You’re really serious here ? This dendro had never seen a core-based exploration contour set ? You cannot imagine how depressing that is …
BTW it’s interesting to see strip bark for Engelmann spruce in the Luckman article – even if Luckman doesn’t use that name. When you see multiple sigma differences in individual cores (as in Tree 31 or the Luckman illustration), a proper data analyst would report and confront the problem. Such differences in an individual core are not caused by temperature and need to be accounted for. Dendros cannot assume that a miracle occurs and the differences cancel out.
We don’t know what Luckman did. We do know that Graybill made no attempt to exclude such cores in his chronologies.
Interestingly, there are some Yamal trees that show the same sort of pattern. There are two cores starting in 1803 – they have the sort of divergent pattern that is reminiscent of strip bark. Or maybe the Type 1 leaning disturbance.
Re: Steve McIntyre (#35),
1803? So there *must* be a CO2 source dating to that year? 😉
Re: bender (#36),
P.S. You’ll want to keep an eye on Cruel Mistress on this CO2 fertilization business.
Re: Steve McIntyre (#35),
Au contraire, we know Graybill specifically focused on such cores in his research. Steve, perhaps you forget your prior discussion of this? 🙂 You quoted Graybill:
100% of the Graybill tree markers we found were at least partially stripbark trees. The closest thing we found to a Graybill Whole bark tree: our #047, Graybill 84-52, which happens to be an archived ITRDB sample, #ALM131. (Click on the link for the image gallery of this tree.) As you can see, it is not exactly a stress-free tree:
Which side did Graybill core? We have no clue. Unfortunately, only one core of this tree was archived, although up to four cores of other trees are in the ITRDB. (The code: ALM131 = Almagre, Tree #13, core #1)
(BTW, it’s clear Graybill tagged 30+ trees on Almagre, and cored many if not most of those. Now that we know where several of them are, we may be able to find most of the rest, if they still exist. I don’t know how or if Lamarche tagged trees in 1968, so finding those trees may be impossible.)
If that’s supposed to be a joke, it isn’t funny.
The presentation is fascinating for longtimers here.
The Briffa Divergence Problem of the late 20th Century:
All the Hockey Sticks in a row:
The Amazing lack of correlation between high latitude tree series and tropical sea sediment proxies:
Steve: Perhaps, but let’s keep comments related to strip bark.
I’ve sometimes wondered if the complete configuration of a tree’s internal growth ring pattern and structure could be non-destructively examined and documented using a technology similar to MRI.
snip
Steve: let’s stick to strip bark issues please,
Steve Mc: …”dated to an advance of the Dome Glacier in 1846. Now look at the date where the growth pulse at Almagre started – the late 1840s!”.
and
“Brunstein’s article on Almagre (AAR 1996 by recollection) reported that trapper records said that the 1840s had huge snowfall”
What puzzles me was the statement in the beginning: “the tremendous growth pulse in the surviving part of the trunk immediately following the glacier scar.”
I don’t think a tree could survive a “glacier scar”. May be you mean that strip bark is basically caused by extreme snow conditions and glacier advances occur during such times?
Re: Larry Hulden (#43),
Larry, Luckman was the one who said that the tree in the above illustration was “scarred and tilted” by the glacier. Luckman’s caption:
I do not have personal information on the tree versus the glacier. However Luckman (AAR 1993) contains the following photo and caption stating that a subfossil tree had been identified in situ and had been tilted by a glacier. These particular glaciers are confined to pretty steep valleys.
Can someone please explain to me exactly what is a glacial scar?
Is it a wandering glacier sideswiping the tree on the way down the hill?
Is it a proud and sturdy tree that stays alive and growing as a a flacier ploughs past?
In my little world of civil engineering, trees and glaciers do not coexist (or not for very long).
In horticultural terminology, we usually call the rapid growth around cambium damage ‘callus tissue’
A miracle just happened.
Dear All,
Please do not take this the wrong way, but the depressing amount of ignorance on this current thread makes it hardly worth the effort to respond. Two quick points:
1. The linear aggregate model is a purely conceptual model – something ideal for teaching to undergraduates to highlight all the environmental factors that can affect tree growth.
2. w.r.t. scarred trees (fire, glacier, avalanches etc) – one would never use tree-ring data from a scarred tree for a dendroclimatic reconstruction – or at the very least one would use a measured radius where the rings were not affected by the accelerated growth around the scar.
There is a wealth of literature on these issues.
Rob
Re: Rob Wilson (#51),
Rob, at Almagre, we saw Graybill tags on “scarred” trees; data from these trees was included in Graybill’s Almagre chronology. Here is the tag on Tree 31 discussed in this post.
For example, pictures of Tree 31 discussed in this post are online here. Here is a picture from the north (other perspectives are in the album.)
Re: Rob Wilson (#51),
Rob, if it’s true that one would never use data from a scarred tree, how do you reconcile the stated preference for stripbark trees?
Re: Rob Wilson (#51), What you meant to say is that one SHOULD NEVER, not that one “would never”. As you note there is a wealth of literature on this and one should never use a damaged tree. What this means of course is this. The very first step in performing a proper QA on ANY collection of cores would determine through independent means whether or not the core came from a damaged tree or not. I am going to resist the urge to rise to your comment about the ignorance displayed on this thread. What I will comment on is the lack of understanding about the QUALITY ASSURANCE process. As you note every dendro knows to avoid damaged trees. every dendro knows this. they are not ignorant. Every dendro knows to take at least two cores. So, the very first thing that I must do to check the work of a dendro is to see if they followed their proceedures. Clearly you would agree with this. I must check, did they sample damaged trees. Did they take two cores? If the study in question does not provide evidence for these two very basic QA tests, then the data is junk. It’s as simple as that.
Re: steven mosher (#73),
One of the ways of assuring users that QC standards have been complied with is to archive photos of the tree in question. It’s very easy to do and Pete Holzmann did this within minutes of our Almagre sampling.
Another check that seems useful to me is to graphically compare ring widths from two cores in the same tree. If the trees are near-circular, then the two cores should match closely. If they deviate strongly as in the Tree 31 example, then one should check for scarring that might not have been attended to in the original sample.
I did a spot check on this latter test with a couple of Andrea Lloyd cores and she confirmed from her field notes that the cores were strip bark. I presented this example in my AGU 2007 poster.
Re: Steve McIntyre (#74), Thanks. I had thought of the photo approach, especially for new sampling. I think this underscores the importance of bringing proxies up to date. The complaint about the lack of metadata is a legitimate concern. It should drive a certain class of scientists to focus on updating the proxies and the metadata. WRT checking cores against each other, it looks like comparing two cores would be a good first check to rule out a damaged tree. If you only have one core, then there is an interesting question on the table. How do you rule out
mechanical damage? especially if you have a six sigma responder. And if you can’t rule out mechanical damage, then what do you make of the results?
Re: Steve McIntyre (#74), Also, the procedure you suggest is quite testable.
Compare 2 cores from a tree and look for the deviation from circularity. how reliably can you pick out damaged trees?
let me put it a different way you wrote ” if the cores match closely” quantitatively what are we talking about here?
Since you have cores and photos and good log notes can you take the cores you have from your field study,
compare the cores from a tree to see if they “match closely” and actually pick out the damaged ones with any kind of skill? That’s an open question.
Re: steven mosher (#79),
Calling Dr. Wilson …
(Say, where’s Rattus and his linkies?)
Re: bender (#80), I think you see the clash of certain cultures here. Let me do my best impersonation of a particular point of view here.
hehe.
Anyways bender I think it might be a interesting exercise to compare the differences between two rings take from the
same tree. On the assumption that two different sides of tree see the same “climate” and on the assumption that all the trees in the tree ring database are damage free, I would think this would give you some kind of insight.
Re: Rob Wilson (#51),
As an ‘outsider,’ one striking aspect of the general topics covered by ClimateAudit and RealClimate is the polarization of many of the most most informed participants. So, kudos to Dr Wilson for contributing at this forum.
The beginning of Wilson’s comment…
…came to mind just now when reading Razib Khan’s The other denialisms post at Scienceblogs. Razib wrote, “serious problems emerge when our intuitive prejudices push themselves into the scientific domain.”
When contemplating depressing amounts of ignorance, Dr Wilson might consider Razib’s footnote —
My limited experience (Tiljander proxies) is that ClimateAudit regulars write in such a way as to make their reasoning plain: ‘unpacking their intutions’ in plain view. RealClimate regulars often do not. If one strays from the fold (e.g. by modifying “varve series” with “upside-down”), the conversation retreats to the realm of etiquette–snark, justifications for edited and snipped blog comments, and the like. It’d be good to see more engagement on the technical issues that underlie reconstructions.
Steve,
see:
Sheppard, P.R., Means, J.E., Lassoie, J.P. 1988. Cross-dating cores as a nondestructive method for dating living, scarred trees. Forest Science 34:781-789.
depending on the number of scars, it is possible to measure a “clean” radius from a scarred tree.
The Luckman example and extreme growth forms are apples and oranges I am afraid.
Rob
Re: Rob Wilson (#54),
Apologies in advance for my ignorance (I don’t have academic paper access to read more than the abstract)… wasn’t that research intended to demonstrate measurement of dates (of scarring events) rather than climate signal represented by ringwidth patterns?
Isn’t it apples and oranges to compare the ability to extract date info vs climate info?
Re: Rob Wilson (#54),
Rob, thanks for commenting. From my perspective, the Luckman example and the Almagre Tree 31 example look apples and apples. Almagre Tree 31 is nowhere near being a White Mountains “extreme growth form”. It is half scarred and half bark. I’ve got some experience in my life in trying to interpret structures from core. In my opinion, the Almagre Tree 31 cores would in cross-section look like the Luckman crosscut. I don’t think that Almagre Tree 31 and Luckman are “apples and oranges”.
In any event, even if this understanding is incorrect, it is surely not incorrect in a massively trivial way and I would appreciate any light that you can shed on this.
Re: Rob Wilson (#54),
According to Salzer et al there is no such thing as “extreme growth forms”; all forms are equally useful for dendroclimatic reconstruction. But you are suggesting that form matters. There appears to be some disagreement here. Who’s wrong? Or are both of you wrong in different ways?
Re: Rob Wilson (#54), Rob–the reason this matters is that bcp are a key ingredient in hockey stick reconstructions, and SM has shown that the particular bcp used are stripbark AND the Yamal few good men (a second key series used in multiple recons with hockey stick shapes) exhibit a pattern of growth very similar to a damage response. Metadata do not exist to show that damaged trees are not used at Yamal.
To bring some context to this discussion, here are links to panoramic images that show some of these trees in context.
Graybill BCP tag #84-36
Graybill BCP tag #84-65, archived as ALM241AA
Rob, with all respect, please explain how the latter example reconciles with your statement that scarred trees would never be used in a reconstruction. Obviously, I’m missing something important here.
Re: MrPete (#58),
The Graybill data for tag 84-65 (ALM241) has extremely low growth in the 1840s – the 1840s is the “loudest” signal in the Almagre record – and has a very sharp increase in growth in the mid 19th century.
We don’t have drill core for this tree as it was located after our permit ended. However, as noted elsewhere, Tree 31 had a post-1840s growth pulse, the cross section of which has relevant similarities to the Luckman cross section.
In very cold climates on hillslopes, 2 things can cause trees to lean: snow pack above the tree, and sliding soil. In either case, the tree will form reaction wood and the rings will be distorted. The effect is opposite in conifers and broadleaf trees.
http://resources.metapress.com/pdf-preview.axd?code=m78r76646322h373&size=largest
The scarring (callus) appears to impart physio-chemical responses. The demonstration in the link indicates that scanning such as done on tree rings could include whether the ring in question was from callusing response. It states “Generally, there is a “cause and effect” relationship. I think a scientific procedure could and should be implemented. Another indicator from the Luckman is the apparent “6 sigma” increase in ring width. Between these two, I think even a core could be taken from the Luckman example and repeatably demonstrated to be callus from scarring.
Re: John F. Pittman (#61),
Have tree researchers studied the physiologic responses to “active” wound repair? This is not in my league, but some indirect comments from the animal world: There are two different physiologic states relating to wound/injury repair. One is the actual repair process where there are unusual local responses such as cell migration, exaggerated replication/cell division, and varying forms of subsequent differentiation. An expanded/new set of physiologic processes and controlling signals are in place. Then following formation of the scar, there are the after-scar effects related to altered structure. Plus the exaggerated after-scar growth observed by dendros may relate to feedback disturbances; i.e., delays in turning off the repair responses and associated signals, and the returning to a “normal” state (it will never be truly normal). Size of the injury should be a mitigating factor. The reasoning behind not excluding scarred trees seems illogical to me; such trees are not “normal”.
Re: Dave L (#71), http://www.epidna.com/showabstract.php?pmid=6223902&redirect=yes&terms=callus+plant+hormones+pinus
A botantist in physiology could confirm if this article is appropriate. There does seem to be independent confirmation of the changes in pinus to callus growth. All I remember from botany is that there is accelerated growth to cover scars or foreign objects. So, not only would glacier, or ice be a candidate, but scarring from rocks, landslides, etc, eliciting a growth response.
Incredible.
Steve actually tracked down the Greybill trees, and can personally verify that they are strip bark trees which give radically different results depending on the orientation of the core.
Steve’s observations are confirmed by the Ababneh chronology which gives a radically different result from Greybill despite using the same set of trees.
Greybill’s own observations appear to support Steve.
But Rob Wilson, with no personal knowledge of the site, declares that “one would never use tree-ring data from a scarred tree for a dendroclimatic reconstruction”, and then accuses others of ignorance.
snip –
Steve: Pete Holzmann tracked down the Graybill trees, not me. But I did instigate the project and get the permit. The Ababneh chronology is for Sheep Mt – that’s a different story, but one that’s in the news with the new Salzer-Hughes study.
Re: Jason (#62),
Dendros, even one as alert as Rob Wilson, don’t always read what I write.
My comments were entirely about the Graybill bristlecone chronology at Almagre about which I have knowledge. I used the Luckman cross-section to illustrate a situation at Almagre that I believe to be analgous – and I remain of that opinion despite Rob’s oracular statements otherwise.
In this particular case, Rob seems to have elbows up about the Jasper chronology and to be implicitly refuting the idea that he and Luckman might have used scarred trees at Jasper. However, I made no comments or statements about the construction of the Jasper chronology in this post.
It’s reassuring to hear that they didn’t use scarred trees. This protocol isn’t mentioned in Luckman and Wilson 2005. Nor did they archive any measurement data (this is not Rob’s fault.)
Well, there is much to say in response to Dr. Wilson’s drive-by. (With so little content it can hardly be called a lecture.) To start, SHOW ME where someone has measured multiple radii around one of these disturbed disks and determined that there is, in fact, a section that is NOT affected by the disturbance. I will guess that the effect is omnipresent, just less exaggerated the further the radius is located from the scar.
“…climate change is the factor causing the growth spurt,” says Greg Wiles, who reconstructs past climates using tree rings at Wooster College in Ohio.
I’d put more faith in him if he worked at Jeeves College.
I sent Rob a follow up email
He replied:
I don’t know why any reader would interpret my post as criticizing Luckman’s procedures – as it was about bristlecones. However, readers obviously did interpret Rob’s comments as supposedly refuting my comments about bristlecones – which he now clarifies that he cannot comment on.
Re: Steve McIntyre (#69),
Your description of the discussion is accurate, but it leaves out important information. Your head post points out Graybill did not work by the standards set by Luckman. Readers took that information and mentally extended the possibility other dendros do not work by Luckman standards either. Rob Wilson enters the conversation to say:
Wilson’s statement is obviously untrue, since Graybill is used in many major reconstructions. I think Wilson is projecting his own high standards onto the entire field of dendrochronology, but the facts do not support him. Craig Loehle points out in Comment #66 that no metadata exists to show Yamal does not have similar problems.
Slide 8 of the PDF touches on the desirabiltiy of Replication within trees, within the site, and between sites in Dendro studies.
Among the points, the presentation prefers 2 cores per tree (perhaps to Steve’s point above); additionally it prefers chronologies of more than 15 trees at a given site; and, replication between sites all in an attempt to emphasize the common signal while reducing “Noise” (from local effects I assume).
Does Briffa’s Yamal study appear to comply w/ any of these points?
Apologies if someone has already suggested this, but has anyone “reversed” the dendro process, (i.e. taken the data and tried to reproduce an image of the core)? It might be revealing to actually see what a core would have to look like to produce certain data.
Here, I thought that “strip bark” meant that some of the bark had been “stripped” off, not that only a “strip” remained.
Same result, but in one case the glass is half empty, while in the other it is still half full!
Re: Hu McCulloch (#76),
The whole range exists of course. The Almagre metadata spreadsheet includes a column “Strip Degrees @ ht” that can be sorted…
Of course, the highest recorded strip-amount value was 320 degrees. At 360 you’re talking a Dead Tree with no current rings. (Our goal, beyond the Starbucks Hypothesis, was recent data updates, so dead trees were of little interest.)
It is curious that all the professors out there whose work is being criticized, whose job it is to explain things to students (who are “ignorant”), seem unable to come here and explain their work. How hard could that be? And they could reach thousands, not just 25 at a time.
Re: Craig Loehle (#81), you should know that a professors job is not to teach. That’s the TAs job.
Here is a comparison of two Yamal YAD cores both starting in 1803. One of the cores (YAD061) drinks Dos Esquis.
Both of these cores are used in the new Briffa chronology and by definition (Rob’s) within-tree disturbances have “necessarily” been allowed. Both are the same age, so the A_t is the same. Thus everything should cancel out except for truly random error. And yet somehow all doesn’t look right.
Re: Steve McIntyre (#83), I’m thinking that everything should NOT cancel out.
I’m thinking that autocorrelation is going to play a role. I’m betting that if you compare a bunch of these that you will
see that the mean of RW1-RW2 is not going to be zero.
Re: Steve McIntyre (#83),
Stay thirsty, my friend!
Re: Steve McIntyre (#83),
They both drink Dos Equis. One in 1860, the other in 1960.
The tree above at the top of the thread has new growth over a damaged area and new bark. From the outside
looking at this tree you would only see a protusion, not a barkless area. I think it would be easy to avoid
coring the obvious barkless area but can you guarantee that a core taken from an area of the tree that has bark
is not hiding an old scar under it? Second question, does the increased growth in the scar area reduce the ring
widths at the non-scarred area in the same year? In some conifers like the example above that have nice circular
trunks, you may be able to see the bulge and avoid that area. In the bristlecone pines, which have a trunk cross
section that is chaotic, not so sure if you can see that. BCPs in the White Mtns are scarred by avalanche, lightning
and rock slides among other things. I am 2 hrs due south of Sheep Mtn and plan to take a drive up there next spring
when they open the place back up to have a look around….camera in hand.
Re: Erasmus de Frigid (#84),
EdeF, it’s not simply avoiding the scarred area. The dilation occurs in the surviving area.
Re: Erasmus de Frigid (#84),
Other than the unfortunate result of a cut-down tree, is there any reason why a dendro couldn’t take cross-sections from trees that appear externally undamaged and then apply some fancy high-tech method using “lines” from the centre-to-edge and than calculating some running mean around the entire circumference (a bit like the minute-hand of a clock). This might give a better growth representation and reduce any “erroneous” data from single or even double cores and prevent personal bias from creeping in, even with the best of intentions. cheers from a long-time reader, first-time replier.
/bite_tongue
The graphic below shows the ratio of YAD061 to YAD041, both of which start in the same year. The YAD061/YAD041 ratio is almost perfectly linear (in biological terms) since 1940 or so. (The R2 for the trend since 1940 is a huge 0.72.) How the hell can you get such a perfect trend?
We’re not seeing the two cores go up and down in unison, we’re seeing one core dilate – a word that we used in projective geometry classes at university – relative to the other, with the amount of dilation increasing linearly.
It is IMPOSSIBLE for (1) both the cores to have been measured perpendicularly; (2) both the cores to have been Luckman-screened to remove within-core disturbances.
Re: Steve McIntyre (#85), Think of it this way. A tree doesn’t know the right response to damage. Trees that underrespond to damage die. So, there may be a built in ( via evolution) tendency to over respond to damage. If the dialation is too small or not prolonged enough the tree will die. pretty and round and dead. If the tree over responds to damage and dialates for a longer period then its chances of survival go up. Ugly but old.
dr. bender is that a sound intuition?
OT: Just in case anyone is in any way interested in my plan to tabulate all of the different studies, their data, methodologies, etc., it’s not happening anytime soon. I was laid off last week, and have pretty much committed myself to starting a new business. Horrible timing, given our economic situation. But I have some pretty cool ideas, and need to be in action about them now.
So, I’m buried in building prototypes, writing business plans, buying and building new domains, etc. I’m sure this is being met with a collective shrug of the shoulders, but my apologies to anyone who might have been interested.
It is so sad that such a huge amount of energy has to be expended by so many talented people analysing research such as this to show the world it is wrong.
snip – too much venting
The reaction to tilting in trees is formation of “compression wood” on the underside of the stem in softwood species (“tension wood” on the upper side for hardwoods). This “reaction wood” is shown in the picture below the quoted statement by the darker and wider rings, and could easily be the cause of a “hockey stick.” However, it would usually be very easy to see this on cores. A whole grove of trees can be tilted by high winds and snow. I wonder whether cores showing these abnormalities are routinely “thrown out.” I also wonder if leaning trees are ignored.
(Apologies if someone else pointed this out; I didn’t read all the comments).
I don’t think one needs to worry quite as much as some people do in this thread about some disturbances to tree growth.
Idiosyncratic disturbances to individual trees should not cause any particular problems for any reasonable statistical approach. These disturbances will make the estimates slightly less precise than if one could control for them explicitly, but the results will still be unbiased. For example, leaning trees is not a per se problem if they lean more or less randomly. It will all cancel out. Similarly, tree scarring from snow will not be a problem if it is idiosyncratic – although a glacier going through a whole stand is another matter. So pictures of any individual tree showing an annomaly is not necessarily a problem. I don’t think you do need to find the one perfect tree provided you sample enough trees randomly.
The problem comes where there are systematic disturbances. For example, specifically sampling strip-bark or scarred trees is a systematic choice that could well affect the results. Systematically sampling long-lived trees may be another. These systematic disturbances are also equivalent to the ‘site-wide’ distrubances mentioned in the post. But even this is not necessarily a problem – until one wants to ascribe meaning to the results. Absent additional information it will not be possible to separate an estimate of site ‘climate’ from site-wide ‘disturbances’ – but if you interpret ‘climate’ broadly your results can still be accurate. For example, saying the ‘climate’ for tree growth in a particular area was poor during a period when there was an insect infestation seems a reasonable statement. (“The temperature was low during a period of insect infestation” does not.)
How might one control for site-wide distrurbances? Get lots of sites and allow these errors to cancel out, just like the errors for tree-specific distrubances can cancel out in a site. One simple way of doing this, and a way that has been used by some here, is to average chronologies from different sites. The site-specific effects cancel out and you should be left with some indication of global ‘climate’. This ‘climate’ indicator may still not be temperature, but it will be better than methods that, instead, heavily weight particular sites and don’t allow the site-specific effects to be mitigated. Steve has had a fair bit to say about this in the past.
You should even be able to test for systematic disturbances in one site’s data if you have enough additional data. For example, let us suppose that trees at a particular site grow assymetrically. If you have data on the direction the core is taken you could control for that. If trees on one side of a stand grow differently from those on the other, data on the location within the stand could be used to control for that. But you need the data, otherwise you are just guessing.
Re: JS (#101),
I strongly disagree. It’s all about (1) sample size and (2) whether the disturbances are random enough in time to get ironed out by the central limit theorem. I think that rarely, in treeline studies, are these two conditions satisfied. When they occur at the end of s series you’re obviously not satisfying (2). That’s the concern here. Disturbances initiated in, say, 1803 and 1846 masquerading as post-industrial 20th century climate change.
Re: bender (#102), I don’t think we do disagree. It’s about (1) sample size and (2) whether disturbances are random enough to get ironed out. My position is merely that one doesn’t need a set of perfect trees to be able to get results. One also doesn’t necessarily need the central limit theorem to kick in if one doesn’t need to make normality assumptions. Normailty might be nice, but isn’t essential. The key is truly random sampling, after that there are many robust statistical techniques that can cover over a multitude of sins – provided you properly use them.
… and even truly random sampling isn’t essential if you have enough metadata to identify and control for sampling problems. (Labour economists deal with these sorts of problems all the time – they have been very creative dealing with sampling problems.)
Have to agree with Rob. The level of ecophysiological ignorance here on this thread (and pretty much this entire blog) cannot be overstated. Scarring is not at all the same as formation of stripbark: A scar is caused by mortality of cambial meristem with no loss of either photosynthetic area or root mass, therefore one would expect positive growth response to occur (evolutionary response to regain sapwood area). Strip bark trees have lost either root mass or photosynthetic capacity that lead to cambial dieback. Read some books!
Re: denialdepotter (#106), it doesn’t really matter for the point at hand that scarring or strip-barking have different mechanisms. What matters is that at some point in the tree’s life a change in its growth has occurred, unrelated to climate, and that subsequent growth, and any relationship it has with climate, will be very different from previous growth. To incorporate either into a chronology will lead to biases unless they are controlled for. Strip-bark, scarring, insects, fire – all disturbances – so what if they have different mechanisms?
Re: denialdepotter (#106),
Re: JS (#107),
Denialdepotter,
I (and presumably most of this thread’s readers) understand that you distinguish scarring of the cambium from the formation of stripbark. Also that scarring will likely be followed by thickening of rings, wheras stripbark processes will likely cause ensuing rings to be thinner. This is helpful background, thanks.
I also understand the general point raised by JS earlier in the thread, and repeated in #107: that local disturbances of all sorts add signals to the dendro record, on top of whatever the climate-derived signals may be. Can local non-climate-caused signals restricted to one or a few trees be distinguished? What about stand-wide events, such as JS’ example of an insect infestation?
During the climate-reconstruction process, what should Best Practices be on this point, with respect to tree selection, coring, metadata collection, data archiving, and statistical treatment? Is there widespread agreement on this topic? As you see it, how rigorously and how transparently are these Best Practices adhered to?
I’d be very appreciative of any thoughts you’d care to offer.
Re: denialdepotter (#106),
You are correct that stripbark processes lead to cambial dieback. But what is the result in the remaining area? Enhanced growth.
I’m sure you’ve read all the books you need to read. But perhaps you can grant that folks here are looking a bit deeper than you imagine?
Re: denialdepotter (#106),
Tell me soemthing I don’t know and haven’t said already, o learned one.
credit to Rob for continuing to contribute to CA discussion.
For CA regulars, it is important to understand the extend to which CA has bee stigmatized by its scrutiny of the Team. The presumption in academic circles is now that any reference to academic work in a CA post is, automatically, critical and damaging: has often been the case with the consistently suspect Team work. So there is a presumption that any post by Steve will be critical, whereas, this particular post was, in my reading, far from critical of Luckman and certainly not of Wilson. But do try to understand the sensitivity: there are good dendros out there, but the Team has expended a lot of energy attempting to demonize CA. In addition, certain fields in academia have embraced blogs, physical geography is not one of them. In Luckman’s own department, blogging is frowned upon, indeed the only active blogger in geography at UWO is the one writing this post! Most academics struggle with communicating their ideas in mass forums and to wide audiences. So the stock answer is “read some books” rather than, “let me explain this to you”. At the same time, CA has a very disparate audience, many who post regularly and are very well informed, many are newbies who ask newbie questions. Again, the present focus in academia is all things graduate, undergraduate teaching and speaking to non-academic audiences is not widely sought after, praised nor rated in “merit” determination. What we don’t do well we dismiss as unnecessary.
Steve has been suggested as a guest speaker at the University of Western Ontario to the geography department on numerous occasions and to other bodies on campus: there is no institutional will nor desire to invite him. My conclusion is that the halls of academia are far more closed minded than most realize. It is not they don’t know, it is that they don’t want to know!
I have to agree with Rob Wilson – most of this discussion reveals a fundamental lack of understanding and experience with the subject. There has been a great deal of work done with the bristlecones over more than 60 years including felling and stem analysis. Much of the remnant wood used for dendro work is from sections, not cores. People interested in climate reconstruction always try to avoid anomalous growth – surge or suppression – when measuring rings for chronology development as it is considered “noise”. The strip-bark form is NOT scarring i.e. wounding, but represents adjustment for energy balance. Idiosyncratic growth anomalies are corrected for by sample and site replication. Dendrochronologists have been aware of these issues for a hundred years, have studied them and done their best to recognize growth anomalies and mitigate their effects when interested in the influence of common environmental factors on growth (as is the case in climate reconstruction) or to search for them if they are interested in stand or tree disturbance – fire, flood, geomorphology, etc.
Text below is copied from a post of mine at that evil site “RC”:
If you are interested in the thread where this was posted here is the link: http://www.realclimate.org/index.php/archives/2009/11/a-treeline-story/comment-page-3/#comment-143383
Dendrochronologists that I know love their subject, study it intently, do the best possible work they can for the love of and excitement of discovery – not to promote some nefarious “agenda”. It is hard, careful meticulous work that requires attention to detail. Most enjoy fieldwork and have a bit of awe for the objects of their study. How not to be awed by an organism that can live for hundreds or thousands of years, rooted in spot, unable to avoid the elements – wind, snow, ice, drought, cold, heat – take what comes and survive – truly amazing.
Please show a little respect for the trees and those who study them.
Re: CB (#113), As you note work on bristlecones uses sections and not cores ( mostly). In order to document that dendros have analyzed these cross sections correctly can you please point me to the records ( say photographs) of these cross sections? I am open minded about granting the people who do this work respect. Please show me the data to test your hypothesis that they are worthy of respect.
Re: CB (#113),
If the anomalies are synchronous then no amount of replication is going to help. Why do you think the Graybill chronologies uptick?
.
And what do you make of the non-use of Ababneh’s chronology in climate reconstruction? It’s got everything that previous chronologies have but no wicked uptick at the end.
.
And since you’re so knowledgeable, tell us what causes these upticks in some trees, not others.
Re: CB (#113),
It’s a welcome development to find a practicing, professional dendrochronologist posting on this thread. Per DarveJR’s comment #117, I hope you consider taking McIntyre up on his offer to run a thread devoted to the issues you raise.
It’s a bit discouraging to see you describe RealClimate.org as “evil.” Obviously that’s not your opinion, but a sarcastic appreciation of what you think readers here must believe.
This is certainly not my own view. Rather, RC is distinguished by being not useful. Because of its heavy traffic, overtly partisan posters, and moderation policies, it is difficult-to-impossible to have an informational exchange, and to have particular questions addressed by knowledgable people in a position to discuss them, if they so chose.
Re: CB (#113), welcome to CA.
I think you will find an overwhelming level of respect (and relief) from most commenters here when an open-minded, non-agenda-driven scientist shows up. You may also experience a little initial suspicion from some because the experience of the proprietor of the blog and the commenters with climate scientists has often been of the type so recently exposed in the CRU emails.
The truly sad thing to me is that the large majority of scientists, who follow the evidence rather than lead it, find their own work and motives suspect because of the misbehavior of a few. Unfortunately the few usually seem to have a rather high profile.
And when those few circle the wagons and are successful at suppressing dissent the idea that science itself is self correcting becomes suspect.
IMO that is the value of a place like this. I hope you’ll hang around and contribute some of your expertise.
The people here are pretty much like scientists in general; most are open minded and looking for the truth whichever way it goes, but there are some with an agenda. Please don’t apply the same thinking to CA that you justifiably object to being directed toward the majority of scientists who act professionally.
Steven – the conversation here might take a while and I am busy, but willing to try. Proper and complete answers take a bit of time and it is late now. I will say quickly though that the growth changes that you and bender seem concerned about are not subtle and are easily recognized as disturbance related (visually or in time series plots) and not due to climate. Series exhibiting these characteristics will be eliminated or truncated if climate reconstruction is the objective. Standard procedure, tree rings 101, not controversial.
Bender, I am not familiar with all the discussions going on here – you have to provide me with some context. What Graybill chronologies are you referring to? Almagre bcp? I don’t happen to care for that material for the simple reason that it it is very “noisy” – full of surges, etc. I believe the site was considered interesting because of its geographic location and the age of the trees, remnant material etc, but I don’t know that it was ever used for direct climate reconstruction. Val LaMarche made use of the frost rings dates from the site though, and the general characteristics of the data are typical of temperature limited high elevation tree ring sites.
Who is Ababneh? What reconstruction?
Re: CB (#116), CB, many of us here appreciate that many dendro-climatologists are knowledgeable and dedicated to good science in an area of fairly specialist knowledge. But the issue here is fundamentals, fundamental challenges of the bases upon which the current work has been built up. This is in addition to the evidence of deep corruption now surfacing – but not unconnected either. If there had been more openness, you would already know about Ababneh’s work.
Folk like myself smelled a rat that something was fishy in the whole of Climate Science. We have been motivated by the desire for justice and integrity in Science, and thoroughness and transparency in the climate science issues that affect us all so deeply. In the circumstances, all we could do was to try to sort out the truth for ourselves, even when unqualified in specialist areas. For instance, I discovered that the Yamal treering records seem to bear no relationship to any of the surrounding temperature record stations’ records. That to me was a fundamental piece of evidence challenging at least those Yamal treerings as proxies for temperature. It also challenged the conclusions of “unprecedented global warming” in which the Yamal Treerings were given a degree of significance they had not earned.
Re: CB (#116),
Ababneh? Read the blog! Reconstruction? Mann et al. 2008. Do a little reading before making pretenses at authority. Thanks for playing.
CB: It seems to me that neither party is in possession of all the facts. Quite clearly there is great opportunity here for clarification and discussion. Steve offers anyone the chance to author and maintain editorial control of a scientific piece on his blog and I’m sure he’d be happy for you to present whatever arguments you feel need presenting.
.
There is a jaded view of dendro work here because of “sloppy” and obstructive practices by some of your peers and those who use their work. In particular, it appears that some of the problems in the work have been “brushed under the carpet” in order to present a “better message”. When these practices are called on, obfuscation has taken the place of candid clarifications. It would be great to hear how these problems are being tackled by proper scientists to clear up any misunderstandings evoked by others.
.
One thing to bear in mind. The majority of people here (and Steve keeps a lid on any that don’t) will treat you as you treat them.
bender, humor me, ok? I don’t really wish to read through every post here. On the other hand the moderator requested posts be on topic so maybe just pointing me to the appropriate thread or discussion will suffice.
AMac, a sense of humor please, otherwise this may get a bit dull.
DaveJR, thanks for the welcome. I am well aware of the prevailing opinion here, not so subtle after all.
Steve M, nice pics of the bcp at Almagre posted on Picasa. I haven’t visited the site myself.
Hello Brian, don’t rush to judgment – how do you know I’m ‘open minded”? LOL
Lucy, your question (off topic) and the material posted at the links will require some time for me to respond as I have not followed this particular debate closely. I do believe that the issues have been competently addressed elsewhere, but I will be glad to have a look over the next couple of days.
For people interested in the basics of tree-ring science and is various applications I recommend the following references (probably need to check local library – not sure what Google Books has online, but I expect these are still under copyright protection):
Tree Rings and climate by H. C. Fritts http://www.blackburnpress.com/trerinandcli.html
A Slice Through Time by M.G.L. Baillie http://www.amazon.com/Slice-Through-Time-M-G-L-Baillie/dp/0713476540 (a portion is available through GB)
Dendroclimatic Changes in Semiarid America by E. Schulman http://openlibrary.org/b/OL6217038M/Dendroclimatic_changes_in_semiarid_America.
There is a bibliographic database for dendrochronology here: http://www.wsl.ch/dbdendro/index_EN
My specific expertise is forest disturbance ecology and my work has involved the collection and analysis of many thousands of samples including wood from North and South America, North Africa, and Russia. I am not a ‘dendroclimatologist’ per se, but as dendrochronology by its nature relies on the climatic signal in tree ring series for dating control and as material from many sites that I have collected and/or dated has been used for climatic reconstruction I am familiar with many aspects of this application including its strengths and weaknesses.
Re: CB (#123),
This is not a bulletin board. It’s Steve McIntyre’s lab notebook. He’s read all these books. They don’t answer the questions Stve is asking. Read the blog. Example: Search the blog for “Fritts”. Thanks for playing.
Re: bender (#124),
I get 123 hits for Fritts on the blog. I doubt CB has time to read all of them, but at least it should do a good job of proving Steve knows the literature.
It’s probably also valuable to let CB know that the Graybill / Idso studies were almost all on live trees with a stripbark growth pattern and thus there were no sections taken
Re: CB (#123),
Excellent point. After reading #176 I’d have to say I did rush to judgment.
You have presented a lot of general info that is pretty much straight out of a forestry 101 textbook.
When first presented with the name Ababneh, you said essentially “what’s an Ababneh?”
When it was pointed out she is a PhD in dendro you said “Oh dat Ababneh, yeah I know her” as though there are 10 or 15 Ababnehs hiding behind every tree and you just weren’t sure which one was being referred to.
Now when pressed to provide anything other than generalities you appeal to some unnamed authorities that it couldn’t possibly have been 5 degrees warmer in the past and point us to “the literature” which you claim has been dismissed.
It hasn’t been dismissed. The conclusions based on it are being questioned so you get defensive and basically say when pressed for an answer to the question”you just don’t understand, we’re the experts, trust us, there’s a preponderance of evidence”.
Anyone else feeling like Charlie Brown trying to kick the football?
Actually, I haven’t noticed any questions from Mr. McIntyre on this thread Bender – other than some confusion about the strip-bark growth form and injury mediated callus formation.
I made some suggestions about basic literature because given the level of discussion they seemed appropriate. If those discussing dendrochronology here had really read and absorbed that material the discussion wold have a rather different character.
I have had a look at one of the threads discussing the Graybill and Idso paper on CO2 and … what to say. When questioning takes on the form of an inquisition or is based solely on the desire to find fault or discredit it cannot really lead to new insight as the mind of the questioner is merely looking for validation of preconceptions. When the questioners lack a fundamental understanding of, or appreciation for, the subject of study they will necessarily take many blind turns and tread over a great deal of already turned soil.
Graybill and Isdo proposed a hypothesis, or rather, refined one from LaMarche et al 1984, but failed to prove it. I am sure you have all had a look at Salzer et al 2009?
I noted a comment to the effect “why don’t those dendros get off the AGW cart and just go back to whatever they were doing before” clueless.
Re: CB (#126),
Yes, but then Mann took their “cherry-picked” data and used it as if it were just ordinary tree ring data. Was this wise?
Re: CB (#127),
And just what blog have you ever been to which wasn’t that way? You’re being very general here. Either give us a link to a properly run blog, or provide a specific post from one of the threads you’ve read which you’d like to criticize. Then we can have a discussion.
Re: CB (#126),
BTW, you might want to go to Wikipedia and peruse Sturgeon’s Law, “Ninety percent of everything is crud.” This is an open blog. Anyone can post anything, although Steve limits some topics and does his best to try keeping things on topic with his head post (which, BTW, is what you should be concentrating on).
Re: CB (#126),
D’yuh think there’s a chance he might be a tad busy at the moment?
Re: CB (#127),
Maybe you concentrate on the ones with the interesting discussion?
Look, you can choose to participate as you please, but you’ll recall your very first post gave us all the impression that you thought you knew it all and were going to enlighten us poor unfortunate boobs. Then, with tail between legs, you admit you’ve never even heard the name “Ababneh”, and started asking people to bring you up to speed. Now, you focus on the noise and pretend that’s all there is.
As I said, participate as you wish, but recent events should tell you that if you miss what’s going on here, you’ll miss a lot.
I have read a few more threads, really sorry I did. Full of paranoid comments, wild speculations and assumptions about people and circumstances the posters know nothing about and thus reaching foolish conclusions…. with small bits of interesting discussion mixed in by a small minority of participants… What to say? I don’t have the answers you desire.
Re: CB (#127),
The blog is what you make it.
Re: CB (#127),
CB, thanks for the tree ring references. Coming into a controversial subject (I’m new to this), it’s helpful to get a sense of what believers view as solid ground.
You find too much chaff and not enough wheat at CA. Fair enough. In my experience, hotheads generally don’t stick with technical discussions, and that provides a guide to higher S/N, and to raising S/N. That may not work for you. Another approach would be to comment at Jeff Id’s “the Air Vent,” a lower-traffic blog where the host is a layperson with statistical training who is engaged in understanding dendro’s contributions to paleoclimate reconstructions.
One month ago, my confidence in the “paleoclimate community” was pretty high, based on my general feelings about the physical sciences. Actual interaction with an issue and some community members came in the form of learning about the Lake Korttajarvi varve series. Specifically, their mistaken use in Mann et al (PNAS, 2008), McIntyre’s calling attention to this error, and the circle-the-wagons response to the problem.
In this affair, the conduct of your community — academic and government researchers, authors, peer-reviewers, editors, prominent pro-AGW bloggers — has been generally disappointing. Well below the standards of any other specialty within the physical sciences (I hope).
Varve proxies aren’t important per se. It’s a “canary in the coal mine” thing. The philosophy and conduct of science are important. Facilitating and rationalizing clear-cut and obvious mistakes, acceding to silence, moving the argument to focus on scorning opponents: all par for the course in paleoclimatology, it seems.
So, your comments garner some sympathy, but not as much as they would have, a few weeks ago.
I assume you do a conscientious job in your own work. Notwithstanding that, your profession as a whole has some severe structural problems that it has yet to acknowledge, much less address. I’d suggest that the paranoid comments and wild speculations that you’ve found on this blog shouldn’t rank all that high on dendrochonology’s To-Do list.
Re: AMac (#135), I think some of your angst is misguided.
.
Many of us here (including the proprietor of this blog) understand the issues with proxies and that the vast majority of the researchers investigating the proxies themselves are honest and forthright. The difficulties you raise concerning funding and the fact that the science takes quite some time to be developed are ones we agree with. Many of us – including Steve – have repeatedly expressed our desires for updating existing series and developing new ones. Many of us – including Steve – have repeatedly expressed our desires that additional work be done to improve the interpretation of existing proxies, develop methods to handle the uncertainties, and discover better proxies. Like you, we want to see that work continue.
.
What you seem to be missing (if I read your complaints correctly – and if I am not, please correct me) is that our frustrations are not primarily concerned with the research done on the proxies themselves, but rather the irresponsible use of those proxies in reconstructions that have a great deal of political influence. For one particular example, let us examine the response of the proxies to temperature. The response characteristics of many of these proxies are not fully known. The divergence problem is just one of many issues here. There are additional problems of multiple collinear signals (CO2, precipitation) during the calibration period, problems that modern proxies are located in regions where temperature stresses can be expected to dominate but past proxies are not necessarily temperature stressed, and gauge repeatability and reproducibility problems with the measurement of the proxies themselves. Yet these proxies are being used to make claims about global temperatures (when coverage within the northern hemisphere is rather poor and coverage outside the northern hemisphere is almost nonexistent) with uncertainties of less than one degree celcius.
.
To me, this is astounding. My current job involves a good deal of pyrometry, and I can tell you in no uncertain terms that it is difficult to measure the temperature of a fully-instrumented 10 cubic foot hot zone of a furnace to within 1 degree celcius. To make claims that we can use proxies with unknown response curves to tell temperatures to within a degree spanning back 2,000 years is simply not credible.
.
For an example that is a little closer to home, let us consider the following. Let us assume that we discover no-kidding thermometers with known response curves located in 40 different areas in the NH, 28 of which are from the United States. This should sound familiar to you. Let us assume that these thermometers yield temperatures that stretch back to 1400 AD. Would you seriously argue that, on the basis of actual thermometers meeting these conditions, that you know what the global temperature was in 1400 AD to less than 1 degree?
.
If so, then let’s close down all but 40 GHCN stations, because the extra data is obviously superfluous and a waste of money.
.
Assuming that you do not think that such a sparse network of observations can yield a global temperature accurate to within 1 degree celcius, then what are we to make of a proxy network of equivalent size and geographic distribution that not only does not respond consistently to temperature over a timespan of only 100 years but also has unknown contributions of temperature, precipitation, and fertilization over the timeframe they are used to reconstruct?
.
That is the issue. The issue is not the dendros themselves, or the folks studying varves, or boreholes, or oxygen isotopes. The issue is how that base research is used.
.
That is why the “pebbles” matter. We’re not talking about some qualitative usage of proxies. We are talking about using proxies to obtain temperature estimates that rival that of actual thermometers. We are looking for differences in global temperature that require that kind of precision. So yes, the pebbles do matter. They matter quite a bit.
.
It is difficult to manufacture thermocouples that will read accurately within 1 degree celcius. These are devices that are explicitly designed to respond to temperature. You seem to expect us not to criticize use of proxies to draw conclusions that require a degree of accuracy and precision that is difficult to achieve with engineered temperature measurement equipment.
.
That, my friend, defies belief.
CB I am waiting for you to show your dendrochronologists stuff here at this thread even though the subject is dendroclimatology. I am layperson trying to get my head around the measurements of tree rings and the reproducibility and accuracy of that process. I link to a thread below where I have taken replicate samples for the Yamal and Schweingruber chronologies and differenced them. (My calculations are towards the end of the thread so you need not search the entire length of it.) In the latter series I simply used the (TRW1-TRW2)/((TRW1+TRW2)/2) in order to remove any residual effects from using a standardization method, such RCS, and at the same time normalize for the slower growing older tree rings.
.
http://www.climateaudit.org/?p=7489
.
My questions for you are two:
1. What is said in the dendro literature about replicate sample differences and where might that be linked? Do you have any insights?
2. What is said and shown in the dendro literature about proxy series confidence intervals? Do you have any insights on this?
.
What I have found is variations within replicate TRW cores from the same tree and same year that would make any proxy series so constructed have wide CIs from this variation alone. I have presented the Yamal series with CIs and found that, when looking at this series in that manner, it is difficult to impossible to make any statements about comparisons of the modern time period to those in the past – with any reasonable amount of certainty. Would you care to comment on these findings?
Re: Kenneth Fritsch (#131),
Search the CRU emails for “Ed Cook” and “confidence interval”. Fun to contrast his approach vs the Team’s.
Automatically assuming someone has read through several years worth of conversations and should understand the local shorthand is a bit presumptuous. Actually I do know many of the people frequently mentioned here, including Ms Ababneh, and have worked with some of the material that is the subject of various discussions, have visited a number of the specific collection sites, etc.
As I said above, if the participants of this blog/journal/whatever, including specifically Mr McIntyre, are mostly engaged in turning over pebbles looking for evidence to support their latest pet notions there really isn’t much to say.
The truth about the scientific work that is the main subject of the tree-ring related discussions here is both more boring, in some respects, and more interesting than you would believe. There is a reason that some things make it through to peer reviewed papers and others do not. There is a reason that this type of research takes time to bear fruit. These reasons have nothing to do with conspiracies or attempts to cook up specific results – sorry to disappoint, but it just isn’t so. The work is hard, time-consuming, careful and meticulous. It requires funding – meaning the process of getting proposals written and submitted………and……waiting to see if they are funded (many are not). Recruiting and then helping students get through their academic program and writing the results of their work up as thesis, dissertation, papers, etc.
The work on the bristlecone has been ongoing for more than 50 years and has involved the work of a large number of people over this period. Every question that I have seen raised here has been thought about, discussed, debated, fought about, both amongst those doing the primary research and within the interested scientific community that debated the published papers.
One of the most personally amusing things to read here (hilarious really) is the use of ‘Team’ to describe a certain group of researchers. You people really have no clue. Many of the criticisms raised here are very naive, or simply stupid. Red herrings and a complete waste of time.
The fact that people have cited Hal Fritt’s work in discussions says nothing at all about their understanding of his large body of research.
As for “recent events” let me say this. I came across an incident described and discussed on one of the threads: Mr. McIntyre was attempting to get some data from The Tree-Ring lab at the University of Arizona. He had difficulty and assumed his IP was specifically being blocked. Rather than contacting the sys admin for help to resolve the issue – or simply inquiring regarding the nature of the problem – he assumed nefarious intent, managed to get the data he desired via other means and brought legal action against the University for withholding data. It was a technical problem that affected others as well, took some time to track down and resolve, and had absolutely nothing to do with him. What an egotistical dolt.
You all might do well to consider that many of your dearest assumptions are simply wrong, unfounded, totally without merit. Rather than assuming the worst about people and their motives, people whom you know very little about – and since you have clothed them with your own misperceptions you couldn’t see the truth if it was spelled out in front of you – if you took the trouble to step back, take a breath, and examine the evidence for what it is (a good long look in the mirror might help a bit) without the terrible dark glasses you all seem to be wearing you might get a completely different perspective.
Sorry for the off-topic post.
Re: CB (#134),
Our comments crossed.
None of your observations would cause me to alter my own remarks.
For what it’s worth, I agree with you that there’s “too much” rancor and noise at ClimateAudit. On the other hand, it’s a high-traffic blog, and Anyone Can Play.
Also for what it’s worth, there are a number of open-minded and knowledgeable individuals who host and regularly comment here, who are willing to cordially engage in substantive conversations, and provide insights into some very technical and confusing areas. For me, that makes it worth the effort to read selected threads. YMMV, as they say.
I’ve been looking for a blog with similar virtues on the pro-AGW side of the fence, with no success. Cheerleading takes up too much bandwidth, and hosts are too willing to trim skeptics’ comments to fit the preferred cut of their own jib. Any suggestions?
Re: CB (#134),
You do know, of course (hate to be presumptuous) that the “Team” assigned that name to themselves.
Well, I guess that explains these questions.
But anyway, since you’re familiar with “Ms” Ababneh, perhaps you can answer Bender’s question:
Re: CB (#134),
You are apparently unaware that the term “hockey team”, of which ‘the team’ is a shortened version of, originated with Michael Mann.
Re: Greg F (#140)
Greg, the comment stands. Very funny. It is one thing for Mann, most likely tongue in cheek, to do so, but the way it is used here with such fear and loathing is quite amusing. These people have egos, as well as considerable expertise, and attempting to back them into a corner together where they must defend themselves and each other just makes it harder to make any headway with the underlying science. I think there are better and more productive ways to expend energy for all involved. This group is not a monolithic cabal, but a number of people with convergent interests that have cooperated on some joint research projects.
Re: John M (#139)
John, sorry, use of her full name would have helped, no way for you to know that. As for the question “why not use her chronology?” I am unable to give a full answer in such a forum. The Salzer et al paper represents the summary of all available data and the most complete analysis of it WRT the questions they addressed. It represents about 10 years of hard work and rests upon all the considerable previous work on the subject. It would be more fruitful to examine that paper and the conclusions of its authors rather than waste a lot of time second guessing their published statements. They made their data available, they are good scientists, not hacks. Give it your best shot.
The fact that they didn’t use a particular grad students unpublished reconstruction is a red herring. It was work they were intimately familiar with, if it had something important to offer as it stood they would have. That is all I can offer about it – choose to believe what you will.
Some of the previous papers on the subject were agenda, rather than primarily science driven in my opinion, and suffered for it (this refers to personalities rather than larger scientific questions – pettiness and personal competition). LaMarche died in 1988 and Don Graybill in 1993 and it took a while for this research line to bear fruit again.
Re: AndyL (#138)
Andy, of course I agree.
Re: Ryan O (#143)
Friend Ryan, your post gets more to the heart of the matter and I cannot provide a quick and simple answer (not to imply that an answer I provided would necessarily put the matter to rest for you). It is an important topic for discussion, central really, and lets pursue it. Please accept my apology that I can’t drop everything right now to write a long and full reply, but I do promise to offer what I can as I can. I am not a statistician nor a climate modeler, so don’t expect me to argue from those perspectives. Statistics and models properly are tools, as is dendrochronology itself in some respects, and I do have strong opinions about their proper uses and strengths an limitations. I would be most happy to have a conversation about this subject and also to discuss site and tree selection, and climate response – questions that seem to be of interest here.
best to all on thanksgiving eve.
Re: CB (#144),
Thanks for your response.
I wouldn’t call our references to “the team” fear and loathing. It’s more like sardonic irony, you know, like calling the executive team at a mis-managed company the “brain trust”.
As far as far as Ababneh’s “unpublished” work, isn’t this it?
But anyway, I’m willing to strap on the muzzle and let you and Ryan have a constructive discussion.
Happy Thanksgiving to all.
Re: John M (#145)
John, I see that she published a proceedings paper discussing climate variability and human settlement patterns – ok, fair enough. Doesn’t affect the substance of my answer above. No need to ‘strap on a muzzle’ for my sake.
Re: bender (#149)
Bender – as far as I am aware I’ve never tortured your cat or insulted your mother – what’s with the attitude?
As for your question about the Salzer et al paper, I think it an excellent piece of work that well summarizes the best current understanding of growth trends in the high elevation bcp stands. I have tested their basic finding with independent data from a different site (over the past ~1,000 + yrs) and it gives the same result. Their conclusion that the trend cannot be primarily CO2 related is logical and sound – how could water-stressed trees a few hundred meters down slope, for which WUE should be very important, not show fertilization effects that their neighbors do? The treeline stands are temperature limited, treeline is a temperature mediated characteristic. The frost damage that occurs in adult trees only within upper treeline sites is a temperature related phenomena. While their conservative physiology – retention of needles for one to several decades – provides a buffer to year-to-year changes evident in the high autocorrelation of ringwidth series, the critical factors limiting growth are growing-season length and mean temperature over that period. Growth trends are temperature mediated. While intense drought does occasionally limit growth in these stands, a drought response is not characteristic of the upper sites. Annual growth increment at the upper and lower sites is poorly correlated between them. Their respective dating characteristics are distinctly different.
The strip-bark growth form is ideal for examining growth trends for the reasons stated above:Re: CB (#113)
All standardization is a mathematical kludge necessitated by growth geometry and sampling method (radial increment at two points – it would be preferable to know the volume of tissue added each year and its allocation amongst the various tissue types). It is a kludge because the mathematical assumptions underlying the method chosen can only be approximations hoping to account for the dictates of tree biology and form in reasonable fashion (e.g. negative exponential curve) or simply designed to limit non climatic trend in the resulting indices without reference to tree biology or form (e.g. cubic smoothing spline). RCS (regional curve standardization) is a well-intended attempt to avoid some of the standardization issues by producing a standardization curve that is similar to ‘site index’ measures typical in forestry applications. It attempts to characterize radial growth increment potential at a given site as a function of tree age. Departures from this “average potential” are assumed to be related to fluctuations in transient environmental factors (temperature, precipitation, etc) rather than those fixed by site location – latitude, local geography, soil characteristics and the like that are assumed to be invariant over the time period of interest. It requires a large sample set to develop a robust (sorry, but its a fine adjective when used appropriately) growth curve and the underlying assumptions do not apply equally well to all sites. The fact that its assumptions are based directly on tree biology and site conditions rather than abstract mathematics is a good thing.
Strip bark tree geometry allows a more direct measure of growth that is not subject to statistical artifacts produced in the process of standardization – radial increment measured in mm..
out of time at the moment.
Re: CB (#150),
I don’t understand. Why would anyone here think the growth trends are signs of CO2 fertilization? The only thing CO2 fertilization has to do with the subject is that Graybill and Idso took the stripbark samples in an attempt to find it. That’s why their samples are overloaded with stripbark. But the cause of the stripbark growth spurt has more to do with the fact that if there’s only a strip of cambium in use there is no radial component to speak of. Each years growth will go into building the wood as a same-length strip while, presumally, the crown and-or root mass will be increasing for many years and thus the thickness of the wood each year will increase. Eventually the “corners” will wrap around there will be a new radial component, and the new vegetative growth will max out. So we should first find thinner rings, then increasing ring widths and finally a tapering off back to normal growth. The yearly differences in growth will be temperature related, and if there is an actual temperature trend it will be subsumed into the overall growth trend, but IMHO, most of the ringwidth trend will be biological and geometrical.
BTW I’m writing the above with ringwidth in mind, but if it’s max density being measured/used, we need to look at the theory of maximum density vs temperature to see how it might apply.
Re: CB (#150),
I think that there is the same problem with the temperature explanation. Temperature drops with elevation by less than 1 degree per 100 m. Accepting the temperature explanation would imply that trees at the tree-line *are* limited, while trees 150 m below the tree-line (that experience temperatures by only about 1 deg C higher than the treeline) are totally not. This would mean that the response function of tree growth to temperature in P. longaeva is flat in a wide interval of temperatures (where trees are totally unresponsive) and then rises abruptly at the temperature of the treeline. Is there any evidence for that?
Another issue is the comparison of absolute tree ring width between the treeline trees and trees below the treeline. I could not find mentioning of this problem in Salzer et al., would be great to have a clue. The question is — do the increased tree ring widths of the treeline trees rise above the unchanged ring widths of trees below the treeline? How do these relate?
It would be strange, in my opinion, if the treeline trees stressed by low temperature would develop wider tree rings than the presumably non-stressed trees living 150 m below the treeline develop at higher temperature. That would be an argument against the temperature explanation.
Re: CB (#150),
Is this standard not adhered to a U of A?
http://en.wikipedia.org/wiki/Doctor_of_Philosophy
Anyway, my “muzzling” comment was to clear the way for you, Ryan, and people with a more detailed knowledge of dendro and statistical analysis to have a constructive exchange. I have a PhD in Chemistry, can follow the gist of scientific argument, and can usually decipher when someone’s blowing smoke (particularly wrt to “peer review” and appeals to authority), so my BS detector sometimes leads me into less constructive engagements.
Re: John M (#153) “Is this a standard not adhered to a U of A?”
Yes it is. BTW, if your ‘BS detector’ quivers then by all means pipe up.
Re: Dave Dardinger (#151)
Dave, I believe your assumptions are incorrect and/or incomplete.
This was proposed by Lamarche et al 1984 as an explanation for the growth trend that had been observed at the high elevation sites. Lamarche was interested in the high elevation sites for historical reasons (he had done erosion work at the Mt Washington site), and because of his interest in ‘frost rings’ that had been linked to post (volcanic) eruption cooling.
Cores and remnant wood from strip bark trees has been used to construct chronologies from these sites since they were first identified as being of interest more than 50 years ago. No special discrimination or distinction was applied to this very common growth form except that, as the oldest trees are always in this configuration, and they were often specifically targeted for other reasons.
Latewood density issues do not, of course, apply to the chronologies in question. Technical considerations/limitations have precluded the construction of bcp maximum latewood density (MLD) chronologies. Perhaps this will change in the near future if other means of extracting the density parameter not dependent on xray measurement are proven effective and reliable. Most of the existing density chronologies have been produced in European labs that have budgets to support the necessary equipment and technical staff required for this work.
While the strip of live cambium does vary in width over time, the corners do not “wrap around” to re approximate a whole bark form. The edges of the strip are exposed to the elements and expand or contract slowly (over decades to centuries). As photosynthetic area of the crown (and presumably root area as well) increases in favorable periods there will be more resources available to produce wider ‘rings’ (or more properly xylem layers). There is no intrinsic constraint on live crown area for trees of this size other than the resources the organism is able to devote to this process (except for factors like average stand height and their effect upon exposure at the individual tree level). Attributes like sapwood area are directly related to the mass of live crown that needs to be supported.
Because the roots, stem, and crown are an integrated system they vary together rather than as independent elements as you seem to suggest in your post.
Does this speak to your concerns? Or are we talking past each other?
Re: Anastassia Makarieva (#152)
Re: NW (#154)
Ok, reasonable question.
There is a marked difference between trees growing at the upper and lower limits WRT their climatic response as reiterated in the Salzer et al paper. Lower forest border trees have higher mean sensitivity and lower autocorrelation than trees at the upper limit. These differences are not subtle and anyone who has worked with the material itself would recognize the differences without resorting to statistics and response functions/correlations with climate variables. The lower boundary is dictated most directly by precipitation/evapotranspiration. The upper bound is clearly temperature mediated. As I mentioned above upper treeline sites are subject to damage by late season frost, those at the lower limit are not.
I don’t believe that the shape of the transition has been specifically studied for bcp, but researchers confronted with the this have always exploited the upper and lower limits of the distribution in order to maximize the response (signal) of interest. Trees in the intermediate zone have a mixed response (precip/temp) that becomes impossible, with curent methods, to disentangle. See the Fritts reference above for a generic answer to this question WRT dendrochronology as a whole that is based on a large body of work on the subject.
I am sorry I don’t have data handy to run some quick numbers for you but I could do this within a week or so. I expect this type of comparison is in the published literature but I can’t point you to a specific paper without doing a literature search. Lamarche has a paper in the ’70s that used this dichotomy of response to look at climate variability – that might be a place to start.
In the White Mtns. (CA) Methuselah Walk is considered to be a typical lower forest border site with a strong precip signal, Campito Mtn. (developed by Lamarche and cited in several papers by him) is considered intermediate, and Sheep Mtn. a high site.
As an aside, the stands used by dendrochronologists are rocky with little ground cover making them unsuitable for grazing by sheep, etc. especially relative to the nearby flats and meadows.
Moisture limited trees as a group tend to grow more slowly (have narrower rings) than trees that are not. For the bcp stands, again, I could come up with some numbers if you are interested.
Bottom line for me – we are not talking about subtle differences in response, but fairly dramatic ones that are easily recognized.
Anastassia, interesting website BTW. I’ve only had time for a quick look.
Re: CB (#155),
My apologies. On re-reading my question, it occurs to me I butchered the english, and may impact how your answer is read.
I’d like to restate my question as “Is this standard adhered to at U of A?”
Is your answer still yes?
Thanks.
Re: John M (#156)
Yes
Re: Ryan O (#157)
Ryan, I haven’t forgotten. There have been quite a few inter-related questions and I am attempting to provide substantive answers. There are some further up the thread than yours that I haven’t responded to yet. I took your question to be even a little more general/broad than your restatement above – and discussing it fully could take quite a little time and more than a single post. It is a little unfortunate that I am in the position of the lone gunman facing the militia here. My expertise does not cover the full range of issues under discussion – I am happy to offer my take, but cannot provide detailed answers to all aspects of the subject at hand.
ok…
Having done a short run through a number of threads here (including this one) I disagree – in fact many of the basic methods and assumptions of producing proxy records from tree-rings are being disputed. Small point though.
You need to be specific and depending what you refer to/want to discuss there is quite a bit that could be said. I made a start at addressing the relationship of temperature to tree growth at upper treeline. Did you have a problem with that as far as it went?
This is an important/interesting issue. It has come up for the bcp record only because a MLD based temperature reconstruction was being directly compared to it. A reasonable conceptual hypothesis for it is based on limiting factors – summer temperatures were a primary environmental parameter controlling cell wall thickening in conifer latewood at some sites but due to environmental changes, e.g. warming, other factors have become more important and obscured this relationship. I assume that some experimental work to examine this problem is planned or ongoing – typically such studies take several years to complete, and often a series of studies (different site types, species, variety of methods, etc.) is required to gain a fuller understanding.
The issue for proxy based reconstructions, as you note below, is whether similar changes occurred in the past. Proxy reconstructions assume that the basic controlling factors haven’t changed over the period of interest. How can we know this? Independent lines of evidence that validate the reconstruction are an important test. Other similar records should show the same basic features – at a minimum. For dendrochronology this means that a network of tree-ring sites with a similar environmental response should show the same basic patterns within a given climate region.
I know that this brings up something that has been discussed here WRT Keith Briffa’s decision not to use Fritz Schweingruber’s data to bolser his Yamal reconstruction. This is a tangent that could generate a whole separate discussion, but the short answer is that just because there is data available from some tree site doesn’t say anything about it’s appropriateness or suitability for a particular task. Tree selection, site selection, and sample size are competently dealt with in Hal Fritts’ Tree Rings and Climate linked above. [I, frankly, have not paid much attention to Briffa’s reconstruction, as my work is not directly related to it. However, I would assume his paper describes his methodology, selection criteria, standardization method, etc. As this work seems to have generated so much interest I will have a look at it in the next week or so.]
For those who might be interested Fritz’s website is here: http://www.wsl.ch/staff/fritz.schweingruber/
Other independent types of proxy records, historical accounts and records, etc. should corroborate prominent features of a regional tree-ring based proxy record. I would not consider conclusions based on a single record from one site to be verified and the longer the time span of interest the more important this becomes. A single record is a site history if it lacks independent context. The Yamal reconstruction, specifically, was not produced in isolation – it is not the first to reconstruct temperatures in this region, so its general applicability can be judged within the context of related work.
Back to divergence – there is a great diversity of evidence now available to suggest the the last half of the 20th century, and now the first decade of the 21st, is substantially different than the previous 5k years. The divergence problem fits within this context. We don’t have perfect knowledge, of course, but neither are we blind in this respect. Neither is the available tree-ring record universally afflicted with this particular problem.
http://www.agu.org/pubs/crossref/2007/2006JD008318.shtml
http://www.ingentaconnect.com/content/bsc/gcb/2008/00000014/00000010/art00017;jsessionid=2t008muuqrb88.alexandra
As to the question of whether the divergence issue might affect the proxies in the past – prior to the instrumental record – we can look at the Salzer paper and see that they conclude from their data that periods equivalent to the past 60 years have not occurred during the past 2,000 + years. While this is not conclusive evidence for the entire NH, it is important and not contrary to the view that divergence should not be an issue for that period.
There are ways to recognize and deal with these problems and any search through the literature will show both acknowledgment of them and solutions proposed and/or applied to particular studies where appropriate. You cannot make such a blanket statement except as a rhetorical device. Uncertainties are acknowledged and highlighted in much of the relevant work.
WRT a specific work the Salzer et al paper addressed the issue of competing signals (CO2 and precipitation) showing that the material they were using had very little precip (mainly limited to single very dry years) response and ruled out CO2 as a significant competing factor that might cause the rise in ring width.
An anecdote here: I was doing a tree-growth monitoring study examining the relationship of growth to precipitation in an area with a bi-modal precip regime. I had instrumented trees with mechanical recording devices that provided a direct measure of tree stem expansion. In collaboration with another person we added a set of electronic measuring instruments to the study trees that were set up with a datalogger. We hoped to eliminate the mechanical devices that required more personal attention (taking measurements, changing charts, etc.) At one point the electronic instruments showed a step change that was not reflected by the mechanical recorders and the electronic equipment designer asserted that the mechanical devices must be in error because they did not show the step change. Turned out to be an issue with the electrical ground circuit.
Measurement of complex systems is not a simple task – agreed.
Trees, as organisms adapted to a specific range of environmental conditions, necessarily respond to their environment by buffering their physiological systems within the tolerable range of external conditions to which they are adapted, and then by migration to new suitable habitat when conditions become intolerable – if they can. It is possible to characterize the tolerable range based on the physical evidence provided by current species distributions and infer the location of past suitable habitat based also upon physical evidence. One can then logically conclude that the prior distributions existed withing the tolerable limits known for the species. If the tolerable range WRT temperature seems small to you perhaps this is because describing ‘suitable environment’ with a mean value is a simplification that, while it does have some utility, is used as a stand-in for a suitable range of temperatures over the length of a growing season.
Reducing the issue to a cartoon noting only the potential problems can make the task seem intractable. For a look at a network of proxy data that is well verified the Cook et al PDSI grid is a good example:
http://www.ncdc.noaa.gov/paleo/usclient2.html
http://www.ncdc.noaa.gov/paleo/pdsi.html
I am very familiar with certain portions of this data, and while the variable of interest is not temperature perhaps it can serve as an example. The gridded data is very robust for the period 1700 to present over most of its extent. Within this period it is possible to resolve fairly detailed questions about regional differences in climate (with respect to drought). As one goes back in time the spatial resolution decays because the underlying data becomes increasingly sparse. I believe the current length is about 2,000 years and for the earliest portion it is impossible to resolve fine scale questions – in fact only the most general questions can be informed by the sparse grid of data points available and some regions are white space – no estimate possible – and errors of estimation are large for the available points. This example is for drought reconstruction and the precip contribution has high spatial variability.
Temperature has less spatial variability and fewer sites would be needed to produce reasonable estimates, but the same principles apply. The fewer the sites contributing data the more general the questions that can be addressed till with only a single site there there is only a point represented and we have lost all spatial resolution and context for evaluation. On the other hand it is possible with a few widely distributed points to examine general questions, and the more that is known about the relationship of those points to the information provided by a dense grid the better context we have for evaluating the information they provide. The more other types of proxy data we have the better context we can create. The efforts of the paleoclimate community to provide this context is very broad and covers a wide range of proxies.
In some respects organisms and their environmental response is to be preferred because the response is integrated and because said organisms are widely distributed. I think you are stuck on ‘degrees Celsius’ and the narrow view of ‘trees as thermometers’ (obviously they are not) and cannot see that 1 deg. C is a stand-in for much broader scale changes in ranges and durations.
The ‘pebbles’ reference had to do with looking specifically with the intent to find fault with a very narrow focus. I would reiterate that the research community has a great interest in improving both the underlying data available to answer questions and the analysis of existing data with respect to new information and perceived problems or shortcomings with previous work. An objective look at the published literature demonstrates both the awareness of the issues being discussed here and attempts to address them where appropriate/applicable.
Please see above.
The site is unresponsive – due to heavy traffic I assume – and I am aware that there have been a couple of additional responses to the earlier posts. Please have patience I will respond. I hope the text above has been helpful.
I will make an additional comment here about RCS (regional curve standardization). Above I gave a brief description of the method theory by comparing it to a measure of ‘site index’ or site potential. To expand just a bit: the idea that one might construct a single curve for standardization purposes that captures the average potential radial growth with respect to age for trees of a given species growing on a particular site has an inherent logic and appeal to dendrochronologists. Other solutions have the disadvantage of choosing an arbitrary form that may have a generic justification (e.g. negative exponential – fixed volume layers added to a cone or cylinder) or be simply completely data dependent (cubic spline or polynomial) with no biological assumption. Additionally preserving trends longer that the series length is impossible with the methods.
In my estimation the ideal application of this method is in situations where site conditions can be assumed to be stable, the stand has an open character so that between tree competition is minimal, and a large population of samples of varying age is available for a reference period for which the curve will be constructed. It will then be possible to compare growth during other periods with that reference period. Large sample depth during other periods is not then required (a robust curve to remove growth trend is already in hand) except in relation to the general coherence of the tree response to the environmental parameter of interest – if the trees are ‘noisy’ then a larger number of samples will be required to extract signal than if the signal is more coherent. This second issue of coherence would apply to all time periods.
Shade tolerant species and sites with closed stand conditions are poor candidates for this method.
Because there will always be an ‘oldest tree’, and overall a smaller number within the oldest age class, the curve estimate will be less robust toward the tail – but unless the behavior of the oldest trees is typically erratic this issue should be of minor importance.
Re: John M (#156)
Yes
Re: Ryan O (#157)
Ryan, I haven’t forgotten. There have been quite a few inter-related questions and I am attempting to provide substantive answers. There are some further up the thread than yours that I haven’t responded to yet. I took your question to be even a little more general/broad than your restatement above – and discussing it fully could take quite a little time and more than a single post. It is a little unfortunate that I am in the position of the lone gunman facing the militia here. My expertise does not cover the full range of issues under discussion – I am happy to offer my take, but cannot provide detailed answers to all aspects of the subject at hand.
ok…
Having done a short run through a number of threads here (including this one) I disagree – in fact many of the basic methods and assumptions of producing proxy records from tree-rings are being disputed. Small point though.
You need to be specific and depending what you refer to/want to discuss there is quite a bit that could be said. I made a start at addressing the relationship of temperature to tree growth at upper treeline. Did you have a problem with that as far as it went?
This is an important/interesting issue. It has come up for the bcp record only because a MLD based temperature reconstruction was being directly compared to it. A reasonable conceptual hypothesis for it is based on limiting factors – summer temperatures were a primary environmental parameter controlling cell wall thickening in conifer latewood at some sites but due to environmental changes, e.g. warming, other factors have become more important and obscured this relationship. I assume that some experimental work to examine this problem is planned or ongoing – typically such studies take several years to complete, and often a series of studies (different site types, species, variety of methods, etc.) is required to gain a fuller understanding.
The issue for proxy based reconstructions, as you note below, is whether similar changes occurred in the past. Proxy reconstructions assume that the basic controlling factors haven’t changed over the period of interest. How can we know this? Independent lines of evidence that validate the reconstruction are an important test. Other similar records should show the same basic features – at a minimum. For dendrochronology this means that a network of tree-ring sites with a similar environmental response should show the same basic patterns within a given climate region.
I know that this brings up something that has been discussed here WRT Keith Briffa’s decision not to use Fritz Schweingruber’s data to bolser his Yamal reconstruction. This is a tangent that could generate a whole separate discussion, but the short answer is that just because there is data available from some tree site doesn’t say anything about it’s appropriateness or suitability for a particular task. Tree selection, site selection, and sample size are competently dealt with in Hal Fritts’ Tree Rings and Climate linked above. [I, frankly, have not paid much attention to Briffa’s reconstruction, as my work is not directly related to it. However, I would assume his paper describes his methodology, selection criteria, standardization method, etc. As this work seems to have generated so much interest I will have a look at it in the next week or so.]
For those who might be interested Fritz’s website is here: http://www.wsl.ch/staff/fritz.schweingruber/
Other independent types of proxy records, historical accounts and records, etc. should corroborate prominent features of a regional tree-ring based proxy record. I would not consider conclusions based on a single record from one site to be verified and the longer the time span of interest the more important this becomes. A single record is a site history if it lacks independent context. The Yamal reconstruction, specifically, was not produced in isolation – it is not the first to reconstruct temperatures in this region, so its general applicability can be judged within the context of related work.
Back to divergence – there is a great diversity of evidence now available to suggest the the last half of the 20th century, and now the first decade of the 21st, is substantially different than the previous 5k years. The divergence problem fits within this context. We don’t have perfect knowledge, of course, but neither are we blind in this respect. Neither is the available tree-ring record universally afflicted with this particular problem.
http://www.agu.org/pubs/crossref/2007/2006JD008318.shtml
http://www.ingentaconnect.com/content/bsc/gcb/2008/00000014/00000010/art00017;jsessionid=2t008muuqrb88.alexandra
As to the question of whether the divergence issue might affect the proxies in the past – prior to the instrumental record – we can look at the Salzer paper and see that they conclude from their data that periods equivalent to the past 60 years have not occurred during the past 2,000 + years. While this is not conclusive evidence for the entire NH, it is important and not contrary to the view that divergence should not be an issue for that period.
Re: John M (#156)
Yes
Re: Ryan O (#157)
Ryan, I haven’t forgotten. There have been quite a few inter-related questions and I am attempting to provide substantive answers. There are some further up the thread than yours that I haven’t responded to yet. I took your question to be even a little more general/broad than your restatement above – and discussing it fully could take quite a little time and more than a single post. It is a little unfortunate that I am in the position of the lone gunman facing the militia here. My expertise does not cover the full range of issues under discussion – I am happy to offer my take, but cannot provide detailed answers to all aspects of the subject at hand.
ok…
Having done a short run through a number of threads here (including this one) I disagree – in fact many of the basic methods and assumptions of producing proxy records from tree-rings are being disputed. Small point though.
You need to be specific and depending what you refer to/want to discuss there is quite a bit that could be said. I made a start at addressing the relationship of temperature to tree growth at upper treeline. Did you have a problem with that as far as it went?
This is an important/interesting issue. It has come up for the bcp record only because a MLD based temperature reconstruction was being directly compared to it. A reasonable conceptual hypothesis for it is based on limiting factors – summer temperatures were a primary environmental parameter controlling cell wall thickening in conifer latewood at some sites but due to environmental changes, e.g. warming, other factors have become more important and obscured this relationship. I assume that some experimental work to examine this problem is planned or ongoing – typically such studies take several years to complete, and often a series of studies (different site types, species, variety of methods, etc.) is required to gain a fuller understanding.
The issue for proxy based reconstructions, as you note below, is whether similar changes occurred in the past. Proxy reconstructions assume that the basic controlling factors haven’t changed over the period of interest. How can we know this? Independent lines of evidence that validate the reconstruction are an important test. Other similar records should show the same basic features – at a minimum. For dendrochronology this means that a network of tree-ring sites with a similar environmental response should show the same basic patterns within a given climate region.
I know that this brings up something that has been discussed here WRT Keith Briffa’s decision not to use Fritz Schweingruber’s data to bolser his Yamal reconstruction. This is a tangent that could generate a whole separate discussion, but the short answer is that just because there is data available from some tree site doesn’t say anything about it’s appropriateness or suitability for a particular task. Tree selection, site selection, and sample size are competently dealt with in Hal Fritts’ Tree Rings and Climate linked above. [I, frankly, have not paid much attention to Briffa’s reconstruction, as my work is not directly related to it. However, I would assume his paper describes his methodology, selection criteria, standardization method, etc. As this work seems to have generated so much interest I will have a look at it in the next week or so.]
Re: CB (#165), No problem. A few things.
The reason, though, is because ridiculously small uncertainties are applied to the reconstructions. If MBH 98 had uncertainties of +/- 5 degrees or more, investigating these details would be irrelevant. Again, I stand by the statement that the big problem is the uncertainty ascribed to the reconstructions. The discussion on the proxies is in that context.
Nope, not at all. I will note, however, that much of the analysis you talk about is qualitative in nature. Unless I see a response curve, with uncertainties . . . or a model, with uncertainties, statements like the following:
demonstrate at most a qualitative relationship. I cannot use this information to assign a +/- degree C accuracy to the signal extracted from the rings. I also would be remiss if I failed to mention that this is still an assumption. For living trees, we know what the temperature is in the grid (or do we . . . heh heh . . . sources of uncertainty everywhere) and yet correlation between tree rings and local growing season temperature is often poor (coefficients of 0.27 ain’t that good, regardless of what the p-value is). Therefore, there are other factors that affect ring growth (some of which may simply be the method of standardization). The confidence with which you seem to imply that we can extract this temperature signal is, to my thinking, unsupportably high. Let me challenge you to blind test to show me a stand of trees that can predict the local temperature to within a degree. 😉
Agree in principle, but we here already knew that. What I am interested in exactly what those basic controlling factors are and what range they each must stay in for the proxy reconstruction to be valid. Once you have determined that, I am next interested in the evidence that shows that the fossil trees remained within those bounds. The peer-reviewed literature is replete with qualitative statements such as what you have written above, and also replete with qualitative assessments that these (generally unspecified) “controlling factors” have remained the same. I have yet to find a single one that puts any numbers to this.
Until you can put numbers to it (and show reasonable evidence that those numbers are realistic), I will refuse to believe. This is because I have a metric crap-ton of experience with metrology and pyrometry (I have set up 2 metrology labs for my division, which is a $1 billion in annual sales medical device company). Qualitative statements and 50 cents will get you a cup of coffee in my lab – but it won’t get you a calibration sticker. In my business, qualitative statements can kill people. Perhaps climate science ought to look at it in the same way.
That wasn’t a cartoon. That was MBH 98. Those weren’t problems – those were the actual conditions of the study.
I agree with this statement and what you write after it. I agree that trees are not thermometers – that they are indicators of several confounded climactic factors.
That, of course, is my whole point (and the point of many on this site).
Trees are not thermometers. Your statement that my focus here is myopic is a non-responsive answer. Our concern here is that trees are being used as thermometers without sufficient justification, and this usage has significant political impact. In my opinion, while the information you present is interesting, you deliberately avoid the one thing that we both apparently agreed earlier was the “heart of the matter”.
It is utterly preposterous to reconstruct past temperatures using proxies such as these and claim an accuracy of +/- 1 degree.
😀
Re: Ryan O (#168),
When using SI units, I was told the proper spelling is “metric crap-tonne.” I believe the same source told me that a metric buttload could be substituted by fitting it to the metric crap-tonne curve, but you either have to hard code in a correction factor (standard practice is to do so without comment), or you have to truncate the metric buttload plot at the end year of the curve fitting period to avoid the divergence problem.
Re: Ryan O (#168)
Friend Ryan, lose the smilies, they don’t become you.
Ok Mr. metrologist, now I can better understand your disadvantage in the current discussion and the source of some of your misconceptions. Tell me now, how is it you choose your doctor? Is it his ability to measure your temperature to an accuracy of +/- 0.001 deg C? Is it because he has a degree in statistics and can describe accurately your statistical probability of dying from a procedure he is recommending? Or are there some other qualities that might possibly be important for you?
Your training has clearly not enhanced your ability to understand the subject of the discussion here. As I mentioned in the anecdote above, the electrical engineer – who is a very smart man, very talented – was willing to reject the direct evidence presented of the error in his equipment and argued with me at some length about it. He never was able to admit the source of the problem, even after fixing it.
The evidence for accelerated climate change is all around us. It has been and is being studied by thousands of scientists from many different fields and perspectives. I am sure you don’t need me to point all this out to you – yet you choose to hide behind your thermometer. Ok. You want your doctor to predict the future and tell you whether you are going to die or not – with appropriately tiny error bars no doubt.
I see you have misunderstood my point about variation in temperature as a stand in for the suite of conditions that is convenient to call “growing season temperature (degree-days is a better descriptor)” and retreated to the “give me a treemometer that reads temperature to +/- 1 deg C or I won’t believe any evidence that is presented”. The bounds of possible past variation are constrained by what we know about climate over the past 2k years. It simply cannot have been 5 deg warmer (or cooler) during that period. The preponderance of evidence provides some basic constraints to allow reasonable scaling. The past fluctuations in treeline and growth rates in upper treeline stands are direct evidence of temperature mediated climate change. That they are calibrated with the only widely available measure collected by humans (often from less than ideally located stations with short records) – well that’s life, you work with what you have. Does the current understanding represent perfect knowledge? Clearly not. Does the fact that four treeline sites, some hundreds of miles apart, show high coherence in basic features and trends over that past 1,000 years, sites where ‘temperature’ is the most limiting factor have anything to do with temperature variation in the western US over that period?
Pressing me for statistical measures and characterizations of work done by others cannot be too fruitful as I can only point you to the literature which you seem to have largely dismissed. I think you need to get out of your lab a little more often and smell the roses.
Re: CB (#176),
I thought the purpose of this blog was not to ‘dismiss’ the literature but to audit it. The litertiure is not ‘dismissed’ but taken very seriously. If papers published in prestigious journals contain elementary mathematical errors that seriously affect their conclusions, why is this not a serious matter?
Re: CB (#176), Hm. Not sure about the tone of this one. I had thought we agreed that use of dendro series as thermometers in reconstructions was the “heart of the matter” (your words, but I agree). You have spent a good deal of time qualitatively explaining how trees may contain a temperature signal . . . but the whole point of my comment and most of the threads on proxies at this site is that folks in the peer reviewed literature are assigning quantitative values to the supposed temperature signal, with unrealistically small confidence intervals. When challenged on that point, you resort to anger – which doesn’t bode well for you being able to convince anybody of anything. 😉
Thanks! I like them, too.
On the contrary. I rather think my training has helped me understand the issues far more clearly than I would have been able to otherwise. I recognize that qualitative explanations – while often intellectually satisfying – often lead one astray when taking or interpreting measurements. If one is drawing quantitative conclusions, one needs to establish the problem in a quantitative way.
Absolutely the wrong interpretation. You don’t get to use outside constraints to reduce your calculated confidence intervals. If the confidence intervals exceed your constraint, then the correct conclusion is that you have no reliable information about that period – not that you can simply reduce the CIs to some value that is scaled in some way to an amorphous constraint to which you cannot assign an actual value (you can only make reasonable guesses).
So when Steve McI calculates Brown & Sundberg style CIs for MBH and finds CIs that exceed 20 degrees C for certain time periods, he is not implying that it could have been 20 degrees warmer during those times . . . he is showing that the math cannot resolve the temperature during those times with more precision than +/- 20 degrees. In other words, the reconstruction gives us no usable information.
As you have dropped some good nuggets of information so far, it would be my hope that you are able to overcome your frustration and continue to contribute in a constructive manner. I intend nothing personal with my remarks.
Re: Ryan O (#186),
All I have seen from CB is air.
Re: CB (#155), This is all very nice information, but as yet you have not addressed what I stated (and you seemed to agree) was the central issue of the debate: the accuracy that is claimed when using dendro chronologies to reconstruct past temperatures.
😉
Re: CB (#155),
Re: Ryan O (#157),
I understand Ryan’s question about accuracy. My question was about range, and I’m not sure of CB’s answer to it. Could be my lack of knowledge and/or grasp of CB’s answer, but I’ll elaborate my question a bit just to make sure it was understood.
As near as I can make out, the rate of change of temperature with height is something between 0.5 to 1.0 Celsius per 100 meters, depending on whether the air is saturated or not and what the current temperature is; so let’s call it 0.8 C/100m for kicks. When I look at various anecdotal stuff from dendrochronoligists, I find things suggesting that a simple change of 150 meters downward (from upper treeline) is enough to eliminate observed growth surges at treeline. So this looks like pretty simple math: 150 meters X 0.8 C/100m = 1.2 C is the approximate temperature change that can be registered by the upper treeline (since at 150 meters below, where it is 1.2 C warmer, the trees don’t respond to the temperature surge with a growth surge).
Therefore, I find it hard to see how a *living* upper treeline of very old trees can measure any temperature changes over their age greater than 1.2 C (ok, make it 1.5 C at most with unsaturated air). Many of the climate reconstructions I see have temperature ranges (over say 2000 years) that are at or near this upper bound. So, I don’t see why I shouldn’t suspect that both the MWP and (perhaps) today’s temperatures are being censored by a “short ruler” that cannot measure temp variance outside of a small range.
Am I completely wacked out, or is there some sense in this?
Re: NW (#158),
I think Ryan O’s questions are similar to mine in that we would like to know the accuracy and reproducibility of using tree rings and densities to measure temperatures. I judge that that that explanation should start with the core replications on the same trees and at the level of tree ring width measurements so as at this point to avoid the issue of standardization.. That is were I started and found some surprisingly wide ranging variations.
.
CB has done the job of introducing us all to the elementary aspects of dendrochronology and it seems very much in line with what I as a layperson have garnered over a short period of interest. I am wondering as we move on to the finer points of dendroclimatology if we are venturing into areas where CB has less experience and basic knowledge. Perhaps a simple explanation by CB of the differences between the chronology (CH) and climatology (CL) part are in order at this juncture of the discussion. From my layperson perspective I see the CH part with less controversial algorithms used for dating tree rings, less uncertainties in the results and having the advantage of being the longer practiced branch.
.
I have only looked at the Yamal and Schweingruber series for variation of replicate core samples and at CIs for the Yamal series. I noticed that Yamal had few replications while Schweingruber had many. I also found that Schweingruber’s duplicates had rather incomplete overlap of tree ring ages, i..e. fewer than Yamal as tree rings common to the replicate samples.
.
I also noted that while Briffa, in a reply to Steve M’s use of the Schweingruber series adjusted the sample count for the duplication in that series. I did not see that adjustment made for replicate samples in the Yamal series. My point being here that most of these published series results have an accompanying core count, but that the replication rates are important additional information required – as is showing statistically how these replicates are handled.
The strip bark issue though perhaps very separate from the issue of other types of traumatic tree events with regards to tree ring widths leads back to this issue of replicate core samples and what they might reveal as to reproducibility. As a practical matter the wide variations in tree ring measurement from the same tree would appear to encourage the use of replicated sampling, assuming it can be shown that the sampling is showing random variations and not biases. I am not at all sure this is what I see in the few articles on the subject that I have read. Perhaps CB can help with that question also.
Re: NW (#158),
I thought the issue was at what point down from the tree line that the tree (some species) loses its thermometer capabilities and becomes more sensitive to precipitation. If a tree has stayed at the same relative position to the tree line for its entire life or life to the point of coring, altitude should not be a factor other than determining how much of its signal is temperature.
.
My question would be whether the basic physiology of the tree that makes it a thermometer is affected by changes over its life that relate to the tree line but would not be manifested or readily detected by the actual change in tree line because that process might lag the effects.
Re: NW (#158)
NW and Ken, a short answer is that the investigators need to combine evidence of tree-line fluctuations with estimate of past temperature changes to model the spatial extent of the upper treeline zone in order to avoid this problem (by ensuring that only material that was within this zone at any given time is used).
In the current study they make extensive use of remnant material that came from above current treeline and believe that overall they have avoided this problem to the extent possible.
Re: Kenneth Fritsch (#161)
The temperature response of trees is dictated by the principle of limiting factors. The temperature response declines down slope and transitions through a mixed response to one dominated by precipitation. Trees that live for thousands of years, as you point out, could grow in a spot that was within more than one zone over its lifespan (see above for a possible solution). Practically speaking it is possible to recognize distinct characteristics of growth that give some indication of the type of response to climate that is being represented.
Re: CB (#150),
Re: Anastassia Makarieva (#152),
A temperature response function of this sort–a threshhold function–would also have some potentially strong implications for using tree growth as a thermometer. True, there could be genotype variation amongst the trees as to the critical value of temperature (call it C*(t)) below which tree t registers temperature in its growth. But even then your expected population-at-treeline response function would be had by integrating over a population of tree-specific threshold temperatures C*(t) (t indexes tree types, not time…there would be an F(t) c.d.f. at treeline.). The result would almost certainly be a highly nonlinear population response function to temperature variations, pretty flat for sufficiently high or low temps with something like an s-shape connecting between those sufficiently high and low temps.
Does that sound right CB? Or are Anastassia and I missing something?
Re: CB (#144), I honestly look forward to any comments you may have. Like you, I only have a limited time to devote to this myself. I am interested in what you have to say, and I imagine others are as well.
.
As we are most likely going to discuss things that are not specific to strip bark, I also pose an open question to the moderators if this is something that we should take to Unthreaded?
Re: Ryan O (#146),
Moderators: Would a separate thread devoted to this topic be practicable?
Per Ryan O’s comments @ #143: I have transcribed a brief talk by Norman Wolmark, the Chair of the National Surgical Adjuvant Breast and Bowel Project on the history of treatment of breast cancer(!). Perhaps surprisingly, some of the social and statistical issues that this field struggled with may be quite relevant to paleoclimate reconstruction. I’ll try to get it online after the holidays.
Happy Thanksgiving.
Re: CB (#134),
The researchers in question are the ones who decided to refer to themselves as the “Team”. I don’t know if that makes it any less hilarious, but I thought you should know.
Re: CB (#134),
Please, Judith Curry I need you here to bear witness. CB, I assume I am correct in your being a visiting scientist.
.
And, CB, when you have sated your need to give us the what for, I would still be interested in your views on the questions that I put to you in my post above.
Steve M. I referred your behavior in the previous post as that of an egotistical dolt. While I don’t apologize for the comment, rest assured that a fair number of the folks on your red list are capable of similar behavior. Human failings are a characteristic of the species.
Ken, I will have a look at the thread and will be happy to contribute if I have something to add.
A request to the regulars here (Bender, Mosh, Kenneth F, Dave D etc):
Please be patient, give CB some space to breath, and ask him specific questions. He has expertise in his area, and topics which are widely discussed here may not be so clear to him. Behave nice, and hopefully he will stay around, and both sides will learn. (Apologies for sexist assumption).
.
A comment to CB:
Your sentence beginning “Rather than assuming the worst about people and their motives…” applies to both sides of this debate, who hold strong and probably mistaken opinions of the other side.
Correction to the above. The correct link was to CB: Re: CB (#134), not AMac.
CB, instead of talking around Salzer et al. 2009, why don’t you give us your read of that paper? Lots of good questions asked there amidst the noise. Why not give us your answers? We love to learn.
I somehow left out one further question I had of CB on replicate sampling and that would be whether he is familiar with the criteria used to reject samples for use in chronologies and how frequently this would be required.
test – comment in reply to Ryan O and John M not appearing
I seem to have run afoul of some combination of site traffic and software designed to prevent the embarrassment of double posting with a rather lengthy reply to Ryan O’s questions in post #142. I will try appending it below this text and apologies in advance to the site moderators if the original attempt does eventually appear.
Re: NW (#158)
NW and Ken, a short answer is that the investigators need to combine evidence of tree-line fluctuations with estimate of past temperature changes to model the spatial extent of the upper treeline zone in order to avoid this problem (by ensuring that only material that was within this zone at any given time is used).
In the current study they make extensive use of remnant material that came from above current treeline and believe that overall they have avoided this problem to the extent possible.
Re: Kenneth Fritsch (#161)
The temperature response of trees is dictated by the principle of limiting factors. The temperature response declines down slope and transitions through a mixed response to one dominated by precipitation. Trees that live for thousands of years, as you point out, could grow in a spot that was within more than one zone over its lifespan (see above for a possible solution). Practically speaking it is possible to recognize distinct characteristics of growth that give some indication of the type of response to climate that is being represented.
Re: John M (#156)
Yes
Re: Ryan O (#157)
Ryan, I haven’t forgotten. There have been quite a few inter-related questions and I am attempting to provide substantive answers. There are some further up the thread than yours that I haven’t responded to yet. I took your question to be even a little more general/broad than your restatement above – and discussing it fully could take quite a little time and more than a single post. It is a little unfortunate that I am in the position of the lone gunman facing the militia here. My expertise does not cover the full range of issues under discussion – I am happy to offer my take, but cannot provide detailed answers to all aspects of the subject at hand.
ok…
Having done a short run through a number of threads here (including this one) I disagree – in fact many of the basic methods and assumptions of producing proxy records from tree-rings are being disputed. Small point though.
You need to be specific and depending what you refer to/want to discuss there is quite a bit that could be said. I made a start at addressing the relationship of temperature to tree growth at upper treeline. Did you have a problem with that as far as it went?
This is an important/interesting issue. It has come up for the bcp record only because a MLD based temperature reconstruction was being directly compared to it. A reasonable conceptual hypothesis for it is based on limiting factors – summer temperatures were a primary environmental parameter controlling cell wall thickening in conifer latewood at some sites but due to environmental changes, e.g. warming, other factors have become more important and obscured this relationship. I assume that some experimental work to examine this problem is planned or ongoing – typically such studies take several years to complete, and often a series of studies (different site types, species, variety of methods, etc.) is required to gain a fuller understanding.
The issue for proxy based reconstructions, as you note below, is whether similar changes occurred in the past. Proxy reconstructions assume that the basic controlling factors haven’t changed over the period of interest. How can we know this? Independent lines of evidence that validate the reconstruction are an important test. Other similar records should show the same basic features – at a minimum. For dendrochronology this means that a network of tree-ring sites with a similar environmental response should show the same basic patterns within a given climate region.
I know that this brings up something that has been discussed here WRT Keith Briffa’s decision not to use Fritz Schweingruber’s data to bolser his Yamal reconstruction. This is a tangent that could generate a whole separate discussion, but the short answer is that just because there is data available from some tree site doesn’t say anything about it’s appropriateness or suitability for a particular task. Tree selection, site selection, and sample size are competently dealt with in Hal Fritts’ Tree Rings and Climate linked above. [I, frankly, have not paid much attention to Briffa’s reconstruction, as my work is not directly related to it. However, I would assume his paper describes his methodology, selection criteria, standardization method, etc. As this work seems to have generated so much interest I will have a look at it in the next week or so.]
For those who might be interested Fritz’s website is here: http://www.wsl.ch/staff/fritz.schweingruber/
Other independent types of proxy records, historical accounts and records, etc. should corroborate prominent features of a regional tree-ring based proxy record. I would not consider conclusions based on a single record from one site to be verified and the longer the time span of interest the more important this becomes. A single record is a site history if it lacks independent context. The Yamal reconstruction, specifically, was not produced in isolation – it is not the first to reconstruct temperatures in this region, so its general applicability can be judged within the context of related work.
Back to divergence – there is a great diversity of evidence now available to suggest the the last half of the 20th century, and now the first decade of the 21st, is substantially different than the previous 5k years. The divergence problem fits within this context. We don’t have perfect knowledge, of course, but neither are we blind in this respect. Neither is the available tree-ring record universally afflicted with this particular problem.
http://www.agu.org/pubs/crossref/2007/2006JD008318.shtml
http://www.ingentaconnect.com/content/bsc/gcb/2008/00000014/00000010/art00017;jsessionid=2t008muuqrb88.alexandra
As to the question of whether the divergence issue might affect the proxies in the past – prior to the instrumental record – we can look at the Salzer paper and see that they conclude from their data that periods equivalent to the past 60 years have not occurred during the past 2,000 + years. While this is not conclusive evidence for the entire NH, it is important and not contrary to the view that divergence should not be an issue for that period.
There are ways to recognize and deal with these problems and any search through the literature will show both acknowledgment of them and solutions proposed and/or applied to particular studies where appropriate. You cannot make such a blanket statement except as a rhetorical device. Uncertainties are acknowledged and highlighted in much of the relevant work.
WRT a specific work the Salzer et al paper addressed the issue of competing signals (CO2 and precipitation) showing that the material they were using had very little precip (mainly limited to single very dry years) response and ruled out CO2 as a significant competing factor that might cause the rise in ring width.
An anecdote here: I was doing a tree-growth monitoring study examining the relationship of growth to precipitation in an area with a bi-modal precip regime. I had instrumented trees with mechanical recording devices that provided a direct measure of tree stem expansion. In collaboration with another person we added a set of electronic measuring instruments to the study trees that were set up with a datalogger. We hoped to eliminate the mechanical devices that required more personal attention (taking measurements, changing charts, etc.) At one point the electronic instruments showed a step change that was not reflected by the mechanical recorders and the electronic equipment designer asserted that the mechanical devices must be in error because they did not show the step change. Turned out to be an issue with the electrical ground circuit.
Measurement of complex systems is not a simple task – agreed.
Trees, as organisms adapted to a specific range of environmental conditions, necessarily respond to their environment by buffering their physiological systems within the tolerable range of external conditions to which they are adapted, and then by migration to new suitable habitat when conditions become intolerable – if they can. It is possible to characterize the tolerable range based on the physical evidence provided by current species distributions and infer the location of past suitable habitat based also upon physical evidence. One can then logically conclude that the prior distributions existed withing the tolerable limits known for the species. If the tolerable range WRT temperature seems small to you perhaps this is because describing ‘suitable environment’ with a mean value is a simplification that, while it does have some utility, is used as a stand-in for a suitable range of temperatures over the length of a growing season.
Reducing the issue to a cartoon noting only the potential problems can make the task seem intractable. For a look at a network of proxy data that is well verified the Cook et al PDSI grid is a good example:
http://www.ncdc.noaa.gov/paleo/usclient2.html
http://www.ncdc.noaa.gov/paleo/pdsi.html
I am very familiar with certain portions of this data, and while the variable of interest is not temperature perhaps it can serve as an example. The gridded data is very robust for the period 1700 to present over most of its extent. Within this period it is possible to resolve fairly detailed questions about regional differences in climate (with respect to drought). As one goes back in time the spatial resolution decays because the underlying data becomes increasingly sparse. I believe the current length is about 2,000 years and for the earliest portion it is impossible to resolve fine scale questions – in fact only the most general questions can be informed by the sparse grid of data points available and some regions are white space – no estimate possible – and errors of estimation are large for the available points. This example is for drought reconstruction and the precip contribution has high spatial variability.
Temperature has less spatial variability and fewer sites would be needed to produce reasonable estimates, but the same principles apply. The fewer the sites contributing data the more general the questions that can be addressed till with only a single site there there is only a point represented and we have lost all spatial resolution and context for evaluation. On the other hand it is possible with a few widely distributed points to examine general questions, and the more that is known about the relationship of those points to the information provided by a dense grid the better context we have for evaluating the information they provide. The more other types of proxy data we have the better context we can create. The efforts of the paleoclimate community to provide this context is very broad and covers a wide range of proxies.
In some respects organisms and their environmental response is to be preferred because the response is integrated and because said organisms are widely distributed. I think you are stuck on ‘degrees Celsius’ and the narrow view of ‘trees as thermometers’ (obviously they are not) and cannot see that 1 deg. C is a stand-in for much broader scale changes in ranges and durations.
The ‘pebbles’ reference had to do with looking specifically with the intent to find fault with a very narrow focus. I would reiterate that the research community has a great interest in improving both the underlying data available to answer questions and the analysis of existing data with respect to new information and perceived problems or shortcomings with previous work. An objective look at the published literature demonstrates both the awareness of the issues being discussed here and attempts to address them where appropriate/applicable.
Please see above.
The site is unresponsive – due to heavy traffic I assume – and I am aware that there have been a couple of additional responses to the earlier posts. Please have patience I will respond. I hope the text above has been helpful.
I will make an additional comment here about RCS (regional curve standardization). Above I gave a brief description of the method theory by comparing it to a measure of ‘site index’ or site potential. To expand just a bit: the idea that one might construct a single curve for standardization purposes that captures the average potential radial growth with respect to age for trees of a given species growing on a particular site has an inherent logic and appeal to dendrochronologists. Other solutions have the disadvantage of choosing an arbitrary form that may have a generic justification (e.g. negative exponential – fixed volume layers added to a cone or cylinder) or be simply completely data dependent (cubic spline or polynomial) with no biological assumption. Additionally preserving trends longer that the series length is impossible with the methods.
In my estimation the ideal application of this method is in situations where site conditions can be assumed to be stable, the stand has an open character so that between tree competition is minimal, and a large population of samples of varying age is available for a reference period for which the curve will be constructed. It will then be possible to compare growth during other periods with that reference period. Large sample depth during other periods is not then required (a robust curve to remove growth trend is already in hand) except in relation to the general coherence of the tree response to the environmental parameter of interest – if the trees are ‘noisy’ then a larger number of samples will be required to extract signal than if the signal is more coherent. This second issue of coherence would apply to all time periods.
Shade tolerant species and sites with closed stand conditions are poor candidates for this method.
Because there will always be an ‘oldest tree’, and overall a smaller number within the oldest age class, the curve estimate will be less robust toward the tail – but unless the behavior of the oldest trees is typically erratic this issue should be of minor importance.
I don’t seem to be able to successfully post substantive replies at the moment – will try later.
For those who might be interested Fritz’s website is here: http://www.wsl.ch/staff/fritz.schweingruber/
Other independent types of proxy records, historical accounts and records, etc. should corroborate prominent features of a regional tree-ring based proxy record. I would not consider conclusions based on a single record from one site to be verified and the longer the time span of interest the more important this becomes. A single record is a site history if it lacks independent context. The Yamal reconstruction, specifically, was not produced in isolation – it is not the first to reconstruct temperatures in this region, so its general applicability can be judged within the context of related work.
Back to divergence – there is a great diversity of evidence now available to suggest the the last half of the 20th century, and now the first decade of the 21st, is substantially different than the previous 5k years. The divergence problem fits within this context. We don’t have perfect knowledge, of course, but neither are we blind in this respect. Neither is the available tree-ring record universally afflicted with this particular problem.
http://www.agu.org/pubs/crossref/2007/2006JD008318.shtml
http://www.ingentaconnect.com/content/bsc/gcb/2008/00000014/00000010/art00017;jsessionid=2t008muuqrb88.alexandra
As to the question of whether the divergence issue might affect the proxies in the past – prior to the instrumental record – we can look at the Salzer paper and see that they conclude from their data that periods equivalent to the past 60 years have not occurred during the past 2,000 + years. While this is not conclusive evidence for the entire NH, it is important and not contrary to the view that divergence should not be an issue for that period.
There are ways to recognize and deal with these problems and any search through the literature will show both acknowledgment of them and solutions proposed and/or applied to particular studies where appropriate. You cannot make such a blanket statement except as a rhetorical device. Uncertainties are acknowledged and highlighted in much of the relevant work.
WRT a specific work the Salzer et al paper addressed the issue of competing signals (CO2 and precipitation) showing that the material they were using had very little precip (mainly limited to single very dry years) response and ruled out CO2 as a significant competing factor that might cause the rise in ring width.
For those who might be interested Fritz’s website is here: http://www.wsl.ch/staff/fritz.schweingruber/
Other independent types of proxy records, historical accounts and records, etc. should corroborate prominent features of a regional tree-ring based proxy record. I would not consider conclusions based on a single record from one site to be verified and the longer the time span of interest the more important this becomes. A single record is a site history if it lacks independent context. The Yamal reconstruction, specifically, was not produced in isolation – it is not the first to reconstruct temperatures in this region, so its general applicability can be judged within the context of related work.
Back to divergence – there is a great diversity of evidence now available to suggest the the last half of the 20th century, and now the first decade of the 21st, is substantially different than the previous 5k years. The divergence problem fits within this context. We don’t have perfect knowledge, of course, but neither are we blind in this respect. Neither is the available tree-ring record universally afflicted with this particular problem.
http://www.agu.org/pubs/crossref/2007/2006JD008318.shtml
http://www.ingentaconnect.com/content/bsc/gcb/2008/00000014/00000010/art00017;jsessionid=2t008muuqrb88.alexandra
Re: CB (#170),
How do you know the divergence is unique to the late 20th century? Tree ring history is written by the survivors. You study the dead as well?
For those who might be interested Fritz’s website is here: http://www.wsl.ch/staff/fritz.schweingruber/
Other independent types of proxy records, historical accounts and records, etc. should corroborate prominent features of a regional tree-ring based proxy record. I would not consider conclusions based on a single record from one site to be verified and the longer the time span of interest the more important this becomes. A single record is a site history if it lacks independent context. The Yamal reconstruction, specifically, was not produced in isolation – it is not the first to reconstruct temperatures in this region, so its general applicability can be judged within the context of related work.
Back to divergence – there is a great diversity of evidence now available to suggest the the last half of the 20th century, and now the first decade of the 21st, is substantially different than the previous 5k years. The divergence problem fits within this context. We don’t have perfect knowledge, of course, but neither are we blind in this respect. Neither is the available tree-ring record universally afflicted with this particular problem.
http://www.agu.org/pubs/crossref/2007/2006JD008318.shtml
http://www.ingentaconnect.com/content/bsc/gcb/2008/00000014/00000010/art00017;jsessionid=2t008muuqrb88.alexandra
As to the question of whether the divergence issue might affect the proxies in the past – prior to the instrumental record – we can look at the Salzer paper and see that they conclude from their data that periods equivalent to the past 60 years have not occurred during the past 2,000 + years. While this is not conclusive evidence for the entire NH, it is important and not contrary to the view that divergence should not be an issue for that period.
There are ways to recognize and deal with these problems and any search through the literature will show both acknowledgment of them and solutions proposed and/or applied to particular studies where appropriate. You cannot make such a blanket statement except as a rhetorical device. Uncertainties are acknowledged and highlighted in much of the relevant work.
WRT a specific work the Salzer et al paper addressed the issue of competing signals (CO2 and precipitation) showing that the material they were using had very little precip (mainly limited to single very dry years) response and ruled out CO2 as a significant competing factor that might cause the rise in ring width.
An anecdote here: I was doing a tree-growth monitoring study examining the relationship of growth to precipitation in an area with a bi-modal precip regime. I had instrumented trees with mechanical recording devices that provided a direct measure of tree stem expansion. In collaboration with another person we added a set of electronic measuring instruments to the study trees that were set up with a datalogger. We hoped to eliminate the mechanical devices that required more personal attention (taking measurements, changing charts, etc.) At one point the electronic instruments showed a step change that was not reflected by the mechanical recorders and the electronic equipment designer asserted that the mechanical devices must be in error because they did not show the step change. Turned out to be an issue with the electrical ground circuit.
Measurement of complex systems is not a simple task – agreed.
Trees, as organisms adapted to a specific range of environmental conditions, necessarily respond to their environment by buffering their physiological systems within the tolerable range of external conditions to which they are adapted, and then by migration to new suitable habitat when conditions become intolerable – if they can. It is possible to characterize the tolerable range based on the physical evidence provided by current species distributions and infer the location of past suitable habitat based also upon physical evidence. One can then logically conclude that the prior distributions existed withing the tolerable limits known for the species. If the tolerable range WRT temperature seems small to you perhaps this is because describing ‘suitable environment’ with a mean value is a simplification that, while it does have some utility, is used as a stand-in for a suitable range of temperatures over the length of a growing season.
Reducing the issue to a cartoon noting only the potential problems can make the task seem intractable. For a look at a network of proxy data that is well verified the Cook et al PDSI grid is a good example:
http://www.ncdc.noaa.gov/paleo/usclient2.html
http://www.ncdc.noaa.gov/paleo/pdsi.html
I am very familiar with certain portions of this data, and while the variable of interest is not temperature perhaps it can serve as an example. The gridded data is very robust for the period 1700 to present over most of its extent. Within this period it is possible to resolve fairly detailed questions about regional differences in climate (with respect to drought). As one goes back in time the spatial resolution decays because the underlying data becomes increasingly sparse. I believe the current length is about 2,000 years and for the earliest portion it is impossible to resolve fine scale questions – in fact only the most general questions can be informed by the sparse grid of data points available and some regions are white space – no estimate possible – and errors of estimation are large for the available points. This example is for drought reconstruction and the precip contribution has high spatial variability.
Temperature has less spatial variability and fewer sites would be needed to produce reasonable estimates, but the same principles apply. The fewer the sites contributing data the more general the questions that can be addressed till with only a single site there there is only a point represented and we have lost all spatial resolution and context for evaluation. On the other hand it is possible with a few widely distributed points to examine general questions, and the more that is known about the relationship of those points to the information provided by a dense grid the better context we have for evaluating the information they provide. The more other types of proxy data we have the better context we can create. The efforts of the paleoclimate community to provide this context is very broad and covers a wide range of proxies.
In some respects organisms and their environmental response is to be preferred because the response is integrated and because said organisms are widely distributed. I think you are stuck on ‘degrees Celsius’ and the narrow view of ‘trees as thermometers’ (obviously they are not) and cannot see that 1 deg. C is a stand-in for much broader scale changes in ranges and durations.
The ‘pebbles’ reference had to do with looking specifically with the intent to find fault with a very narrow focus. I would reiterate that the research community has a great interest in improving both the underlying data available to answer questions and the analysis of existing data with respect to new information and perceived problems or shortcomings with previous work. An objective look at the published literature demonstrates both the awareness of the issues being discussed here and attempts to address them where appropriate/applicable.
Please see above.
The site is unresponsive – due to heavy traffic I assume – and I am aware that there have been a couple of additional responses to the earlier posts. Please have patience I will respond. I hope the text above has been helpful.
I will make an additional comment here about RCS (regional curve standardization). Above I gave a brief description of the method theory by comparing it to a measure of ‘site index’ or site potential. To expand just a bit: the idea that one might construct a single curve for standardization purposes that captures the average potential radial growth with respect to age for trees of a given species growing on a particular site has an inherent logic and appeal to dendrochronologists. Other solutions have the disadvantage of choosing an arbitrary form that may have a generic justification (e.g. negative exponential – fixed volume layers added to a cone or cylinder) or be simply completely data dependent (cubic spline or polynomial) with no biological assumption. Additionally preserving trends longer that the series length is impossible with the methods.
In my estimation the ideal application of this method is in situations where site conditions can be assumed to be stable, the stand has an open character so that between tree competition is minimal, and a large population of samples of varying age is available for a reference period for which the curve will be constructed. It will then be possible to compare growth during other periods with that reference period. Large sample depth during other periods is not then required (a robust curve to remove growth trend is already in hand) except in relation to the general coherence of the tree response to the environmental parameter of interest – if the trees are ‘noisy’ then a larger number of samples will be required to extract signal than if the signal is more coherent. This second issue of coherence would apply to all time periods.
Shade tolerant species and sites with closed stand conditions are poor candidates for this method.
Because there will always be an ‘oldest tree’, and overall a smaller number within the oldest age class, the curve estimate will be less robust toward the tail – but unless the behavior of the oldest trees is typically erratic this issue should be of minor importance.
For those who might be interested Fritz’s website is here: http://www.wsl.ch/staff/fritz.schweingruber/
Other independent types of proxy records, historical accounts and records, etc. should corroborate prominent features of a regional tree-ring based proxy record. I would not consider conclusions based on a single record from one site to be verified and the longer the time span of interest the more important this becomes. A single record is a site history if it lacks independent context. The Yamal reconstruction, specifically, was not produced in isolation – it is not the first to reconstruct temperatures in this region, so its general applicability can be judged within the context of related work.
Back to divergence – there is a great diversity of evidence now available to suggest the the last half of the 20th century, and now the first decade of the 21st, is substantially different than the previous 5k years. The divergence problem fits within this context. We don’t have perfect knowledge, of course, but neither are we blind in this respect. Neither is the available tree-ring record universally afflicted with this particular problem.
http://www.agu.org/pubs/crossref/2007/2006JD008318.shtml
As to the question of whether the divergence issue might affect the proxies in the past – prior to the instrumental record – we can look at the Salzer paper and see that they conclude from their data that periods equivalent to the past 60 years have not occurred during the past 2,000 + years. While this is not conclusive evidence for the entire NH, it is important and not contrary to the view that divergence should not be an issue for that period.
There are ways to recognize and deal with these problems and any search through the literature will show both acknowledgment of them and solutions proposed and/or applied to particular studies where appropriate. You cannot make such a blanket statement except as a rhetorical device. Uncertainties are acknowledged and highlighted in much of the relevant work.
WRT a specific work the Salzer et al paper addressed the issue of competing signals (CO2 and precipitation) showing that the material they were using had very little precip (mainly limited to single very dry years) response and ruled out CO2 as a significant competing factor that might cause the rise in ring width.
An anecdote here: I was doing a tree-growth monitoring study examining the relationship of growth to precipitation in an area with a bi-modal precip regime. I had instrumented trees with mechanical recording devices that provided a direct measure of tree stem expansion. In collaboration with another person we added a set of electronic measuring instruments to the study trees that were set up with a datalogger. We hoped to eliminate the mechanical devices that required more personal attention (taking measurements, changing charts, etc.) At one point the electronic instruments showed a step change that was not reflected by the mechanical recorders and the electronic equipment designer asserted that the mechanical devices must be in error because they did not show the step change. Turned out to be an issue with the electrical ground circuit.
Measurement of complex systems is not a simple task – agreed.
Trees, as organisms adapted to a specific range of environmental conditions, necessarily respond to their environment by buffering their physiological systems within the tolerable range of external conditions to which they are adapted, and then by migration to new suitable habitat when conditions become intolerable – if they can. It is possible to characterize the tolerable range based on the physical evidence provided by current species distributions and infer the location of past suitable habitat based also upon physical evidence. One can then logically conclude that the prior distributions existed withing the tolerable limits known for the species. If the tolerable range WRT temperature seems small to you perhaps this is because describing ‘suitable environment’ with a mean value is a simplification that, while it does have some utility, is used as a stand-in for a suitable range of temperatures over the length of a growing season.
Reducing the issue to a cartoon noting only the potential problems can make the task seem intractable. For a look at a network of proxy data that is well verified the Cook et al PDSI grid is a good example:
http://www.ncdc.noaa.gov/paleo/usclient2.html
http://www.ncdc.noaa.gov/paleo/pdsi.html
I am very familiar with certain portions of this data, and while the variable of interest is not temperature perhaps it can serve as an example. The gridded data is very robust for the period 1700 to present over most of its extent. Within this period it is possible to resolve fairly detailed questions about regional differences in climate (with respect to drought). As one goes back in time the spatial resolution decays because the underlying data becomes increasingly sparse. I believe the current length is about 2,000 years and for the earliest portion it is impossible to resolve fine scale questions – in fact only the most general questions can be informed by the sparse grid of data points available and some regions are white space – no estimate possible – and errors of estimation are large for the available points. This example is for drought reconstruction and the precip contribution has high spatial variability.
Temperature has less spatial variability and fewer sites would be needed to produce reasonable estimates, but the same principles apply. The fewer the sites contributing data the more general the questions that can be addressed till with only a single site there there is only a point represented and we have lost all spatial resolution and context for evaluation. On the other hand it is possible with a few widely distributed points to examine general questions, and the more that is known about the relationship of those points to the information provided by a dense grid the better context we have for evaluating the information they provide. The more other types of proxy data we have the better context we can create. The efforts of the paleoclimate community to provide this context is very broad and covers a wide range of proxies.
In some respects organisms and their environmental response is to be preferred because the response is integrated and because said organisms are widely distributed. I think you are stuck on ‘degrees Celsius’ and the narrow view of ‘trees as thermometers’ (obviously they are not) and cannot see that 1 deg. C is a stand-in for much broader scale changes in ranges and durations.
The ‘pebbles’ reference had to do with looking specifically with the intent to find fault with a very narrow focus. I would reiterate that the research community has a great interest in improving both the underlying data available to answer questions and the analysis of existing data with respect to new information and perceived problems or shortcomings with previous work. An objective look at the published literature demonstrates both the awareness of the issues being discussed here and attempts to address them where appropriate/applicable.
Please see above.
The site is unresponsive – due to heavy traffic I assume – and I am aware that there have been a couple of additional responses to the earlier posts. Please have patience I will respond. I hope the text above has been helpful.
I will make an additional comment here about RCS (regional curve standardization). Above I gave a brief description of the method theory by comparing it to a measure of ‘site index’ or site potential. To expand just a bit: the idea that one might construct a single curve for standardization purposes that captures the average potential radial growth with respect to age for trees of a given species growing on a particular site has an inherent logic and appeal to dendrochronologists. Other solutions have the disadvantage of choosing an arbitrary form that may have a generic justification (e.g. negative exponential – fixed volume layers added to a cone or cylinder) or be simply completely data dependent (cubic spline or polynomial) with no biological assumption. Additionally preserving trends longer that the series length is impossible with the methods.
In my estimation the ideal application of this method is in situations where site conditions can be assumed to be stable, the stand has an open character so that between tree competition is minimal, and a large population of samples of varying age is available for a reference period for which the curve will be constructed. It will then be possible to compare growth during other periods with that reference period. Large sample depth during other periods is not then required (a robust curve to remove growth trend is already in hand) except in relation to the general coherence of the tree response to the environmental parameter of interest – if the trees are ‘noisy’ then a larger number of samples will be required to extract signal than if the signal is more coherent. This second issue of coherence would apply to all time periods.
Shade tolerant species and sites with closed stand conditions are poor candidates for this method.
Because there will always be an ‘oldest tree’, and overall a smaller number within the oldest age class, the curve estimate will be less robust toward the tail – but unless the behavior of the oldest trees is typically erratic this issue should be of minor importance.
Re: CB (#172), Re: Ryan O (#174), Re: CB (#176),
Perhaps me not being a native speaker prevents me from understanding stylistic subtleties, but in my view CB in your last response you might be reacting to a non-existent antagonism — Ryan O.’s post did not contain anything attacking you or any other dendrochronologists or even the climate change per se (by the way, as you will probably agree, global climate change is not equivalent to global warming). On the contrary, if I do not misunderstand the blog’s policy, your participation is very welcome and received with attention. Personally, I have found the information you provided very helpful to my current interest in the subject. At the same time, I support Ryan O.’s concern about the need for quantitative arguments; I find personal attacks unacceptable in a scientific discussion; finally, from my experience I think that exposure of one’s research to specialists in different fields can be extremely productive, to say nothing about the fact that as scientists are working for the humanity, speaking with educated human beings the scientist is speaking with the ultimate consumer of his/her work, which entails both sides to a mutual respect.
To me, the fact that old trees — some of the oldest on Earth — started growing fast at the treeline (and not elsewhere) is an intriguing fact deserving serious attention. From all I have learnt, claiming that it is temperature that made the trees grow faster, is premature. It cannot be taken as an explanation, just as a hypothesis. This hypothesis should be tested to be compatible with other firmly established evidence. By forcefully focusing on temperature, we can overlook other important evidence.
I have two concerns, one of which I have already expressed.
1. If trees at the treeline are limited by temperature, they should have lower growth rates than the trees below. After this presumed temperature-induced growth acceleration, the treeline trees might not develop ring widths wider than the trees below the treeline.
2. If only treeline trees can react to temperature, then it becomes of paramount importance to establish that these trees have remained “the treeline trees” for all the 2000 years period they are used as temperature proxies. First of all, since these trees are alive, this means that the treeline has not been lower than it is now. Second, if the treeline was higher at some point back in time, then the present-day treeline trees would not react to temperature at that time and cannot be used as temperature proxies.
I do not know what happens to trees if they suddenly become the treeline (e.g., due to the dieback of the upper strip of the stand). Perhaps they would show a sort of a release response? Due to increased water availability at the treeline? I simply do not know. What about winds and a possible change of circulation patterns? From what I know about boreal ecosystem, trees at the coastal treeline, for example, when exposed to predominantly northern cold winds like these, show a very clearly articulated response with all foliage developed at the southern part of the tree and only bare stems facing the wind. I would imagine that if the dominant wind changes its direction, one might see active growth at the northern part of the tree.
Re: Anastassia Makarieva (#177),
thank you for your 2nd concern.
this is exactly what i’ve been trying to understand & wondering if i was missing the obvious & being stupid.
Looks like it was a problem with one of the links I had in the text – the above #165 and #166 are a single post in reply to Ryan O.
CB: “growing season length” is not a physiological mechanism. What is the mechanism by which trees respond to temperature and what is your proof the temperature response is (a) linear, (b) not confounded by a moisture response?
.
P.S. You should ty searching CA for “growing season length” and see who’s name comes up first.
I am confident that, before AD1850, stripbarks do contain a record of decadal-scale growth season temperature variability. I am equally confident that, after that date, they are recording something else.
CB – “The evidence for accelerated climate change is all around us.”
Against what is the rate of climate change defined as accelerating? Apologies if you have already explained this – I’ve not followed the whole thread.
Re: curious (#181),
If you think temperatures of the past have large uncertainty, the rates of past warming are even more uncertain. Consider, for example, Briffa’s statement that AD1032 was the coldest year on record. Smack in the middle of the MWP.
Bender – Agreed, that is my layperson’s understanding but CB referenced “evidence which is all around us”. My attention was caught by CB’s dismissive and IMO antagonistic response to Ryan’s on point posts and hence I thought I might have missed something upthread.
Anastassia – I haven’t been following the whole thread but this caught my eye:
“I would imagine that if the dominant wind changes its direction, one might see active growth at the northern part of the tree.”
I think trees do respond to mechanical stresses of wind and I think there have been posts at CA mentioning this – simple beam strength requirements mean it will be in the tree’s “interest” to develop a thicker section inline with the applied force. A while ago I posted a question asking if there are any studies of tree rings in boughs or branches as they will be subject to a constant gravitational component. I didn’t see a response but I only follow at a distance. One of the dendro papers I looked at did refer to this phenomenon in trunk sections – I think they called it “compression growth” – (I think it was a paper Bender flagged up and I could probably find the reference). Sorry if this is stuff you already know or if it has been covered above/elsewhere.
Ryan O, I am very interested in obtaining what you have laid out as the pertinent aspects of the concerns that most of us here have about dendrochronology and dendroclimatology with respect to temperature proxies. It is the quantitative aspects and the CIs that we can place on chronologies. I think most of us accept and comprehend the qualitative rendition and state of affairs that we have received from CB.
.
What is the most basic part of these quantitative aspects? I say those aspects start with the tree ring measurements and the reproducibility of that measurement in replicate cores. Can we ask CB to start with those aspects and then move on from there to higher levels of quantification?
.
In passing I noted CB’s reference to tree ring/densities series having good coherence and that is something that I have recognized for some time, i. e. the series indicate that trees, that are within a reasonably close regional area, are reacting to the same something . This observation has been made by Rob Wilson also. Again though this is unfortunately a qualitative issue and, by example, one can see in the Yamal and Scweingruber series one that the bigger issue is not the matching of the peaks and valleys (they do match reasonably well) but the amplitude of those gyrations which tell different stories for the more current time period.
.
I would hope that CB tells us early on whether he can or wants to eventually provide us with any views on the quantitative aspects of dendroclimatology. I am a lot older than you Ryan and time is of the essence. (Don’t do smilies, but that is just me).
Re: Kenneth Fritsch (#188),
CB is only a little different from Tom P. Playing at authority. Dodging the hard questions. I can spot them from a mile away. Don’t get yor hopes up.
.
Ok CB, prove me wrong. Teach me something I haven’t already lectured on here at CA.
Re: Kenneth Fritsch (#188)
Ken, quantitative discussion might require specific data. The both the theoretical and practical aspects of quantitative methods are covered and summarized by Fritts 1976 Tree Rings and Climate. Standardization (curve fitting) summarized in Cook and Kairiukstis Eds. 1990 ‘Methods of Dendrochronology: Applications in the Environmental Sciences’
‘Coherence’ has both qualitative and quantitative aspects. People doing the primary work with dendro material say “the material crossdates well” to mean that the the year-to-year changes in growth match well from one sample to the next and patterns can be easily recognized across multiple samples. There is a high signal to noise ratio. This can easily be quantified if the material is measured for ring width or latewood density – for example. Trees can vary a great deal in their signal properties both within and between sites, that is, for various reasons (microsite, disturbance history, etc) some trees will have more of the target signal in their growth than others. At the site level, some sites – most random forest sites – will have relatively low ‘signal’ and high ‘noise’. Competition, lack of stress from the climate parameter of interest, stand disturbance (insects, fire, blowdown, disease, etc) – these factors will obscure climate response. Researchers interested in precip reconstruction will choose sites often near the arid limit of the species being sampled. Trees for which moisture availability is the primary factor limiting growth. Likewise trees growing at their (cold) temperature limit are sought for the purpose of reconstructing temperatures.
Latewood density is related to growing season temperature in climatic regions where cool summer temperatures may limit the cell wall thickening process that leads to higher density values. Forest disturbance and stand dynamics could interfere with signal coherence I would assume, but I don’t work with this parameter myself.
The basic measures of core-to-core and tree-to-tree coherence are usually characterized by simple correlation. You can see a typical example here in the text below figure 2:
Click to access KMB09_QuaternaryRes.pdf
Cores/trees are often rejected from a sample set because they exhibit signs of injury or disturbance. Leaning trees are cored at 90 deg. to the axis of lean to avoid compression or tension wood. Branches are not used except for special purpose studies. Open stands free from disturbance and the effects of competition are preferred.
Early researchers like Schulman selected only a few of the best samples for analysis, in part because measurement was a slow and laborious process and all calculation was done by hand or with a slide rule. Standardization curves were individually fit by hand.
Modern measuring equipment and statistical packages have allowed researchers to process relatively large quantities of data… BUT… quantity is no substitute for quality.
Pre-whitening by auto-regressive modeling is a step beyond this. I can discuss the conceptual basis for such methods and have opinions about their use to produce reconstructions.
Re: CB (#193),
Go away now.
Re: CB (#193),
As far as I can tell, I have only one ax to grind, and it is the mutually agreed upon “heart of the matter”. That ax is using trees as thermometers with accuracies that rival real thermometers in reconstructions. Without meaning to be rude or flippant, it seems to me that you are now deliberately avoiding this by discussing generalities and claiming that I am somehow making the conversation lose focus. Forgive me for being blunt, but how is avoiding the “heart of the matter” and redirecting the conversation to discuss the generalities – which is what you are presently doing – anything but losing focus? I understand that you seem to feel quite comfortable discussing the generalities, and perhaps you are not as comfortable discussing the specifics, but remember you were the one who claimed that the CA threads were chock-full of irrelevant criticisms and details. I find myself somewhat mystified that when I attempt to bring the conversation back to the “heart of the matter”, that you then feel free to claim that I am the one causing a loss of focus.
Then let’s discuss specific data. I assume you have some, since you entered into this discussion by stating that the concerns you’ve seen at CA amount to nothing more than overturning pebbles to support our pet theory. Since our pet theory happens to be that there is no legitimate way by which dendro series can be used to extract a temperature signal for thousands of years in the past to an accuracy of within +/- 1 degree celcius, it would seem to follow that you have said specific data at the ready.
[Moderator – if you have a chance to delete the redundant posts it would be appreciated – make the reading easier. Sorry for the mess]
Re: Ryan O (#186)
Then please accept my apology.
I agree, and I was not implying that confidence intervals could somehow be magically reduced in such a fashion. The MBH papers were not the subject of the current discussion either – the conversation can’t proceed effectively if it lacks focus and specifics – please don’t bring every ax you’re grinding to the table at once. My reaction was to your overgeneralized critique.
Logic and sound reasoning must underpin arguments – whether qualitative or quantitative. I agree that statistical characterizations of quatitative estimates are an important tool and must be appropriately applied to be meaningful.
I was saying that utility of proxy data to understand past climate was central to the discussion – a more general point.
A point I was trying to make about temperature in relation to tree-growth was that mean station temperature (monthly, seasonal, annual) was an inadequate measure to characterize the temperature influence on growth. Length of the growing season is roughly correlated with mean temperature and ‘degree days’ is perhaps a more biologically suited measure. Calculating such measures for high elevation sites (many forest sites really) over any length of time usually involves interpolation from available station data or use of algorithm generated estimates like PRISM. These clearly are not ideal solutions and accuracy of estimates varies from place to place. Snow melt and soil temperature are other related components.
You can see the statistical contributions of temperature and precip measures to bcp ringwidth via correlation in the Salzer et al figure 4.
From the tree data perspective coherence of the common signal in a data set gives a measure of the strength and coherence of external forcing in determining growth
Re: Anastassia Makarieva (#177)
Anastassia, why do you assume temperature limited trees would necessarily have smaller rings/lower growth rates than moisture limited ones? I have seen very slow growth rates in larch from northern Siberia, but as you can see in the Salzer et al figure 4a the growth rates of the high elevation trees are ~0.4mm on average compared to ~0.2mm for the lower, precip limited site.
I tried to address part of your question 2 above – about the position of the trees and remnant wood relative to treeline. I said that the samples came from at and above current treeline so the authors attempted to take that concern into account. I also said that an improved method, in my estimation, would be to model the treeline fluctuation through time so that a better segregation of the data could be achieved. I believe that such an attempt is being considered and the work has funding to continue for a few years.
As for wind effects – I cannot personally vouch for the extent of ‘flagging’ as an indicator of wind stress at all the sites – I have visited only Sheep Mountain. The trees there have some shelter from their position on the slope of the mountain. I have seen pictures of flagged trees on Mount Washington – for instance the one that appeared on the Nature cover for this article: http://www.nature.com/nature/journal/v307/n5947/abs/307121a0.html
Evident wind stress:
http://www.panoramio.com/photo/27762735
http://www.panoramio.com/photo/11295925
Such factors could add some noise to tree growth (reaction to mechanical stress similar to gravity induced ‘reaction wood’) however reaction wood in a sample is easy to recognize, and samples showing such effects are avoided when measuring material for climate reconstruction.
Re: CB (#191),
re-
I tried to address part of your question 2 above – about the position of the trees and remnant wood relative to treeline. I said that the samples came from at and above current treeline so the authors attempted to take that concern into account. I also said that an improved method, in my estimation, would be to model the treeline fluctuation through time so that a better segregation of the data could be achieved. I believe that such an attempt is being considered and the work has funding to continue for a few years.
this would be common sense approach surely!!
‘being considered’. this should be basic methodoligy, what are you on about (more grant money maybe).
are you telling me dendro’s do not do this as part of their survey.
Re: dougie (#203)
Dougie, I think you misunderstand. All of the current material used in the study is geo-referenced. An additional site was rejected because they did not have the samples mapped. What has not been done is to develop a model of climate fluctuations that can retrodict the upper treeline zone. This requires reasonable temperature estimates and would need to be verified based on what is already know about the fluctuations from mapping and dating wood.
There has been a longterm interest in mapping treeline fluctuations. LaMarche had his Campito Mtn. site surveyed and all wood and trees mapped and tagged in the 1970s. There is a continuing effort to map and tag wood in the Methuselah Walk area, but this has more to do with an attempt to locate material from 8-9,000 years ago and connect a floating segment with the main 8,600 yr chronology.
Re: CB (#204),
ok, i bow to your comments, lost me there!!
been trying to find yamal paper quote which talks about tree line being further north 800-1000 yrs ago, can’t find it!! help anybody.
if true, does this bother you CB, or is ok in your world?
CB —
Thanks again for coming and addressing dendro-related issues, and for providing links to what you view as the literature’s “best of” summaries. Given the apparent state of the dendrochronology and climatology professions, it shows considerable integrity that you’re here. I’m read-only on the methodology you describe, but I (along with other lurkers, I’d wager) am learning.
Further thoughts off the topic of strip bark analysis here.
CB – Thanks for the photos in 191. Re:
do you have any references for this? I’d be especially interested to see anything which looks at correlation between ring growth patterns in branches and boughs and those from the same specimen’s trunk. To my mind this could be a starting point for a calibration curve for mechanical stress?
Also I’d still like to hear your thoughts on the “evidence for accelerated climate change” which is “all around us”. Thanks
This blog’s in the toilet. So sad.
[CB — it looks like you had two “sets” of replicated posts above? I’m assuming the last in each set is the best one. I’ll try to work it out and delete the extras]
CB, you wrote:
Before I comment, a bit of personal introduction: I’m an engineer w/ scientific training, married to a field biologist/ecologist (marine specialty transplanted to Colorado so she’s learning new arenas 🙂 ). I happen to live within eyesight of Almagre so it isn’t too difficult to get up there.
First, a comment on the “scarring” discussion. I think two things got conflated, confusing the issue. Clearly, the stripbark trees are themselves damaged by some mechanism. On the other hand, nothing suggests that the living tissue has been scarred (in the way that frost/etc cause scarring.) Some onlookers may have been confused by this but I assure you those who have been here a while are not confused about that. What we *are* interested in is the stripbark response mechanism.
Second, I have two comments/questions on your statement:
1) I’m interested in your statement about Almagre stripbark “surges”. The way you said it suggests that such surges are not found at other stripbark BCP sites? We found rather extreme ‘surge’ anomalies when we collected data at Almagre, as Steve noted in the head post. It would seem the response curve is heavily dependent on exactly where in the sb portion of the tree one takes the sample. Are you suggesting these surges are uncommon in sb bcp? When archived data sets often contain only one sample per tree, how can anyone verify that assertion?
2) Sadly, Almagre data is a part of some reconstructions AFAIK. I certainly resonate what you’ve said about appreciating the trees. Unfortunately, the sense we get from some published climate scientists is that they take these proxy data sets and insert them into their analytical meat grinders with little if any appreciation or understanding of the underlying physical reality… and thus produce chopped beef results.
(One of the reasons I took so many photos on Almagre, and published both the galleries and panoramas, was to give at least some ability to visualize what’s going on for those who have never been to such a site.)
You also wrote:
3) Since bark-stripping tends to be an “event” on a given date, which significantly reduces the amount of live tissue to be supported, how do you adjust for these disturbances? I haven’t yet seen this accounted for.
Thanks for your time and interest.
Re: MrPete (#199)
Mr Pete, thanks for the help with the thread:
162 keep
163-66 delete
167 keep
168-73 delete
What a mess.
Surges like those are not related to climate but to disturbance response. They are not synchronous between trees and the data is very ‘noisy’. Treatment with a very flexible cubic spline would remove much of their effect – and also any trend the series might have. I am saying that material with those characteristics would normally be rejected for climate studies because the influence of individual tree variation would dominate a resulting chronology and that attempts to remove it during standardization would also remove any trends of interest. Sites are usually over-sampled in order to provide sufficient usable material. The weaker the signal of interest the more samples required. Standard procedure is sample two cores/tree from opposite sides. Strip bark trees may have only a single good sample point – center of the strip – and two would be redundant – extra work and damage for no significant gain.
I agree that this is a problem – more common than it should be, and not confined to tree ring proxies or necessarily ‘climate scientists’. I like the Salzer et al paper because the lead author collected material, visited all the sites, worked up a good bit of new material and is intimately familiar with the whole data set and is aware of its strengths and weaknesses.
Re your question 3 – the strip-bark form can be event mediated, but more typically in the Great Basin bcp is the result of progressive dieback. I have seen this occur also in Pinus flexilis, Juniperus scopulorum and Juniperus occidentalis – trees that can also live 1-2,000 years. If an event mediated strip stabilizes in form then there might be a depression in growth for a few years as a result of injury, but the subsequent growth should be ‘normal’
event mediated strip form in foxtail pine:
http://www.panoramio.com/photo/11300738
bcp stands, White Mtns. Ca:
http://www.panoramio.com/photo/12181403
Ryan O, I don’t have private access to the data used for temperature reconstruction. This is not my area of work and while I have professional opinions about the subject, it is not an area I typically focus on. I have worked with the material and some of the data from Sheep Mountain and have also worked with foxtail pine from the Sierras in Ca. Your focus has been on the temperature reconstructions, not mine. My work involves the study of disturbance and the production of primary data sets for use mainly by other researchers – most often drought reconstruction.
Are we agreed that trees at some sites can be sensitive to growing season temperature and reflect this in changes in ringwidth or latewood density? Do you agree with the basic conclusions in the Salzer et al 2009 paper?
Hal Fritts worked for many years to develop a tree growth model based on realistic physiology and environmental inputs. It grows conifer ‘tree rings’ a cell at a time to produce rings that could be compared with actual samples – ring width, cell sizes and wall thickness, etc. It is called PRECON and information and the model itself can be found here:
http://www.ltrr.arizona.edu/~hal/
It was considered overly complex and did not necessarily produce realistic results. The world is a big and varied place – hard problem. Interesting work.
A different model was developed by Vaganov, Shaskin and Hughes – you can read about it here in “Growth dynamics of conifer tree rings: images of past and future environments:
http://tinyurl.com/ydfl3fo
I think you might find it interesting.
Re: CB (#202), From your last post, I have three takeaways. You can correct me if I am wrong.
.
1. You have no intention of addressing the accuracy issue with using dendro series in temperature reconstructions because you do not have access to the data.
.
2. You have no idea how to place numerical uncertainties on the output of a calibration of tree rings to temperature.
.
3. You have no idea how to place numerical ranges on the “basic controlling factors” that would ensure that such a calibration remained valid over the entire timeframe of the reconstruction, nor do you have any idea how you would obtain numerical estimates for the actual conditions of these factors.
.
This will probably sound harsh, but there is no nice way to put this.
.
You come onto this site and post several times about how the proprietor of this blog and many of the commentors prefer to stroke their own egos and focus on irrelevant minutia that is immaterial to the conclusions of the papers we discuss. You speak as if the issues we raise (confidence intervals, calibration methods, standardization techniques, the divergence problem, CO2 fertilization, strip bark issues, et al.) are ones with clear resolution in the literature (which turns out not to be the case – and, well, we already knew that).
.
Yet the one key issue – the ONLY reason you see strip bark even MENTIONED on this site – concerning the ability to use trees as thermometers, you have nothing quantitative at all. Do you really think there would be anything dendro here if the chronologies weren’t used to make statements about past temperatures with a ridiculous level of certainty?
.
You seem to think that we should show more respect to dendros. I have an alternate proposal. What should be happening is that until #1 – 3 above are quantified, dendro-based temperature reconstructions should be tossed into the garbage can because they can neither be shown to be truly temperature reconstructions nor can the confidence intervals be shown to be at all realistic.
.
I have nothing against dendros; the field is pretty interesting. I understand that the vast majority are honest and hard working scientists. Given its potential to be able to tell us quite a bit about our past, I’m all for additional funding and work to bring the chronologies up to date, generate new chronologies, and figure out better means of processing them.
.
With that being said, I have zero patience for dendros who think that qualitative arm-waving is sufficient to explain how and why these chronologies can be used for multi-millenial temperature reconstructions that purport to show global temperatures within a degree.
Re: CB (#202),
The Salzer paper does not deserrve to be in PNAS. It does nothing to clarify the cause of the uptick in the highest elevation samples. It makes no sense for you to posit that it is “growing season length” when samples jsut a few metres downslope exhibit a completely divergent pattern.
.
And please reply to this:
Re: bender (#180),
[I notice that ClimateGate provided a bit of a distraction to what was a rather fruitful thread… and I never had a chance to follow up on CB’s comments…]
Re: CB (Nov 29 17:10),
You wrote:
and
I may be ignorant of some underlying reality, but both of these observations cry out for more explanation.
1) Why do you say injury elsewhere causes “a depression in growth” in the remaining strip? Our direct observations at Almagre were that growth is not depressed but increases in the remaining live strip when other bark is removed. Visibly clear as a growth pulse in the rings for event mediated stripping.** Presumably this is also true for progressive dieback?
2) We did some multiple sampling only a few inches apart on stripbark bcp’s, and found a wide divergence in growth. If there’s only a single good sample point… how can we demonstrate that the single point is any good? How can we correct for the mechanical impact on growth of the dieback/stripbark mechanisms?
3) If these ancient trees tend to go through non-climatic dieback/stripbark processes, and if growth tends to increase in the remaining live area, wouldn’t these combine to form a generalized “recent growth surge” signal among upper treeline — ie most-exposed — tree stands? How can we disaggregate this “mechanical” signal from climate-related factors? This seems a potential general confounding factor in upper treeline studies of all life forms affected by mechanical stripping/dieback processes.
[** This effect makes sense intuitively as well; AFAIK various studies have been demonstrating hormonal (?) communication among various parts of a tree regulates growth in response to such external factors.]
A proposal for improving the signal to noise ratio of CA…
What if we distinguished two kinds of threads:
“Open” – all comers, beginner questions welcome
“Pro” – no beginner commentary (perhaps a “side” thread can be created for peanut gallery discussions)
These would be helpful for visitors looking for the meat of the blog.
CB, I am not a stupid man nor am I a patient man, so I’ll give this one more shot.
.
I see little direct analyses and discussion of the reproducibility of measurements of replicate tree wing widths in the dendro literature. I have completed some difference series analyses of replicate core samples from the Yamal and Schweingruber series. Below I have linked the thread at CA where these analyses were posted and below that the pertinent tables and charts from that analysis including the Yamal series with my chart showing confidence intervals.
.
Do you care to comment? Or provide any links to where dendros discuss specific cases of replicate samples and the results obtained?
.
http://www.climateaudit.org/?p=7489
CB, I have post in moderation because it has a number links. In the meantime I want to remind you that I linked you previously to a thread where I did differencing of replicate tree cores and asked for a comments from you and any links you might have to papers where the reproducibility of replicate cores is discussed.
.
http://www.climateaudit.org/?p=7489
CB, if you need tree ring data from the Schweingruber and Yamal series, my post in the thread linked above contain links to both. If you cannot find the specific links let me and I will post them.
CB as a follow up here some links.
.
The Schweingruber series is here:
.
ftp://ftp.ncdc.noaa.gov/pub/data/paleo/treering/measurements/asia/russ035w.rwl
.
Yamal should be here but it is intermittent for my access of late:
.
http://www.cru.uea.ac.uk/cru/people/melvin/PhilTrans2008/YamalADring.raw
.
There are of course large repositories of raw tree ring data of which I assume you must be aware. Steve M has many linked and collated sources of tree ring data that unfortunately timeout for me when attempting to link of late.
CB, here is a large repository of raw TR data and chronologies – some chronologies have no corresponding raw data but many do. Rob Wilson’s TR data show replicate cores which appear to be designated by coring direction north, south, east and west.
http://www.ncdc.noaa.gov/paleo/indextree.html
Is CB coming back with some good stuff to answer the questions he was asked, or was he just another driveby whose dog ate his homework?
The former I hope.
Hughes, Oct 10, 2003: A further problem arises from the observation that radial increment may increase rather dramatically in the period after most of the bark dies back, but of course we don’t know when that was.
http://www.eastangliaemails.com/emails.php?eid=368
Bump. The miraculous thread.
18 Trackbacks
[…] I’ve prefaced my comments in the past with disclaimers about my non-science background, but this one is too rich. Apparently, if you drill straight into a tree, you get really skinny tree rings, and if you drill […]
[…] climate skeptic blog Climate Audit posted a link to the hacked files with a note reading: “A miracle just happened.” Schmidt noted that, “four downloads occurred from that link while the file was still […]
[…] climate skeptic blog Climate Audit posted a link to the hacked files with a note reading: “A miracle just happened.” Schmidt noted that, “four downloads occurred from that link while the file was still […]
[…] at the climate skeptic blog Climate Audit posted a link to the hacked files with a note reading: “A miracle just happened.” Schmidt noted that, “four downloads occurred from that link while the file was still there […]
[…] at the climate skeptic blog Climate Audit posted a link to the hacked files with a note reading: “A miracle just happened.” Schmidt noted that, “four downloads occurred from that link while the file was still there […]
[…] short while later, another user, likely the same RealClimate foreign user, posted a comment to an article on the blog of Steve McIntyre (ClimateAudit) at 7:24am EST from a Russian IP address. […]
[…] uploaded to realclimate in the very early morning (Eastern). At 7.24 am Eastern (5.24 blog time), a comment by “RC” was placed at Climate Audit saying that “A miracle has happened” with a hyperlink […]
[…] in the process. At this time none of us knew of the upload to realclimate.org or had noticed the “a miracle just happened” post on climateaudit.org. We believed we were exclusively sitting on The Pentagon Papers of our […]
[…] tid efter skrev en anden bruger, sandsynligvis den samme ukendte bruger, en kommentar til en artikel på Steve McIntyres blog (ClimateAudit) kl. 7.24 am (13.24 dansk tid) fra en russisk […]
[…] bender’s final comment on the thread says it all: Bender Posted Nov 29, 2009 at 9:30 AM | Permalink | Reply […]
[…] necessary-reading Neverending Audit website, I came across the thread at CA on stripbark: Miracles and Strip Bark Standardization and the infamous CB. It's a […]
[…] problems of strip bark standardization were being discussed in the thread where Climategate was first mentioned – a thread which contains relevant illustrations of the […]
[…] including several from your star performer Michael Mann, were released by the hacker/leaker, alluding to this very cartoon: “A miracle just […]
[…] to the file (and I confirm that I had not previously noticed the significance of the comment here at 5.24 am blog time (7.24 am Eastern) where the name of the poster “RC” (identified as […]
[…] sent me off on my latest “detour” which I began at Steve McIntyre’s Nov. 16/09 Miracles and Strip Bark Standardization” which he had whimsically concluded […]
[…] Durban che inizia domenica, come due anni fa per quello di Copenhagen. Prima l’annuncio del miracolo da quelli che equiparano misconduct scientifica e pedofilia. Poi l’indirizzo del server russo […]
[…] di Durban che inizia domenica e con le stesse modalità, un certo “RC” annuncia un miracolo sul blog di Steven McIntyre, “revisore del clima” e consulente di aziende […]
[…] di Durban che inizia domenica e con le stesse modalità, un certo “RC” annuncia un miracolo sul blog di Steven McIntyre, “revisore del clima” e consulente di aziende […]