Marcott et al have posted their long-promised FAQ at realclimate here. Without providing any links to or citation of Climate Audit, they now concede:
20th century portion of our paleotemperature stack is not statistically robust, cannot be considered representative of global temperature changes, and therefore is not the basis of any of our conclusions.
Otherwise, their response is pretty much a filibuster, running the clock on questions that have not actually been asked and certainly not at issue by critics. For questions and issues that I’ve actually raised, for the most part, they merely re-iterate what they already said. Nothing worth waiting for.
They did not discuss or explain why they deleted modern values from the Md01-2421 splice at CA here and here. Or the deletion of modern values from OCE326-GGC300 as asked here.
Nor do they discuss the difference between the results that had (presumably) been in the submission to Nature (preserved as a chapter in Marcott’s thesis).
Nor did they discuss the implausibility of their coretop redating of MD95-2043 and MD95-2011 as discussed here.
Instead of dealing with actual questions, for the most part, their “FAQ” filibusters about questions that no one was asking in connection with this study.
Eventually, they get to a question about their redating:
Q. Why did you revise the age models of many of the published records that were used in your study?
But here too they filibuster. They spend 3.5 paragraphs pontificating about radiocarbon recalibration – a topic that was not an issue. Indeed, on several occasions, including a comment on a similar filibuster by William Connolley, I clearly distinguished between radiocarbon recalibration and coretop redating, taking issue only with coretop redating. Nonetheless, Marcott et al adopted the same filibuster that Connolley had used.
When they eventually get to coretop redating, they say only:
In geologic studies it is quite common that the youngest surface of a sediment core is not dated by radiocarbon, either because the top is disturbed by living organisms or during the coring process. Moreover, within the past hundred years before 1950 CE, radiocarbon dates are not very precise chronometers, because changes in radiocarbon production rate have by coincidence roughly compensated for fixed decay rates. For these reasons, and unless otherwise indicated, we followed the common practice of assuming an age of 0 BP for the marine core tops.
This is nothing more than a re-iteration of the statement in their SI that I had quoted in my Core Tops post:
Core tops are assumed to be 1950 AD unless otherwise indicated in original publication.
This is not an answer or justification of what they did. I urge interested readers to re-read the Core Tops post and more generally the Marcott sequence.
Nor did they directly answer questions about the questionable uptick. They pose the question for themselves as follows:
Q: What do paleotemperature reconstructions show about the temperature of the last 100 years?
They give a “short” and “technical” answer. First they stated:
A: Our global paleotemperature reconstruction includes a so-called “uptick” in temperatures during the 20th-century. However, in the paper we make the point that this particular feature is of shorter duration than the inherent smoothing in our statistical averaging procedure, and that it is based on only a few available paleo-reconstructions of the type we used. Thus, the 20th century portion of our paleotemperature stack is not statistically robust, cannot be considered representative of global temperature changes, and therefore is not the basis of any of our conclusions. Our primary conclusions are based on a comparison of the longer term paleotemperature changes from our reconstruction with the well-documented temperature changes that have occurred over the last century, as documented by the instrumental record. Although not part of our study, high-resolution paleoclimate data from the past ~130 years have been compiled from various geological archives, and confirm the general features of warming trend over this time interval (Anderson, D.M. et al., 2013, Geophysical Research Letters, v. 40, p. 189-193; http://www.agu.org/journals/pip/gl/2012GL054271-pip.pdf).
Although they now apparently concede that “the 20th century portion of our paleotemperature stack is not statistically robust, cannot be considered representative of global temperature changes, and therefore is not the basis of any of our conclusions”, this was definitely not the impression left by the authors when the article was published, when it was hailed as supposed confirmation of the Hockey Stick.
Andy Revkin, an extremely experienced reporter, certainly didn’t interpret Marcott et al as taking the position that their reconstruction wasn’t “statistically robust” in the modern portion: he titled his article “Scientists Find an Abrupt Warm Jog After a Very Long Cooling”. (Note: see Revkin’s fresh commentary here).
Further, their present admission that their reconstruction has no statistical skill in its modern portion completely refutes the previous claim that it had somehow vindicated Mann’s Hockey Stick, the distinguishing characteristic of which was supposed to be its “statistical skill” in reconstructing temperatures since 1850 (its calibration and verification periods.) If as Marcott et al now admit, their reconstruction is “not statistically robust” during the Mannian calibration and verification periods, it cannot “confirm” the Mann Hockey Stick.
later in the article, they take another run at this question as follows:
A: Here we elaborate on our short answer to this question above. We concluded in the published paper that
“Without filling data gaps, our Standard5×5 reconstruction (Figure 1A) exhibits 0.6°C greater warming over the past ~60 yr B.P. (1890 to 1950 CE) than our equivalent infilled 5° × 5° area-weighted mean stack (Figure 1, C and D). However, considering the temporal resolution of our data set and the small number of records that cover this interval (Figure 1G), this difference is probably not robust.”
This statement follows from multiple lines of evidence that are presented in the paper and the supplementary information:
(1) the different methods that we tested for generating a reconstruction produce different results in this youngest interval, whereas before this interval, the different methods of calculating the stacks are nearly identical (Figure 1D),
(2) the median resolution of the datasets (120 years) is too low to statistically resolve such an event,
(3) the smoothing presented in the online supplement results in variations shorter than 300 yrs not being interpretable, and
(4) the small number of datasets that extend into the 20th century (Figure 1G) is insufficient to reconstruct a statistically robust global signal, showing that there is a considerable reduction in the correlation of Monte Carlo reconstructions with a known (synthetic) input global signal when the number of data series in the reconstruction is this small (Figure S13).
Again, this is adding little to nothing to what was known more or less on day one. In Marcott’s original email to me, he had referred to the above paragraph in the article, arguing that the above sentence (which stated that the difference between reconstruction versions was not robust constituted a statement that the reconstruction itself in the modern portion was not statistically robust. I certainly do not read the original sentence this way and do not agree that it says what the authors now contend that it says. Nor does it clearly and unequivocally say what the authors now say:
the 20th century portion of our paleotemperature stack is not statistically robust, cannot be considered representative of global temperature changes, and therefore is not the basis of any of our conclusions.
Nor do the above caveats of non-robustness properly deal with the dependence of their 20th century uptick on their deletion of 20th century data points from critical series and the importation of earlier data points into the 20th century by unjustifiable coretop re-dating.
As a closing point, in Simonsohn’s widely publicized discussions of academic fraud last year e.g. here, Simonsohn stated that compliant articles ought to disclose “failed” calculations, as well as the ones reported (see Simmons et al 2011 here.) In the SI to Marcott et al, the authors reported on a wide variety of sensitivity studies, most of which will be of little interest to any reader.
However, they conspicuously did not report the results prior to the coretop redating that have been preserved in Marcott’s thesis and were presumably the ones in the article rejected by Nature. This version lacked the distinctive Marcott uptick. Under Simonsohn’s standards, disclosure of this “failed” calculation would have been obligatory. Comparing the two results shows sensitivity to Marcott core re-dating – a topic that ought to have been brought to the attention of reviewers by Marcott et al themselves. The authors are then free to argue for the merits of their coretop redating.
Although the authors now argue that the coretop redating doesn’t “matter”, it clearly did cause a difference between the present results and the results submitted to Nature and clearly did “matter”.
This is not to say that this issue affects “everything” in the paper, which, as the authors now argue, dealt with other issues. However, if these other issues are the ones that really “matter’, as they now say, then they should not have embarked on the coretop redating that resulted in a spurious uptick.
I think that their decision to take guidance from Realclimate was exceedingly unwise. It would have been better to simply respond to questions in a natural way when they were initially raised.
Update 1:51 pm) Andy Revkin has a comment on the FAQ here, pointedly commenting:
there’s also room for more questions — one being how the authors square the caveats they express here with some of the more definitive statements they made about their findings in news accounts.
254 Comments
Reblogged this on Climate Ponderings.
Yet what are the odds that the IPCC cites Marcott’s bogus claim of support for Mann without mentioning the lack of robustness, except perhaps buried in supporting material?
Revkin:
Seems a fair question to me too.
That’s Revkin acknowledging the issue while saying the least possible about it. He sees this as making the team – which he is a member of – look bad.
By April Fool’s Day Revkin has to be leading the way in facing up to his own responsibilities for this PR disaster for both science and mass media with:
and:
Those are in the two threads linked to by Steve above. As Revkin says, there’s more below these prominent inserts – much more.
Re: Richard Drake (Apr 1 02:57),
although I find Revkin one of the best mainstream journalist covering these issues, I do not feel sorry for his situation. In Finnish: Niin makaa kuin petaa (You’ve made your bed and you’ll have to lie in it).
A brilliant comment to the New York Times man on 18th March Jean, which I recommend to others. His reply wasn’t too bad by climate standards (a low bar if ever there was one) but didn’t include a yes-or-no answer to this:
Not bothering to answer you should by now be a scandal. Not asking the authors a bigger one. And their not having done the deed, before being asked, way bigger than that.
But in saying that I am expecting maturity. Remember, we have to talk baby steps with climate science and their media acolytes, it’s all they’re capable of. Fortunate that no policy maker would take their output seriously until they mature, isn’t it?
Talking of infant-size reactions, Michael Tobis takes a very long time to say “This site also got it wrong, sorry.” (Sixth paragraph from the end, second sentence, in case you missed it.) As for the rest, it reminds me of nothing as much as the little boy told by determined parents that he must remain seated at the table who says in a fury “I may be sitting down on the outside but I’m standing up on the inside.”
At one time, men took responsibility for their actions and words. Now, at all costs, most of these scientists can not utter the words: “We we wrong.” How very unfortunate.
I will never understand why revkin himself cannot seem to think to ask those questions. He seems equally satisfied whether he is being served Vichyssoise, or a badly-made, Cold potato soup.
Steve: he covers a lot of topics and can hardly be expected to know the nuances of proxy reconstructions. He comes from a perspective but he also tries to report fairly.
.
That was my thought: mere re-iteration.
One RC commenter asked (reasonably):
To which Steig, petulant as always, responded:
Unfortunately, they haven’t yet tried “rational discussion” as a response to the coretop dating issue (or other such issues.) They didn’t answer the questions or issues about the individual cores. Re-iterating a sentence from the SI is not answering the question.
Asking for info from those who understand these things: Does the core-top redating affect the _other_ important claim made by the paper: That they have a time range of accuracy to within a century or two (or three?), and thus can make statements about the size and duration of the jiggles in the long handle of the hockey stick? I would have thought that messing with the dating of cores should mess with the handle as well.
Steve: dates in the handle come from radiocarbon dates which are not affected by their coretop dating.
Ah – then that would explain the weird graphs in your earlier post, where only the latest points in the sequence are stretched?
This isn’t just a filibuster, they are defending themselves on the grounds that their paper made an incredibly subtle misrepresentation and it’s the reader’s fault for not noticing. Without the closing uptick, the main implication of their reconstruction is that, in the 20th century, we experienced the coldest conditions of the Holocene. With the uptick, we experienced nearly the warmest. The sharp uptick in the instrumental record can only be compared against their reconstruction if they can show their low-frequency proxies are capable of registering such events. If the 20th century portion of their reconstruction does not have the same uptick as the instrumental record, we would conclude that it could have missed similar swings in earlier centuries as well, so the absence of such swings in the earlier part of their graph tells us nothing about the presence or absence of decadal and century-scale warming events. Likewise, an annual or decadal observation from the modern instrumental record cannot be compared against values from their reconstruction, if their reconstruction is not capable of resolving events at that time scale.
But that is precisely what they do in Figure 3 of their paper, and it is the basis of their claim that “Global temperature, therefore, has risen from near the coldest to the warmest levels of the Holocene within the past century, reversing the long-term cooling trend that began ~5000 yr B.P.” Without the uptick in their proxy reconstruction this kind of statement could never have been made. The presence of the uptick in the proxy graph validates their comparison of the instrumental record against the proxy record. By admitting that the uptick is not robust and cannot be a basis for any conclusions they have undermined their own findings, root and branch.
Moreover, they can’t conceal the fact that they defended the robustness of the uptick in their paper. Marcott et al. stated that a RegEM variant reconstruction eliminated 0.6 of the 0.7 C closing uptick, but that due to data limitations the difference between reconstructions was “probably not robust” (p. 1198). In the context this implies that they consider the size of the closing uptick to be insensitive to choice of methodology, in other words that the uptick is robust.
But now they say the 20th century uptick is not robust, cannot be considered representative of global temperature changes, and is not the basis of any of our conclusions. Did they know this prior to drawing Figure 3 and inserting it in their paper, and if so, how could they not have been aware that it would convey a false impression to the reader? Contrary to Revkin, the issue isn’t just the definitive statements they made in the media but the claims they made in the paper itself.
There are some other issues as well. I had a couple of other posts that I was thinking about, but was waiting for their FAQ, which took longer than I expected.
There’s an interesting divergence problem with well-dated alkenones. ON a long time basis, alkenones look like they are a reasonable proxy, but they seem to go down in the 20th century, even cores with well-dated late 20th century samples. I’ll discuss some time.
SM replies to RC within hours. Inverse takes longer.
Re: pouncer (Mar 31 17:14),
As a layperson, I found that what Shaken et all set out to do with their FAQ was accomplished. It was a description of their work described in simple terms that most could understand.
I don’t know why something so simple took so long.
The delay in responding and the posting at Real Climate, leads me to think that there was a lot more going on behind the scenes in developing the FAQ’s/responses and that the intended audience was not those asking the tough questions but rather to assure people like me that all was going along quite swimmingly. I also don’t understand why the RC team has not handed the floor over to the authors. Why do they think we want to read their answers to questions that are more properly put to and answered by the people who actually wrote the paper? I find the “I’ll ask and get back with an update” very silly. Have the team at RC signed on as their press agents?
Even laypersons like me know that the tough questions have still not been addressed and that the authors have had a grand time splashing their non-robust findings around in the media pool since day one.
Re: pouncer (Mar 31 17:14),
Er, that should be Marcott et al.
“…alkenones look like they are a reasonable proxy, but they seem to go down in the 20th century…”
Divergence in the twentieth century is a common feature for good proxies, it’s even probably the best guarantee of their quality.
How many proxies diverge?
Is it remotely possible that it’s the instrumental record that diverges? Just askin.
ghl,
To my knowledge, this is the case of all proxies with good high frequency correlation (MXD, glaciers, snow, TLT). And indeed, the obvious conclusion is that the problem comes from instrumental data.
Isn’t how they deal with the differences in resolution between proxy reconstruction and the instrument record they compare against (not the uptick) described by steps 10-12 in the FAQ? Starting with:
“Estimated how much higher frequency (decade-to-century scale) variability is plausibly missing from the Holocene reconstruction by calculating attenuation as a function of frequency in synthetic data processed with the Monte-Carlo stacking procedure…”
and
“Pooled all of the Holocene global temperature anomalies into a single histogram, showing the distribution of global temperature anomalies during the Holocene, including the decadal-to century scale high-frequency variability…”
I suppose the question becomes is *this* a valid way of dealing with the differences in resolution, since they were not depending on the spurious uptick to draw the conclusion about past vs present temperature.
Ross:
First, thank you for explaining that. Second, wow. Deception bound up in deception. Why is Climate Science like this?
Ross — with respect, what do you feel the correct null hypothesis is for current global mean temperature and it’s change over the industrial era? You suggest that, without the uptick, their result would indicate that we are in the coldest period of the Holocene. And yet the problem with the uptick is not that it should be flat, rather it is that their reconstruction does not provide a reasonable constraint on global mean temperatures during the last few centuries. This is not to mention the fact that the effective smoothing of this set of proxies sounds like it’s on the order of several hundred years.
I would posit that the correct null would be informed by the instrumental record, which does show this uptick (and perhaps more). So in the absence of a robust reconstruction from the proxies for this time period, I cannot see how you conclude that it puts us in the coldest part of the Holocene. The observed warming is clearly outside of the range of their reconstruction where it is robust, whether or not Marcott et al show it.
Ross’s exact words were:
And later:
I read both of these without for a moment thinking that Ross actually believed that “in the 20th century, we experienced the coldest conditions of the Holocene.” Clearly he was stating the implications of the paper, which are now fatally undermined. As are all the headlines it occasioned.
Also, the paper can say too little about the magnitude of “natural variability” within 300 year intervals through the past 11,000+ years, yet the authors (and their flacks) are trying to make it out that recent decades are nearly unprecedented.
The paper cannot address the issue of whether post-1950 temps. are nearly unprecedented or not. Marcott et al. do not provide a study which can rule in or out temp. increases (or declines) over several decades to a century or more. Yet, that is all that the alarm merchants really want to use the paper for.
Skiphil, that’s how I read things as well.
I asked this before on a previous Marcott thread; If the temporal resolution is so low outside the modern era, how can this paper tell us anything about whether modern era temperatures are unprecedented, or even how they compare with the rest of their recon? I can’t see how this paper is useful in any way, unless the goal was to make a big, smooth hump.
And how exactly do you establish the reconstruction (of the rest of the holocene) is robust when the proxies are not capable of properly matching the known instrumental record? The only ‘robustness’ I can see is there is and was no way of resolving any possible variations unless the lasted for several hundred years, and if(!) additionally the proxies back then also would respond to this, unlike in more recent years ..
A Physicist:
Their Figure 1b shows an upside-down U shape that would end at a Holocene minimum, but for the uptick. I’m not saying that’s correct, I’m just saying that without the uptick the graph ends at a cold minimum.
I don’t know what you mean by the “correct null hypothesis about the global temperature.” A null hypothesis is constructed so that if it is true, a statistic formed from the data follows a known distribution, making it possible to reject it based on the p-value of the statistic. It is not necessarily constructed based on what one thinks is true, and in fact is often constructed on the expectation it is false.
I guess what you are saying is that the default assumption is that a valid reconstruction should look like the instrumental record over the 20th century, and I can’t argue with that. If we apply that criterion, the Marcott et al graph gets into a great deal of trouble in Fig S11c where they show the separate terrestrial and oceanic portions. The land portion is monotonically increasing from 1250 to the present whereas the oceanic portion is monotonically decreasing, but for a crazy uptick at the end which we can now discard. Needless to say the instrumental record does not suggest the two temperature series could diverge so sharply over 750 years. Also, in their Figure 1A they compare their global series to the EIV-CRU series, which is a land-only recon. They ought to compare the EIV-CRU series to the brown line in Figure S11c, and I suspect if the two are overlaid they will diverge rather badly.
From my laymans perspective – the authors paper is about THEIR proxy reconstructions, and they make the attention getting claims of recent warming based allegedly on their proxy’s and work.
The instrumental record may coincidentally match or support their warming claims, however they admit those warming claims are “not robust” and largely unsupported by their reconstruction data.
In the end what the instrumental record says should have no real relevance or value to their claims and conclusions. If their aggrandized statements regarding recent warming are based on the isntrumental record there is no need to include the recent data in the paper.
Which is advice they would have been well advised to heed.
I suspect, however, for a CAGW believer, the lure of the “hockey stick” is simply too difficult to ignore 😉
I agree, but ask what is sacrosanct about 1950 to current temperatures that are melded with prior paleo reconstructions? The choice nicely excludes the extremely warm 1920 to 1949 period. What would use of temperatures from 1920 forward do the the hockey stick?
Re: Ross McKitrick (Mar 31 13:56),
It’s worse than that! They arbitrarily “calibrated” the reconstruction by forcing it to have the same mean as the global EIV-CRU (land-only temperature!) reconstruction from Mann et al (2008) over the interval 500-1440AD. I quickly calculated that just by choosing the CPS-CRU from Mann et al (2008) their baseline would shift by 0.27C! Of course, by choosing other reconstructions and/or time intervals you can shift the baseline almost as you wish.
I also wish they would tell how they obtained the uncertainties for the post 1850 part in their spliced version of Mann’s EIV-CRU. I’d just hate it if we end up with another uncertainty mystery besides the infamous MBH99 case…
Somebody… Anybody? (…Nick?)
Can anyone address this…I’m curious about it too. Can someone find the combination-lock code to bringing this up at RC and pass the implicit-claims-of-nefariousness sensistivity test such that it might be addressed instead of deleted?
It speaks directly to one of the FAQs that was posted, though that particular question was re-worded into a version that was never really asked as far as I can tell. And as such, the answer isn’t the same either.
“I also wish they would tell how they obtained the uncertainties for the post 1850 part in their spliced version of Mann’s EIV-CRU. “
Why wouldn’t it just be the published uncertainties of CRUTEM? The ranges seem to match.
Re: Nick Stokes (Mar 31 21:21),
yes, it certainly is a possibility. However, in that case there is no “innocent” explenation for the splicing as they needed to add knowingly those uncertainties to the data obtained from Mann’s webpage. I’m sure you recall:
“However, in that case there is no “innocent” explenation for…”
Haha. Nick certainly walked straight into that one!
“yes, it certainly is a possibility. However, in that case there is no “innocent” explenation for the splicing as they needed to add knowingly those uncertainties to the data obtained from Mann’s webpage.”
hm…. which is potentially why there was no answer for this question?
Lars P. (Apr 1 17:37),
Well, I’m aware of a very plausable (which I believe actually happened) answer (which is not the one Nick proposed) and it is rather “innocent” for the authors. Let’s just say now that it also has some “implications”, so I’d prefer the authors themselves to answer the question 😉
Edit: UC just confirmed that I’m likely right with my guess:
Jean S, can you re-state the question here so that it is squarely in play?
[Jean S: Where did the uncertainties for the post 1850 part (CRU instrumental) come from in Marcott et al Fig. 1A? The CRU-EIV reconstruction is here (corresponding description and Readme.txt).]
Nick Stokes:
“Why wouldn’t it just be the published uncertainties of CRUTEM? The ranges seem to match.”
Too wide in the Marcott version. And it would be difficult to estimate the effect of Mann-smoothing. Here’s HadCRUT3 (NH+SH)/2 21-point-smoothed:
Re: Steve McIntyre (Apr 2 07:05), it’s time to give the answer to the puzzle… let’s see how Mosher would have proceeded had he been up to the task 😉
First, there are some clues in the files I linked to. Notice that Mann has updated his Readme.txt-file at least twice. One of the updates seems to indicate that not only Readme.txt was updated but apparently also the eivrecondescription.xls file. Let’s try to figure out what had changed there. Luckily, this is rather easy task as the original files are still preserved by NOAA and PNAS. Here’s the link to the NOAA file containing the (almost) original content of the corresponding directory.
Let’s compare the eivrecondescription.xls files. In the original files he’s essentially indicating that the reconstruction is ending AD2006 (i.e., essentially “grafting” the instrumental), and that has been changed to correctly reflect the fact that the actual reconstruction ends in 1850. So that seems to be the content of the update, but that doesn’t help us with the uncertainty mystery. Now Mosher would ask: were the above mentioned files the only files that had been updated?
Now check the dates in Mann’s own archive. Notice that all *composite.csv -files (which contain the composite reconstruction series) have also October 2009 date (curiously one day after the last update was supposingly done)! Now check the original *composite.csv files. There they are! Uncertainties are running all the way up to AD2006!
So what likely happened. Mann had screwed up with his CSV-files by extendind his AD1800 uncertainties over to the spliced instrumental (1850-2006). Marcott simply took Mann’s reconstruction with uncertainties and plotted them, but did not download them from Mann’s “official” archive but from NOAA (or PNAS). Mystery solved 🙂
Finally, I was aware of this mistake by Mann already in early 2009 (when it was uncorrected even in Mann’s site), but I had simply forgotten it until this Marcott incident happened. Just to say that nobody thinks I’m a detective comparable to Mosher 🙂
dude I am humbled. awesome detective work.
“took Mann’s reconstruction with uncertainties”
Always a bad thing 😉
If Mosher is Sherlock, aren’t we looking at Hercule S?
My personal item of curiousity here has been addressed by the folks at RealClimate.
See here:
http://www.realclimate.org/index.php/archives/2013/03/response-by-marcott-et-al/comment-page-1/#comment-325934
and here:
http://www.realclimate.org/index.php/archives/2013/03/response-by-marcott-et-al/comment-page-1/#comment-325947
It appears everyone is on the same page as far as what Marcott et all ‘doesnt’ do:
1. It’s not supposed to be used for (nor represent) Medieval/Modern comparisons — the only claims for ‘consistency’ come from a similar pattern of rough warming and cooling periods, the magnitudes are aligned to Mann’s proxy, but could well have been aligned to a different one (the selection of Mann et al, 2008 is appropriate therefore because it’s currently the standard-bearer)
2. It’s not supposed to make a call on 1850-present temperatures — that is left up to the instrumental record (the best data we’ve got for the whole set). So, conclusions drawn from Marcott et al that involve 1850-present are made in light of what is already known, rather than being new in the paper itself.
I believe I am satisfied with these explanations, and I get that there’s not going to be one reconstruction that by itself can take care of the whole of the Hollocene straight through modern times, so it was always going to require different ones with different calibration methods.
It looks like Marcott et al is being positioned for ‘the standard’ for the Hollocene through AD 1000-ish, then Mann et al for 1000AD – 1950-ish, and then the instrument record for 1850-present, with the overlapping periods being the points of refinement and examinations of consistency.
I suppose the question becomes (and has been discussed/debated) whether or not these above points have always been clearly made within the paper and its attendant press releases, and if instead its the blogosphere that flew off the handle with this one.
Re: Salamano (Apr 1 04:25),
http://www.realclimate.org/index.php/archives/2013/03/response-by-marcott-et-al/comment-page-1/#comment-325934
That is simply not true, gavin. By choosing a specific (published) global reconstruction (and keeping the time interval fixed), you can easily shift the baseline by at least 0.3C. The “20th C increase” is someting like max 0.8C, and you should take only half of that to the consideration due to the resolution issue. And there is no “God given” reason why the the interval should be 500AD-1440AD; using other intervals the range increases considerably.
The amount of the baseline shift is nothing more than the difference between mean value over a chosen period between Mann’s EIV-CRU and a chosen reconstruction. In order to get the idea of the possible magnitude, here’s the IPCC 4AR spaghetti diagram (NH only; but you get the idea).
I believe that specific point has been conceded…
http://www.realclimate.org/index.php/archives/2013/03/response-by-marcott-et-al/comment-page-1/#comment-325947
…with the addition of ‘that’s not the point of the paper or the alignment’. You can certainly use any reconstruction that covers that era to align the data (though must defend it worthiness as a standard).
The magnitudes are apparently not what the study is after during that era (or the modern era either), as other papers have it covered. Therefore, the only thing that needs to be shown in order to be deemed ‘consistent’ is that the Marcott et al proxy and its aligned reference both show incremental warming and cooling during the same basic time periods. Are you getting the same reading?
” The “20th C increase” is something like max 0.8C, and you should take only half of that to the consideration due to the resolution issue.”
Marcott et al have defined in the paper what they mean by “20th C increase”. It’s the instrumental mean for 2000-2009 minus the instrumental mean for 1900-1909 (Fig 3). I don’t see what resolution issue should cause that to be halved.
They compare that range with various distributions in Fig 3. I don’t see how differences in the anomaly base matching would greatly impact that discussion.
Nick Stokes (Apr 1 06:11),
C’mon Nick, you know better. If the other data is in a resolution of hundreds of years, you can not compare to it a difference of mean of two decades in another data.
It is already a stretch to say that they could take a difference of the mean of the first half and the second half of the the intrumental record for their comparision. Unfortunately, the instrumental record is so short, and that is pretty much the best available option (although it’s better to take the half of the linear trend, but there is not much difference here).
C’mon again. They should half the distance between vertical gray bars, and then you could change the position of the colored curves, say, half of a degree to the left or right. No impact to that discussion?
“No impact to that discussion?”
Their summary statement is:
“Global temperature, therefore, has risen
from near the coldest to the warmest levels of the
Holocene within the past century, reversing the
long-term cooling trend that began ~5000 yr B.P.”
It’s reasonable to interpret “levels” as a medium term average, so that isn’t particularly resolution-dependent. And yes, shifting the anomaly base offset would change the range – say from “not so near the coldest” to “even warmer” or whatever. But it doesn’t change the basic conclusion, which is that in terms of the way medium-term average temp has drifted over the Holocene, 0.8°C is a lot.
Salamano, you sound as though you want to be convinced and are less skeptical than most posting here. What you report from RC is not very enlightening in my judgment. Tacking an uncertain shorter term reconstruction (with its own set of issues) onto a very low resolution and uncertain longer term reconstruction and then attempting to tie to the instrumental record by way of tacking the intermediate reconstruction to the instrumental record is a very uncertain proposition. Would you agree?
I think I’m looking at it from a different angle, but yes, I guess I’m not as upset about what’s gone on with parts of the Marcott et al sage once I got my questions addressed.
I agree that “Tacking an uncertain shorter term reconstruction (with its own set of issues) onto a very low resolution and uncertain longer term reconstruction and then attempting to tie to the instrumental record by way of tacking the intermediate reconstruction to the instrumental record is a very uncertain proposition” — However…
(a) I don’t think the instrumental record is itself uncertain.
(b) The authors (and the folks at Real Climate) aren’t even saying that Marcott et al is useful for Medieval/Modern comparison (its been manually aligned to the ‘standard’ proxy of the period, it could have just as easily been aligned to a different one with some different results). The only value of this alignment as it relates to Marcott et al is that the two simply can be shown to rise and fall relatively in tandem, indicating they’re picking up relatively the same climate elements. Just about any reconstruction from this time could perhaps claim the same alignment consistency, and as such Marcott could have been tuned to any one of them.
In that light– If Marcott et al made known the non-robustness of the late sample, and they made known the artificiality of the alignment of itself to Medieval temperatures, then perhaps they’ve taken their proxy off-the-hook for the kinds of things we typically examine them for in the AD1000-present. Meaning, it should only be valued for its 11,000ybp – 1500ybp(ish) reconstruction (with its error bars) THEN something like Mann et al (2008) takes over with some overlap at the beginning through to 1950 THEN the instrumental record, with overlap at 1850… Stitching together the reconstruction from the beginning of the Hollocene through to the present using different methods and error bars with each…and each standing or falling with respect to their primary era of focus.
So then, I’m sympathetic to the idea of leaving aside the criticisms of Marcott et al that exist in spots the authors already have conceded. It’s Mann et al (2008), or whatever other proxy you use through that period that has more questions to answer. Then certainly when it comes to 1850-present, I’m fine with accepting the temperature record, so I don’t really care what Marcott et al admittedtly irrelevantly shows. So when it comes down to the scientific goal of understanding temperature over time through the Hollocene, I think we’ve got a graphic that shows different proxies ‘taking over’ for relevant eras in which they are more robust, rather than being spliced or graphed onto each other.
However, I DO find serious issues with the following graphic:
For it implies that one study is the source to go to for the entire above reconstruction in a seamless format. Instead, there should be different colors at different points with different error bars (perhaps even with slight breaks in between). Even with that sort of reconstruction though, legitimate arguments are still possible for those different eras, but with each paper individually taking the heat (so to speak) for each time-window. I do wonder how the error bars and such were calculated for these transition periods, as trying to stitch them all together will have problems at the ends. For example, I do not see the (already admitted) problems in Marcott et al when it comes to 1940 as a reason to question what the temperature record is 1850-present or 11,000 years ago – 2,000 years ago. But whoever created that graphic that Roger Pielke jr, has been railing about DOES need to answer for it, regardless if the actual science is pointing to that type of general temperature reconstruction after examining 3-4 different proxies/studies together. I don’t think it’s enough to say that its been dealt with by Marcott’s admissions when the paper is promoted in that way, with graphics like that one, but by the same token, I don’t see the errors in some of these claims rise enough to the level to doubt things like the instrumental temperature record. Unless I’m missing that more serious meaning, but I’ve been asking.
Salamano,
“…I don’t see the errors in some of these claims rise enough to the level to doubt things like the instrumental temperature record.”
Marcott is just another failure in the desperate attempt to match proxies with instrumental temperatures.
It is now a lot and time to ask the right questions. Especially since we knows that some proxies are quite good.
Salamano, you have provided a rather lengthy discourse on what Marcott shows and does not show, but a few threads back I took the Marcott authors at their word about resolution of their reconstruction and replotted it using a data point for every 500 and 1000 years. I linked to these plots previously and will not link here again since the calculations are simple for anyone interersted to make for themselves. What those plots show is a rise from the beginning of the proxy to a platuea from 9000 to 7000 years before present of 0.3 to 0.5 degrees C depending on the data point separation used and then the grap dives by 0.6 degrees C to present time. That exercise also shows what little influence the recent warming (as measured instrumentally) over a period of some 40 years has on 500 and 1000 year period.
The correct view of Marcott et al says we are headed into an ice age of unknown duration that is being mitigated over the long term average by an as yet unknown amount from AGW. That view would make for a very unique and publishable paper, but with a very different message.
Kenneth,
I actually asked this very same question and made a similar point/question over at RealClimate…
See:
http://www.realclimate.org/?comments_popup=15038#comment-326027
and
http://www.realclimate.org/?comments_popup=15038#comment-326031
The explanation that is given regarding the drop in Marcott et al that finishes the reliable part of the proxy is that the combination of different lenses through which we view the LIA period (using Marcott et al, Mann et al, and the earliest parts of the temperature record) posit that the LIA itself is a bump-down combination of a negative portion of short-term natural variability augmented by a bottoming out of the long-term forcings that were cooling at the time. I still believe, as I’m thinking you might, that this evidence taken together may well also mean that we would have been subjected to a precariously cooler regime (unknown feedbacks included– “snowball Earth” anyone? 😉 until what seems like we got off that track largely(presumably) on anthroprogenic means — natural/physical mechanisms don’t appear to be enough to get the job done quick enough, right?
Fast-forward to 1998, and we’ve got a scenario where natural variability is in phase with Global Warming forces, and we see an explosion in warming– presumably we’re just another strong El Nino away from doing it again considering current AGW force tracking as demonstrated by Tamino. But, my latest personal interest concerns the “New” feedback mechanism of crazy cold anomalies flying around the Northern Hemisphere actually created by the melting icepack. Stefan Rhamstorf (of “Moscow Warming Hole” fame) has been looking at this recently. He points out that the net temperature anomaly across the NH even with this phenomenon in effect is 0-change from the months before… but MY point is that in order for predictions to be correct, the net must continue to keep pushing positive. Else, there needs to be an admission that perhaps the Earth has a propensity to increase cooling feedbacks to at least some extent to mitigate the rise brought on by continuing AGW forces. Deep ocean warming is another one. I know that some climate science/activists aren’t “happy” with such mitigants (and often only work on the warming ones rather than the cooling ones), but it’s interesting to see how the temperature record is currently reflecting them all being out on the table as we get closer to leaving the lower bound of 95% confidence on a very expensive climate modeling effort that is hardly very old. [As a caveat, it is of course reasonable to postulate the relative risk/rewards involved in all feedbacks and warming/cooling knock-on effects for the future, additional sea-level rise predicated on deep ocean warmth perhaps being one of them.
Whoops… sorry about the tangent there.
So Marcott used Mann to show that Mann confirms Marcott confirming Mann?
From what I was seeing, the phrase “confirm” only relates to temperature movement (slopes) during the same reconstruction periods– not to the actual meat-and-potatoes of many folks’ concerns (MWP-modern comparisons, etc.)
By this same token, Marcott et al could have also “confirmed” Loehle et al and whatever other reconstructions that exist during that same period that show general warming prior to the top of the WMP and then a cooling afterward to the LIA (but only differed in their amplitudes and peak maxes).
“It’s worse than that! ”
Hard to keep pace with you, where are we now.. Let’s see
Your experiences in the field become very illuminated by the cynical appraisal in your comments, Ross.
Ross McKitrick raises the very important point:
The corresponding averaging of the temperature record to the same resolution of the proxy was graphically shown by David Middleton at WUWT in: Simple Test of Marcott et al., 2013 Posted on March 11, 2013, especially his Fig showing 140 year averaging.
Ross,
Very clearly expressed. Thank you. +1
“The sharp uptick in the instrumental record can only be compared against their reconstruction if they can show their low-frequency proxies are capable of registering such events.”
No, they recognize the resolution mismatch by making the comparison via an intermediate, CRU/EIV, which is long enough to compare with proxies and has resolution to compare with instrumental. Then they go on to try to remedy the resolution shortfall statistically:
“Because the relatively low resolution and time uncertainty of our data sets should generally suppress higher-frequency temperature variability, an important question is whether the Holocene stack adequately represents centennial- or millennialscale variability.”
How successful they do this is debatable. But they aren’t relying on proxy resolution of the modern period.
Thank you, Nick S, for succinctly admitting that Marcott et al has no point … apart from the schlock, horror of the initial “meeja Hockey Stchick”, of course
… what a wondrous ado, and all about nothing (I admit I’ve shamelessly plagiarized William Shakespeare here)
Nick Stokes,
I don’t understand your comment here.
If the proxies don’t reflect high frequency changes, nothing is going to fix the problem Prof. McKitrick points out (unless and until we know that current temps will remain stable for a couple hundred years). I don’t see how the attempt to scale the proxies to CRU/EIV (or anything else) can give proxies the retroactive ability to record high frequency changes. Even if CRU/EIV “has resolution to compare with instrumental”, that ability cannot be transfered to low frequency proxies via any form of statistical magic.
“…can give proxies the retroactive ability to record high frequency changes”
No, nothing can do that. The proxies don’t need calibration from instrumental. The only thing that needs matching is the anomaly base. We have a whole lot of proxies where the actual temps cover a wide range, but we combined the anomalies. We don’t know, and don’t need to know, the base mean temp.
Same with instrumental. All we need is a single number offset for those two bases. We can get that from their common overlap with the CRU/EIV series, which has enough resolution at both LF and HF to match each.
Nick Stokes,
Ok, but I took your comment above to state your disagreement with the comment from Ross McKitrick which you quoted.
As I understand it, McKitrick’s point is that you can’t make any sensible comparision between current instrument temps and the proxies unless you have reason to believe that the proxies can record high frequency changes or, I suppose, you already know that current instrument temps will remain at current levels (or maybe higher) for a very long time into the future. If we agree the proxies don’t record high frequency changes, then it seems to me we are left with: (a) McKitrick is correct; or (b) we already know what the instruments will show 120 or so years from now. If the answer is (b), that would seem to require a very detailed proof.
It seems such a simple question – asked and re-asked: If the reconstructions cannot be trusted to report the temperatures we think we know by instrumental means, why should we trust them for anything else?
[this is [probably snip-worthy]
The proxies they use only contain (they hope) a low frequency (300 year) temperature signal and therefore cannot be validated against (or compared to) the instrumental record.
I guess you could validate different types of proxies against each other and derive a confidence interval, but even this is probably beyond these guys. At the end of the day the paleo guys are basically just averaging and smoothing stuff.
Steve,
In your comment at 1:45 PM I believe you mistakenly added the final paragraph to your quote of Steig.
Yep. It should be split off typographically.
“Otherwise, their response is pretty much a filibuster, running the clock on questions that have not actually been asked and certainly not at issue by critics.”
Perhaps the term you’re looking for is “straw man”?
If the very excellent term “Straw Man” wouldn’t exist, we would have to invent something to denote this debate tactic. How about a “Hockey Stick Argument”? Or maybe the “Real Climate Fallacy”?
Mosher said on another thread, something to the effect that:
1. Make claim ‘a’ in the paper.
2. Make claim ‘b’ in the paper.
3. Hold a press conference and make claim ‘c’.
Claim ‘c’ is not supported by the paper.
If claim ‘c’ turns out to be “not robust”, the paper stands.
This is basically how this has played out. The paper itself isn’t the source of the controversy; the claims made to the press are. The claims made to the press absolutely depend on the blade. Now that the statements to the press are out in the wilds, walking back the blade is completely safe. Ideally, the press statements about the present being the hottest in 11,300 years, which are unsupported, should be walked back, but they won’t.
Kind of a neat “trick”.
But things are stranger than Mosher predicted, in that it turns out that claim ‘c’ – that the uptick was robust – was implied by the paper, as Ross has explained. Evidence that the uptick wasn’t robust wasn’t robust, so they said. To now say the uptick itself wasn’t robust is to contradict their paper, as well as all the “Hockey Stick confirmed” and “We’re screwed” headlines. I find this layers within layers approach quite hard to describe, including using RealClimate as a firewall to prevent the faithful learning that none of the answers really are, but I don’t think the right word is science.
I just reread Ross’ comment. You’re right. It’s worse than we thought. The paper itself is irretrievable schmutz.
alternative tactic: switch topics fast, gish gallop away. answer a different objection,
claim your opponent is engaged in a neverending audit, blah blah blah..
also known as the “look over there!” gambit …
bit o/t but BRE-X scandal & more spring to my mind for some reason.
Steve has covered this before, but needs rereading -https://climateaudit.org/2005/12/28/kennedy-editor-of-science-on-pbs/
What is a 4th grade Science teacher (or a 12th grade AP physics teacher) going to do if one of her students asks what it means for Figure 1C of the Marcott graph given the recent caveat: “20th century portion of our paleotemperature stack is not statistically robust, cannot be considered representative of global temperature changes, and therefore is not the basis of any of our conclusions.”
How do you respond to questions on how many points on Figure 1C they should consider invalid for making conclusions or inferences?
Would you recommend removing the “0-100” data from the graph to minimize confusion for a typical 4th grader?
Would you recommend doing anything in regards to the graph during a discussions with an AP physics class?
What I am a bit curious about is how this curios-looking FAQ-response ended up at RealClimate? (Or if it has been presented elswhere publicly before that)
It looks an aweful lot like a PR-campaign to defend the already achieved (publication in Science) and if PR and defence against repeated or more criticism RealClimate most certainly is the goto-place.
Other pertinent questions would be if the RC-team was being ‘helpful’ in preparing these FAQ-replies, or if they in part even were material in getting the thing published in the first place …
It appears this particular thread over at RC is more heavily comment-moderated than usual. It is their right, but it may well be that some of the “frequently asked” questions will not get answered here or elsewhere by the principle authors. They may have to be asked in other places in order to receive the hearing desired. Tamino seems to have been willing to take up some of them; perhaps Revkin as well. I typically like RealClimate as a source for this sort of explanatory value, but so far it is a little incomplete.
My personal bit of unanswered concern revolves around the “stack alignment” that makes the max in Marcott et al’s ~ 1000AD mean the same as Mann et al (2008)’s mean. If I’m not mistaken, it makes the “consistency” between the two rather artificial. [Jean S pointed this out originally I think]. If the whole point of the “consistency” is to merely grade as such if it gets warmer in certain places and gets colder in others, then just about ALL published reconstructions do this– even ones that show the MWP as warmer than present.
I also see somewhat of an inconsistency if its said that temperatures have been on a gradual cooling trend for nearly the whole Holocene, only ticking up at the start of the instrumental record. Would not the RWP and the MWP have at least some necessity to show forth, even if it remains below present values..?
In brief, they are unwilling to address candidly questions about their work that have been posed by first rate analysts. Such a stance can only bring shame upon them.
“…will not get answered here or elsewhere by the principle authors.”
Did you mean principal or lack-of-principle authors?
Think of RC as a stonewall behind which Marcott at al are hiding.
Commenters can be snipped and/or ignored. It’s published there so they won’t have to answer questions other than those they choose to answer, however incompletely.
And to continue with the metaphor, Marcott et al are behind the stonewall with there hands over their ears saying, and I’m quoting Pointman, “La, la la. I can’t hear you. I’m not listening.”
It has been my experience that RC has indeed been a place that takes on nearly all “Frequently Asked Questions” within Climate Science, as well as discusses much of the new contributions in the field.
They obviously don’t have to be the only place to tackle any/all of my questions, but there are several blogospheric conclusions (from places I consider credible) that this FAQ, while advertizing thoroughness, tackles more questions that were not actually asked, or rephrases asked questions into a more answerable form.
I think saying things like “some things some folks will never be satisfied with the actual answers” can be more of a cop-out. I’d like to see the answers before not being satisfied, you know? 😉
Which just makes my point above. They’re not trying to defend the paper, they’re trying to defend the unsupported claims made to the press by confusing the paper with the press statements. Defending the paper is properly done in the peer review process. RC is a media outlet. This is a media campaign.
no way will they leave this job to Cook at SkS. for a variety of reasons
Pielke, Jr:
Just a thought, and I may be wrong, but is it not possible that if Mann was a reviewer of the Marcott paper it may have been Mann’s influence behind the decision to respond via Real Climate. Mann’s all in support for this paper suggests more than a passing interest, perhaps even more than just the obvious interest to support the Marcott findings in an attempt to bolster his own hockey stick.
How’s that going, the bolstering of Mann’s hockey stick? I don’t seem to be hearing so much about that these days. 🙂
Posted Revkin’s question at RC….
March says:
31 Mar 2013 at 2:12 PM
Andy Revkin has a comment on the FAQ, pointedly commenting:
there’s also room for more questions — one being how the authors square the caveats they express here with some of the more definitive statements they made about their findings in news accounts.
How about it?
[Response: Can you be specific? The press release they put out doesn’t have anything contradictory. – gavin]
My reply left hanging over the borehole…
Perhaps you didn’t read the PR. the headline states: “Earth Is Warmer Today Than During 70 to 80 Percent of the Past 11,300 Years”.
Above we read:”Thus, the 20th century portion of our paleotemperature stack is not statistically robust, cannot be considered representative of global temperature changes, and therefore is not the basis of any of our conclusions.” Without the uptick there is no basis for the headline.
It’s April fools day downunder. Did you post this a day early?
Now posted but the last line snipped. Fair enough!
Obviously Mann is not an author, but this comment is interesting (quoted by Revkin). I doubt the authors tried to correct him and say that they didn’t mean that:
“The key take-home conclusion is that the rate and magnitude of recent global warmth appears unprecedented for at least the past 4,000 years and the rate at least the past 11,000….
…But, again, the take-home conclusion: the rate of warming appears to be unprecedented as far back as the authors are able to go (to the boundary with the last ice age). And the rate of warming appears to have no analog in the past, as far back as the authors are able to go.”
Re: gober (Mar 31 20:53),
yes, it is fascinating to see what the great climate scientist Michael Mann said was the “key” implication of the paper before the big smokescreen was needed at Real Climate. Are there going to be public corrections of this and all similar remarks that were spread through the media weeks ago? They can’t actually say anything about the rate or magnitude of recent decades compared to the previous 4,000 or 11,000 years, based upon the paper.
The entire logic of leaving a FAQ at RealClimate escapes science itself.
Given the serious issues Steve has raised with that paper, any explanatory essay should have been published as an official corrigendum in Science magazine. A corrigendum itself goes through peer-review, which would put Science on the spot to get it right.
Science senior editor Caroline Ash, and Julia Fahrenkamp-Uppenbrink, senior editor of the London Office, whose responsibilities include climate, have an obvious duty to intervene, to contact the authors, and to require that they submit an analytical update for peer-review and publication.
Posting a FAQ at RealClimate is completely unacceptable.
Also worth requoting pottereaton here (https://climateaudit.org/2013/03/14/no-uptick-in-marcott-thesis/#comment-404659)
The curious thing for me is that Shakun was selling exactly that same story to a Canadian radio station just last week, long after this controversy broke.
And another one, pointed out by Revkin himself in the AP article (http://science.time.com/2013/03/08/recent-heat-spike-unlike-anything-in-11000-years/#ixzz2PAzuLYxm)
The authors seem, in effect, to be refusing to answer specific questions relating to their published work. The next step should be a complaint (not a comment) to the journal. Furthermore, I think the relevant university ethics committees ought also to be alerted.
For the credibility of their work, and for their own reputations, they need to be made to properly answer the legitimate concerns that have been raised.
Better than previously, Robert. But I still disagree with you that meanwhile discussions about this, about why and how, should have been held back ..
I don’t see what this paper contributes without the 20th century portion.
That not only takes away the hockey stick, but it brings into question the responsiveness of the proxies to temperature as well.
And if “not robust” and irrelevant to their conclusions, then why was the 20th century portion included at all?
My thoughts exactly. And I think pointed out by others. Without the uptick, heck, even with it, this offers nothing new.
Indefatigability. For English, an ugly sounding word but it’s the only one which precisely describes your and the other Mac’s efforts. Lordy, am I glad you’re our team …
Pointman
I never ceases to amaze me by how much Real Climate Science (TM) is intellectually bankrupt.
Criticism by “he how must not be named” (“accusations from the usual suspects”) that forced them to react, yet they don’t even have the decency to link these “accusations”? So people would have a chance of finding out first hand what is “accusation” and what is valid criticism? To borrow a term I read from a proponent of RealClimateScience: “They rabidly hate linking to ClimateAudit.”
No word about the radio silence from RealClimate, yet “Readers will be aware of the paper”? I would say that even in their very first words they try to move the pea already.
Talking about the entire holocene, the holocene, and the early holocene in the intro paragraph, while linking to a hit-piece by the NYT that focuses on the blade of the hockey stick? Don’t look there, look here? Let the media focus on the (rotten) blade, we talk about how solid the rest of the stick is?
To paraphrase Bill Hicks:
Misrepresent your study like that, you’re off the scientific roll call forever. End of story. Everything you say is suspect. Every word that comes out of your mouth is like a t*rd falling into my drink.
Tony Mach – since the FAQ responses at Real Climate barely address any of the “accusations” raised at Climate Audit they have no need to link to any posts at Climate Audit. They are merely responding to questions raised elsewhere.
Somewhere.
I must be missing something here.
They compared the *instrument record* to the proxy reconstructed Holocene. The instrument record does show an uptick of warming in the past century.
Smoothing in the proxies means they are low temporal resolution. So they estimated the high-resolution variability that had been smoothed out of the proxies, as described in the FAQ and supplementary materials. That is why they say:
“These [2000-2009 instrument] temperatures are, however, warmer than 82% of the Holocene distribution as represented by the Standard5×5 stack, or 72% after making plausible corrections for inherent smoothing of the high frequencies in the stack”
That 72% comes from comparison to the bold ‘standard 5×5 plus high frequency’ distribution in figure 3.
So their conclusion (1) isn’t dependent on the uptick in the proxy reconstruction and (2) isn’t dependent on comparing the *smooth* proxy reconstruction to the instrument record.
It *is* dependent on how they adjusted the smooth proxy to simulate the proxy + high frequency. Was there work described somewhere that demonstrated they did that wrong? That would seem to be more relevant than this excessive focus on the universally agreed to be “not robust” proxy uptick.
Leo, you cannot compare the values from one data set to those of another, then note the extent to which they appear dissimilar and conclude that a change has occurred in the underlying phenomenon, unless you first show that they are equivalent measures of the same thing. The fact that the instrumental record looks very different from the proxy reconstruction might mean that the climate has changed, but it also might mean that the proxies simply aren’t measuring the same thing as the instrumental record. To account for that possibility requires constructing a statistical calibration model during the overlap interval.
You are correct that much depends on the validity of their argument that they “estimated” the high resolution variability that their underlying data don’t measure. This is an inherently very weak form of argument. They can’t observe the high-resolution variability that, once smoothed, gives their low-resolution proxy values. So they can’t say if their simulation of it is accurate. And if their simulation of it doesn’t look like the instrumentally-measured variability in the 20th century, that might mean the climate changed, or it might just mean their simulated variability isn’t very realistic.
“To account for that possibility requires constructing a statistical calibration model during the overlap interval.”
No, their proxies are independently calibrated, in °C, by laboratory means and other independent observations. So there’s no scale issue in making the comparison. They align the baseline by reference to CRU/EIV which is calibrated to instrumental, but extends over a long period so the proxy mean can be matched.
Nick, the calibration of the individual proxies only puts the input series in some kind of degree units. It does not mean that the output of their method has the same spatial and temporal sampling frame as the CRU/EIV series. And lining up the zero point doesn’t constitute a calibration either, although presumably it is a start.
But if you want to argue that their method is a valid calibration between their reconstruction and CRU/EIV, then you need to address the fact that the latter is a land record. The proper comparison would be the land record from the Marcott et al method (shown in S11c). It trends monotonically upward from 1250 onwards, whereas the CRU/EIV series trends downward over most of that interval. So how can they both be measuring the same thing?
I’m not a scientist, but I would like to know how these Marcott et al. proxies were calibrated:
In Revkin’s interview with Shakun, Revkin quotes Rohde on the time scale and resolution issue. He then asks for Shakun’s source of confidence that there are not 100-year “wheels” (that is what it sounded like) when Shakun speaks over him and describes them as “blips.” Revkin suggests that Rohde is saying their resolution is only at the 400-year level.
Shakun’s response:
Caveat: I’m not a stenographer, but it’s fairly accurate.
The relevant passages start around 2:10 in the interview.
Sounds like an unpolished but accurate description of their Fig 3, where they show the instrumental century change against various Holocene distributions.
Re: pottereaton (Apr 1 00:18),
In other words, they assume (at least) one of the conclusions. Can anyone admit circular logic?
Time for more popcorn.
Mr. Pete: further on in the interview he wanders into a bit of reflexive certitude and has to correct himself after being challenged by Revkin:
(my bold)
Takes place roughly at the 7 to 8 minute mark.
Revkin suggest it (kind of dispassionate), Shakun parrots it eagerly (at 2:00):
“Suuuuper-hockey stick! Yeah right!”
Yeah right indeed. A hockey stick on steroids.
(And I can’t stand people who smirk and nod like Shakun does. Revkin is almost delightfully “down to earth” in comparison. And I was kind of ignoring the videos, and now I wish I hadn’t seen them.)
Here’s the transcript of that part of the conversation:
That’s a lot of kinda/sortas. I don’t want to know what they kinda did – I want to know what they did.
I saw Andy Revkin’s interview with Elevator Shukan and I don’t think he was fooled then.
================
Bingo. Andy puts the Elevator Shakun video at the end of his DotEarth column, underscoring his hysterical performance. I think Andy’s a little upset.
=============
Re: kim (Mar 31 18:37),
Yes, it was most interesting that Revkin placed the Shakun interview video at the end of the new post….. because Shakun’s wild arm waving certainly gives a different “message” about the paper than the new party line in the FAQ.
Oh, will they call Revkin a “fossil fuel shill” for pointing out the inconsistencies (and worse) of the Marcott authors?
they are pulling out all of the stops. There are folks busy this Eastern Sunday in an unprecedented offensive on the credibility of Steve and Andrew Revkin. Pretty obviously coordinate, totally despicable and pathetic. This type of thing would not be needed if the science was real.
Eric:
This is better than Superbowl Sunday
Haha. Given the rumbling of the zamboni, what’s the harm in laughing out loud?
Ouch! But thanks, both a ya.
====
At some point Revkin has to get tired of being embarrassed and recognize that his really cool access to the smartest climate scientists is instead just him being used by a bunch of bumbling scientist activists to broadcast their agenda. Perhaps that day is approaching.
My sense from the couple of additional comments from Revkin is that he knows that the Marcott et al response is totally inadequate and its inadequacy raises additional questions as to the authors professional integrity when dealing with the media. I expect Roger Pielke Jr will hit this aspect very hard.
Gavin provides a response to questions left unanswered….
31 Mar 2013 at 3:35 PM
Re #11. Can you be specific? Steve’s post doesn’t have anything untruthful.
Here are a few unanswered questions posed by Steve McIntyre.
[Response: Some suggestions below, but hopefully one of the authors can chime in if I get something wrong. (some editing to your comment for clarity). – gavin]
They did not discuss or explain why they deleted modern values from the MD01-2421 splice at CA
[Response: I imagine it’s just a tidying up because the three most recent points are more recent than the last data point in the reconstruction (1940). As stated above, there are not enough cores to reconstruct the global mean robustly in the 20th C. This is obvious in the spread of figure S3. Since this core is quite well dated in the modern period, the impact of this on the uncertainty derived using the randomly perturbed age models will likely be negligible. – gavin]
Or the deletion of modern values from OCE326-GGC300.
[Response: Same as previous. – gavin]
Nor do they discuss the difference between the results that had (presumably) been in the submission to Nature (preserved as a chapter in Marcott’s thesis).
[Response: You are jumping to conclusions here. I have no idea what the initial submission to Nature looked like, and nor do you. From my experience working on Science/Nature submissions, there is an enormous amount of work done to take them from a first concept to the actual submission. In any case, with respect to the thesis work, the basic picture for the Holocene is the same. – gavin]
Nor did they discuss the implausibility of their coretop redating of MD95-2043 and MD95-2011.
[Response: The whole point of the age model perturbation analysis (one of the important novelties in this reconstruction) is to assess the impact of age model uncertainties – it is not as if the coretop date is set in stone (or mud). For MD95-2011, I understand that Marcott et al were notified by Richard Telford that the coretop was redated since the original data were published and that the impact of this on the stack, and therefore the conclusions, is negligible. – gavin]
Perhaps Marcott et al can provide the answers.
[Response: I will ask, and if there is anything else to add, I’ll include it as an update here. (Just as an aside, please note that I am more than capable of reading McIntyre’s blog, and so let’s not get involved in some back and forth by proxy. I am not interested in playing games.) – gavin]
Considering all the “games” Gavin & co. have played through the years, including uncredited borrowings from ClimateAudit, not to mention countless failures to engage issues in an honest and timely manner, this is enough to choke me:
I gather then that we’re not supposed to consider Dr. Schmidt’s answers/elaborations as proxies for the authors’….
I appreciate Gavin Schmidt’s input on this. But if he doesn’t like proxies, maybe he could ask the authors why in the world they _won’t just answer questions_? It does not reflect well on them or their enterprise. I’ve mentioned this before, but why not again: Why should people like Steve McIntyre have to waste their time trying to reverse-engineer fuzzy papers? Provide complete information on how the results were derived.
‘notified by Richard Telford’, ::grin::, who reads ClimateAudit.
=====================
So does the more than capable gavin, ::grin grin::
====================
Am I right in thinking that Steve Mc’s past attempts to post comments on RealClimate have been blocked?
A reason he started Climate Audit, as I recall. And haven’t they always hated that second word.
Steve; CA was started in response to RC, but not because they blocked my comments. It is true that they subsequently blocked comments, even when the comments were on their critique of my own work. In response, I wrote a post that they vehemently objected to: Is Gavin Schmidt Honest? But the questions raised in that post were valid. Even for comments that were eventually posted, they would delay approval, sometimes for a couple of days, and then slip them in – a technique that they smugly discussed with Jones in the Climategate dossier.
ISTM the story here is still all about divergence between proxies and temperatures. The proxies fail to match up very well to the thermometer records, hence the various forms of legerdemain to make them look better (disclaimed in Marcott et al’s case through the other side of the mouth). But if they diverge when we’re able to see, why wouldn’t we expect them to diverge when we don’t have comparison thermometer data?
Hence the controversy is not really about an uptick at the end–the real issue is the “shaft” of the stick, as always. That’s what’s needed to make today’s climate “unprecedented” and “alarming.” Harping on the uptick is counterproductive because it makes it sound like one is doubting the thermometer record.
The stronger explanatory stance is to say “We already knew from the superior thermometer data that temperature went up over the last x years. The paleo stuff had nothing to add to that. In fact, the paleo data don’t even pick up this recent increase unless they are manipulated by arbitrary re-dating, which is the “trick” employed by Marcott et al. So the paleo data also quite likely missed past increases that occurred long before human CO2 emissions had any effect, such as the MWP, meaning that today’s warming is quite likely not that different from what happened a few centuries ago.”
Reblogged this on Climate Daily.
Despite the paper has fully blown up. the authors do not commit to their duty of cleaning up, correcting or retracting. Instead, they just rely on the continuation of the current dysfunctional system. Very disturbing.
If the modern period is not robust, which seems obvious to me and likely was to the authors, then it seems that the two questions are how well were the proxies calibrated to modern instrumental temperatures, and to what extent could the modern period of warming have occurred during the proxy periods without showing up in the reconstruction?
It seems that the proxies don’t really overlap the instrumental period in any meaningful way. In that case they couldn’t be directly calibrated. It sounds like (and I haven’t yet read the paper in detail) they calibrated against a different reconstruction, which had previously been calibrated against the instrumental period. I wonder how you could calculate the error bars for that? I wonder if they show how robust it was to selection of the reconstruction to calibrate against. Ross indicates that they just chose one and don’t discuss that.
It would require a simulation to see how the modern warming would affect there proxy reconstruction if it had occurred in a previous time period. A very interesting simulation would be to see how a 2C warming over a two hundred year period would look.
“It sounds like (and I haven’t yet read the paper in detail) they calibrated against a different reconstruction, which had previously been calibrated against the instrumental period.”
The proxies are individually calibrated independently, and then anomalies created from a common period, and then a recon. The CRU/EIV combination is then used as an intermediate to align the mean to instrumental. I don’t think that has big error implications.
“To compare our
Standard5×5 reconstruction with modern climatology,
we aligned the stack’s mean for the interval
510 to 1450 yr B.P. (where yr B.P. is years
before 1950 CE) with the same interval’smean of
the global Climate Research Unit error-in-variables
(CRU-EIV) composite temperature record (2),
which is, in turn, referenced to the 1961–1990
CE instrumental mean (Fig. 1A).”
Isn’t this exactly what I said? They align the mean with the CRU reconstruction. That is essentially the calibration step. As they say that is one step away from calibrating with the modern instrumental record. That kind of indirection has to have some significant error implications.
Re: Nicolas Nierenberg (Mar 31 22:46),
I guess you already noticed that the “CRU reconstruction” is nothing but Mann’s EIV-CRU reconstruction. As I said earlier, just by picking the CPS-CRU recon from the same Mann’s paper changes the baseline by 0.27C. Not to speak of other reconstructions or other time intervals.
No, it isn’t the calibration step, as normally understood. It doesn’t set the scale; it doesn’t do anything for individual proxies. It matches up two different anomaly bases for aggregate measures.
They seem quite dependent on the Mann et al. reconstruction. They also use it to determine the level of variability as if it was an instrumental record. I would think they might have tried seeing how sensitive their results were to the choice of different reconstructions as any given reconstruction has strengths and weaknesses. Certainly it isn’t clear that the Mann et al. reconstruction represents the variability of temperature accurately.
“To examine the sensitivity of our main conclusions to this missing variability, we use the 390 Mann et al. reconstruction (2) to add the amount of high-frequency variability exhibited in 391 global temperature over the past 1500 years to the Holocene stack (1) as white noise and (2) as 392 red noise. Our aim is to determine how much this missing variability may widen the Holocene 393 temperature distribution.”
Nick, you are right but I’m sure you recognize that the realignment is particularly problematic when you consider their oft-repeated conclusions in the paper. For instance, from the abstract:
If we naively “assume” that the authors didn’t mean to imply that their now silly uptic was representative of recent temperatures, we are only left with a comparison of the reconstruction including the specific offset of this graph to global temperature records. If the offset is basically a dart throw as Jean S and others have shown, what does that mean for the conclusions?
In the best possible light, the standard “unprecedented” conclusions are not supported by the paper. I can think of other ways to state the same thing though.
I’m sure you guys have thought about this more than me, but the mean of the period in the Mann reconstruction has to have an error range. I didn’t see anything in their paper that specified how large that error would be. So you have to add that error in some manner to the error in their reconstruction during that period when you do the centering/calibration. I’m not sure in what manner these errors would combine but it needs to be added to their own error bars for the reconstruction in the first place if you want to compare to the modern instrumental record at all. (Which even has error bars, but much smaller of course.)
Nicolas, there are interesting aspects to their “uncertainty” estimation that are still undiscussed, though Jean S and I have alluded to some issues. Oddly, their “uncertainty” about temperature is at a minimum about 5500 years ago. This is an artifact of their methodology.
Therefore the precision of this statement. “These temperatures
are, however, warmer than 82% of
the Holocene distribution as represented by the
Standard5×5 stack, or 72% after making plausible
corrections for inherent smoothing of the high
frequencies in the stack (6) (Fig. 3).” Seems highly suspect to me. They are treating the Mann et al. reconstruction as having the same frequency variation as an instrumental record, and they aren’t adding any error term for the mean of the Mann et al. reconstruction. (Let alone questioning whether they should look at multiple reconstructions during that period, or what there error might be when only looking at that period.)
Jeff,
I’m sure from the context that that particular statement (“1500 years”) refers to other people’s recons, not to their ‘uptick’. I don’t think the anomaly base alignment is a dart throw; Jean S has shown variations of order 0.3°C. But I think M et al, in the discussion which is related to Fig 3, to the proper contrast. They’ve looked at the Holocene, and showed a pattern of variation, warmer originally, then declining to recent. And they say the instrumental rise is quite large in comparison, which is a reasonable observation.
Nick,
I guess assuming that the relative range of temperature changes in their reconstruction is accurate, you could say that the recent instrumental increase is a significant percentage of that and therefore large. But, of course there could have been temperature changes much larger than that during the period they cover that would have been completely filtered out.
Also now that I am looking more closely at their graph I wonder what the real odds are that the lowest temperatures in 12,000 years just happened to have occurred 100 years ago? Right before the modern age of CO2 increases? A coincidence like that would make me pretty suspicious without a physical explanation. What would have accounted for a nearly 1C drop in temperature over the last 7,000 years?
Nicolas
“..without a physical explanation…”
They do talk about that on p 1200, para beginning
“The general pattern of high-latitude cooling in both hemispheres opposed by warming at low latitudes is consistentwith local mean annual insolation forcing associated with decreasing orbital obliquity since 9000 years ago (Fig. 2C).”
Nick, so this is pretty funny to me. They reference Huang 2004 as their source for decreasing insolation over an 11,500 year period to the present. I found the paper from S Huang 2004 GRL. It only covers 1500 to the present and claims increasing temperatures and insolation during that period. See figure 4.
This has all the correct page numbers etc. I have no idea how this could be a reference for an 11,500 year period.
Click to access 2004GL019781.pdf
Yes, there is clearly a numbering error here. Huang is being cited in connection with Fig 1e.
Yes, and it completely falls outside the error bars for their reconstruction. In fact two of the four reconstructions that they plot clearly have no correlation with their reconstruction. Again this brings into question for me the size of the error bars and the overly confident comparison to the instrumental temperature period.
I think the insolation results were computed using this interesting site.
Reference
Nick, do you really think so? Laskar isn’t in their list of references, and that site seems to be using scales of millions of years. What would make someone think their figures are useful over an 11,500 year period? I don’t see anything in the paper that discusses using it at such short durations, and the web site seems to anticipate much longer time frames.
It seems that you have either gotten bored, or agree with me that the insolation results are unlikely to be from the Laskar paper/site. Interesting to have a completely missing reference like this.
Nicolas, I’ve been following your and Nick’s conversation and also found it very interesting that over at Realclimate, Chris Colose asked a related question (that Nick sort of answered) but Gavin has completely ignored. Like the dog that didn’t bark…
Link to Chris’s question: http://www.realclimate.org/index.php/archives/2013/03/response-by-marcott-et-al/comment-page-2/#comment-326110
72
Chris Colose says:
1 Apr 2013 at 10:03 PM
I have a (different, probably easier) question regarding the Marcott Study.
I was never under the impression that direct changes to obliquity (compared to precession for example, which is of course modulated by obliquity) was the main driver of Holocene climate change, in particular the high-latitude seasonality. This is mentioned in their text on a few occasions. Did they get this right?
And Nick’s response: http://www.realclimate.org/index.php/archives/2013/03/response-by-marcott-et-al/comment-page-2/#comment-326146
77
Nick Stokes says:
2 Apr 2013 at 4:55 AM
Chris Colose #72
I don’t know. But I found they have messed up the references for their insolation sources. I think they meant to refer to Laskar et al 2004. And used this site for the calculations.
“What would have accounted for a nearly 1C drop in temperature over the last 7,000 years?”
Look at S26 of Marcott SM. When they remove the North Atlantic sites, the drop disappears, and the recon for global is ‘exactly’ the same as 30N-30S. They seem to use this info to point out what is wrong with the models. AMO? So what are the North Atlantic sites which seem to be critical to their recon? And have any of them been redated?
Nicolas,
Sorry, I didn’t get bored, I just missed that you’d replied. The web-site works – I did a calc that would have given a latitude band to the requisite resolution.
The thesis references Laskar – that’s how I found it. Laskar gives formulae that are applicable over short ranges as well as long.
So I see Gavin says the following: “The whole point of the age model perturbation analysis (one of the important novelties in this reconstruction) is to assess the impact of age model uncertainties”
That’s bang on, and I’ve commented here about Marcott’s perturbation algorithm and analyzed its decisive contribution to the uptick. Hope this was noted.
Steve: I’ve noted your interesting comments about perturbation and will try to work it through. Can you post up code for what you did, to save time for others
OK, but I use a proprietary language, so I’ll give the logic & a little generic code. First a quick discussion about perturbations:
Perturbations are a substitute for complex statistical analysis, particularly for avoiding Kolmogorov-Smirnov testing. The idea is that you “actualize” (Marcott: “realize”) your standard errors by manifesting them as data, then subsequent processing already includes all the error-fuzz, all done. So you fuzz your data into the error ranges, e.g. your datum says 1500AD with standard age error of 150 years, so you write 1000 records into the range 1350AD-1650AD — taa daa, your data now includes the standard error.
Now this is why I’ve been stamping my feet, because Marcott et al have manually set the standard age error for 1950AD to zero. Wow, 1000 permutated records, all dated 1950AD, from just 1 datum! After this processing, all data are collected into 20-year bins, and the 1940AD (ie, 1930AD-1950AD) bin is toweringly dominated by its own data, with very little donated from earlier dates. There are two end effects operating simultaneously to isolate the 1940AD bin: (1) age error = zero, and (2) data bleeding in from only one side, ie, earler times, nothing from 1960AD+. So initial temperatures are well-preserved in the 1940AD bin after perturbing.
OK, here’s some simple logic to do it. Note I haven’t got the temperature uncertainties so I haven’t perturbed the temperature, but that is just a systematic (per-proxy) effect which is unimportant to the outcome. Also I’m not geographically (5×5) weighting.
(1) Get the data, use Marcott’s temperatures, convert the 73 proxies from temperature to anomaly, and then add an offset to each proxy so that its anomaly over 4500-5500BP is zero. You’ve done that already.
(2) Get the earliest and latest date for each proxy. You won’t be permutating beyond those dates.
(3) Permutate each datum as follows: (RANDOMIZED is a re-seeded random # between 0 and 1)
if age_error is missing then age_error=0
for i=1 to 1000:: agep = age + age_error*(RANDOMIZED – 0.5):: output:: next
(you’ll need to retain the i value to do it Marcott’s way)
(4) Bin agep into 20-year age bins. Be careful that negative ages round the same as positive ages. Now remove any age-bins which are outside the earliest and latest dates for each proxy, see (2).
(5) Get avg temperature for each i-proxyID-bin combo, now sort that by i-bin and get the avg temperature for each combo. You now have 1000 temperature stacks for each of the 73 proxies.
(6) Sort all of it by bin (that is, the 20-year bins). Get avg temperature per bin. You now have the final temperature stack.
The above code weights each of the 73 sites equally, whereas Marcott weights by the rules of the 5×5 grid. That will account for differences — following are Marcott’s & my first 20 bins:
age Marcott my
bin anom-C anom-C
— ——- ——
10 0.60 0.59
30 -0.10 0.32
50 -0.26 0.33
70 -0.26 0.06
90 -0.19 -0.04
110 -0.16 -0.07
130 -0.15 -0.12
150 -0.25 -0.03
170 -0.22 0.11
190 -0.22 -0.12
210 -0.37 -0.33
230 -0.35 -0.35
250 -0.37 -0.29
270 -0.35 -0.23
290 -0.33 -0.21
310 -0.32 -0.07
330 -0.30 -0.04
350 -0.29 -0.06
370 -0.27 -0.04
390 -0.24 -0.04
410 -0.22 0.03
Not exactly the same as Marcott’s, but he does that 5×5 geographical weighting and maybe some other stuff. Also, even Marcott cannot exactly reproduce his output because it’s the result of a random process, so it’ll be a little different each time. Makes one wonder how many time he threw the dice before he got a roll that he really liked. Whew — enough from me, cheers.
Steve: thanks very much for this. Contributions like this from commenters are hugely appreciated.
Errm, that should be
for i=1 to 1000:: agep = age + age_error *( 2 * RANDOMIZED – 1):: output:: next
Not a huge difference though, output starts:
age Marcott my
bin anom-C anom-C
— ——- ——
10 0.60 0.62
30 -0.10 0.34
50 -0.26 0.33
70 -0.26 0.05
90 -0.19 -0.04
110 -0.16 -0.02
130 -0.15 -0.15
150 -0.25 -0.13
170 -0.22 0.06
190 -0.22 -0.04
210 -0.37 -0.30
230 -0.35 -0.30
250 -0.37 -0.26
270 -0.35 -0.22
290 -0.33 -0.12
310 -0.32 -0.15
330 -0.30 -0.10
350 -0.29 -0.14
370 -0.27 -0.09
390 -0.24 -0.08
410 -0.22 0.03
430 -0.25 -0.16
450 -0.23 -0.12
“(3) Permutate each datum as follows: (RANDOMIZED is a re-seeded random # between 0 and 1)
if age_error is missing then age_error=0
for i=1 to 1000:: agep = age + age_error*(RANDOMIZED – 0.5):: output:: next”
If I understand rightly, you’re using the age_errors from col H of the spreadsheet. I understood how they did it differently, though it may have much the same effect. Looking at line 83 of the SM, they say they perturb the age models. To me, that suggests they perturb the carbon dating age points (cols L,M), and then interpolate from the perturbed age points. That might make a difference near the endpoints, tho not at BP 10.
For us statistically-challenged types, can you (NZ Willy or Nick Stokes or anyone else) explain how the perturbation analysis would result in core-top re-dating (as Gavin appears to suggest by raising perturbation analysis in response to the core-top re-dating critique)? As I (think I) understand from NZ Willy’s walk-through, the analysis is intended to address uncertainty due to age error by simulating values for the period spanned by the ‘age error’ range. And I (think I) understand that age errors are a facet of carbon-dating, such that the re-calibration of the proxies might have resulted in age-error values greater – perhaps significantly greater – than the values derived in connection with the original publication of a given proxy series.
But even assuming newly-derived age errors for a given depth of a given core that are hundreds of years wide, I don’t understand how perturbation analysis could result in a determination that the originally-reported core-top dates were off by so much – where ‘so much’ appears to my lay eye to be an amount greater than the reported age error values (using NZ Willy’s apparent approach rather than Nick Stokes’). And even if the newly-derived age error values were thousand-year spans, the result of a perturbation analysis would seem to be the simulation of core-top values across the entire span, not the re-dating of the core-top value from one end of the span to the other.
Steve: my understanding of their perturbation analysis (and this is with the caveat that I havent even started to parse it) is that they assume that the coretop dates are properly dated. The impact of coretop redating by itself can be seen by contrast to the Thesis figures.
I think you’re right, Nick, it seems they age-perturb the age control points (these are age-uncertainty minima in each proxy profile), but don’t interpolate in between, rather they random-walk the points in-between with a maximum departure (from the staight interpolation) of 150 years. This can be coded but not trivially, and the bear is that the randomness prevents full repeatability — so Marcott’s output can’t be reverse-engineered in the traditional sense.
Important point for “uptick” hunters: Note Marcott’s SM line 66: “The majority of our age-control points are based on radiocarbon dates”. So which ones are not — well, the core tops, of course, with their age-uncertainty set to zero. So these (age=10) are used as perturbation control points, and the neighboring ages of 30,50,70, etc are random-walked away from there which means that their perturbation widths are quite small, because you can’t random-walk far away in the y-axis until you have gone some distance along the x-axis in this schema. This means that my own algorithm, above, *overstates* the perturbation width in the modern era. That will be why Marcott shows more of an uptick from 30BP to 10BP than I do. Dare I say this perturbation method looks like it was designed to preserve that uptick? Oh, I guess I already have.
dsmader
” I don’t understand how perturbation analysis could result in a determination that the originally-reported core-top dates were off by so much “
I don’t think that’s the perturbation analysis. I think it’s just physical reasoning.
If you look at the 1000-yr change, MD95-2043, Cacho et al had C14 age points at 14 cm and 54 cm (and more) at which they assigned ages 1980 and 3216 years. They then extrapolated to 0 cm using a uniform deposition rate assumption, giving age 1008 yrs. But any uniform deposition rate at 0cm implies 0 yrs.
Now it might be that 0 cm is not the actual surface, But another core from the same (Cacho) group M39-008 (also deep water) starts at 5 cm, and in fact seems to be set to age 0 at 0 cm.
If you look at the original MD95-2043 paper, it’s looking for sudden change events over the last 50000 years. They aren’t particularly concerned about the last 1000. In 2006, when they wanted to compare with some other cores, some authors including Cacho used re-dated MD95-2043 (tho not, I think, the coretop).
NZ Willy says: “even Marcott cannot exactly reproduce his output because it’s the result of a random process,, so it’ll be a little different each time.”
If this is true, it strikes me as incompetence. Pseudo-random generators are “seeded”, typically with a large integer or a longish character string. The same seed used again will produce the same sequence of random numbers; being able to reproduce a run, exactly, is useful for debugging, for investigating possibly spurious results, for archiving results, and for publishing reproducible results.
Am I wrong in this? Have I missed something?
Some packages have functions that use the datetime to build a seed, when the seed isn’t explicitly provided. If you are coding for reproducible research, you supply the seed, and you can get the same results with every run. But if you don’t want reproducible results, they’re never mandatory.
jonhl: agreed! But I would think “if you don’t want reproducible results” you would publish in the Journal of Irreproducible Results rather that in the journal Science…
I would second the comment/request of dsmader. Additionally, is there any indication in the field that perturbation analysis gives a more accurate picture than the K-S method mentioned?
K-S is difficult to use & interpret, and referees don’t like to see it. “Accuracy” is a misnomer in the analysis of small data which is K-S’s forte. It’s an interesting thought — what if Marcott had not perturbed, but instead gave K-S confidence levels on his output. We would all be blinded by the science. Oops, must not give them ideas… 🙂
Sunday night humor…. At least Mann doesn’t think that Revkin is a “fossil fuel shill”…. Funny to see the notorious Michael Mann trying to rein in some of his more strident friends in the Alarmist camp. No one should ever want a character reference from the integrity-challenged Michael Mann. Someone went ballistic because Revkin stated he thought the release of the Marcott et al. (2013) FAQ on Easter Sunday was “irksome” — so to some ppl Revkin is a “fossil fuel shill” merely for a critical remark about the timing of releasing long-awaited info over a holiday weekend.
no one needs Michael Mann as a character reference
At the end of the Twitter tiff Roger Pielke Jr wasn’t too impressed:
Roger Pielke Jr. @RogerPielkeJr 1h
I’m looking closely at Marcott hockey stick paper, press release, media comments … Another ugly episode for climate science. Blog coming.
This is excellent news. I expect Roger to be forthright on the misleading PR that the authors generated around what is looking more and more like a piece of pedestrian work dressed up for PR purposes.
Hey, I called it right…
You did. The article’s called Fixing the Marcott Mess in Climate Science and says this near the beginning:
They’re good steps but even when they’re all said and done (and I don’t wish to cast aspersions by implying any doubt that they will be) I think Kim put it best on Climate Etc on 19th March:
Pielke won’t call it a lie but untruth would have done just as well. It’s vital that they do for the ‘Scythe’ what they signally failed to do for the Stick. The name was always highly suggestive.
Half the reason I’m of the belief that CO2 doubling sensitivity is sub-catastrophic is that my personal best take on the facts known to me points me there. But the other half – and what is relevant to this thread – is that I know how people behave when they believe the facts are on their side, and I know how they behave when they don’t. The behavior of the CAGW community is so thoroughly and so consistently of the latter sort, that if the world actually IS headed to catastrophic warming, it will be THEIR fault the political action wasn’t taken to head it off. If you truly need to convince me of something, you must first behave like you yourself have faith in your facts. If you do otherwise, the fault is only yours that I am not swayed.
Exactly – behaviour always gives the game away. In a controversy like this the truth is always out there in front of everyone. That’s what’s so fascinating about this whole AGW business.
It is sad, but wholly expected, that the FAQ does not deal honestly with the Q. How could it, when Steve has caught them with both hands in the cookie jar in an absolutely irrefutable way.
The more interesting question is, what to do about it, where and when. RC is certainly not the appropriate forum given a Science publication. Perhaps OSU, which might be concerned for its academic reputation. Perhaps IPCC, which should be (but probably is not) concerned for theirs. Perhaps the MSM looking for an angle; recent UK developments come hopefully to mind.
The issue isn’t here–Steve settled that quite nicely, more than once. It is ‘there’. Some of you all need to take it to them ‘there’, as others of us are beginning to do.
Regards for Easter
I sensed an obtuse, diversionary post of the type published at RealClimate would be the response of Marcott et al when Clark wrote to Revkin saying:
Frequently Asked Questions are what the average person might ask the authors of the paper–questions with answers that people with any knowledge of the subject can answer easily for themselves. Superficial questions like “Q: Did you collect and measure the ocean and land temperature data from all 73 sites?” The questions being asked by Steve and others on this blog are not those kind of questions. Their questions are incisive go to the core truth of the paper and challenge its validity. For authors/scientists, to not deal with these kinds of questions forthrightly is not only cowardly but flagrantly deceitful.
It is hard for me to see any worthwhile recourse to the Marcott et all paper other than the formal submission of a rebuttal paper. The faulty claim and emphasis occurred in the paper itself and no amount of post factum commentary on non-robustness goes even half way to rectifying the situation. As OldWeirdHarold says:
“The paper itself is irretrievable schmutz.”
My vote is for a formal rebuttal with the alternative of well-founded blog based critique being commendable but limited in its impact.
My vote is for a formal rebuttal…
Well go ahead and write one; I wonder what it would look like.
Peer review could be improved by publishing the comments or suggestions made by the anonymous reviewers. They should be encouraged to write in a fashion that preserves their anonymity if this concerns them.
In some cases reviewers are quite happy to come forward and discuss their comments and the reasons for them.
RC sure picked an interesting day to send us all down the rabbit hole.
This is simply hilarious. Real Climate:
“Our view is that the results of the paper will stand the test of time, particularly regarding the small global temperature variations in the Holocene. If anything, early Holocene warmth might be overestimated in this study.”
Run Marcott! Run Shakun! Is it Halloween already? The Upside Down Mann is gonna get ya!
I mean really? Really? This is their answer? We’ve all seen those movies where the “bad guy” blusters and obscures, blatantly not answering the question and obviously misdirecting. The audience practically wants to yell the truth at the screen. They want to see justice served. Those are movies. This is real life. No one is supposed to actually behave like that bad guy in real life, least of all educated scientists. Marcott et al know they are playing coy games and being unresponsive and something in their moral make-up is letting them be okay with that.
Yes, I understand the seductive nature of “noble cause corruption”, I know that Michael Mann and others are whispering in their ears, telling them they are up against devious “fossil fuel shills” and so on. Even if Marcott et al really believed that to be true, despite the lack of any evidence, how far does one have to fall from grace to sink to such gamesmanship that integrity and truth are acceptable casualties?
Pielke, Jr. weighs in with a scathing blog article:
http://rogerpielkejr.blogspot.com/2013/03/fixing-marcott-mess-in-climate-science.html
Pielke, Jr., linked above, concludes:
Better yet,
Doesn’t that sum up 95% of the problem and the other 4.5% SkS…
The FAQ are also posted at Marcott’s web page here:
http://proglacial.com/Public%20Links/FAQ%20-%20Holocene%20Temperatues.html
The FAQ text is in an image so that search engines will not find it.
The date of revision for the FAQ at the bottom of the page is incorrectly given as 3/29/12 which is only one year off. It too needs to be “redated”.
The date at RC is given as 3/31/13 so the FAQ may have been published two days earlier at Marcott’s site.
Marcott & al’s conclusion that modern temperatures are greater than ‘x%’ of the Holocene is not supported by the data. Their reconstruction does not have the resolution to support such a claim. There could be hundreds of spikes in the actual temperatures from the Holocene much warmer than at present and they would simply have averaged them out.
Marcott & al would need to average out modern temperatures to the same resolution as their reconstruction before any conclusion can be drawn. This would drop modern temperatures at the endpoint by about 0.4 C (assuming a 0.8C rise over 120 years), which would make modern temperatures unremarkable in comparison to the past.
Marcott & al cannot simply compared the magnitude of two series if the resolution is not matched and obtain a meaningful result. Yet this is what they have done in their conclusion, which suggests that their conclusion is flawed.
Since Marcott et al ends in 1950 and the IPCC has already concluded the temperature increases pre 1950 are due to natural causes, Macrott et al cannot be showing us anything about the human influences on temprature.
Rather, it must be concluded that what Marcott et al is showing is only the natural variability in temperature and if any weight is to be given to the uptick, it is that it shows that at higher resolutions there are significant temperature spikes due to natural causes.
Some salient points from the above:
1. Marcutt’s original thesis showed no 20th century uptick, but his published article did.
2. The uptick arose from re-dating core tops, without presenting any evidence that this was justified, and from deleting data.
3. The uptick would not exist if the article had used the same data and methods for the last century-odd as for earlier periods. Inter alia, earlier periods are century-plus scale averages; later periods are short-term. This means trends and rates cannot be compared between periods.
4. The article nevertheless implied that the large sharp 20th century uptick was itself robust in the context of the article, since when a different procedure produced a different result, the difference was written off as not robust.
5. Accordingly, Marcott himself used the uptick at time of publication to claim that we had “never seen something this rapid” as recent warming.
6. Now that 1-4 have been pointed out, Marcott claims, contra 5, that the uptick is “not statistically robust” and “cannot be considered representative of global temperature changes”.
7. Marcott also claims that 4 did not mean what it said, but the opposite, i.e. 6.
8. Despite 5, Marcott further claims that the uptick “is not the basis of any of our conclusions”.
Without making judgments re motivation or honesty, one might well ask:
A. If the 20th century uptick was not robust, why was it included in the study?
B. Why were conclusions drawn on the basis of the uptick, including by Marcutt and Shakun, and including in the FAQ?
C. If the paper did not claim exceptional warming, why did Marcutt claim it in public?
D. Imagine the paper had left out the uptick: would the rest of the reconstruction have been published, given its lousy temporal resolution?
E. The now renounced claims of exceptional warming rest on data mishandling and exploiting the inevitably sharper trends inherent in short-term data. If this was not a deliberate attempt to mislead, then why now deny earlier statements (cf. 7 and 8 above)?
Valuable overview of some central issues David. I’d only question the ‘lousy temporal resolution’ in question D. It’s not reasonable to expect more, as Paul Dennis made clear on 19th March responding to Richard Telford:
Note he answers your question D in the negative for Science and Nature. It’s a good question, in my book, just not the final phrase. I think ‘given its other flaws’ would do fine.
That’s a very good summation.
An excellent summary.
Could we also add to 3) that, without the (statistical processing that gave the) uptick, people would have wondered why there was no recent warming in the reconstruction, and therefore questioned the validity of the reconstruction in earlier times as well?
People, perhaps. Ockham meanwhile would probably have think that if there is no uptick, well, is that there is no uptick.
Steve:
But Steve it surely does confirm Mann’s Hockey Stick since it has found yet another way (using non-dendro proxies) to produce a sudden warming using statistical slight-of-hand that Mann would be proud of. It certainly matches the statistical skill (approximately zero to a zillion places).
The only ways to produce Hockey Sticks are biassing the proxy selection and/or the altering the weights of the proxies. The result appears to be the sole reason for the prestidigitation in the first place.
What I’d like to know is whether the editor of Science is as impressed with the excuses or whether he or she would want the authors to issue a Corrigendum or withdraw the paper until the issues are resolved.
Reminds of the burglar who dragged a stolen safe home through the snow and was surprised when cops knocked on his door a few hours later…That no one would compare to the thesis and they can pretend it isn’t there? Really?
Pielke Jr on Twitter, and don’t miss is new blog post on Marcott et al.
I would think in a more rational world we might be celebrating the recent Marcott paper, with its statements of uncertainties, its arbitrary re-dating of cores and its presentations of easily viewable individual proxy series, as a major revelation of the limitations on the use of available proxies in making long term temperature reconstructions. As example of these limitations we have the following two statements from Marcott et al:
“The results suggest that at longer periods, more variability is preserved, with essentially no variability preserved at periods shorter than 300 years, ~50% preserved at 1000-year periods, and nearly all of the variability preserved for periods longer than 2000 years (figs. S17 and S18).”
“We took the 5° × 5° area-weighted mean of the 73 records to develop a global temperature stack for the Holocene (referred to as the Standard 5X5 reconstruction) (Fig. 1,Aand B). To compare our Standard 5X5 reconstruction with modern climatology, we aligned the stack’s mean for the interval 510 to 1450 yr B.P. (where yr B.P. is years before 1950 CE) with the same interval’s mean of the global Climate Research Unit error-in-variables (CRU-EIV) composite temperature record (2), which is, in turn, referenced to the 1961–1990 CE instrumental mean (Fig. 1A).”
The first statement unequivocally puts a major limitation on the periods of time – modern to historical – that can be validly compared – no variability at 300 years periods and only a half at 1000 year periods is preserved in the reconstruction. The Marcott paper also references the individual proxies for viewing and shows, without comments from the authors, just how incoherent the trends are in those proxies and, as a result, how difficult it is to visualize that stacking these proxies is going to produce a meaningful result.
The second statement shows how uncertain the splicing of the instrumental/CRU_EIV record to the Marcott reconstruction (given for the sake of argument here that it is a reasonably valid thermometer) really is if one notes the uncertainties in the record and reconstruction and the fact that a 940 year period variability is only preserved by 50%.
After SteveM’s analyses of the dating changes in the top cores in Marcott et al and the Marcott authors’ lack of a reply with reasons for the changes we are left with apparent arbitrary dating changes without which we would have the familiar divergent reconstruction towards the end of the series.
After listening carefully to the Marcott authors and those attempting to find something in their reconstruction that will reinforce the consensus view of AGW, one can only conclude that there is no there there.
The third statement excerpted from Marcott below introduces a Monte Carlo exercise to estimate CIs for the reconstruction that I thought from my first read of the paper might be worth pursuing. I note that NZ Willy and SteveM are interested in a detailed look at that process.
“We account for chronologic and proxy calibration uncertainties with a Monte Carlo–based randomization scheme (6).”
Kenneth. If a rebuttal paper is sent to Science magazine (or Nature) — which seems to me by far the best course of action — then your observations need to be included.
“Without providing any links to or citation of Climate Audit..” Steve, I know you weren’t expecting any such thing. Remember Gergis et al making the “independent” discovery of the terminal deficiencies of his own paper the day after you analyzed the paper. All I can say is the speed with which they respond to your work tells all that your stuff is must-reading daily for these guys. Some one with the same initials as you, SuperMan, never got any respect or credit from the joker and his mob, but we honest, hardworking citizens are always grateful.
Although its clear the caveats no longer square,
RC elaborations have yet to clear the air.
Addressing questions unasked,
unconnected answers have masked
the issues that Steve laid bare.
The original thesis rejected by Nature
lacked a common feature, incurvature.
This led to Marcott contemplating
“commonly” practiced core redating.
Yet, if such a task isn’t specious
Why leave it absent from the thesis?
Their answer in convenient youtube form: http://www.youtube.com/watch?v=bAQv6KTfQow
Well, dissapointing the FAQ but was anybody expecting more?
Not acknowledging the corrections, not acknowledging the authors of the corrections, not addressing the most important questions, not linking to skeptics sites and in no case to climate audit.. How could they, this would give credit and potentially open a dialogue…
Well in a way Marcott did confirm (what we knew about) Mann and not only that.
And Yosh shows here the results of the new climate icon from Marcott’s diagram:
http://www.bishop-hill.net/blog/2013/4/1/introducing-the-scythe-josh-230.html
If the portion of the graph for the 20th century is not robust why even include it?
Perhaps it was for entertainmentor illustrative purposes only.
BTW, Tamino responded to your plagiarism charge in more depth here when I asked him to clarify: http://tamino.wordpress.com/2013/03/22/the-tick/#comment-80544
Re blogged this on http://orach24463.wordpress.com/2013/04/02/4-1-2013-6_46_10-pm/
There is a message in Marcott that I think many have missed. Marcott tells us almost nothing about how the past compares with today, because of the resolution problem. Marcott recognizes this in their FAQ. The probability function is specific to the resolution. Thus, you cannot infer the probability function for a high resolution series from a low resolution series, because you cannot infer a high resolution signal from a low resolution signal. The result is nonsense.
However, what Marcott does tell us is still very important and I hope the authors of Marcott et al will take the time to consider. The easiest way to explain is by analogy:
50 years ago astronomers searched extensively for planets around stars using lower resolution equipment. They found none and concluded that they were unlikely to find any at existing the resolution. However, some scientists and the press generalized this further to say there were unlikely to be planets around stars, because none had been found.
This is the argument that since we haven’t found 20th century equivalent spikes in low resolution paleo proxies, they are unlike to exist. However, this is a circular argument and it is why Marcott et al has gotten into trouble. It didn’t hold for planets and now we have evidence that it doesn’t hold for climate.
What astronomy found instead was that as we increased the resolution we found planets. Not just a few, but almost everywhere we looked. This is completely contrary to what the low resolution data told us and this example shows the problems with today’s thinking. You cannot use a low resolution series to infer anything about a high resolution series.
However, the reverse is not true. What Marcott is showing is that in the high resolution proxies there is a temperature spike. This is equivalent to looking at the first star with high resolution equipment and finding planets. To find a planet on the first star tells us you are likely to find planets around many stars.
Thus, what Marcott is telling us is that we should expect to find a 20th century type spike in many high resolution paleo series. Rather than being an anomaly, the 20th century spike should appear in many places as we improve the resolution of the paleo temperature series. This is the message of Marcott and it is an important message that the researchers need to consider.
Marcott et al: You have just looked at your first star with high resolution equipment and found a planet. Are you then to conclude that since none of the other stars show planets at low resolution, that there are no planets around them? That is nonsense. The only conclusion you can reasonably make is that as you increase the resolution of other paleo proxies, you are more likely to find spikes in them as well.
Good analogy and well explained.
Yes, very helpful and a way out of the ‘stuckness’ of much of the debate at the moment.
Seems like an excellent analogy, but is it feasible to increase resolution in proxy records of the distant past?
I have been struck from the beginning (of the Marcott discussion, reading the paper and SI etc.) by the fuzziness of the pre-instrumental proxy data. How can 300+ year resolution in noisy datasets tell us about any possible e.g. 30 yr or 100 yr spikes (or declines) in temps.??
It seems that the Marcott study flattens out the proxy data, assumes little natural variability, and then the authors and their flacks want to draw comparisons from high frequency 20th century instrumental data.
We don’t know what will be possible with proxies in the future, all we know is the resolution possible now. But if history shows us anything it is that science has the ability to surprise.
the question of resolution is an interesting one. However, for now, I’d prefer that readers not over-pursue this topic based on analogies until we get some answers on questions asked to date.
Steve – this is an issue of ethics. Were I in your position I would contact the journal and also put the authors on notice that, should answers about their methodology not be forthcoming, it will be pursued as a complaint about ethics.
A grey beard on an ethics panel may not follow the arguments too well. He would, however, see that specific questions hadn’t been answered. Not all grey beards on university ethics panels are like Gleik.
” Rather than being an anomaly, the 20th century spike should appear in many places as we improve the resolution of the paleo temperature series.”
No, that’s where the analogy breaks down. The resolution wasn’t improved; the spike was due to end effects (“not robust”). Other parts of the range would not have end effects, or at least not nearly so concentrated.
C’mon Nick…
The comparison that the warmers are making is between high resolution instrumental data and proxy data with 300+ year resolution. They say the paper is still useful for understanding current temperature swings because you can compare proxy to instrumental.
Re: Nick Stokes (Apr 2 01:20),
I agree, it does appear the spike is a result of the methodology and is not supported by the data.
I’m not suggesting by this analogy that Marcott’s math is correct. Rather what I’m saying by analogy is regardless of whether the math is right or wrong, the conclusion itself doesn’t follow because the inference from low to high resolution only works in one direction and Marcott has the direction reversed.
What I showed by the analogy is that even if the spike was rock solid mathematically, it does not show there is anything unusual about the 20th century. Rather, the only valid conclusion that one can draw from the spike is that it indicates that we are likely to find spikes as we increase proxy resolution.
Nancy Green, thanks for providing the elegant re-focus.
Well, I’ll weigh in with thanks, Nancy; your analogy helped me a lot. Steve had already seen with much finer resolution, though, likely beyond what he’s illustrated.
============
Re: Nick Stokes (Apr 2 01:20),
By analogy what Marcott et al has done is to hold up a high resolution picture of a star and say “see this star is different because it has planets around it; none of the low resolution pictures of other stars show any planets”. What I’ve tried to explain is that this is false logic.
What Steve, you and others have done is to look at the high resolution picture under the microscope and you’ve found that what Marcott says are planets are not planets at all. Instead, by analogy they appear to be dust specs on their equipment that they neglected to clean properly. As well, you’ve found that Marcott failed to mention that there were other high resolution pictures that didn’t show the planets. As a result in the FAQ Marcott has admitted these are likely not planet after all.
Re: Nick Stokes (Apr 2 01:20),
What Marcott in the FAQ is now claiming is that it makes no difference if the planets are in fact dust specs, that there are other high resolution photos that show the planets around the star, thus this star is different because none of the low resolution pictures of other stars show any planets. What I’m showing is that this argument is false.
Alternatively, one could take the 150 year modern instrumental record and reduce it to a single point average representing the entire period. That would compare to what the proxies provide.
“Marcott et al: You have just looked at your first star with high resolution equipment and found a planet. Are you then to conclude that since none of the other stars show planets at low resolution, that there are no planets around them? That is nonsense. The only conclusion you can reasonably make is that as you increase the resolution of other paleo proxies, you are more likely to find spikes in them as well”
I think, Nancy, you are overlooking how scientists and advocates like Marcott view this issue. They start more or less from the proposition that the modern warming period is unnatural (and many here, I judge would agree to some extent or another) and thus they would want to say looking at the first highly resoloved star with planets which being somehow thought to be unique by other independent means that it says something about that uniqueness. The point that the more skeptical would make is that given some unnatural character with the current warming could we use the Marcott paper’s analyses and evidence to rule out that previous warming episodes were as intense or more so than the modern one. To that end we would have to say, and wonder why the Marcott defenders cannot say it, is that the resolution is not even close to that required to make that comparison. As a matter of fact, looking at the modern era under the resolution limitations of the Marcott reconstruction, one might conjecture that the unnatural modern warming is preventing us from so quickly falling into an ice age of unkown duration.
What would be hilarious is to plot the marcott reconstruction, and then overlay an unsmoothed vostok record.. which as I recall spikes above the modern era.
Imagine the howls
All I could find on a quick search was Central Greenland.
The scale is eyeballed, but close enough to get the effect.
OK, I went back and did a Vostok/Marcott version, with a cleaner Marott
http://img543.imageshack.us/img543/6617/marcottvostok.jpg.
Vertical scale is very close.
Horizontal is approximate.
And here it is with the baseline shifted (a guess).
Ouch. Scythes aplenty. Dangerous world. And strangely not.
Excellent. The overall pattern argued by Marcott seems unremarkable.
Others have claimed that the paper is brilliant. Does anyone have any idea as to what makes the paper brilliant? How does it help GCMs?
awesome.
That would be a good one for Anthony to post.
Is it correct to apply area weighting to grid shapes such as a 5 x 5 when temperature is the parameter? I think not. One can understand the process easily for rainfall, which can be expressed as mm per square meter of ground, & is additive, so that you can pour daily rain gauge water into a measuring jar after several days and get the same total as day-by-day.
Temperature, on the other hand, is the temperature of the air above the ground and its measure does not depend on the area of the ground below. It is not additive. You cannot add daily temperatures and reconcile them with an independent later confirmation like rainfall can be. (In theory you can, but you need different instruments).
An analogue might be height above sea level of the ground surface. If this was gridded and averaged, the accumulated result for a region should not need area weighting. Indeed, this analogue might lead to a double error when temperature grid cells were area weighted and the lapse correction for levelling to a fixed altitude (if used) was based on area weighted altitude calculations, instead of point locations where stations were sited.
……………………
In the Marcott paper, is there another source of error if some proxies are calibrated for part of a year, for example against summer temperatures, when the core temperature graphs are for annual periods? Can one make a valid annual reconstruction using some data based on (say) summer temperatures, other proxies on annual? Sometimes yes, sometimes no?
……………………
WUWT is displaying http://wattsupwiththat.com/2013/03/30/the-pitfalls-of-data-smoothing/
This is relevant to this thread, with discussion of removal of high frequencies by smoothing and the unreliability of certain statistics procedures that are then commonly used on smoothed data. I’ve added a small essay from 2 years ago questioning whether correlation coefficients can be used on averaged data (yearly, monthly, weekly, daily). In general, they cannot.
Click to access GHS%20on%20chasing%20R%2c%20extended.pdf
If my essay is correct, and I presume that Willis is correct, then there are fundamental questions still hanging over the temperature data used for proxy reconstructions in the Marcott paper. Some temperature reconstructions use the correlation of temperatures between pairs separated by known distances to adjust the raw value of ‘anomalous’ sites and to infill missing data. I question whether such well-known temperature reconstructions are correct (still, after all these years).
…………………..
This is old news, but the whole outcome of Marcott et al depends on the thermometer era and the measured trends. More correctly, it depends on the early part of the record. It will be interesting to see how proxies are calibrated in new work where the temperature has changed very little in many parts of the world for 15 years. The high frequency texture of a reconstruction is enhanced by a steep temperature trend in the calibration process. OTOH, a proxy calibrated at constant temperature reveals a measure of noise. Trend and noise are blurred in the Marcott paper, to a remarkable degree given that neither was known for the majority of the period studied.
I am sure there is plenty of rich material to be discussed in the scientific realm here but I think a layman two cents is worth popping in as a sorbet to clear the pallet for moment here 🙂
When you see headlines literally saying “We’re screwed” spun off a paper, and it later turns out it has no basis in truth, then I have to ask – Can anyone show me that that any of the same defenders today raised the slightest objections to any of this spin?
Until then, for me, all the most finely honed scientific defence about what Marcott et al actually said just goes in the sophistry bin.
So, do I get it right? The study showed that the current warming is unprecedented for 11500 years, but it really did not because of it’s proxies’ low resolution, but it did not have to because we know that the spikes did not exist (as there is no mechanism for that) in the past and the current warming, the spike, though a product of dubious procedures and “not robust” in the study (but this does not matter because we know from instrumental records that there’s a spike ), is unprecedented, because we know it is (and we know why and we already know that it’s gonna continue)… So what exactly did this “excellent study” add to the knowledge when all that it was supposed to add, it did not but we already know all of this without the study. Arrghhh…
A poster on the FP story called windy2 put this link to a rather neat graphical illustration of what Marcott can say about 20th century warming.
That picture shows what I as a layman understood at the start, but says it better than I ever could. Does it strike anyone here as too outrageously simple?
Re: tlitb1 (Apr 2 10:53), Although as someone pointed out to me on BH the graphic is a bit kind about the tick existing at all in their data, I think it is certainly good for illustrating the slight of hand over the comparison of high resolution to smoothed.
A three year old wants to know what the jagged teeth are in the lion’s mouth.
==================
Ross has a summary in this morning’s Financial Post (Canada):
http://opinion.financialpost.com/2013/04/01/were-not-screwed/
Just took a look at Pielke’s post, and realized, once the uptick is removed from Marcott’s graph, it plainly shows that we are headed into a new ice age.
Tamino just published a post in which he tries to prove, if I interpret him correctly, that temperature spikes in the Holocene comparable to what we’ve experienced in the past century are unlikely and then concludes with this:
Link above doesn’t work. Here’s the url:
fyi, commentors at Tamino and RealClimate are excited about a “Three spikes” test Tamino has done to see if a temp. spike of 0.9C would be seen in the Marcott study had it really been there in the Holocence:
Shakun has posted a reply at Realclimate
http://www.realclimate.org/index.php/archives/2013/03/response-by-marcott-et-al/comment-page-3/#comment-327407
Shakun’s “explanation” is a non-explanation. He appears to have been paying attention to CA commentary since he’s not aware of the bom spike in MD03-2421. He said:
Nope. They used data after 1940 for numerous series in the calculation of their 1940 value. For example, consider MD95-2011 and MD95-243, where they used “1950” data (0 BP) in their calculation of 1940. If they wanted to cut off their reconstruction at 1940, fine. But that doesn’t justify the blanking of MD03-2421 data after 1940.
This is typical a Gavin Schmidt-type answer. Totally unresponsive.
Nor does it explain what they did with OCE326-GGC30.
Is there a way that the blanking out of the data past 1940 on MD03-2421 ends up being no big deal? It doesn’t sound like it should matter much if they didn’t bother with an already-declared-to-be-non-robust section of data, regardless if they included it anyway. Their concentration was primarily the pre-Mannian section of the Holocene, and then used the rest as more of a ‘ball-park’ kind of consistency.
What does the last clause in the last sentence it Shokun’s response mean:
I believe their saying that their research, taken together with the instrument record, provides the conclusion that the recent rapidity with which the temperatures have risen is also remarkable in that the rise itself is equal to a large portion of the range in which we find the entire Holocene.
In other words, implying a stable, peaceful, comfortable, adaptable range gently bumping up and down throughout 11,000 years…then bottoming out at the Little Ice Age, only to rapidly reverse upward to cover roughly the same temperature span, but in 1/100th the time– implying a volatile, un-natural, “alarming”, and “un-adaptable” trend.
I had another question out there for the statistically elite folks out there…
…I see the proxy mean and its fluctuations, and I get that to stitch together the entire Holocene temperature record it requires at least 3 proxies (Marcott et al, Mann et al or your one of preference, and the instrument record), each with their own error bars. I believe the transitional periods between the various proxies are more or less critical for cross-validation, yes? [with the exception of the more or less understood decay/dating parts of early/mid Marcott and of course the temperature record….basically leaving Mann et al with the most work to do].
…Tamino has done some good work to demonstrate that a spike the likes of we’ve seen recently in the temperature record, even in a grid resolution of 100-300 years, might still likely have been caught in the reconstruction, and that if there could have been such a spike posited it would have to come with some physical basis.
…My question is Do the error/uncertainty bars within a proxy reconstruction permit a viewer to infer any sort of temperature line through it as at least possible, or does one still have to hold fast to the mean? I wonder if it’s at least possible to sneak a good spike in there undetected, but I do think the proxy and the requisite physical basis might preclude matching the magnitude/speed we’ve just seen.
Physicist Clive Best says he has debunked Tamino’s last post (looks plausible though I can’t judge). Also notes that Tamino has blocked him! The notorious “open mind” of Grant Foster aka Tamino strikes again….
Maybe the fiasco surrounding the Marcott Mess is best captured by the insightful remark made in about 1939 by the British engineer, George Callendar:
“Climatology is a difficult subject. By long tradition the happy hunting ground for robust speculation, it suffers much because so few can separate fact from fiction”.
22 Trackbacks
[…] […]
[…] off to Prof. Roger Pileke Jr, who has followed Steve McIntyre’s lead in standing up for science in the face of continued lies and scientific fraud. Here are some […]
[…] But more courageous readers take it further: […]
[…] is great value in such an analysis of pre-20th century temperatures. And there can be no doubt there will be continuing debates and discussions about the paper’s methods and conclusions. But one point that any observer should be able to […]
[…] The Marcott Filibuster […]
[…] Marcott FAQ at Real Climate has been causing comment over the Easter weekend, particularly at Climate Audit. Roger Pielke Jr also has an excellent post, aptly titled Fixing the Marcott Mess in Climate […]
[…] The Marcott Filibuster […]
[…] the weeds over this particular study, please take a look at Steve McIntyre's analyses over at his Climate Audit site and Tamino's take over at his Open Mind […]
[…] weeds over this particular study, please take a look at Steve McIntyre’s analyses over at his Climate Audit site and Tamino’s take over at his Open Mind […]
[…] UPDATE: I missed this long discussion and lots of comments yesterday from Steve McIntyre et al … The Marcott Filibuster […]
[…] by peopleneedpower Re blogged this from Climate Audit […]
[…] is great value in such an analysis of pre-20th century temperatures. And there can be no doubt there will be continuing debates and discussions about the papers methods and conclusions. But one point that any observer should be able to clearly […]
[…] The Marcott filibuster at ClimateAudit […]
[…] Siamo alle solite. Una ricerca, solida o debole, interessante o banale, non importa come sia, l’importante è che possa in qualche modo sostenere la causa dell’AGW, nella fattispecie facendo anche risorgere dalle ceneri in cui è giustamente finita una delle pagine più brutte della ricerca scientifica degli ultimi anni, l’Hockey Stick di Michael Mann, con il sostegno compatto delle truppe di Real Climate, il blog delle vestali del clima che ospitano il documento di risposta senza fare alcun cenno alle critiche sollevate. […]
[…] Avoiding the issue of explaining the flawed research on climate change. […]
[…] en un análisis de las temperaturas anteriores al siglo 20o tal. Y no puede haber ninguna duda habrá continúan los debates y discusiones sobre los métodos y las conclusiones del artículo. Pero un punto que cualquier observador debe […]
[…] en un análisis de las temperaturas anteriores al siglo 20o tal. Y no puede haber ninguna duda habrá continúan los debates y discusiones sobre los métodos y las conclusiones del artículo. Pero un punto que cualquier observador debe […]
[…] rainbows while others keep pointing out the fairytale. Just look at the crap Marcott is pulling. The Marcott Filibuster Just imagine how much we might already know if these fools weren't wasting all this time on […]
[…] Judith Curry skriver vidare att sagan om Marcotts och medförfattares artikel har blivit verkligt intressant över weekenden (författarnas frågor och svar på RealClimate publicerades under förra weekenden). Förutom blogginlägg och artikel av Roger Pielke Jr respektive Ross McKittrick citerar hon Steve McIntyres blogginlägg speciellt det senaste The Marcott filibuster. […]
[…] continue to do so. This time Shaun A. Marcott and friends moved data around by 100′s of years to hide the same […]
[…] https://climateaudit.org/2013/03/31/the-marcott-filibuster/ […]
[…] delle truppe di Real Climate, il blog delle vestali del clima che ospitano il documento di risposta senza fare alcun cenno alle critiche sollevate. E così, Roger Pielke Jr, che non è scettico, ma non è neanche una vestale, ci va giù […]