A couple of days ago, Greg Laden published a response from Malcolm Hughes to my recent Sheep Mountain article. In today’s post, I’ll show that the “response” was both unresponsive and absurd.
In my article, I had compared recent (“out-of-sample”) values of the Sheep Mountain chronology as shown in a detailed figure in Salzer et al 2014 to the corresponding values of Graybill’s Sheep Mountain chronology, a series discussed on numerous occasions here and in our 2005 articles, since it was the most dominant series in Mann’s North American PC1, which in turn was the most dominant series in the Mann et al 1998,1999. (Indeed, the contribution of all proxies other than bristlecones to the Mann et al reconstruction was indistinguishable from noise,)
In making this comparison, I emphasized the importance of out-of-sample testing as a means of validating a proxy reconstruction. I showed that, after 1980, the bristlecone chronology declined dramatically, while NH temperatures went up.
In response, Laden wrote as follows:
Climate science denialist Steve McIntyre has also weighed in on Salzer et all’s research. His post is truly mind numbing, as he treats Salzer et al as a climate reconstruction paper, and critiques it as such, but the paper examines the methodology of tree ring proxy use and the ecology of tree rings. McIntyre shows the same figure I show above (Figure 5 from that paper) and critiques the researchers for failing to integrate that figure or its data with Mann et al’s climate reconstructions. But they shouldn’t have. That is not what the paper is about. Another very recent paper by the same team is in fact a climate reconstruction study (published in Climate Dynamics) but McIntyre manages to ignore that.
First, I reject the idea that it is somehow an error, let alone a “mind numbing error” to take tree ring chronology information from one article on Sheep Mountain and compare the information to earlier articles. If the authors neglected to do the comparison themselves, I’m entirely entitled to do so.
Nor do the results shown in the Climate Dynamics paper change one iota of my criticism. So it was very misleading for Laden and Hughes to imply that it does. I’ll show this step-by-step.
First, here is the chronology shown in Salzer et al 2013 (Clim Dyn).
Figure 1. Salzer et al 2013 Figure 3 Top Panel.
The above figure is considerably smoothed. Hughes, Salzer and coauthors did not archive the chronology from this figure, but did archive (August 2014) a temperature reconstruction that is a smoothed linear transformation of the above chronology. I estimated the change in scale from the graphic and show that the chronology in the figure and the archive of the smoothed reconstruction match up to linear transformation, other than a somewhat puzzling tail shown in the graphic, but not present in the archive.
Figure 2. Comparison of linear transformation of archived reconstruction (yellow) to Salzer et al 2013 Figure 3 detail.
In the next figure, I’ve overplotted onto the Salzer et al 2013 figure in a similar style as I had used for the Salzer et al 2014 figure. In the left panel, I’ve shown the Graybill Sheep Mountain chronology as used in Mann et al 1998. On the right, I’ve shown the Graybill Sheep Mountain chronology and the Salzer et al 2014 South chronology both in red, comparing them to the Salzer 2013 smooth (black-yellow) and the scaled HadCRU NH temperature that is the “target” of the Mann reconstruction.
Some points are obvious regardless of cavils from Laden or Hughes. Under the presumed relationship between temperature and the Graybill Sheep Mountain chronology, values of the chronology should be nearly triple their long-term mean. However, this hasn’t happened. There are two effects here.
First, even the Graybill chronology (to 1990) went dramatically down after 1980. While the Salzer chronology doesn’t show the dramatic decline of the Graybill chronology, this is because it doesn’t have the same dramatic increase. The Graybill chronology in 1980 is dramatically higher.
Third, the values shown in the archive are highly smoothed, diminishing the information on post-1980 performance. In addition, the archived version of Salzer et al 2013 ends in 2006, while the information shown in Salzer et al 2014 ends in 2010. This may or may not be relevant to the puzzling discrepancy between the downtick shown in Salzer et al 2013 Figure 3 and the archived data.
Figure 3. Annotated version of Salzer et al 2013 Figure 3 top panel. Left. 1000-1980 showing Salzer 2013 chronology (black-yellow) compared to Graybill chronology to 1980, as used in Mann et al 1998. Right – 1875-2010 detail. The Graybill chronology is shown up to 1990 together with the Salzer et al 2014 South chronology as digitized. Magenta- HadCRU NH scaled to the mean and standard deviation of the Graybill chronology over 1902-1980 calibration period.
The Mann reconstruction, either intentionally or by accident, ends in 1980, pretty much exactly at the top of the market of the Sheep Mountain chronology and did not show or discuss the divergence between the bristlecones and temperature that had already occurred in the 1980s. Instead, Mann has repeatedly asserted that his proxies tracked temperature to 1980 and were not impacted by the post-1960 divergence problem of the Briffa reconstruction. Laden repeats this talking point as follows:
Note that there wasn’t a “divergence problem” in Mann et al in the sense of Briffa et al. Mann et al match the observational record very well through 1980, which is the end of the calibration interval (owing to the fact that many proxies drop out after 1980). This is something else the deniers tend to get wrong; they try to conflate the Briffa et al post-1960 divergence problem Mann et al’s hockey stick work. There is no such issue with that work, in that there was no detectable divergence through the end of the calibration interval.
I, for one, have never conflated the issues of the Briffa reconstruction with the issues of the Mann reconstruction and indeed have, on numerous occasions, observed that the inconsistency between these two tree ring based reconstructions was an issue that specialists ought to have addressed more squarely long ago. (And perhaps would have done so had the discrepancy not been concealed in spaghetti graph comparisons which hid the decline in the Briffa reconstruction.) However, if the key proxies in the Mann reconstruction suffer from a post-1980 divergence problem that is even more severe than the post-1960 Briffa divergence problem, I don’t see that this offers any comfort to those pointing to non-divergence up to 1980. Quite the opposite.
In his peroration in Laden’s post, Malcolm Hughes stated:
Back in 1999 we (Mann et al) made the best available choices with the information and data we had. Now, more than 15 years later, with a Bristlecone Pine record that extends back 5000 years, the original results hold up remarkably well.”
I am as aware as anyone that it’s easy to criticize people with 20-20 hindsight. However, even at the time, there were many caveats attached to the validity of stripbark bristlecone chronologies and it is far from evident that the data met the assumptions of the methodology.
But even if that was the “best” that they could do at the time, the proxy has not performed out of sample. End of story. It is beyond ludicrous to say that a proxy that has gone dramatically down since 1980, while NH temperatures have gone sharply up has performed “remarkably well”. Those are the words of grant-seekers, not scientists.
Climate Audit 
140 Comments
I think you are missing the point entirely. You need to understand that any proxy is subject to internal variation, external effects that make it less sensitive (or entirely useless), etc. This is true of proxies, and it is true as well with instrumental data. If your thermometer breaks, you don’t use the data from it any more.
Also, my post was not a response by M.H. to you. It was a post about recent research placed in the broader context of tree rings, and even more broadly, proxies, in which I had a couple of quotes from some of the researchers.
greg do you seriously think that tree ring are the equivalent of thermometers?
Can you declare with any certainty that trees were functional and accurate proxies fro temperature too years ago?
of course you can’t.
Greg, are you seriously claiming that the Sheep Mountain chronology performed “remarkably well” as a temperature proxy after 1980?? If you’d put your life savings in 1980 into a fund indexed to the Sheep Mountain chronology, would you say that it performed “remarkably well”? Of course, you wouldnt.
I am highly aware of the limitations of various proxies. One of the difficulties that is systemically ignored by the multiproxy jockeys over the years is that ex post data snooping runs into the same methodological problems as “systems” to predict the stock market. The analogy is not complete, but there are elements.
In stock market systems, people propose relationships which seem to work in sample, but fail out of sample. The new indicators are proposed and they too fail out of sample.
I see the same sort of problem in Mannian-style proxies, such as the bristlecones or Kaufman’s muds. I don’t preclude the possibility of developing newer and better proxies. For a while, I thought that alkenones might do that, but they have their own frustrations.
However, performance such as we;ve seen from Sheep Mountain disqualifies it as a valid proxy and it would be best if specialists just conceded the point rather than look foolish.
Data snooping is virtuous, if it is for a noble cause.
He’s not saying it performed remarkably well, exactly the opposite. He’s saying it performed ‘remarkably well’ (where did you get that quote not in Laden’s comment?) until then.
Steve: Hughes says: “Back in 1999 we (Mann et al) made the best available choices with the information and data we had. Now, more than 15 years later, with a Bristlecone Pine record that extends back 5000 years, the original results hold up remarkably well.” I interpret that as meaning to state that their belief that the proxies chosen 15 years earlier, including and in particular Sheep Mountain, “hold up remarkably well” – a conclusion that I do not believe to be justified. If he wanted to say that the bristlecones had diverged from temperature over the past 15 years, then that would have been easy enough to say.
The phrase “hold up remarkably well” could allude to the mbh reconstruction as other reconstructions have extended a ‘temperature history’ further back in time. Or perhaps it alludes to bristlecone pines’ correlation to other temp reconstructions further back in time?
Request of Steve or other readers – As I mentioned in previous posts, several years ago, I located a table / data set/report for the bristlecone pines a listing by year going back to the 1400’s (or earlier) which listed by year, the difference in temp from the base year in a format similar to: 19×0 +1.20, 19×1 +1.09, 19×2 +1.28. However due to my definiency in internet search skills, I am unable to locate the file or a similar data file. Can anyone point me to a source data. Thanks
one other point – in your article, you raised Salzer et al 2013 implying that it would refute the claims made in my blog article. It is evident that there is nothing in Salzer et al 2013 that refutes the blog article. Are we agreed on that?
“If your thermometer breaks, you don’t use the data from it any more.”
If your model breaks, you don’t use it any more. Unless you’re a climate scientist.
Mmmmh, thermometers are fragile, models even more. Trees themselves are more resistant.
The argument is sophistry. The apt comparison for alleged tree ring response to temperature is to the response of mercury to temperature. Broken thermometer, broken tree..so what? Given Mr. Laden is not a climatologist I think this argument here is an authentic meme from developed by the dendros or related pr firm peeps to rationalize the attaching of (massaged) thermometer data to cropped “proxy” data. What else do the dendros got to keep their share of the scare-money flowing? The non-treeline stuff from Australian Climate Science Ltd. is too whack.
“I think you are missing the point entirely. You need to understand that any proxy is subject to internal variation, external effects that make it less sensitive (or entirely useless), etc. This is true of proxies, and it is true as well with instrumental data. If your thermometer breaks, you don’t use the data from it any more.”
Ok – I think I’m now in agreement that things are worse than we thought.
Laden: ” If your thermometer breaks, you don’t use the data from it any more.”
Forgive me, but isn’t that exactly the point that Steve McIntyre makes? The proxies as dervied from Sheep Mountain and the bristlecones, and the Briffa divergence problem all argue that the tree ring proxy data thought to be good thermometers have shown not to be, when taken in their entirety.
Ergo, the proxy thermometer IS broken, so why would Laden be arguing that it has not? Nothing Laden is saying makes sense, wrapped around such logic and evidence.
Craig,
The defect in thinking is improbably common, as are the attacks on those who recognize the defect.
Greg, you’re saying that certain trees represent you perfectly when their data happens to match your hypothesis, but they (almost miraculously) stop representing you when their data does not match your hypothesis. Then you offer NO explanation of the actual tree-based process change that flips the tree from one mode to the other. Natural variability, you say? What causes the natural variability? Did some Tolkien-esque Ents perhaps convince the trees to change? What stops it from being natural variability all the way down? It appears to me that you can’t even state your hypothesis.
All of which makes you the very model of a modern post-normal climatologist.
gregladen,
“I think you are missing the point entirely.”
Hold that thought until you next face a mirror.
Greg: the problem is prior to 1900, how you do know if the “thermometer” broke (worked)? How do you know if it was just right by randomness over the 1900-1980 period? (i.e. a broke clock is right twice a day)
That is the problem now with saying it broke after 1980. You then have to concede the possibility it was broken for many periods in the past.
Exactly. And this is the exact problem that D’Arrigo and Briffa and others recognized back around 1990 – that if the proxy=thermometer doesn’t work in the present, then ALL data reconstructions connected to tree ring proxies become suspect.
The upshot SHOULD be that dendroclimatology has lost its fundamental underpinnings.
Briffa and D’Arrigo were quaking in their boots (and still may be). And Briffa, knowing how big of scientific misconduct it was, resisted Mann’s dictate to “Hide the decline” right up until the end, when he shut his mouth and did it like Mikey told him to do.
Greg – you make the following statement in your blog article – “The tree rings from certain sites seem to properly reflect temperature variability up until around 1960, and after that, the usability of the signal from that proxy can be reduced. This pattern has been noticed in a number of different tree ring records; the phenomenon is widespread enough that it has a name. It is called the “divergence problem.”
However, based on the proxy reconstructions of the bristlecones pines that I have seen, they did not track temps during the heat wave of the 1930’s very well. So 10 years out of sample, plus the post 1960 period adds up to approx 50 years not tracking temps very well which is approx 40-50% of the time period in which available instrumental record is available. Can you point to any data that shows better correlation of temps during both the 20’s and 30’s.
Honestly, I thought it was a joke when somebody first told me that hide-the-decline was merely about “the well-known divergence problem”. That sounds like a parody, like when we used to say that our answer and the right answer “differed only by a constant”.
Exactly. It was a correlation that they THOUGHT held, but has shown that it doesn’t.
In Laden’s own words, “If your thermometer breaks, you don’t use the data from it any more.”
To me this says that tree rings as a proxy for temps aren’t good now, and probably weren’t EVER – there was a “sort of correlation” that they jumped on, but one that they all should be good scientists about – and simply stop using it.
Are you trying to say that bristlecones used to be good thermometers but broke 1980? If so why?
Observer effect? On steroids…
Laden…at least you don’t use your sneering, insidious “denier” quips here. Why do you use such invective?
It’s a bonding thing.
“If your thermometer breaks, you don’t use the data from it any more.”
That’s not the analogy you needed here
The trees did not “break” 35 years ago
Instead, as our host makes “remarkably” clear
You should have used one like this, posted below:
“If your thermometer turns out to have been broken, you don’t use the data from it at all.”
==============/ Keith DeHavelle
OT – nice to see you back Keith 🙂 hope all is well(as can be)
Thank you; I am as well “as can be”
But widowed and wheelchaired; much happened to me:
http://level-head.livejournal.com/625504.html
==============/ Keith DeHavelle
Very moving Keith.
“If your thermometer breaks, you don’t use the data from it any more.”
Incredible. Just……incredible.
“If your thermometer breaks, you don’t use the data from it any more.”
Mr. Laden, can we all agree that if nearby working thermometers–those at stations operated by the US Government for example–are not analyzed in light of claims of bristlecone temperature sensitivity, such a study would be inadequate, correct?
Your use of the D word rendered your diatribe moot and pointless. No one forced you to falsely use the word. It was your decision alone.
I think you just said that post-1980 the tree-ring (proxy) thermometer is “broken”. Since we are looking at times before thermometers, we need proxies that are consistent with actual temperatures: a post-19080 or post-1960 break from temperature records says that at a minimum you should reject a solid reflection of tree-ring data for temperatures. At a minimum. Beyond that, you should say some trending is probable but not all; tree-ring proxies are potential reflectors of temperatures, but not reliable.
If this had been said in the beginning, nothing more would have been done. Including the Mann-Stynn lawsuit. The dispute is about claiming certainty – and using it for political and financial purposes – when that certainty was not true. And known by Mann et al, if not in 1998, then shortly afterwards, despite the continued use of the Hockey Stick.
Yep. As Richard Feynman says in his class on the Scientific Method said, if the test you derive from your guess (hypothesis) produces evidence/data that does not agree, then your guess/hypothesis is WRONG.
Dendroclimatology in its entirety is a hypothesis guessing that the tree rings are proxies for temperatures. That SEEMED to be the case – with some of what Laden refers to as “internal variation, external effects that make it less sensitive (or entirely useless), etc.” – the “wow” in the correlation during that 1930s) – indicating some weakness in the results.
By 1990 Briffa and others saw that weakness becoming a SERIOUS non-correlation.
Anyone arguing NOW that there is any correlation – Hey, folks, it doesn’t work! Give up!
Oh, but they can’t DO that, because it will ALL come tumbling down if they do.
All Laden can do is point at the time when it DID correlate and ignore the post-1960 failure to correlate. What kind of logic is that? That when the guess/hypothesis isn’t right anymore, that it is okay, because it USED TO?
Somebody teach that man the scientific method, PLEASE.
“You need to understand that any proxy is subject to internal variation, external effects that make it less sensitive (or entirely useless), ”
So Greg Laden is conceding that tree ring proxies are entirely useless. If they’re useless, why is he using them?
So what “broke” the trees on Sheep Mountain in 1980?
PS You should look up the logical fallacy “begging the question”.
Greg Laden,
On a previous thread a few months ago you upheld the method of tacking a modern temperature record at the end of a proxy series like Mann has done so notoriously (and without giving any notification of the use of such a technique).
Have your views on such methods changed or do you still adhere to them?
After reading Mr Laden’s article and his response to you, I am underwhelmed. It strikes me that the most apposite thing to say is he went into a battle of wits unarmed.
Knife. Gunfight.
Tree-ring. Thermometer.
“You need to understand that any proxy is subject to internal variation, external effects that make it less sensitive (or entirely useless), etc. This is true of proxies, and it is true as well with instrumental data. If your thermometer breaks, you don’t use the data from it any more”
The critical question to answer for any proxy is how well did it perform as a “thermometer” over a long period of time in the past. The only test available to answer that question is to determine how well it performs over a short period of time in the present. This is done by comparing the proxy temperature predictions to the known temperature measurements over that short time period. If it tracks well during the entire present time period, it is possible that it may have tracked that well in the past, but not certain. If it only tracks well for a portion of the present time period due to internal or external effects that change it’s sensitivity, it is then reasonable to assume that it likely only tracked well for a portion of the past for the same reasons. It would then be considered unreliable as a thermometer because you have no way to know when it didn’t track well in the past. I believe that Steve is just proving by this test that the Sheep Mountain chronology is a broken thermometer because it only tracks well for a portion of the present time period. It cannot then be relied upon to measure temperature in the past.
Duh, but with an adder:
Sheep Mountain, as one proxy location, should also give dendroclimatologists doubts about using ANY tree rings. That OTHER tree ring sites ALSO have the divergence problem (non-correlation) should have at least SOME of them asking, “WAIT A MINUTE! Maybe this proxy isn’t good, after all.”
That they do NOT – how is anyone supposed to have a rational discussion with someone who does not ask that question?
Greg Laden wrote, “At around 1960 the ability of the tree ring data to represent regional temperature declines and the tree rings become useless. Prior to that time the data should be used. After that time the data should be discarded.”
http://scienceblogs.com/gregladen/2014/09/22/steve-mcintyre-misrepresents-climate-research-history/
What’s strange to me about much of climate science is that hypotheses are asserted as facts – it is manifest in the paleoclimate work about the past climate as well as the modeling work about the future climate. I think it is POSSIBLE that certain select tree rings were ‘effective’ temperature proxies until some date in the 20th century, though I have not seen any science PROVING the multiple layers of hypotheses in that statement. As many here observe, an excellent test of the hypothesis is out of sample performance, for which the hypothesis has failed.
Dear Uncle Ebenezer,
As I wrote last time my tulip bulbs market predictor was doing wonderfully well but unfortunately it all crashed out of sample recently. Can I have an advance on my next 5 yrs stipend to get it all back?
Your loving nephew
Mortimer
Seriously? It could be time to bring back the no-longer-host-approved “It’s not — it’s —” tagline.
Sooo…. from what I read from Laden’s publication, all critiques of this proxy are irrelevant and nobody had anything useful or reasonable to say about these obviously perfect thermometers? The fact that they don’t seem to correlate or even loosely track actual “temperature” is something only a climate “denialist” would consider.
I lost IQ points from that read, and I didn’t have any left to spare.
Jeff,
Your problem is a serious lack of faith. It would surely all make sense if you just believed hard enough. So get on that unicorn, snort a full dose of that magic pixie dust, squeeze your eyes tight and ride with the fairies all the way to the land of climate alarm enlightenment.
Otherwise, you’re stuck with nothing but facts and logic. And they will never get you to that promised land.
Perhaps the issue is treemometers know when you are examining them for ecological studies versus temperature: Clearly dead tree-samples are self aware.
Otherwise the criticism by Laden that McInyre “treats Salzer et al as a climate reconstruction paper, and critiques it as such, but the paper examines the methodology of tree ring proxy use and the ecology of tree rings” doesn’t make even the slightest sense.
Normally I would consider the examination of “the methodology of tree ring proxy use and the ecology of tree rings” a study of the metrology of tree rings. Obviously that directly impacts any down stream applications of tree rings, including, for example, temperature proxies.
Here’s a video celebrating the “remarkable” performance of Sheep Mountain stripbark bristlecones as a temperature proxy after 1980. A Climate Audit reader tries to explain the poor out-of-sample performance.
Oldie but Goodie Steve 🙂 you old “Climate science denialist”
well done Greg,thanks for another BS/PR tag to remember you for.
Funny and to the point. Which is surely why any dehumanisation implicit here is just fine, unlike Greg Laden’s earlier?
Actually a serious question. I consider the opening up on “climate science denialist” on this thread and the zamboni sparing Mosher on the last, allowing us to address some of the politics, as something akin to Climate Audit Christmas Specials. Not that I’ve given my view on either, apart from the flip comment that denialist language is really a way of bonding with others with whom one wishes to feel one agrees. Or perhaps not so flip. Bonding. Christmas. Perhaps I’ll come back to it.
The monkey’s are always good fun but it is truly disappointing that their tongue-in-cheek antics are an accurate analogy of Laden’s argument.
It’s like arguing with the advocate lawyer. There is always an argument, whether it makes sense or not is not really relevant to the existence of the argument, and we shouldn’t take the fact that the argument is made as an indication of character of the lawyer making the argument. After all, it is simply the lawyer doing his job and it is the jury’s job to judge the veracity.
In this case, it is supposed to be scientific discussion, we are supposed to all be able to admit when truth is truth and in the case of the out-of-sample data, the truth is that the trees don’t match temperature. They don’t. These were previously preferred trees, chosen from many other trees by multiple processes for their previous match to temperature, and their newly updated data revealed divergence from temperature which precludes their use as thermometers. We know it, Laden knows it and even the effing monkeys in the video know it.
Laden’ knows full well that the processes to “chose” the right trees is deeply in question. He also knows that the now obvious answer, basically kills the use of this data in dendrothermometry. It doesn’t just call past studies mildly into question. So his fatuous, flatulent argument, is Fake, False and other completely synonymous F words we don’t use in science blogs, falls flat…
I don’t know if that comment is particularly helpful. It was fun to write, and I do believe it, but maybe a bit of scissor action is in order.
Jeff: “In this case, it is supposed to be scientific discussion, we are supposed to all be able to admit when truth is truth and in the case of the out-of-sample data, the truth is that the trees don’t match temperature. They don’t.”
As I said a few minutes ago, if someone insists that something that is wrong should still be used, how in the HECK are we supposed to deal with such idiocy?
Truth is truth. And wrong is wrong. Laden is supposed to be a scientist and yet he is arguing that Mann’s work should still be used when the data used is no longer reliable (not true)?
WOW…
Jeff Id (10:35 PM):
It also got this reader thinking about the original video. With the music turned off and the graph put the correct way up the monkeys go silent and stationary. That’s what was needed from the IPCC and science academies and didn’t happen. That’s actually far from funny. I’d vote for the zamboni not to used in this case. If I had a vote. 🙂
Jeff Id,
I’m not sure Laden knows. In this world of filter bubbles, both internal and external, a lot of evidence is simply not heard and logic went the way of Jimmy Hoffa.
The world we live in is truly surreal when it comes to our old fashioned notions of critical analysis and logical deduction.
Ring chimered models parade in the heat,
To think that I saw it on Bristlecone Street.
========================
There is a nice write up of the correlation problem by the Financial Times’ Tim Harford here
http://timharford.com/2014/11/finance-and-the-jelly-bean-problem/
The jelly bean problem is to find what colout jelly bean produces acne, when you have an almost limitless supply of colours….
milep –
Ironically your link about jelly beans ties in well with my comment far below at 12:13 PM, about tree rings being proxies for both temps and rainfall.
One could also add in the immediate micro-climate sunshine, slope, availability of nutrients, etc., as a nice analog for the jelly bean analysis.
Found link for cartoon
Woops, let’s try again
http://xkcd.com/882/
Mikep – I hope you didn’t overlook this cartoon: http://xkcd.com/556/
” After all, it is simply the lawyer doing his job and it is the jury’s job to judge the veracity.”
That’s very close to the analogy I use in my Intro science class to explain the scientific method, although I use a politician instead of a lawyer.
A politician is in search of data to support a predetermined conclusion, while a scientist (should) be in search of a conclusion that is in best agreement with the data at hand.
I have some very apt quotes from climate scientists gleaned from reading CA that I use to nicely illustrate the former…
Steve Garcia says: “As I said a few minutes ago, if someone insists that something that is wrong should still be used, how in the HECK are we supposed to deal with such idiocy?
Truth is truth. And wrong is wrong. Laden is supposed to be a scientist and yet he is arguing that Mann’s work should still be used when the data used is no longer reliable (not true)?”
Crimestop means the faculty of stopping short, as though by instinct, at the threshold of any dangerous thought. It includes the power of not grasping analogies, of failing to perceive logical errors, of misunderstanding the simplest arguments if they are inimical to Ingsoc, and of being bored or repelled by any train of thought which is capable of leading in a heretical direction. Crimestop, in short, means protective stupidity.
This typical Greg Laden “inability to argue as an adult” faire:
Somehow being a critic of a particular aspect of climate science magically transforms you into a “climate science denialist”.
LOL. Just LOL.
This is the sort intellectual munchkinism on Greg Laden’s part that usually requires chemical intervention to achieve. But I will take Greg Laden’s word (if he chooses to give it), that he was able to reach this intellectual low-point without artificial aid.
I agree. Greg Laden knows damn well that Steve is not a “climate science denialist,” and for him to say so is indicative only of his inability to make a cogent argument without including an ad hominem slur; which, as we have sadly seen over the years, is par-for-the-course from his compatriots in this field.
Carrick, earlier this morning, I wrote Laden taking exception to the ad hominem as below.
Ironic isn’t it. I would argue that Laden’s post does not even rise to the level of “mental laziness”. He doesn’t perform even the most rudimentary fact checking before throwing out his verbal garbage and name calling. To wit the ridiculous comment about Steve not wanting people to know about the CO2 attributed trend correction in MBH99. The ignorance continues in the first comment of this thread:
I haven’t stopped giggling since I have read this. It is kind of like making a comment to RomanM that ‘you need to understand that valid summary time series statistics depend on having a stationary mean.’
I’d not seen this level of candour before: …”or entirely useless”…!
Looks to me like a potent “external effect” might be scrutiny. Probably best to avoid that if you don’t want to upset these sensitive proxy thingummyjigs.
Steve, just curious – did you write Greg an e-mail, or did you make a make it as a comment to his recent post on his website ?
IF the latter, it isn’t there now . There are only 7 replies: 2 yesterday, 5 today . . .
Given that when people who make indefensible claims seldom if ever rise to defend them I don’t expect Laden will respond. And he is surely sufficiently aware of his own unimportance to the science to not risk a cushy position defending rubbish opinions. Opinions are like pie-holes; everybody has one.
Surely this is clear. We are not dealing with a population of trees that track temperature well. Look at any collection of these proxies; they looks like a whole lot of noise. From that larger collection of noise, some of the sets of trees are going to look like they are tracking temperaure. That doesn’t mean they really are – that could happen by random chance too.
How to tell the difference? You look at more data, out of sample. If those sets of trees do poorly out of sample, that is evidence that the fact that they did well in sample was just random chance; they aren’t really proxies at all. If they continue to do better than the rest out of sample, that is evidence that they are really temperature proxies.
I know all this is obvious, but Laden doesn’t seem to understand it so I thought someone should say it clearly.
I have had the same issue at ATTP when I posted there; most people did not understand that weighting proxies by agreement with instrument temperatures is not allowed without further validation.
miker613 –
You are deescribing the entire Divergence Problem, which dates back to before 1990 – and was discovered orginally by the dendroclimatologists themselves. And THEY are the ones who initially had (and I believe still do) a problem with seeing what they were seeing. And they DID go out and check other sites – lots of them. If they had checked them and found Sheep Mountain (or other sites) unusual, they would have said so LONG AGO.
Ergo, the Divergence Problem EXISTS, and it isn’t going away anytime soon.
I’ve been on this for ever since shortly after the Climategate emails, picking up on it in 2010. I said then that this problem is the doom of dendroclimatology. I think that this current discussion is just a point along the way.
Oops! I meant:
“I think that this current discussion is just a point along the way toward that doom.”
It’s not often that en entire scientific field is tossed in the dumpster of history, but this may be one of those times.
Everyone is hoping that Laden will show and furnish some entertainment. I imagine that it is not to be. Too bad, for he is very good entertainment, indeed.
I’m sure this question has been asked before. What percentage of Mann’s temperature reconstructions is determined by tree ring proxies? What about average of all temperature reconstructions that use tree rings?
Craig Lohle:
I’m reminded (yet again) of one of the Almagre trees. A simple demonstration of how a storm can cause the Strip-Bark phenomenon.
Seems at least slightly possible that following such a major storm, trees just might have a significant “growth pulse” in response, that could last even for 100+ years before growth returns to normal.
But I’m sure such things are fully accounted for in the proxy models. (Not.)
“I think you are missing the point entirely. You need to understand that any proxy is subject to internal variation, external effects that make it less sensitive (or entirely useless), etc. This is true of proxies, and it is true as well with instrumental data. If your thermometer breaks, you don’t use the data from it any more.”
Since I don’t follow Laden’s blog, I wouldn’t know if he’s taken objection to the use of lake sediments contaminated by modern activity, i.e. “broken thermometer”, not to speak of fixing it by turning it upside down. But since Steve “need to understand” how this works, I suppose he has written whole dissertations about it.
If you fellas aren’t careful, you are going to give Mr. Leaden the idea that his intellect and manners are not appreciated here.
“Back in 1999 we (Mann et al) made the best available choices with the information and data we had.”
So the “best available choice” was to grossly overweight bristlcones in order to swamp all the other proxies combined? A statistical thumb on the scale was the “best available choice”? I guess it depends on the objective one is trying to achieve.
Amazing.
I believe that anyone who thinks that tree rings are reliable thermometers, no matter how noisy, is an idiot. Sorry, but there is no nice word that fits. Damn fools, all of them.
Steve,
You might want to check if it is really true that “there was no detectable divergence through the end of the calibration interval” by comparing the proxies to the thermometers on a year-by-year basis.
This comparison could be done by looking at the first differences of each series. A simple plot might be enough. Looking at the correlation of the first differences would be a simple statistical test. (I’m sure there is a more sophisticated test, but I don’t know it off the top of my head.)
If the proxies really “work” over the calibration interval, and if there is “no detectable divergence,” the proxies should follow the ups and downs of the thermometers each year as well as follow the general upward trend of the thermometers. But, if the analysis is merely cherry-picking proxies that happen to have an upward trend over the calibration interval for whatever random reason, then we would would expect the first differences to “diverge” and not be highly correlated.
This seems like an obvious test. It is hard to believe nobody has done it.
Part of an email received recently, h/t Phillip:
“It’s actually quite funny that you can reconstruct past temperatures from stalactites, ice cores, boreholes, mud from the ocean floor, tree rings etc., but someone using a thermometer and best practice for the times can’t be trusted! It’s not as if you can’t go out, hang a thermometer in similar conditions to those used in the past, and compare your readings to modern practice and make an accurate conversion table for any adjustment needed.”
From Australia’s Bureau of Meteorology on temperatures: “Observations made before 1910 may have used non-standard equipment.”
We need a similar statement, different years, for temperature dendroclimatology.
Although, what is a non-standard tree?
A brief Greg Laden update, slightly o/t. May be of interest.
http://nigguraths.wordpress.com/2014/12/20/the-greenpeace-nazca-lines-selfiegate-and-greg-laden/
Steve: nice spotting.
Hadn’t spotted the dual roles of Sven Teske Shub. There are dodgier people than our ‘fancy PhD from Harvard in Archaeology and Biological Anthropology’. Thank you for this.
Ps
Damn I found it 🙂
Technical issues aside, how does Mr. Laden expect anyone to take him seriously when he begins his critique with an ad hominem (“Climate science denialist Steve McIntyre….”)?
unfortunately, for some folks, you are not taken seriously unless you do use the ad homs.
Several years ago I happened across the website of a group who used tree rings as rain gauges. I asked how I could tell whether to use a tree as a rain gauge or a thermometer. Their answer was, as far as I can remember, mere obfuscation. I suppose they really meant “whatever suits our pursuit of research grants”.
If I were a serious climate scientist and even a serious advocate for immediate mitigation on AGW, I would not be very happy with Laden’s non-reply to SteveM’s out-of-sample finding on a tree ring proxy. As a serious reader and analyzer of the temperature reconstructions, I would only call attention to this non reply as a means of attempting to elicit a better response from the non-replyer and/or from the climate science community at large.
Perhaps if I heard a reply from someone in the community who would attempt to define the difference between in-sample and out-of-sample testing in statistical terms and explain how that difference could affect the validity of proxies used in temperature reconstructions I would feel better. I would guess that Laden would see no problem with post facto proxy selection and truncating a proxy where he thinks it has gone bad. If the those in the community truly want to discuss the analyses and criticism of the reconstructions, they have to address these issues head on – otherwise the discussions are a waste of time.
It seems quite an open-and-shut case. Laden seems to be trying to muddy the water. I think the whole argument can be distilled down to two main points:
1) Surely it is universally accepted that biological growth rate is a product of many environmental factors and therefore any “biometric” is a measure of the collective effect of these.
2) Even if you can produce a statistical model that appears to isolate the relationship between one of these environmental factors and the biometric it is imperative that the model be tested using the “out-of-sample” approach outlined by SMcT.
The proxy can only be validated by out-of-sample testing otherwise you’re just claiming validation via the very data you used to optimise your model. This is why everything Laden says is just arm waving.
I haven’t seen anybody jump on this part of Laden’s first comment:
“This is something else the deniers tend to get wrong; they try to conflate the Briffa et al post-1960 divergence problem Mann et al’s hockey stick work. There is no such issue with that work, in that there was no detectable divergence through the end of the calibration interval.”
Who the hell CARES about Mann’s work not including the period of non-correlation? If later evidence/data/research show Mann’s work to be faulty in the data he chose to use, then Mann’s paper is WRONG. In this case, the faulty data is the supposed correlation between tree-rings and temps. If that is wrong, then every part of Mann’s paper about the pre-1980 proxy record is WRONG.
This makes Mann’s paper no more correct than the pre-Galileans or pre-Copernicans were about their theories. Widely accepted theories have been proven wrong in the past, and some will be proven wrong in the future. Mann’s (Hockey stick) is one in the present that – based on HIS work – the Divergence Problem prove wrong.
Oh, some day someone may prove the Hockey Stick to be true – but it won’t be Mann’s work that proves it. Mann’s paper is already only fit for the dustbin of history. BECAUSE it does not NOW include the Divergence Problem. What it included in 1998 doesn’t matter anymore.
Greg Laden, Do you understand the above comments? You have given no indication that you do. For the future I suggest it is to your own serious benefit that you grasp the key point being being made here.
You would also set a better example by not engaging in name calling (denier) and guilt by association of those you may not agree with..
Just saying.
Mmm still in moderation after 24 hours that will teach me for sharing pearls of wisdom 🙂
Greg Laden refers to broken thermometers and it is good. There is, here, a sort of consensus on the impossibility of drawing useful temperature information from trees. This concensus is based in part on a review of used methodologies and secondly on the nature of the proxy. If methodological criticisms seem to me entirely founded, those most basic on tree physiology looks blurred with confusion.
There is not one proxy, but three. Criticism are based on the best known: ring width. Summer wood densities of standing trees form the second proxy. To my knowledge, no demonstration has been made of any inferiority of this proxy over instrumental data for the measurement of changes in regional temperatures.
The third proxy, summer wood densities extracted from sub-fossil trunks is unfortunately probably very difficult to use due to the instability of some components.
The divergence problem is primarily related to summer wood densities of standing trees (Briffa et al. 1998). This divergence does not appear in 1960, but in the early twentieth century.
phi: “There is not one proxy, but three. Criticism are based on the best known: ring width. Summer wood densities of standing trees form the second proxy.To my knowledge, no demonstration has been made of any inferiority of this proxy over instrumental data for the measurement of changes in regional temperatures.”
phi – please can you clarify your double negative? Are you saying that summer wood density of standing trees is a proven regional temperature proxy and,if so, please can you supply references? Can you also clarify the relationship between your terminology of “summer wood density” and “MXD” (maximum density I believe?)
Apologies if this is known and accepted by all – I don’t follow so closely these days. Thanks.
Curious,
My expression was a bit vague but indeed I think to MXD (maximum latewood density).
I do not claim exactly that something is proven but that there is no demonstration of the superiority of instrumental data over MXD for evaluation of changes in regional temperatures.
In fact, several clues suggest that MXD are more reliable.
Two examples:
1. TLT as control data.
2. Glaciers melting as control data.
phi – thanks for responding. Your interesting plots reminded me I’d seen your discussion with Jeff et al at theairvent a while back but I didn’t have time to follow or digest it. I just had a quick read of it now – a good constructive thread for those interested:
Without wishing to head too far OT, IIRR there were other threads where you discussed this work too – have you progressed (or published) things since then? Craig’s point re: upper response limit seems reasonable to me and IMO Kenneth’s final table showing some proxies trending positive whilst others trend negative merits a response. Apologies if this was covered in the thread – as I say I just did a quick read.
Curious,
The discussion at The Air Vent covers several threads and I do not remember all the details. There are also interesting things here on CA about Briffa 2013. Remain obviously many issues to be clarified; the covered season is small especially in high latitudes, very warm years are apparently poorly rendered etc. In my humble opinion, the consistency is still outstanding and the general pattern pretty clear.
I have not published but I have summarized a part of the problem in French here: http://www.skyfall.fr/?p=1399
I find the idea that MXD is a better thermometer than ordinary temperature sensors to be frankly bizarre and really not a tenable idea:
Tree ring proxies are well known to respond both multi-modally and in a nonlinear fashion, are typically an indicator only for one season of the year, and the spatial sampling is often atypical both in spatial coverage and in elevation of the proxy compared to mean elevation of the region they are supposedly sampled.
Ordinary temperature sensors are much more linear, are true indictors of temperature, well studied, the physics relationship between the sensor response and temperature well characterized and understood. They work for all seasons, and the coverage, while nonuniform for many regions (like the West Coast of the US for the case considered in this blog post), is over-sampled.
Kenneth, I think you’ve nailed the issue, which I think amounts to a restatement of the “loss of loss frequency information” by tree rings proxies that I comment on occasionally. As I’ve suggested on other threads, if you lock the multi-decadal behavior of tree-rings to “true” temperature proxies (which typically have much worse temporal resolution), perhaps you can improve the temporal resolution of the reconstruction. This is an idea that was first proposed by Moberg (2005), I believe.
Still, it’s not obvious how you disentangle the multimodal behavior of trees. Unlike the arguments given by phi, it’s not obvious it’s possible to do this
Carrick – I’ll try and read a bit of phi’s French version over Xmas. I think it possible that:
“I do not claim exactly that something is proven but that there is no demonstration of the superiority of instrumental data over MXD for evaluation of changes in regional temperatures.”
is intended as a critique of the adjusted surface station datasets – ie “instrumental” vs. “proxy” data. The divergence between surface and satellite records would make the same point and, of course, a satellite is an instrument too.
phi?
moderator – please can you check this thread? I saw a response from phi on this point but can no longer find it. Thanks.
curious, yes I understand his complaint is with the adjusted record. I happen to think arguing that you shouldn’t fix known systematic problems with your raw data to not be very tenable though.
It’s worth noting there’s currently a divergence between different satellite measurements. I trust UAH more over the period they were using the AQUA satellite, but that’s a subjective thing.
While I don’t think the surface temperature record is perfect, there are enough stations that virtually any hypothesis, including urban heat island effect can be be tested (I like Mosher and Hausfather’s work there). As far as I’m concerned, if people are going to criticize the surface temperature record, they need explicit testable hypotheses.
A closer look at what phi shows here and has shown on other occasions is the phenomena that I like to call the tree ring proxy dilemma. Tree rings can have reasonably good high frequency (as in annual) correlations with temperatures from the instrumental record while at the same time having very poor low frequency relationships with temperature (as in decadal trends). Tree rings for that matter are good at marking volcanic events as frequently pointed to by the dendroclimatologist, Rob Wilson. The problem to which I refer to above becomes evident when comparing near neighbor proxy responses to these events where the amplitude can varying considerably from proxy to proxy. This same problem can be seen when the individual proxy and tree ring responses to a temperature from the same location can correlate well on a high frequency scale, as measured by r, and yet have very different trends over decadal periods of time.
The tree ring dilemma arises for some parties doing reconstructions in that, on recognizing that time series such as tree ring responses with reasonably long memories from autocorrelations and/or long term persistence can have spurious trends and series ending trends that would “by chance” correlate with temperature trends, they will use or suggest using detrended correlations between tree ring response and temperature. Unfortunately such a correlation will not guarantee that the proxy gets the decadal trends correct and often times does not. The divergence problem can be an example of this dilemma.
I suspect some of this gets lost on those who do not appreciate how two time series with a high correlation in r can have very different trends. I should show this in a following post.
The link below to some individual tree ring proxy data from a Briffa reconstruction shows very well the point I made above about correlation, r, not being a good predictor of decadal trends.
Kenneth—please see my comment immediately above yours. I managed to reply at the wrong level.
By the way here’s a pdf to the pay-walled article I referenced:
Click to access Franke_etal_NatureCC_2013.pdf
Carrick, I hadn’t noticed that article before. YOu have to be careful in conclusions that are drawn from it. They report on 56 “temperature” sensitive and 128 “precipitation” sensitive proxies, derived (mostly) from the Mann et al 2008 network (described as “state of the art”)with some extras.
They say:
In their SI, they add:
They do not provide information on which proxies were screened out because of contradictory interpretation by authors.
They say:
of the 56 proxy records in their temperature network, 41 are Mann et al 2008 proxies. of the Mann et al 2008 proxies, 24 are gridded MXD series. It is my recollection that the gridded MXD series as used in Mann et al 2008 used a variation of hide-the-decline: that Mann deleted post-1960 values of the actual gridded data and replaced it by infilled data. Two of the 41 are Greenland ice core d18O series. 15 are ring width series, only two of which are stripbark: ca529 (Timber Gap Upper) and nv513 (Mt Washington).
Franke et al state reasonably enough that precipitation-sensitive proxies should not be used in temperature reconstructions – a criticism that has been made on many occasions here about both Mann et al 1998-99 and Mann et al 2008. This has been regularly fobbed off with armwaving invocations of teleconnections, so it’s nice to see Franke and coauthors catching on to the problem.
I double checked and confirmed this recollection. Mann et al 2008 truncated Briffa’s gridded MXD versions in 1960 and replaced them with infilled data from 1960 to 1998. This may not impact the spectral properties, but worth recalling given publicity subsequent to Mann et al 2008 about using tricks to hide the decline in MXD data.
Oy vey.
Frank et al: “We exclude records if the interpretation of the proxy as temperature- /precipitation-sensitive made by the original authors does not agree with the correlation screening results.”
Forgive me, but excluding data that does not agree with some “industry standard” is not, to my mind, acceptable science.
It is especially egregious when they don’t even say which – or how many – specific datasets are excluded (argument noted by Steve M) and what the specific exclusion reasons were for each one. They could not have all been excluded for exactly the same reasons. The vague “does not agree with the correlation screening results” is completely inadequate.
Scientists are SUPPOSED to include – for honesty’s sake as well as good scientific principle – the fact that the authors have looked for evidence to falsify their conclusions and have dealt with such possible rebuttal evidence head-on. To exclude some because of non-agreement with THEIR chosen (cherry-picked) screening parameters is politics, not science.
Basically, this seems to be an admission of cherry-picking.
WHY is this so prevalent in climatology? Their internal politics must be maddeningly stifling to honest inquiry. I wonder how they can sleep at night.
There are 12 bristlecone/foxtail series that pass their screening tests. All except one were in the precipitation table, including influential series in Mann et al 1998-99 and Mann et al 2008. One series ca529 appeared in both dsets, though they said that no such series occurred.
Hi Steve,
I just noticed it today myself (and am still absorbing it),, so thanks for the quick comments.
This sentence immediately caught my eye:
It’s my impression that this is an understudied problem.
This is also true for other proxies (Brandon Shollenberger had at least one example with respect to Ljungqvist, I don’t have time right now to dig it up).
The handwaving argument by Mann and others is that precipitation proxies can act like temperature proxies, if there is a correlation between precipitation and temperature. While it’s not an unreasonable assumption, it’s totally non-obvious to me the relationship would be stable over time (so it’s a problem for out of sample data).
I remember the problem with MXD as used in Mann 2008. I don’t remember the exact details of how this was done, but if my memory serves me, Mann did not state in his paper that he had performed this manipulation.
Carrick: “The handwaving argument by Mann and others is that precipitation proxies can act like temperature proxies, if there is a correlation between precipitation and temperature.”
Wow, that assertion doesn’t sound right. I’ve followed Steve M fairly well on this topic, and I’ve never heard him say anything to that effect.
I myself have mentioned on several occasions how I DO NOT think precip AND temps can both be extracted from the same proxy data – as well as arguing that the attempt by both biologists and dendroclimatologists to use the same data is evidence that NEITHER is useful.
Perhaps you misunderstood me and then also assigned that misunderstood argument to Steve M?
Steve Garcia, Steve McIntyre is clearly aware of the problem, as you can see from the last paragraph of his comment a few entries above:
The assertion I commented upon is in fact what is typically made when people attempt to justify the use of precipitation proxies as temperature proxies
So I’m not sure what you think I’m misunderstanding in this case.
Carrick:
There is more than one example with Ljungqvist, though I can’t recall all of them given he’s been an author on several reconstructions which use somewhat different data sets. The one you probably have in mind though is the Dulan proxy, which has been discussed on this site before. In regard to it, Ljungqvist told me:
This is a somewhat strange argument as the paper he sent me to support his claim showed half of the periods the Dulan series covers were warm and dry in the proxy’s area. That would seem to indicate even if there is a relationship between temperature and precipitation there, it is not stable and cannot be used to justify using the proxy as a temperature proxy.
One other thing which stuck out to me is Ljungqvist (and his co-author) were pleasant and responded promptly. Until I pointed out the paper contradicted their position. Then I didn’t get any more responses. It might just be coincidence.
Carrick –
Acknowledged, but I see the two statements as being 100% opposite:
You assert this: ““The handwaving argument by Mann and others is that precipitation proxies can act like temperature proxies…”
And then you quote Steve M as saying: “Franke et al state reasonably enough that precipitation-sensitive proxies should not be used in temperature reconstructions – a criticism that has been made on many occasions here…”
Based on the clear language of those two, it sounds to me that you are asserting that Steve M says that precip proxies CAN act like temp proxies, but then you quote him arguing the opposite. Steve M does not say at all in that, that precip proxies can act as temp proxies.
I don’t think I misunderstand Steve M, but it appears that perhaps you do.
Steve Mc: I think that you’ve misunderstood Carrick.
Steve Garcia:
Okay at least I understand your confusion.
But perhaps you can explain to me and possibly others how “The handwaving argument by Mann and others is that precipitation proxies can act like temperature proxies”, which clearly addresses the arguments made by Mann and others (without naming who they are) can be construed to say that “Steve M says that precip proxies CAN act like temp proxies”.
I never said Steve McIntyre was one of the “others” who did and given that I was directly responding to statements by Steve McIntyre which as you note said the contrary, I don’t know how you could reasonably come to the conclusion that I intended to lump Steve McIntyre’s in with Mann’s.
For what it’s worth, the assertion that ” precip proxies can act as temp proxies” is not the sort of thing that I say or am likely to say. Nor is Carrick remotely likely to attribute this sort of statement to me. So some misunderstanding has cropped up – I suggest leaving it at that.
Brandon Shollenberger:
I don’t actually expect the relationship between temperature and precipitation to be stable, though it’s possible there are regions of the Earth where it might be.
I suppose we could conjure up scenarios where you use long-duration accurate temperature proxies (which are limited in temporal resolution and have poor spatial sampling of the Earth’s temperate field) are used in conduction with higher resolution precipitation proxies to produce a higher resolution spatiotemporal reconstruction, which will work as long as the relationship between temperature is stable over the time scale associated with adjacent measurements from the long-duration accurate temperature proxies.
But I’m pretty sure nobody is doing anything that complex at the moment. I think Mann in particular is just assuming the relationship to hold and never really attempts to make any justification of it.
Carrick and Steve M –
Ueah, that is my bad. Steve, don’t laugh, but somehow Mann’s name in “The handwaving argument by Mann” somehow got into my brain as YOUR name. I know, what a hoot, right?
A blind spot in my vision? A climate dyslexia thing? I don’t know.
Steve M, I agree, completely, that you are the least likely person to argue that precip proxies can act as temp proxies”.
So it is turning out that we are all in agreement about the precip-vs-temp interpretations of proxies, including Brandon.
My misreading got this going, and it was not misunderstanding but only misreading ON MY PART. Mea cupla and apologies to all.
We are on the same page, and I am glad that my thinking LONG AGO is being borne out as valid. (I wasn’t the first person to think it, but I did come to it on my own, so I am pleased that other good minds see it that way, too.)
Brandon –
You said
This deserves a response, big time.
This is pretty much exactly what we all should have expected – that essentially on average half of the warm periods would be wet and half dry. And (I presume) half the cold periods, too.***
“…and cannot be used as a temperature proxy” – how can anyone come up with any other conclusion, given such a 50-50 correlation of warm with wet? If warm is 50% of the time dry, then tree-ring growth is stilted – but would APPEAR as a cold period, based on the assumption of dendroclimatology.
To me, that would negate dendroclimatology in its entirety, barring some extraordinary evidence to the contrary AND a damned solid explanation for why the data in the paper Ljunqqvist sent should be tossed out.
*** I would NOT expect that full-bore ice ages would correlate with this 50-50 issue. In my other area of interest, I run cross a lot of information about the ice ages, and THOSE extreme periods DO find cold and dry together consistently. Thus warm and wet DO only come in non-ice-age conditions – but that does not at all mean that warm periods cannot also be dry part of the time. But severe cold DOES correlate to DRY. For example, Antarctica is presently a desert, technically speaking, due to its low annual precipitation levels.
But all of that has no connection during conditions that are not ice age, that are “somewhat normal” and have “somewhat normal” wet-dry regimes.
To try to correlate any of this to a temperature precision approaching tenths of degrees of variations sounds foolish in the extreme.
Steve Garcia—it’s all good. Now that the miscommunication is cleared, up I’m glad we can move on to more substantive questions. Alas I lack the time to study them myself, but looking at regional scale correlations between temperature and precipitation over the period where there is surface instrumental data would be interesting. The direction I’d like to see pursued is:
Is there a stable relationship? Or does the relationship vary over time and if so, what is the time scale associated with that?
The Dulan area of China is desert or near-desert. Curiously it is relatively near Thompson’s original Dunde ice core.
There is very interesting speleothem d18O data from China covering very long periods in high resolution, a set of data that ought (in my opinion) to have been specifically noted up in the IPCC assessment, as opposed to so much focus on more reworking of bristlecones etc. I’ve been meaning to write up the Chinese data for a long time.
In the Holocene optimum, the Chinese monsoon was very strong and resulted in very depleted O18 values in speleothems and undepleted values in the LGM. This raises a very interesting question for the interpretation of Lonnie Thompson’s Dunde ice core, which does not (as interpreted) contain the prominent Holocene Optimum of the nearby speleothems, but does contain very depleted O18 values downhole, which Thompson interpreted as LGM.
Steve Garcia:
I see no inherent problem with saying there is a stable relation between temperature and precipitation for a given area. I suspect that is true for some areas. The problem is it has to be demonstrated, not assumed.
For instance, suppose you had proxies in an area you knew to be good measures of temperature. Suppose you also had proxies in an area you knew to be good measures of precipitation. Now suppose one of these proxies extended back only 400 years while the other extended back 1,000 years. It seems reasonable to me to compare the proxies and see if they have a discernible relation. If so, I could see using that relation to “extend” the shorter proxy back in time based on its relation to the longer one.
It would be a not insignificant amount of work, and there are a number of nuances and caveats that’d need to be dealt with. As far as I know, people aren’t even trying. The Ljunqqvist (and Christiansen) reconstructions didn’t dig into data quality. Instead, they just used proxies which had already been used by other people in the field. That’s understandable, but wrong. Had they actually addressed data quality issues, at least half a dozen of their proxies would have been discarded.
One thing I found strange is they chose their proxies by using the (let’s be honest) cherry-picked sample used in other papers in the field. They then did sensitivity testing by removing (I think two) proxies from their data set and seeing what effect it had. They found their results were robust. The obvious response is: Well, yes. If your data is cherry-picked, you will likely get the same results no matter what subset of it you use.
Brandon –
“I see no inherent problem with saying there is a stable relation between temperature and precipitation for a given area. I suspect that is true for some areas. The problem is it has to be demonstrated, not assumed.”
Yeah, that is one of the “DUH” things about science – that until the idea is demonstrated with evidence and is not falsified, it is not science but simply hypothesizing/speculating. At best one might push it as philosophical, but that should never be mistaken for actual science.
“It would be a not insignificant amount of work, and there are a number of nuances and caveats that’d need to be dealt with. As far as I know, people aren’t even trying.”
That is more or less what got me on this subject in the first place – my curiosity got me looking for the work that backed up the idea that ONLY anthropogenic CO2 emissions were causing the apparent rise in global temps. A simply thing, right? You want to assign blame/credit for something to a singular forcing, so what is the first step in that process? To devise a test or tests that will eliminate all other possible causes – probably one by one. Lo and behold, no one back in the 1990s or before ever were “even trying”. In other words, the science wasn’t there to back up the assertions, no matter HOW reasonable the assertions sounded. (Liberal that I was – and still am – I thought that there would be all SORTS of such evidence in papers all over the place, since we could see evidence in the cities of emissions of SOME sort almost every day; instead I found that NONE of that process of elimination had ever taken place. The assertion was accepted as fact, without anyone “even TRYING.”) That was when I smelled a rat.
If they aren’t even TRYING, then the system is broken, because Mann’s work and that of those around him and Hansen should have been met with proper skepticism in the first place. No one ever applied the “Extraordinary claims require extraordinary proof” skepticism to the warmist assertions. Shame on the system that it took until Steve M to just try to audit some of the works, including Mann’s. The journals did not do their due diligence (as Steve M showed that Mann did not). Mann wasn’t the first person to do bad science. There are supposed to be checks and balances to prevent garbage science from getting out into the world.
Have a Merry Xmas! I had a few minutes and thought I’d jot this down. Now it is back to our late dinner party. (I wonder if they missed me?)
‘“…and cannot be used as a temperature proxy” – how can anyone come up with any other conclusion, given such a 50-50 correlation of warm with wet? If warm is 50% of the time dry, then tree-ring growth is stilted – but would APPEAR as a cold period, based on the assumption of dendroclimatology.’
I’ve kind of vaguely followed this discussion in the comments, but really I’m having trouble telling what you-all are talking about. So what if warm and wet are not correlated? If tree-ring growth depends on both, then the wet would average out in the long term, and the tree-ring growth would show a temperature signal.
miker613 –
“If warm is 50% of the time dry, then tree-ring growth is stilted – but would APPEAR as a cold period, based on the assumption of dendroclimatology.”
Absolutely.
That is pretty much the point, Mike, of skepticism toward that idea.
Add in the small values of increases and decreases that are “measured” (asserted) by the tree-ring proxies, it obviously only takes a slight drop in precipitation to mimic a slightly colder year or decade. But which is it – a drop in precip or a drop in temps? They may claim to be able to tell the difference, but in reality, they are assuming constant precip. Otherwise they have no basis at all for the claims of tracking temps. Separating precip ups and downs from temp ups and downs is a fool’s errand, and pretend science.
I myself WOULD accept such claims – but only if they actually show exactly how they can separate out one signal from the other. and then back it up with further corroboration in REAL data – more than once or twice.
Until that time, nah, their assumptions have led them down a blind alley. They WANTED to find evidence that showed that humans are damaging the climate, and anything that suggested it they bit on – leaving their scientific minds behind, alas.
Science is not about the data/evidence. Science is about the attempt to interpret what the data/evidence is telling us, and then compile those interpretations into gestalts that we would maybe call understanding. THAT sometimes means that we have to make assumptions about various factors. But then those assumptions have to be really tested, to make sure they are not misleading our interpretations.
Dendroclimatology, as you are seeing, is based on an assumption that may or may not be true. We here are arguing that the assumption is false, by not been properly isolated and put into the proper context. The main complicating factor is precipitation – HOW do they defend their assumption that precipitation is not affecting the tree-rings and bolluxing up the works?
All the handwaving in the world will not cover up the realities of what they are leaving out, in their rush to convict AND EXECUTE their perceived murderer of the climate – CO2 emissions. Their entire focus is a Rush to Judgment, to give the jury (the world at large) the bum’s rush, and get that jury to shut down most of or all industry and rid the Earth of vehicles. The long term of it all is that, sooner or later, precip will be seen as complicating the tree-ring data so much that the claims of tree-rings as proxies for temps will fall apart.
I myself think that that time is NOW, though I may stand alone on that. The Emperor truly has no clothes on. In its eagerness to not look the fool, every citizen of the world who has bought into the global warming thing has jumped on the bandwagon. That is the rush-to-judgment bandwagon, and it is going to go off the rails. It is just a matter of time. Because the science it is based on is faulty.
You see that right now, it seems. Make up your own mind. Or at least tell them to put up or shut up – Ask them how in the tree-ring record do they separate the precipitation signal from the temperature signal.
And THEN make up your mind.
Merry Christmas to all!
Steve Garcia, I think you’ve commented only on my quote of your own words, not on my reply to them.
miker613 –
Yes, I only pasted in my own quote. My bad on that.
As to your comment, “So what if warm and wet are not correlated? If tree-ring growth depends on both, then the wet would average out in the long term, and the tree-ring growth would show a temperature signal.”
No.
If half the time the wet and dry average out, then we would not SEE the wet and dry periods, and that is important. Take the 1930’s Dust Bowl. If all we had was tree-ring data on that period, dendroclimatologists would read it as a cold period, wouldn’t they? (And the biologists – who use tree-rings as precip proxies – would see it as a dry period, but miss the hotness.)
And then take the Little Ice Age – we KNOW it was colder, especially in northern Europe. But it was also quite wet for most of that time (but not all). But we only have one or two temperature records from (part of) that time. If we did not have ANY temp records from then, we would look at it as cold, but would not have any idea about the wetness. And the wetness, with whatever boost in tree-ring growth, would be seen by dendroclimatologists as tending toward warm periods.
Now, go back in the totally pre-instrument time, to before the MWP or the Roman Warm Period – at some time when we see centuries of narrow tree-rings. When we see such a period, do the narrow tree rings mean cold? Do they mean dry? Cold AND dry? Warm and dry? Cold and wet? Whoever interprets such tree-rings – do we know which bases they use to interpret it all? So far, the dendroclimatolgists don’t ell us specifically in each paper, so years from now will researchers reading those papers know which interpretations were being used in our time?
…How long of a period does it take for “average in the long term” to actually average out? 2 months? 5 years? 10 years? 50 years? 300 years? 1000 years? And when we average out over some long periods, what short-term signal is lost? Who chooses the starting and ending points for periods of such averaging? (Trust me, those points are used by both sides in the current date, to tell very different stories.) And, regardless of what we find out about long term averages, do we then know which length of time is going to tell us anything that is significant to us as a scientifically based civilization trying to understand what goes on and what doesn’t go on, now and in our future?
We aren’t even ASKING such questions, much less getting any understanding out of it. Basically, IMHO, what we should be doing now is tabulating data and NOT working so darned hard to interpret it as meaning anything at all. Not yet. There are times for gathering data and times for garnering meaning from that data. They are not necessarily the same times. Premature conclusions based on being in the middle of the data collecting are dangerous things. They can lead to ideological inertia that can be hard to back away from. (Look at the Clovis Barrier in anthropology – for 70 years it stifled the field, and the field lost several decades and is only now beginning to back out of that blind alley. Now no one has an answer to how long humans have been in the Americas. That is a sad state of affairs, all because of a premature conclusion that was then very strongly defended – plus all the contrary work that was blocked and ridiculed.)
Until the precip-vs-temps tree-ring signal is fully understood, we should be keeping an open mind about what the tree-rings mean. The Divergence Problem is perhaps only the first of the signs that temperature does not track with tree-rings as well as we need a proxy to do. At the least, it should be giving everybody pause.
Steve Garcia, everything you said is right, but – that’s what normal variation is like. No temperature proxy is very good. Graph any set of proxies, you’ll see a bunch of noise and no apparent signal. That’s because of zillions of other things going on. Probably precipitation is one of those, but there are loads of other things too that the experts can list. So what does one do? One averages over time, and hopefully the noise averages out and the signal is left. That’s why we do linear regression, or whatever other method is used – to separate noise from signal. Not perfect, but that’s what you do when you have noise you can’t understand or can’t measure.
When you’ve done that, did it work to find the signal? I don’t know; ask a statistician. You need some way of validating it. You certainly can’t afford to just pick the proxies that match 20th century temperatures or such; that’s a way to guarantee that every set of proxies will show a signal whether it’s there or not.
But anyhow the fact that you’ve found one confounding factor (precipitation) is not in itself a guarantee that proxies don’t work; handling confounding factors is part of the business.
miker613 –
Mostly, what you said is right, too. The most pertinent points, I think, are
“No temperature proxy is very good.” and
“…handling confounding factors is part of the business.”
In climate science, the fact is that those important people – connected to the IPCC – are not dealing with those confounding factors at all (that is in any way apparent, anyway). The effect of this is that none of the confounding factors is getting addressed and are being left out altogether. I don’t know about you, but to me, not even addressing them is negligent. The principle of “due diligence” dictates that all factors need to be accounted for, CONSCIOUSLY, to the best of their ability and effort. Pretending that they don’t exist (as it appears is the case) is really, truly, bad science. That is my opinion.
The Divergence Problem is SCREAMING that other factors are either not being taken into account and SHOULD be, or that tree-rings aren’t good proxies in their own right. When the proxy data values are being presented at 0.1° precision levels, this borders on either stupidity, blindness, willful carelessness, or scientific misconduct. Yes, certainly tree-rings show SOME temperature signal – but without isolating that signal (if that is even possible), then that signal, as presented, can’t be anything more than imaginary.
One part of what you said is that I would argue against:
I am pretty sure one can only do a viable linear regression on data that is known to be properly vetted – and ISOLATED – for the signal you are attempting to understand. You cannot use linear regression to separate proper data values from non-proper data. The separation is another process altogether. The data has to be purely what you are working with – and NO other signals included. Otherwise, the regression is just doing b.s. on garbage data. And then nothing but garbage can come out of it. It IS literally, then, garbage-in-garbage-out. (Maybe I am wrong on this; Steve M or others could tell you if I am right nor wrong. But I really don’t think so: You first must have good and isolated data.)
The other thing I’d disagree with is:
The wild variations in climate data is not necessarily “noise”. The graphs you see in various places have lots of ups and downs, but that doesn’t make them noise, per se, just variations of the data. Noise is extraneous (coming from other sources), while those ups and downs are real. The question in climate graphs is, though, “Do those ups and downs really represent what they think they represent (temps)?”
If some of the signal AS SHOWN in the graphs, at the individual data point level, are including other signals (precip/site nutrients, competition from close-by trees and vegetation, etc.) instead of purely temp influence, then the data isn’t necessarily good data. It is tainted.
So, basically, it comes down to “What is good data?” and “Does the tree-ring data represent what it is said to represent?” There is no doubt that the measurements are real. People took cores, and the cores are competently measured to good precision. But do they actually represent temperatures?
If there is ANY precip signal represented in the cores, then everyone should be asking how much precip is in them? With Brandon’s revelation that half the warm years were dry and half wet, it appears that that question would seem to be answered or be brought into dispute: TOO MUCH of the tree-ring signal is precip for temperatures to be extracted from the tree-rings reliably. ANY precip signal makes the temp signal suspect. Especially when resolved to 0.1°C.
Statisticians have been separating the effects of correlated variables on measured response variables for many years so, in principle, this should be possible in the area of paleoclimatology if properly implemented. High quality data properly selected in sufficient quantities and appropriately analyzed could indeed result in reasonable information about the past climate.
IMHO, many of the methods used by the climate community contain inherent shortcomings which produce flawed estimates. Selecting proxies which “show a temperature signal” using correlation with contemporary records can induce statistical bias in estimated temperature series as discussed in the CA post here. Composite Plus Scale methods can be strongly affected by differences in proxy behaviour between calibration and non-calibration period with no means of identifying such proxies since much of the time the publications do not appear to examine how each proxy relates to the final reconstruction. It is also not necessary here to point out the foibles of using those exotic home-grown untested procedures whose properties have not fully been understood.
Despite this previous history, it does not mean substantial improvements are not possible. For example, why not start with all proxies and calculate a reconstruction which must relate to a proxy in the same way outside the calibration period as the proxy does to the temperature record during the calibration period? Proxies would self-select themselves to have a higher impact on the end result due to their uniformity of behavior and specious proxies which happen to match current temperature accidentally would have a correspondingly reduced effect. This methodology would work best on on multiple proxies of the same type in the given local region.
There are ways which one can try to extend this sort of methodology to proxies simultaneously affected by two (or more) variables such as precipitation and temperature to get reconstructions of each but that is a matter which would require a longer detailed description and it is time to visit with the Christmas turkey… 🙂
Comments on those two graphs:
Top graph: The invalidity of HADGRUT3 – versus UAH – I will assert is showing the post-adjustment values, not the raw data. There are many examples out there of individual stations adjusted up to and above 2.0°C upward. This graph seems to reflect that. Thus, phi, if your argument against temperature data is in regards to HADCRUT3 and those other datasets that depend on it, this playing with the data seems a likely suspect as to why it does not track well with the satellite data. Comparing the HADCRUT fudges with straightforward satellite data does little more than tell us that the HADCRUT fudges are improper and perhaps scientific misconduct. Unfortunately climatologists attach serious belief to the HADCRUT output, while dissing AUH as often as they can. Back around 2004 or so, in fact, there was a concerted effort against UAH’s satellite data, in which they eventually found a very slight mis-calibration – and then they have seemed to consider UAH satellite data as not worthy of consideration, while continuing to accept HADCRUT3 as gospel.
Bottom graph:
1. This graph seems to only show that a.) Suisse and HACRUT3 are almost indistinguishable, b.) MXD tracks well with ADJUSTED Suisse, and C.) MXD tracks well with Hutt et al 2009. What does that mean, other than that MXD does not track well with HADCRUT3?
2. The legend says that the solid red line is “Suisse corrigé de 1.5°C par siècle depuis 1890” which translates as “Swiss corrected 1.5 ° C per century since 1890”. Yet the red line begins well before 1890, in what seems to be 1870. (Apparently the Suisse also tie their datasets to HADCRUT3, since the dotted line is nearly identical to HADCRUT3.)
3. As one goes farther back in time, the correlation of MXD with either Hutt et al or with Suisse adjusted down by 1.5C/century begins to fall apart. Going back past about 1953 the “WOW” between the two seems to get big awfully fast.
That last seems to negate any assertions of “better tracking” of MXD to Hutt and the adjusted Suisse data – with the proviso that HADCRUT3 is post-adjustment (meaning that AS PRESENTED it is NOT thermometers themselves, but the fudged values.
Exchanging one adjusted dataset for another (Suisse 1.5C for HADCRUT3) seems to be an argument with a large void in the middle.
I would inform you all – including Greg Laden – that tree-rings are not just used by dendrochronologists to tell age and by dendroclimatologists to tell (they think) temperatures, but also by biologists to tell how much precipitation occurred.
See: Stahle and Cleaveland 1992 “Reconstruction and Analysis of Spring Rainfall in the Southeastern U.S. Over the last 1000 Years” (full paper)(http://www.uark.edu/misc/dendro/PUBS/1992_BAMS.pdf)
See also:
which states: “Ancient baldcypress trees found in wetland and riverine environments have been used to develop a network of exactly dated annual ring-width chronologies from the southeastern United States, across Mexico, and into western Guaemala. These chronologies are sensitive to growing season precipitation in spite of frequently flooded site conditions, and have been used to reconstruct moisture levels [in] the southeastern United States and Mexico for over 1000 years.”
***
Wrap your head around that one for a moment: Two different groups of scientists using tree-rings as proxies for two entirely different things – two different variables output from one indicator.
That is like having two variables X and Y in an equation and trying to come up with actual values for BOTH. That can’t be done. (Which one did what, in any given year? How do they distinguish which year had better rainfall or temperature, and how much of each was affecting ring widths or densities?) What both groups seem to do is to assume the OTHER variable (rainfall OR temps) is constant. Which neither one IS.
So this, we all know, is simply not true. Rainfall varies a LOT from year to year, and so do temps. And as we see with temps in the thermometer age, temps range a lot from decade to decade, too. And as shown from the qualitative accounts from the LIA, MWP and Roman Warm Period at least, the temps range widely from century to century, too. So, the biologists are not working with a full deck. But then, we all are beginning to appreciate that the dendroclimatologists aren’t, either. (Witness the Divergence Problem.)
From the Central England thermometer history, we know that the temps in the latter part of the LIA in at least one place were considerably below the 19th century (if not consistent).
We also know that portions of the world go through droughts and non-droughts. Thus, BOTH rainfall and temps vary widely.
That being the case, neither group – biologists or dendroclimatologists – has a factual basis for their assumption of constancy in the opposite variable from the one they are “measuring” with the tree-rings.
Here is Greg Laden in more detail on the subject three months ago.
http://scienceblogs.com/gregladen/2014/09/22/steve-mcintyre-misrepresents-climate-research-history/
The comments in that post are a fun and informative read.. 2 opposing perspectives both adequately represented!
Hardly. Greg Laden and his supporters clearly show any real grasp of basic scientific principles or logic.
Spurious correlation plus data snooping=?
Post hoc data snooping is just another way of saying “confirmation bias”. After passing through a climate science re-education camp one fully accepts the small addendum to the scientific method “Inconvenient data may be obscured, deleted or otherwise disposed of as ‘dirty laundry’.”
“We have learned a lot from experience about how to handle some of the ways we fool ourselves. One example: Millikan measured the charge on an electron by an experiment with falling oil drops, and got an answer which we now know not to be quite right. It’s a little bit off because he had the incorrect value for the viscosity of air. It’s interesting to look at the history of measurements of the charge of an electron, after Millikan. If you plot them as a function of time, you find that one is a little bit bigger than Millikan’s, and the next one’s a little bit bigger than that, and the next one’s a little bit bigger than that, until finally they settle down to a number which is higher.
Why didn’t they discover the new number was higher right away? It’s a thing that scientists are ashamed of–this history–because it’s apparent that people did things like this: When they got a number that was too high above Millikan’s, they thought something must be wrong–and they would look for and find a reason why something might be wrong. When they got a number close to Millikan’s value they didn’t look so hard. And so they eliminated the numbers that were too far off, and did other things like that. We’ve learned those tricks nowadays, and now we don’t have that kind of a disease.
But this long history of learning how to not fool ourselves–of having utter scientific integrity–is, I’m sorry to say, something that we haven’t specifically included in any particular course that I know of. We just hope you’ve caught on by osmosis
The first principle is that you must not fool yourself–and you are the easiest person to fool. So you have to be very careful about that. After you’ve not fooled yourself, it’s easy not to fool other scientists. You just have to be honest in a conventional way after that.”
Cargo Cult Science
Richard Feynman
From a Caltech commencement address given in 1974