A few days ago, we discussed the unresponsive answers provided to climate scientist Jeff Severinghaus in February 2003 when he inquired about the validity of tree ring widths as proxies due to the inconsistency (divergence) between temperature and ring widths, answers characterized by Severinghaus here as not being a “straight answer”.
In first quarter 2003 (almost exactly the same time as Severinghaus’ inquiry), Soon et al raised almost precisely the same question in Soon et al (EE 2003). The answer of Mann and a long list of coauthors (Ammann, Bradley, Hughes, Rutherford, Jones, Briffa, Osborn, Crowley, Oppenheimer, Overpeck, Trenberth and Wigley), which is the topic of today’s post, took hide the decline to new levels.
As noted above, Soon et al (EE2003) clearly articulated the impact of the divergence problem on the validity of temperature reconstructions:
Strong evidence has been accumulating that tree growth has been disturbed in many Northern Hemisphere regions in recent decades (Graybill and Idso 1993; Jacoby and D’Arrigo 1995; Briffa et al. 1998; Feng 1999; Barber et al. 2000; Jacoby et al. 2000; Knapp et al. 2001) so that after 1960-1970 or so, the usual, strong positive correlation between the tree ring width or tree ring maximum latewood density indices and summer temperatures have weakened (referred to as “anomalous reduction in growth performance” by Esper et al. 2002a). The calibration period of Mann et al. (1998, 1999, 2000a) ended at 1980, while 20 more years of climate data post-1980 (compared to the 80 years length of their calibration interval, 1902-1980) exist. If the failure of inter-calibration of instrumental and tree growth records over last two to three decades suggests evidence for anthropogenic influences (i.e., from CO2, nitrogen fertilization or land-use and land-cover changes or through changes in the length of growing seasons and changes in water and nutrient utilization efficiencies and so on), then no reliable quantitative inter-calibration can connect the past to the future (Idso 1989). Briffa and Osborn (1999) have also criticized the impact of unusual tree growth on the calibration procedure of tree-ring climate proxies (see additional discussions in Jacoby and D’Arrigo 1995; Briffa et al. 1998; Barber et al. 2000; Briffa 2000; Jacoby et al. 2000).
This matter has largely been unresolved, which means that global or Northern Hemisphere-averaged thermometer records of surface temperature cannot be simply attached to reconstructed temperature records of Mann et al, based mainly on tree-ring width, which cannot yet be reliably calibrated, to the latter half of the 20th century.
These issues remain unresolved eight years later, even though Mann et al (2003) was hailed at the time as “discrediting” Soon et al 2003. In fact, Mann et al (2003) did not directly respond to the divergence issue.
Instead, on this point, they relied on the rhetorical effect of an expanded spaghetti graph, showing six reconstructions (MBH99, Jones et al 1998, Mann and Jones 2003, Crowley and Lowery 2000, Esper et al 2002, Mann and Jones 2003), none of which, in the diagram, showed evidence of a convergence problem, reinforcing the argument of Mann et al on the supposed unimportance of the divergence problem: (Note Dec 1: LucySkyWalker reminds me of a prior discussion of this graphic at CA here in which Tim Lambert pointed out that the (yellow) “Crowley and Lowery” series shown here wasn’t actually the Crowley and Lowery series, but MBH99 plus 0.5 sigma.)
Fig. 1.Comparison of proxy-based NH temperature reconstructions [Jones et al., 1998; Mann et al., 1999; Crowley and Lowery, 2000] with model simulations of NH mean temperature changes
over the past millennium based on estimated radiative forcing histories [Crowley, 2000; Gerber et al., 2002—results shown for both a 1.5°C/2*CO2 and 2.5°C/2*CO2 sensitivity; Bauer et al., 2003]. Also shown are two independent reconstructions of warm season extra-tropical continental NH temperatures [Briffa et al., 2001; Esper et al., 2002] and an extension back through the past 2000 years based on eight long reconstructions [Mann and Jones,2003].All reconstructions have been scaled to the annual,full northern hemisphere mean, over an overlapping period (1856–1980), using the NH instrumental record [Jones et al., 1999] for comparison, and have been smoothed on time scales of > 40 years to highlight the long-term variations.The smoothed instrumental record (1856–2000) is also shown. The gray/red shading indicates estimated two-standard error uncertainties in the Mann et al. [1999] and Mann and Jones [2003] reconstructions.Also shown are reconstructions of ground surface temperatures (GST) based on appropriately areally-averaged [Briffa and Osborn, 2002; Mann et al., 2003] continental borehole data [Huang et al., 2000], and hemispheric surface air temperature trends, determined by optimal regression [Mann et al., 2003] from the GST estimates.All series are shown with respect to the 1961–1990 base period
At the time, no one knew about “hide the decline”. Mann et al do not mention anything about deleting adverse data. The Briffa reconstruction labeled in the legend as “Briffa et al scaled 1856-1980”, giving no clue to readers of hide-the-decline. Let’s now look at a magnified version of this graphic, blown up so that we can see how they handled the Briffa (orange) reconstruction. As sharp-eyed CA readers FergalR and haroldw observed, if you squint closely, you can see that the Briffa reconstruction was chopped off before its end.
Indeed, they did not simply “hide the decline”, their “hide the decline” was worse than we thought. Mann et al did not merely delete data after 1960, they deleted data from 1940 on, You can see the last point of the Briffa reconstruction (located at ~1940) peeking from behind the spaghetti in the graphic below:
Detail from Mann et al (EOS 2003) Figure 1. Arrow points to Briffa series peeking out from behind the spaghetti
Had Mann et al used the actual values, the decline would have been as shown in the accompanying graphic:
Figure 3. Re-stated Mann et al (EOS 2003) Figure 1 showing the decline.
Had Mann and his 13 co-authors shown the Briffa reconstruction, without hiding the decline, one feels that von Storch (and others) might have given more consideration to Soon et al’s criticism of the serious problem arising from the large-population failure of tree ring widths and density to track temperature.
Climate Audit 
285 Comments
I can’t imagine any rational, disinterested body – say, a civil jury – buying an argument that this was accidental or careless or just bad math.
“At the time, no one knew about “hide the decline”. Mann et al do not mention anything about deleting adverse data. The Briffa reconstruction labeled in the legend as “Briffa et al scaled 1856-1980″, giving no clue to readers of hide-the-decline.”
Steve, I know you are very careful about not going a bridge too far in alleging malfeasance, but is there any reason to characterize this as anything other than scientific fraud and academic misconduct? Especially if it was hailed as rebutting an issue raised by Soon et al, and was not simply a paper that “happened” to also talk about temperature reconstructions. In this case it would appear they (i) knowingly deleted data that was directly relevant to the issue raised by Soon, (ii) knowingly presented the data in a way that would hide the issue raised by Soon, and (iii) knowingly pretended that their reconstruction did not suffer from the issue Soon had raised and that Soon was wrong.
Note that Briffa et al (Orange) is also nearly the coldest proxy in the LIA period, especially around
1620, and then in the 1800s. 1940 is the peak year
in over 600 yrs of data!
Woodentop at Bishop Hill notes the following said by Michael E. Mann in http://climategate2011.blogspot.com/2011/11/0031txt.html
A knowing ellipsis between friends.
FergalR gets the
Gold on the foremast this time.
So many Mobys.
========
Bad analogy.
Moby won.
The term ‘moby’ refers to the popster of that moniker who advocated that particular form of trolling.
Heh, you’re presuming Moby won’t, this time.
==========
Ajax and Achilles I see on the scene. Who is this interloper Langsdorff?
============
Phil Jones actually posted something very similar in the days following Climategate 1 release.
In an unauthorised attempt to “come clean” he posted what his WMO headline grahic would have looked like without “Mike’s trick”.
At the time I commended the frank and honest response.
It stayed up for about 24h before the damage limitation team presumably saw it and ripped it down again.
I spotted it at that time but foolishly did not keep a local copy.
Steve, didn’t you post a copy of “honest” Phil’s plot at about that time ?
Steve; I remember commenting about the calculation at the time. I;ll look
Looks like you caught a copy of Phil’s coming clean copy:
Good one.
Re: P. Solar (Dec 1 12:24),
the statement (“confession”) from which I linked the graphic is still available here.
Great to have a reminder of pre-Neil Wallis ‘poor Phil’ responses from CRU. The alternative view to the last sentence being put nine days ago:
Decision makers were consistently denied well-presented information, such as a truthful spaghetti graph, showing how uncertain it was that recent temperatures were unprecedented. I’m with FOIA.
Thanks for posting that. I checked back on the original post a few days after it was posted and it got a 404.
Having a chance to look back an review this now it’s amazing. It’s not just “hide” the decline, it’s turn the decline around and make it go upwards! The two other proxies which had no intention of going skywards also a little warming hand from our Phil.
I again commend his frankness in posting what it should have looked like, just a shame it was only after he got rumbled an not when he was asked to advise world leaders on climate.
I’d forgotten just what a blatant fr… this graph was. All three proxies get carefully cropped and blended into thermometer data IN THE SAME COLOUR. Look at the blue line. It’s not even a cut and paste graft, the two datasets have been deliberately mixed and blended together in a way the does not represent either dataset. It’s pure fiction.
I mean you just can’t do that.
My god , if you tried that with stock options you’d get 5-10.
Just to underline this , I mean you’d get _hard time_. You would not get a pat on the back from the judge, comments to the effect that your work and professional “integrity” was beyond question and would you like your old job back, sir ?
May I emphasize now that the still undefended, unjustified statement by Phil Jones quoted above remains one of the more astonishing utterances by a scientist in recent times:
I see that they didn’t just hide/erase the Briffa line after 1940. The removal of it’s values distorts the 2 SD shaded area at the end. When the line is pasted back in it falls below the shaded area!
It smells bad.
Certainly true if they included all values to 2,000.
Would this observation also be true if they included values to 1980 (consistent with the labeling of the graph)?
Actually we are both wrong here. The caption says the grey shading refers only to Mann 1999. Though visually, the effect is to suggest error bars for all of the series.
Steve,
OK some else is going on. The graphic in the PDF from Storch’s copy you referenced yesterday and the one shown (available in a couple of places like Schneider’s website) are different in significant ways. The Briffa plot terminates the same, but some others do not.One big difference is the number of proxies.
Is the copy on Storch’s site a pre-release? It has a bunch of object info contained.
Is this the result of the discussion of the change made after acceptance?
It looks to be the swap of Briffa and Osborn 1999 for Mann and Jones 2003. Discussed in #2530
Is van Storch a reviewer of Mann 03 EOS?
By the 1940 cutoff of Briffa et al. (MXD) is discussed in email 0285
” The only exception is Brifffa et al MXD, where the 1856-1940 period is used instead
(because it starts to diverge downward about 1940 relative to the NH annual mean record).
We also don’t show it after 1940.”
Discussion of line colors weight and style is discussed in both #0285 and #0595
Michael Mann wrote in #0285: “The scaling should be clarifed in the caption. I believe (Scott?) that we’ve scaled the 1856-1980 trends to be equal to those of the instrumental annual full NH mean record, after setting the means equal over the same interval 1856-1980. One can also scale the variance (as you and I did in our submitted GRL article) and the result is basically the same…
The only exception is Brifffa et al MXD, where the 1856-1940 period is used instead (because it starts to diverge downward about 1940 relative to the NH annual mean record). We also don’t show it after 1940. I agree this has to be made very clear in the caption, and Scott should be able to help you guys make sure the caption is accurate.”
The scaling is indicated in the caption, although it says that all reconstructions are scaled, yet the legend implies that only two are. (Phil Jones complained about the “ambiguous” wording of the caption in #0595.)
The caption says nothing about the Briffa series.
UC alert!
I believe (Scott?) that we’ve scaled the 1856-1980 trends to be equal to those of the instrumental annual full NH mean record, after setting the means equal over the same interval 1856-1980.
Here’s a new calibration method, that might be called “Trend Matching”, to add to CCE (the preferred Classical Calibration Estimation that regresses the dependent variable “y” on the exogenous variable “x” and then inverts), ICE (Inverse Calibration Estimation that inconsistently regresses x on y), and CPS (Composition Plus Scale, which matches variances for no particular reason but is often used in climate studies).
Trend Matching actually came up earlier this month in the discussion of lower troposphere trends at https://climateaudit.org/2011/11/07/un-muddying-the-waters/#comment-309105 . Gavin argued (correctly) that this is not an effective way to calibrate, but then Steve replied that it is what Gavin had said he himself was doing in an earlier comment.
This thread on “Snip the Decline” is not the place to resolve this issue, but it could make an interesting future CA post for you or Steve.
In fact the final Eos version of Mann+12 (2003) does not make it clear whether or not Trend Matching was used: “All reconstructions have been scaled to the annual full northern hemisphere mean, over an overlapping period (1856-1980), using the NH instrumental record [Jones et al 1999] for comparison, and have been smoothed on time scales of > 40 years to highlight the long-term variations.” This just says that the means were matched over 1856-1980, but not how the vertical response was determined.
The second paragraph above was intended to be in blockquotes:
Of course, in the case of Briffa et al, they only used the period 1856-40.
I think I mentioned this somewhere here during Climategate 1,
3499.txt
UC (Dec 2 07:30),
Wow, he is indeed scaling such that trends of (Mannian) smoothed series match. Amazing, and they do not even bother to mention this in the articles. And now, UC, you should get back to more important things (like changing dipers) … congrats!
UC tells me it’s a girl! Congratulations!
Oooh, I missed this. UC has a new addition to the world? Congratulations, UC!
Ditto. Reproducibility at its best.
Nicely put. My congratulations to UC as well.
BTW Spence_UK, Jean S and UC are all “original” Climate Audit readers, if I can make such a distinction.
Thanks! Too bad I don’t have infinite time as I’m not retired (4986), so you have to fight without me for a while ( http://www.climateaudit.info/data/uc/climatewars.png )
3499.txt, mike:
I’ve attached a cleaned-up and commented version of the matlab code that I wrote for
doing the Mann and Jones (2003) composites. I did this knowing that Phil and I are
likely to have to respond to more crap criticisms from the idiots in the near future, so
best to clean up the code and provide to some of my close colleagues in case they want
to test it, etc. Please feel free to use this code for your own internal purposes, but
don’t pass it along where it may get into the hands of the wrong people.
Also the continuation is funny … telling a lot about the quality of mike’s work (was there ever a corrigendum issued?).
Dividing numbers in the right order is a foolish and incorrect thing to do.
I agree with Minnesotans for Global Warming that “The Cause” is a key catchphrase for Climategate 2. But in terms of open science this is corker. “Code in the hands of the wrong people” – what Climate Audit’s all about.
Gsvin is claiming it’s a typo
http://www.realclimate.org/index.php/archives/2011/12/unforced-variations-dec-2011/comment-page-1/#comment-221072
“It” being the misleading “1856-1980” caption:
and really embarrassing is me saying “Gsvin is claiming it’s a typo”
lol
very meta
Given that Soon and Baliunas had raised divergence as an issue, hide-the-decline (both enhanced and regular) is not “just a typo”. Mann and Jones 2004 also had enhanced-the-decline together with a deceptive caption (which said that the series was “scaled” to 1856-1940 but did NOT say that it was truncated at 1940. A deceptive caption in itself does not justify enhanced hide-the-decline. In the case of Mann et al 2003, the caption was doubly deceptive. No doubt they intended it to be only singly deceptive, but it wasn’t.
Could they have left it off as an alternative means of hiding the decline? While avoiding more overt forms of misleading people, it is still misleading. Particularly when they are purporting to dismiss Soon and Baliunas’ critique of divergence. And on this point they were quite right.
And as usual, it’s ‘just a coincidence’ that if there were going to be a typo in this article, it would be the exact typo that would support the story that we already know (from their own words) that they were trying to tell.
Gavin is in a hole and needs to stop digging.
RTF
Re: Richard T. Fowler (Dec 3 14:38),
I used to dislike Gavin, but nowadays he’s my favorite Team member. His endless defence of Mann is so hilarious that I think Gavin is seriously challenging the Baghdad Bob as the best spokesperson ever.
I am glad that Gavin’s attention has been brought to this issue. Now he can issue an update the RC article on Jones Mann 2004, with a correction to the graphic displayed from Mann EOS 03.
http://www.realclimate.org/index.php/archives/2004/12/temperaturevariations-in-past-centuries-and-the-so-called-hockey-stick/
What did the other tree ring proxies show during 1940-2000? If they were wildly divergent from Briffa’s, does that call in question the usefulness of dendrology?
Finnish treering studies at http://lustiag.pp.fi/wd2010report.htm don’t show any divergence problems, but they don’t show a hockey stick either, so they have to be excluded from the team studies.
I don’t often quote myself on CA but I am pleased in hindsight with two parts of what I wrote three days ago:
Highlighting added. I agree we should not use dendrochronology to describe the voodoo science of Mann – which, note, nobody has dared to criticise publicly for eight years within the magic circle, including Briffa.
And then Steve McIntyre came along. We are forever in his debt.
I prefer the word “dendromancy” myself.
In Australia we have a very apt term, “drongo’ meaning a ‘not so sharp’ person. Seemingly then ‘drongochronology’ fits the bill nicely. 🙂
Streetcred —
There’s nothing wrong with dendrochronology. In fact, it’s pretty sophisticated and useful. It’s just “drongoclimatology” that’s the problem! 😉
Well, Hu, it’s not a joke. Or a very sad joke. No mirth, IMO. While dendrochronology is surely useful for studying droughts, you know better than I that it has been bastardized severely by the so-called “dendroclimatologists”. Many bastards there, as Steve has exposed. The STUPID concept of tree-ring temperature proxies is so BAD that even the practicioners have abandonded it!
Just like the team’s decision not to focus primarily on the divergence problem upon discovering it, the decision to bafflegab Severighaus makes little sense. Until one realizes that the team is really driven by “hiding the decline” so as not to “dilute the message” of the “cause”.
“…failure of tree ring widths and density to track temperature.”
Case studies show that densities approach very well all proxies supposed to follow temperature (glaciers, TLT, snow etc..), All but stations thermometers. It’s almost time to draw the necessary conclusions.
> “…failure of tree ring widths and density to track temperature.”
> Case studies show that densities approach very well all proxies supposed to follow temperature
> (glaciers, TLT, snow etc..), All but stations thermometers. It’s almost time to draw the necessary
> conclusions.
You don’t mean someone’s been fiddling with the temperature record? >:-0
This is where solid evidence of the UHI effect, and the lack of compensation in the records,
would explain all.
Doesn’t Jones allow somehting like 0.05 Dec C for UHI?
Nial
Maybe the error bars need error bars…
It appears that the colour coding and overlay order has been deliberately chosen to maximise the chances of the decline hiding going unnoticed.
Economy with truth raised to level of artform.
I can’t help thinking that the decline is real – for those north Russian sites: for instance Ostrov Dikson shows a sharp decline 1950 – 1960 and no warming again before ~2000: http://data.giss.nasa.gov/cgi-bin/gistemp/gistemp_station.py?id=222206740006&data_set=1&num_neighbors=1
Could be, here in the SH I compared the CRU2010 data that was released by the Met office in response to CG 1 to both the CRU & Bom data.
The CRU 2010 data has a decline of 0.459 deg / cent from 1950 to 2010.
I did ask the bom for copies of the original observer sheets but was redirected to their website.
But but, weren’t we told that treemometers are teleconnected to the global climate?
Craig Loehle 2009 posited that the divergence problem was due to tree ring growth reaching some limiting ceiling and flattening out. Two things about this: First, that they didn’t just flatten out, but actually took the (steep) decline you show in Figure 3. The second is that if the tree-ring-vs-temperature relationship in the post-1940 period is real, then no one seems to have taken that back in time to see what that suggests about earlier tree-ring data with the same values – such as in the MWP.
Briffa et al 2001 Plate 2 (http://www.ncdc.noaa.gov/paleo/pubs/briffa2001/plate2.gif) shows the regional tree ring graphs, and all but one of them – the flat (e)CAS graph – show much of the pre-1940 tree-ring values above post-1940 (DP) values. This argues against Loehle, in that the post-1940 values are not near the all-time highs, so the limiting ceiling hypothesis cannot hold for the present period, since it did not happen in the earlier periods.
What does this mean? Well, the tree-ring raw data is valid, as measured. It could mean that the dendro-to-temperature calibration might be wrong. It might mean that there are other factors besides temperature that affect tree-rings.
What it does NOT seem to mean is that the linearity of the 1880-1940 period holds true for all periods. S&B’s argument about projecting into the future is 100% valid, at least to get everyone to stop and figure out just what is going on. While papers are being written about the data, few are being written about what is going on in the biology of trees. (E.g., Schweingruber had a study on heightened CO2 and heightened Nitrogen.) While they need to slow down and get empirically sound causes nailed down, with the IPCC and global warming and Kyoto and all going on, The Team seems to be caught in between the DP and trying to buffalo everybody into believing that “there is nothing to see here; move along”.
What I think is going to happen when all is said and done is that the linearity between dendro data and temps will no longer be seen as valid. That doesn’t mean that they can’t use dendro data as proxies, but that it will be a much more complicated conversion. Carbon14 dating gets by on not-quite linearity; they have identified periods where radiation events have skewed the C14 values.
And I think that one outcome of that is that the MWP and LIA will be re-discovered AND that their extremes will be of a higher order: The MWP will be found to be warmer and the LIA will be found to be colder than anyone presently thinks. The data exists, and with some of it higher than tree-ring values at present, it follows that – if they work the post-1940 tree-ring values into the overall picture – the temps back then have nowhere to go but up.
We may finally have an explanation why the Vikings were able to farm in Greenland 1000 years ago.
#0166 (emphasis added):
So, if dendro DOES rule in multi-proxies, what does this mean overall? Was dendro used to calibrate ice cores or varves or corals? If dendro is found wanting, does an entire house of cards come falling down?
We seem to be seeing that dendro data is not linear, that is for sure. Whatever has changed from the seemingly constant linearity pre-1940 needs to be incorporated – if possible – into the conversion from dendro to temps. It may not actually be possible. If so, climatology is in deep trouble.
But if it is not linear in the present (which IS the DP problem), then all past reconstructions and future projections are not worth anything (as S&B suggested) and the IPCC would need to pack up shop and go home while the science gets realigned.
Steve Garcia,
I argued in my 2009 paper that either tree response could flatten out if moisture was not limiting, or decline at higher temperatures. Both are possible.
It seems to me the choice of such thick lines is deliberate to hide stuff. I certainly never make graphs like that.
They have decided that Briffa’s data is bad after (1940, 1961, whichever is convenient) without feeling any need to prove this. Their theory is more real than the real world.
As to the choice of line thickness, in #0285, Michael Mann writes:
“Ou[r] finding has been that using too thin coloured lines makes them indistinguishable to many people. The thicker coloured lines are easier to make out, for people who have trouble distinguishing fine colour differences. So I’d lobby for the thicker lines, using thin lines in a few cases to draw further distinctions (with this many curves, we need to use colour, thickness, and line pattern type as much as posslbe, to distinguish).”
The choice of line thickness seems innocent to me. They did screw up by not mentioning the Briffa truncation in the caption or legend, and certainly the placement of the Briffa line behind other, thick lines makes it [the truncation] difficult to observe.
So what caused Mann to change his mind about making full disclosure?
If it was innocent, then why in every case is the thick red line (instrumental) hiding inconvenient results?
‘every case’ sounds statistically significant. Where’s my Walpole & Myers? It’s gotta be in there somewhere.
==============
Perhaps we should rename this a “linguini graph.”
In his recent blog series Jim Bouldin is developing a fundamental critique of the very possibility of doing long term dendro proxy based temperature reconstructions. fascinating series with much more to come, it seems.
Then he makes critical remarks about arguments over statistical methods in Hockey Team papers etc. when he regards even the most basic methods as lacking rigorous analytical foundations:
Yet, it seems to me as an outsider that Bouldin is far too dismissive of differing levels of analysis. Yes, if his more fundamental criticisms were accepted (or even closely considered) then of course it follows that all methods and studies depending upon such fundamental issues must be on hold or re-scrutinized in light of his more fundamental critique. YET, as long as the entire field is proceeding as though such basics are established, it is hardly irrelevant for McIntyre and other climate auditors to scrutinize all the intermediate steps.
Who is Max Beran?
This is one rich post. It might be some time before people can take in all of it. Keep up the good work.
I don’t want to use the F word, but this seems to be an intentionally misleading response to the question raised – there is a divergence issue in some of the data presented, but this is simply cropped off the chart and then hidden under the other spaghetti.
I wonder what Mann, Jones et al. (with their teaching rather than reseatch hats on) would say if an undergrad presented results in this manner…
“I wonder what Mann, Jones et al. (with their teaching rather than reseatch hats on) would say if an undergrad presented results in this manner…”
Obviously “Well done”
Depends on the estimated impact of the results on “the cause”…
It really could be no more than sloppy graphics work. Graphics are one of the most misused elements in mathematical analyses. Spaghetti graphs are particularly bad since they are generally difficult to read, and to make them understandable, each series has to be individually discussed in the text. They could for instance have avoided all appearance of “hiding” anything by insuring that truncated series were displayed above longer series so the terminations were clear. This is different than splicing a temperature series on to tree-ring measurements and pretending they’re the same.
Read the emails I have listed above. They discuss this figure quite a bit.
You may want to consider 1403 & 0116
Quote Max Beran (to Keith Briffa): “Yes but what about the substantive dendro queries.”
I also don’t know about the ‘F’ word, but I will characterize it as serious research misconduct.
There is no way that is cut at 1960.
1554.txt, about the divergence problem “since ca. 1950” between Cook and Briffa:
date: Wed, 5 Mar 97 16:42:10 EST
from: ???@lamont.ldgo.columbia.edu (edward cook)
subject: The devil …
to: ???@uea.ac.uk
Hi Keith,
The devil made me do it. I have nominated you for a LDEO Climate
Center visiting Climate Scholar. If it comes about, you can of
course tell them (me) to get stuffed! I really think that there
would be keen interest here on your work.
Cheers,
Ed
P.S.
Here is my message to Broecker’s secretary:
Hi Moanna,
Sorry for not responding on that. Bob Dickson would be fine. My
nomination for a CC visitor in the future is Keith Briffa from the
Climatic Research Unit, University of East Anglia. He is doing some
very interesting work with multi-millennial tree-ring records
covering much of the Holocene and is working on understanding the
cause(s) of a very large-scale change in the response of trees to
climate (e.g. over most of Siberia) that has resulted in an anomalous
divergence between temperature and tree rings since ca. 1950.
Cheers,
Ed
PP.SS.
Any more luck with the Kalman files?
Jeff, you’re right…the cut is at 1940.
email 0285.txt
cc: ???@uea.ac.uk, ???@uea.ac.uk, ???@virginia.edu
date: Tue, 10 Jun 2003 08:26:12 -0400
from: “Michael E. Mann”
subject: Re: Figure 1
to: Phil Jones , Scott Rutherford
“The only exception is Brifffa et al MXD, where the 1856-1940 period is used instead
(because it starts to diverge downward about 1940 relative to the NH annual mean record).
We also don’t show it after 1940.
I agree this has to be made very clear in the caption, and Scott should be able to help you
guys make sure the caption is accurate.”
[I doubt the caption was ever made “accurate” to draw attention to the truncation after 1940 of Briffa et al MXD – maybe someone knows the answer?] – Answered in this thread!
yes Briffa is cut at 1940 but the legend claims it goes to 1980
I think the decline in the 1960s-1970s in the tree ring data is probably real. In the global temperature graphs in the early 1980’s, there was a 0.5 deg. decline in temperatures. Today, this 0.5 deg. C decline has magically disappeared, or diminished to 0.1 deg. C. It looks to me that the tree ring data is consistent with the 1980s temperature measurements.
We know that HadCRU has been trying to reduce the 1960-1970s drop in temperatures, to make it look like the overall warming looks like it is steadily rising over the 20th century. It would be ironic that in manipulating the thermometer temperature data, HadCRU invalidated their own tree ring data.
Steve does an excellent job giving credit where credit is due. Willie Soon and Sallie Baliunas deserve abject apologies from Mann et al. Does this explain the almost hysterical response to their papers and the hit put out on deFreitas?
so that after 1960-1970 or so, the usual, strong positive correlation between the tree ring width or tree ring maximum latewood density indices and summer temperatures have weakened
strong
I had a quick look at the cached Mann08 data sets. I correlated both the CRU and HAD NH reconstructed temperature data from 1850-1950 to a number of his ring densities. I fitted the 100 year series starting at the first 100 years, the 2-202, 3-203, e.t.c.
One would, a prior, expect the best r2 statistic to occur at year=1850 or there about’s.
This is not the case.
I suggest the word strong is yet another ‘hide the pea’. One could use the word strong if the temperature calibration vs ring width/density gave the best r2 statistic at the starting year >95% of the time.
If the correlation of the temperature (starting 1850) vs tree rings gives much better correlations at other starting years, then the correlation cannot be strong.
Indeed, I would go so far as to suggest that this type of roving correlation should be used to identify the ‘strength’ of correlation during the calibration procedure.
Steve – you have to be careful with Mann 08 data. In the MXD data, actual data has been deleted and replaced by “infilled” numbers.
Steve, how do you know which is infilled and which is data in that case?
We have a copy of the non-infilled data which is somewhere in Steve’s on-line archive. An easy way though is that you can see the decimal places of the infilled stuff are long.
correction- Probably in the archive.
Try a search on “bodge” for the “briffa bodge” many refs
Martyn, in Ostrov Dikson, which I mentioned above and which is one of the longest-running stations in the Russian Tree Ring Land (it’s on the Arctic Sea coast in the Taimyr province, not very far east from Yamal), summer temperatures dropped by 3 degrees from ~1955 to ~1970. Only since 2000 they’ve had temperatures that are again comparable to those that were common in several multi-year periods in the 40s and 50s.
I just remembered that I made a comment on this on WUWT in this thread:
——
Espen says:
August 23, 2010 at 6:42 am
I once had a look at Ostrov Dikson temperatures (quite close to both Yamal and the Polar Ural sites) to see if there really was any divergence. I downloaded the June-July-August temperatures from GISS for the station Ostrov Dikson and averaged with a 3-year period to remove some of the noise: http://i45.tinypic.com/2ns6jk6.jpg
Made me think that the divergence problem is really a problem with the temperature record, not the proxies…
——
Also see Lucy Skywalker’s excellent “Circling Yamal” articles: http://www.greenworldtrust.org.uk/Science/Scientific/Arctic-Yamal3.htm
That article is excellent, well worth a read. Thanks for the link.
Why don’t the tree rings and temperature graphs overlap so that it is possible to see that they validate each other?
They don’t so it’s basically a study that falsify its conclusions.
Dear Steve
Have you seen this article by Fred Pearce? I was going to say delete “its off topic” but maybe not?
http://www.indexoncensorship.org/2011/11/climate-change-secrecy-freedom-information/
Steve McIntyre is a pernickety Canadian. A retired mining geologist, trained mathematician and amateur climatologist, he has for the past eight years locked horns with the Climatic Research Unit (CRU) at the University of East Anglia, trying to gain access to their data on the history of global temperatures.
He is not (repeat: not) paid by, beholden to or in regular contact with fossil fuel companies or lobby groups trying to undermine climate change science. He is not even a climate sceptic. For years, McIntyre has been asking for CRU’s “crown jewels”, raw data assembled from weather stations round the world that it says proves how much the world has warmed in the past 160 years.
Dear Steve
I have just gone straight into moderation? So this is a test.
I advised you of an article about openess by Fred Pearce in which you take centre stage? The link I got at Climate Etc.
Regards
S
Outstanding as usual. Is there any way you could modify the graph to show what the error bars would be if the truncated portion had been included? I think it might be revealing as well.
In any case, we can see the temperature increase in the last two centuries (even discounting the outlier red curve), even with the decadal-scale fluctuations (relevant to the sense that a possible flattening of the last decade, even though recent years have often been very hot on average.) This trend was supported by the recent study (Berkeley Earth Surface Temperatures Project, true?) with skeptic input, oversight, and funding (the sort of thing that indeed should be done from time to time.) So I don’t understand the continued resistance to at least the “substantial likelihood” that temperatures will rise in coming decades on average, and that much of it comes from increasing CO2.
CO2 is a “greenhouse gas” and adds heat, the only debatable questions is: how much is the “forcing factor.” I could get that much acknowledged even by skeptic Lubos Motl. But if it *isn’t certain* that temperatures will rise or such rise is caused by increasing CO2, then it *isn’t certain* that it wont: ie, it is possible there is a correlation and that constitutes a risk factor. We don’t need certainty to have a risk factor, certainty is a red herring as far as i>action is concerned. Also, much (I know, not all) of what would ameliorate CO2 production is a *good* thing anyway, due to savings and soft landing of diminishing non-renewables. (All this well put by Tom Friedman.) Cheers.
“Fine minds make find distinctions.”
Interesting Neil that you argue that warming beginning 2 centuries ago supports human causation when greenhouse gases were only becoming significantly elevated beginning around WWII. You also invoke the precautionary principle, but note the converse: if there is a non-zero chance that the MWP was warmer than today, then warmer temperatures (say 2 deg) do not end life on earth. Why doesn’t precaution cut both ways?
Craig, EJD; I picked “two centuries” as a convenient vague time frame as I could see at first glance. It is clear that the curve is steeper later and continues in a way we don’t see earlier (see for example http://en.wikipedia.org/wiki/GISTEMP which shows steepening average trend line since 1920, when industrial CO2 was indeed adding more and more to the atmosphere, moving up past 300 ppm and on. There was lots of industry by then and even quite earlier like 1840s. OK, little old me was off in my top of the head estimation, so what. No, not “interesting” at all.)
As for risk factor, you offer a red herring about “end life on earth”: the risk involves damaging events like sea level rise, weather event driving, etc. I question whether the MWP was any worse than now, do you disagree with:
http://en.wikipedia.org/wiki/Medieval_warm_period
“Despite substantial uncertainties, especially for the period prior to 1600 when data are scarce, the warmest period of the last 2,000 years prior to the 20th century very likely occurred between 950 and 1100, but temperatures were probably between 0.1°C and 0.2°C below the 1961 to 1990 mean and significantly below the level shown by instrumental data after 1980. The heterogeneous nature of climate during the Medieval Warm Period is illustrated by the wide spread of values exhibited by the individual records.[12]”
, and see the graph. Nor does their experiencing a warm period mean that in our specific circumstances, we have nothing to worry about. Insurance agencies are already concerned about increasing costs, they can’t afford to be politically anti-correct and have a bottom line to look after.
Furthermore, it looks like the MWP had some adverse affect anyway:
“Review of more recent archaeological research shows that as the search for signs of unusual cultural changes during the MWP has broadened, some of these early patterns (e.g. violence and health problems) have been found to be more complicated and regionally varied than previously thought while others (e.g., settlement disruption, deterioration of long distance trade, and population movements) have been further corroborated.[31]”
Yeah, Wikipedia but the hive mind keeps it on its toes.
Sean, I’m just going by the graph I see at the top of this post, and I don’t see a definite “alternative reconstruction” to go on. You also failed to take into account the Berkeley study, FWIW, as well as the most significant graph of most recent temperatures is essentially based on literal temperature measurements, true?
I didn’t know this had an “unthreaded” category, and would have used it if I had noticed. I note that in most science blogs it is not considered wrong to riff on larger consequences. I note that skeptics visiting pro-AGW threads tend to, not that their example makes it OK here.
Mr Loehle
Who has said 2K temp increase will end life????
It will not. However it will cause changes in weather patterns. In the MWP if you honestly believe in it, life was pretty simple. If your land became infertile due to temperature you could simply up mud-hut and move northwards a few KM.
Are you suggesting that today’s world is still possible?
how do YOU KNOW what 2K increase will bring. I am certain you have no better means of predicting the future than the “Team”.
How do YOU KNOW how much following a precautionary principle will cost. How much does it cost to run a smaller motor? How much does it cost to make your home more efficient? Do you really need a new xxxx every year.
Most people agree that CO2 will raise the temp by 1K (or more). You suggest we are now entering a climate cycle like the MWP so presumably the temperatures will be MWP+1K. Is this safe???
Would it not be better to limit this to MWP+0K. Would this not cause less damage to society?
You are very sure that the MWP and the present are part of a cycle. Would you care to suggest what magically causes this 700year cycle? I can give you cycles, all that you require (except 700years!): http://tinyurl.com/bnpgnfv
I would love to answer you at length but it would be OT and risk snipping. I will just say that I have studied the mid-Holocene warm period and the ecology of this period at length–no disasters or extinctions.
was supposed to be a reply to someone..can’t even remember who.
What does this discussion have to do with this thread?
This thread is about whether trees make good proxies, and whether certain evidence that they don’t was deliberately hidden. Are you making an unrelated point about the “substantial likelihood that temperatures will rise in coming decades on average”? Then your comment belongs on “unthreaded”.
Or are you saying that the hiding of the decline ‘doesn’t matter’ because there is a “substantial likelihood that temperatures will rise in coming decades”. In that case, you have made a more fundamental error. Doing science right matters.
Neil, the reason this all matters is that people use the proxy record for past centuries to argue that the internal natural climate processes yield a relatively stable temperature measure, hence the modern warming trends must indicate a non-natural forcing. That argument hinges on the “uniformitarian” assumption that the relationship between proxies and thermometer records during the period in which both can be observed holds up over historical intervals where only the proxies can be observed. The problem is that over the 20th century we have a type of proxy that appears to follow temperature trends for part of the century but then in recent decades it diverges, such that the proxies fail to register the rising temperatures. That opens up the possibility that when you observe flat intervals in a historical proxy record, it either indicates contemporaneous temperatures were flat as well, or that they were quite high, and the proxies simply failed to record it in the same way as some are failing to do now. In other words the divergence problem puts into question the empirical claim that the modern trends have passed outside the range of natural variability, since it suggests that proxy-based estimates of the range of natural variability may be systematically biased too small.
All these points can be found in the discussions in peer reviewed literature. But as Steve has shown, when it came time to put these uncertainties before public audiences, including policy makers, some scientists deleted or hid the diverging data, creating an artificial impression of uniform proxy-temperature correlation. We already knew this occurred in the TAR, on a cover of a WMO report, and in the AR4. Now Steve has shown that it was done in an EOS article, ironically published in order to gang up on and berate Soon and Baliunas for, among other things, drawing attention to the divergence article in their own article.
Since one of the issues at stake when trying to determine the size of the CO2-related threat is the magnitude of anthropogenic versus natural variability, the issue discussed on this thread is very relevant to the angle from which you are approaching.
Well Ross, it’s lucky that we have so many other non-tree ring proxies that all confirm a hockey stick isn’t it?
Which ones would they be?
glacier length records, marine foram records, borehole temperatures, snowpatch reduction, icecap mass balance changes, changes in sediment supply to rivers, sea ice records. That enough for you? Perhaps you should just accept that global temps are likely higher now than at least for the 1000-2000 years. Not surprising, really, with all the C02 that has been emitted!
I accept data and facts, not unproven assertions. I also accept the fact that the RWP and the MWP were as warm or warmer. That’s the facts.
Writing a laundry list of things that have no relation to hockey sticks do not prove your point. Now, again, I ask, which ones back up your contention of proving the hockey stick. Glaciers retreat and grow (and have been since the formation of the planet). Snowpatches melt and grow. Rivers deliver sediment daily. Sea ice is not constant.
Sine waves are not hockey sticks. So which ones would that be?
You only “accept data and facts” when they support your predetermined position. This isn’t skepticism it’s denial. Isn’t it strange, though, that a whole range of proxies from a whole range of environments all show that the last few decades are almost certainly the warmest in the past 1-2ka?
Ever considered that you might be wrong?
Monty, an even stronger data set: arctic tree line. Clear long term evidence of climate influence.
And… tells us that today is not necessarily special.
Ever considered that YOU might be wrong?
Without getting into policy issues, if history tells us that today’s climate is likely not outside natural variability, the scientific conclusion is quite different than what we’re hearing from IPCC.
I accepted your printed words at face value. I did not create any strawmen to puncture your balloon with. Why do you now pretend to be telepathic and devine what my thoughts are? What am I thinking about now? Give up? The color purple.
I asked you a question. You responded with a laundry list of environmental occurrances, none that were in any sort of pattern. Then you impugn me with your strawman. I have to ask why? Is my question threatening to you or your existance?
As Ross McKittrick has stated, we have tree rings, an upside down data set, and little else. Yet you want me to accept on faith your words? I already told you I accept data and facts. I question all else. That is all.
So please stop with your “Karnac the magnificent” imitation and either respond with an answer or say you either do not want to or cannot.
Carbon dating works for me.
Koch, J., Osborn, G., and Clague, J.J., 2007. Pre-Little Ice Age glacier fluctuations in Garibaldi Provincial Park, southern Coast Mountains, British Columbia, Canada. The Holocene, 17: 1069-1078. http://www.sfu.ca/~jkoch/ho_2007.pdf
Glacier length? It is true that some glaciers are shorter than their maximum lengths which occurred during the depths of the LIA, but did not the melting start in the 1700s? Are you suggesting it is warmer than the depths of the LIA?
Depends which glaciers you are talking about. In some parts of the world glaciers began receding following “LIA” cooling around the end of the 19th century; in other parts much earlier. Like the so-called MWP, there was probably no globally-coherent LIA. A number of glaciers advanced during MWP times (probably increased precipitation) and there’s lots of evidence to show that many glaciers are more recessed than for 5-6ka.
“Like the so-called MWP, there was probably no globally-coherent LIA”
There probably was both. But keep up the mantra. It will provide you with solace.
So tell me why glaciers in parts of Southern Chile receded in the late 19th century, while others receded in the 17th (over 200 years earlier). Still want to argue for a globally synchronous LIA?
Sure – what you are arguing is that the LIA started and ended world wide synchronously. I challenge you to find anyone that maintains that. Even the MWP was not uniform in the years it affected every part of the world. The Stanford study showed that the MWP was only a 150 years in places like China (versus the 300 that is generally agreed to about Greenland).
Today, there are some areas experiencing record cold. And yet if you adhere to the hockey stick – and your contention – that would be impossible. However, if you agree that a world average does not mean uniformity around the world, then you can have record colds even as the globe as a whole warms.
You are arguing against yourself and the data.
“Abstract: Moraine systems of Glaciar Lengua (unofficial name) and neighbouring glaciers of Gran Campo
Nevado (53S) in the southernmost Andes were mapped and dated by dendrochronological means. They
were formed around AD 1628, 1872=1875, 1886, 1902, 1912 and 1941 with the advance in the 1870s being
calendar dated. Recessional moraines within each moraine system correspond to brief standstills or minor
readvances. A significantly older moraine could not be directly dated by dendrochronological methods as
the forest on it was assumed to be second-generation or older. From soil-formation rates, it is assumed
that this moraine was formed at some time between AD 1280 and 1460, a time in which many other glaciers
in Patagonia formed moraines. Overall, fluctuations of Glaciar Lengua show a strong synchronicity to
other glaciers in the Patagonian Andes between 41S and 55S.
This study suggests that Glaciar Lengua and possibly all glaciers of Gran Campo Nevado reached their Holocene maximum during the ‘Little Ice Age.’”
Click to access ho_2005.pdf
You already answered your question. If a glacier can advance because of increased precipitation, it can recede because of a decrease.
Weather patterns are not the same all over the globe in regards to a change of temp. Some regions get more water, some get less.
I personally would not expect glaciers in one area to act the same as another somewhere else during either event.
Including the ones that are upside down? And contaminated by human activity?
Sorry, I couldn’t resist
Interestingly, when I combined only non-treering proxies
Loehle, C. 2007. A 2000 Year Global Temperature Reconstruction based on Non-Treering Proxy Data. Energy & Environment 18:1049-1058.
Loehle, C. and Hu McCulloch. 2008. Correction to: A 2000 Year Global Temperature Reconstruction based on Non-Treering Proxy Data. Energy & Environment 19:93-100.
I got a clear signal of the MWP and LIA than the Team likes to deny. Sure they show warming in recent decades but also show warming from prior to 1940 when GHG became elevated. They also show a warmer MWP which indicates we are not in danger. So what is your point exactly?
Monty,
It is my understanding that no one has yet created a 1000 year or 2000 year reconstruction which shows a hockey-stick shape without relying heavily on one or more of: 1) Yamal; 2) BCP; and 3) “upside down” Finnish lake varve. If this is incorrect and a 1000 yr or 2000 yr temp reconstruction showing a hockey-stick shape has been published which does not rely on one or more of the foregoing proxies please identify the paper(s).
Thanks.
Monty: perhaps what you say is true. The trouble is, the tree ring proxies are the ones that have figured prominently in WMO and IPCC publications, and in the studies used to provide ‘independent’ corroboration of the MBH results. The attempt by Mann to re-establish the hockey stick profile on a non-tree-ring basis foundered on its reliance on an upside-down Tiljander sediment series. If you were to go back over the paleo assessment literature that has been used for public communication and policy advising over the past 15 years and remove the tree-ring based results you would have little left. Now you might, in hindsight, believe that the equivalent case could have been made without any tree-ring studies. Events took a different course, however, and we have to deal with what was done, not with what might have been.
Hi Ross. I agree….but this is a different issue. If we could turn back time, I’m sure that the reliance on MBH would be seen as premature (although much of science behaves like this at times). But the fact that other proxies also show hockey-sticks should tell us something about whether MBH were essentially ‘right’ or not IMO.
There are hockey stick proxies, and non-hockey stick proxies. Tells us vanishingly little. You like hockey, but that doesn’t mean your sport is what was played in the past 🙂
Not sure I understand your comment.
Monty @ 1:01 PM —
You might be interested in a series of comments from earlier today on the Tiljander data series’ use as proxies for temperature, over at the Air Vent. RomanM #46, followed by AMac #49, RomanM #50, Kenneth Fritsch #51, AMac #53.
Thanks. I’ll take a look.
Other proxies do NOT show hockey sticks. The greenland ice core data show warmer MWP and even warmer mid-holocene. The recent warming is not what makes it a hockey stick, it is the lack of past variability and a cool MWP.
Craig: yes, but no-one is arguing that the MWP isn’t seen in Greenland! As you know, it’s probably mainly a North Atlantic phenomenon. Soon and balliunas tried hard to prove a coherent LIA and MWP and made a complete mess of it.
“As you know…” ??? I don’t know that but there might be contrary evidence…
Goni, M.A., Woodworth, M.P., Aceves, H.L., Thunell, R.C., Tappa, E., Black, D., Muller-Karger, F., Astor, Y. and Varela, R. 2004. Generation, transport, and preservation of the alkenone-based U37K’ sea surface temperature index in the water column and sediments of the Cariaco Basin (Venezuela). Global Biogeochemical Cycles 18: 10.1029/2003GB002132.
Maybe the MWP did appear in the southern Hemisphere…
I think one of the points is that it van be a good idea to search for contrary evidence and examine it — instead of refusing to look. Is that not one of the issues that Steve raises?
Monty…
The word you are looking for is “accidentally”…as in “MBH were ‘accidentally’ right”. You can guess that the answer is 42, or you can work it out mathematically…but if you do happen to guess right, it doesn’t mean your methodology is sound. That’s why math instructors require you to show your work.
A graph created using bad methodology that accidentally looks similar to other graphs does not support those other graphs, nor vice versa. Graphs created using bad methodology should be stripped from the discussion.
Ross: re-reading your comment I agree with you. Maybe there WAS too much emphasis placed on MBH. But you must remember that this was ‘new science’ in 1998….attempts like this hadn’t really been done before. No doubt it was flawed and, with hindsight, it shouldn’t perhaps have been used so prominently. BUT, all we know about paleoclimate since then has tended to reinforce their conclusions which means that the central message of MBH was correct. I would argue that the shaft of their hockey stick needed more variability but nothing comes close to the extent or magnitude of present warming. You are arguing that the policy was wrong…I might agree with you but I am arguing that the science was pretty well right.
The hockey stick which will speak will be the one with handle and palette of the same wood.
Re: phi (Dec 1 15:41),
Monty, prior to MBH, there were literally hundreds of published material going back decades documenting the RWP, MWP and LIA. MBH is what was new, and fit the narrative “we must get rid of the MWP”. CG2 brings it to an even better light.
And I would argue it is exactly opposite of your unsupported claims.
http://www.c3headlines.com/temperature-charts-historical-proxies.html
See my previous post for additional links to paleo studies since 1998.
monty fiddles the dial of the radio in his cave, praying for broadcast of victory.
=========
Monty, “[MBH98] was ‘new science’ in 1998….attempts like this hadn’t really been done before.”
It was published despite Michael Mann knowing the reconstruction failed critical statistical tests of validity, knowing that the short centering PCA method inappropriately lofted PC4 into a false prominence, and knowing that the split-bark bristlecone pine series (=PC4) itself was commonly recognized to be an inappropriate temperature proxy (not that there’s a physically defensible appropriate one).
So, it wasn’t “new science.” It wasn’t science at all. And attempts like that hadn’t been done before because most scientists are ethical enough to not knowingly publish false analyses.
One may hope, after the AWG scare has run its course, that such practices will not be done again.
Or that at least others will say “this is the kind of thing the great Dr Frank warned of during the AGW scare” and that will scotch it at inception. But our track record as a species of learning from history isn’t great. (But what other species even tries to write history :))
monty (Dec 1 15:18),
That is not true.
Taking us back to an attempt like this in 1979. Thanks.
It is worth reading also the very next post, where Steve discusses what was actually “new” in MBH98 and concludes:
Yes, Ross, this its the crux.
And given the degree of falloff in the tree-ring data (as shown in Briffa, at least) the amount of “too small” of the bias is anybody’s guess, isn’t it?
And “systematic” is really the operative term. While some may look at it as a difference between the actual tree-rings and the actual temperatures, it is most likely an artifact of the methods of processing the TRW data, the MXD data, and adjusting the instrument data. From the emails it seems clear that the Team never considers that it could be anything but the actual rings and temps. It could never be something they are doing wrong.
“We don’t need certainty to have a risk factor,”
No decent risk assessment claims certitude, that I know of. A risk factor doesn’t need certainty, it needs intellectual honesty.
“All but stations thermometers. It’s almost time to draw the necessary conclusions.”
The only logical conclusion is either than no proxy is tracking temperature except station thermometers, or that all proxies except station thermometers track temperature, or that neither tracks temperature.
It’s a multiple choice, is’nt it? Fortunately, we have some tracks.
– TLT, in principle, are temperatures (http://img708.imageshack.us/img708/1363/anomthn.png).
– It is expected that balance of glaciers melting depends mainly on temperature (http://img38.imageshack.us/img38/1905/atsas.png).
I would add that the density of laterwood can not be simply assimilated to the growth of trees (or tree ring width), the parameters involved are different (http://img97.imageshack.us/img97/8966/dendrod.png).
I would disagree with that. We have a lot of ecological data on which species do well in a particular temperature range in places like Siberia. One could dig trenches along a North-South axis and measure the 14C age and the levels of different plant species. From the change in the pattern of the species one could have a reasonable guess as to the type of climate there was during the 14C age profile.
DocM – If you look at Plate 2 of Briffa et al 2001 (http://www.ncdc.noaa.gov/paleo/pubs/briffa2001/plate2.gif), you’ll see that the post-1940 tree-ring numbers in all but one of the regions is quite a bit below the higher levels of earlier decades and centuries. Doesn’t this argue that the high side of the temperature range was not a factor in the post-1940 period? If the earlier levels were not a high-side limit, then why would (lower) modern levels be?
“It might mean that there are other factors besides temperature that affect tree-rings.”
???? like rainfall ????
How is that climate science continues to insist it is a slight change in temperature that makes trees grow? Trees need water, sunlight and nutrients. When all three are present they grow. Remove any one of these three factors and they stop growing. Nowhere on this list is temperature.
Dear ‘Dear!’ Steve: LucySkywalker posted a link to the EOS paper i believe you are discussing here. I note the plot on the original EOS paper is not the same as your pot shop blowup version and the version. Nothing in the EOS original plot passes over the big fat red line. They are not the same?? Just saying. —- Cheers Johnnnny
Steve – excellent question. It seems that there are TWO versions of this graphic. I downloaded the paper in Nov 2003 and was using the version as published at EOS. Lucy’s link is to a version posted at Hans von Storch’s website in which the order of line plots is different – so that the big red line is on top of all others.
There are many interesting differences between the two charts.
For one, the legends are much less detailed than the ones shown on Steve’s copy.
“Briffa et al Scaled 1856 – 1980” (chart above) becomes simply “Briffa et al.” (in Lucy’s link) and the reference in the caption is “Also shown are two independent reconstructions of warm-season extratropical continental NH temperatures [Briffa et al, 2001; Esper et al, 2002], with no indication of the period covered.
There are more differences than just the order of the lines. Figure 1 from von Stoch’s website contains the Briffa Osborn 1999 Eurasia reconstruction. The one that is published in the EOS Forum contains the Mann Jone 2003 reconstruction.
The von Stoch’s version looks to be the version that was submitted and accepted by EOS. See email 2530 for discussion of the plan to make the substitution to use the Mann Jones 2003 reconstruction.
It is interesting to note that Wikipedia references the von Stoch version in the Hockey Stick discussions.
Click to access Soon.EosForum20032.pdf
Here is the EOS link (to be copy / pasted into the browser.
To be clear, I think that Johnny’s link is not the final version as it appeared in EOS and as excerpted by Steve, but rather a near-final version. Note that only Mann + 11 coauthors are listed, rather than 13 as reported by Steve for the published version. I’ll try to run down the EOS version later in the day.
It’s odd that this is on von Storch’s website given that he is not one of the 11 co-authors. Did he sign on later?
In the storch website version, the legend merely says that this is Briffa et al, without commiting to 1980, while in the version Steve shows, the legend clearly and misleadingly states Briffa et al, scaled, 1846-1980.
See below at 3:08 PM for a link to the final Eos version, with Mann + 12 coauthors (including now Hughes). von Storch never signed on, so perhaps he was just trying to publicize the point the article makes by putting a preliminary draft on his website.
Odd, that it became more damning.
=============
Unprecedented.
To quote LucySkywalker over at WUWT:
‘………….I recommend readers to visit the Eos Forum paper itself, and use magnification themselves to verify Steve’s picture. When you do this, it is, er, mindblowing to see how the incriminating evidence is there, in the paper itself, when we apply a simple forensic technique – attend to the details – that are cleverly made unnoticeable at the ordinary scale.
Note too that the grey is the “uncertainty” attendant on Mann 1999. Note too that there are many other suspiciously truncated lines, whose provenance we can detect by referring to the colour chart. Note too that it is mainly PSEUDO-DATA from models that rises along with the temperature record (which is also vulnerable to recent UHI)…………..’
Good point about the uncertainties being related only to Mann 1999. Visually, it looks like the uncertainties relate to all the lines. I think a couple of commenters above (including me) were confused about this.
I can’t see which other truncated lines Lucy is referring to. Can someone spot them and list them?
One problem when you magnify to the necessary level as Lucy suggests (or as Steve does above) it’s really hard to line up the decades. Can anyone produce a magnified chart of 1950 – 2000 on a grid with decades?
There’s a difference between the cutoffs in the Team EosForum paper cutoffs and the cutoffs Steve shows in the later version here, where Mann & Jones replace Briffa & Osborn for the long records.
More stage-managing of appearance?
Does anyone have the URL for the graph Steve has ACTUALLY used here? Steve quoted the URL I gave above, in the Harvard thread, which does not have the graph actually shown here.
Isn’t Steve’s diagram from the published Eos version, which Mike Mann thoughtfully provides for us at http://holocene.meteo.psu.edu/shared/articles/eos03.pdf ?
The key unresolved question is whether tree ring widths actually do systematically diverge from measured temperature. There are numerous hints in the E-mails that the divergence is actually an artifact of the “calibration” methodology, most likely used to create a flat handle for the hockey stick. Osborn pretty much concedes some of that here (from 0237):
“First, it is important to note that the phenomena is complicated
because it is not clearly identifiable as a ubiquitous problem.
Rather it is a mix of possible regionally distinct indications, a
possible mix of phenomena that is almost certainly in part due to the
methodological aspects of the way tree-ring series are produced. This
applies to my own work, but also very likely to other work.
The implications at this stage for the ‘hockey stick’ and other
reconstructions are not great. That is because virtually all long
tree-ring reconstructions that contribute to the various
reconstructions, are NOT affected by this. Most show good coherence
with temperature at local levels in recent decades. This is not true
for one series (based on the density data). As these are our data, I
am able to say that initial unpublished work will show that the
“problem” can be mitigated with the use of new, and again
unpublished, chronology construction methods.”
Sorry for posting this excerpt again, but until this is resolved by reprocessing the raw data I don’t see much point in trying to find a natural/manmade cause for a phenomenon that may be a data processing artifact.
Briffa never published the “unpublished” studies that “prove” his point about it being a processing problem. It is also NOT true that no other trees show divergence: typically trees within a population, or populations within a region when shown to be “negative responders” or “nonresponsive” to warming are EXCLUDED from further analysis and not included in reconstructions. They insist that these are “bad” trees but it is just cherry-picking. See Wilmking’s work on this.
CLIMATEGATE….the movie starring:
David Beckham as…………………FOIA (doesn’t have to say much)
Balderick (black adder) as………..Phil Jones
Bill Nighy (actor)as……………..Anthony Watts
Daniel Craig as………………….Steve McIntyre
Jimmy Swaggert as………………..Al Gore
……..
Thanks as usual Steve.
The Briffa et al series is also used in Jones & Mann 2004, again truncated at 1940. (Zooming into the J&M 2004 graph finds the Briffa series peeking out behind the others as before).
Unlike Mann et al 2003 (EOS), J&M 2004 uses the correct dates in the label (“Briffa et al scaled 1856-1940”), and the truncation and reasoning is revealed in the caption (“The various other (smoothed) NH reconstructions shown in the enlargement to Figure 5a have been scaled by linear regression against the smoothed instrumental NH series over the common interval 1856–1980, with the exception of the ‘‘Briffa et al.’’ series, which has been scaled over the shorter 1856–1940 interval owing to a decline in temperature response in the underlying data discussed elsewhere [Briffa et al., 1998a].”)
You had discussed the EOS 2003 spaghetti graph in a CA article “Mann on Splices: the Case of Crowley and Lowery” (22 Nov 2005), which is where I stumbled across the Jones & Mann 2004 graph.
Re: oneuniverse (Dec 1 10:28),
You say:
)
I disagree that this discloses the truncation. “Scaling” to 1856-1940 means that they;ve subtracted the mean over this period and divided by the standard deviation over this period, It does not disclose that they truncated after 1940. The language is very sly – it gives them a fallback deniability if challenged, without clear disclosure.
Nor is the absolute decline accurately described as merely a “decline in temperature response”.
I’m in agreement, the truncation is not disclosed in the caption – a careless reading on my part.
Oneuniverse – do you have a link to a pdf version of this file? The figure is on some sites, but the paper is pay walled.
“Climate Over Past Millenia”, Jones & Mann 2004
Google Scholar with a search term “Jones Mann 2004” found a PDF of the paper as the first result. (Including the title in the search term is usually a better idea though). In general, if the search doesn’t return a PDF link on the right-hand side, it’s worth clicking on the “All <n> versions” link next to the result – sometimes one or more of the alternative versions have PDF links.
“no reliable quantitative inter-calibration can connect the past to the future”
Soon came up with a very nice, pithy explanation of the divergence problem here.
The first draft of the EOS article contained the following langauge (2895):
Briffa commented to Mann that his language gave comfort to those who took issue with the divergence problem:
Briffa proposed the following edit:
Steve,
Instead of devising research into the different causes of the divergence at the biological level, which would, in time, allow them to corral the issue, The Team sits around parsing passages so as to mask their ignorance of the cause(s). How that is good science, I don’t know. If they don’t know the cause(s) for the DP, it is their responsibility to determine it and not use the proxy that diverges until they have adjusted the use of the proxy to the results. Instead, they use the proxy when it agrees with them and discard it when it doesn’t, and then cover their tracks with bomfoggery. In their own terms, this is “crap science.”
Steve; your comments are not correct. I’ve posted in the past on the reluctance of “young dendros” to accept armwaving non-explanations of divergence.
I think this is called an outing. Briffa has just been outed.
Cheeers Steve. And glad you responded as now I have your autograph! of sorts and I don’t have to trudge up to Canada.!!! Get used to being famous! There are some people whose autograph I wouldn’t mind getting and it ain’t those twits in the news papers!! BTW—Mucho thankso for everything!
In his March 31 Congressional Testimony, Kerry Emanuel said
“Good, decent hard-working scientists savaged”… yes, like Soon and Baliunas.
“Neither we nor several other investigative panels found any evidence of misconduct”… then again they only interviewed the scientists being investigated.
“we confirmed what was by then well known”… but not because the scientists themselves had been up front about it, but because Steve had been exposing it on Climate Audit starting in 2007.
“poor judgment in constructing a figure for a non peer-reviewed publication”… such as the TAR, the AR4 and EOS?
“only omitted that part of it that was provably false” … and there is no way of knowing which segments of earlier proxy records are false in the same way, yet they are all taken at face value.
“anyone with the slightest interest in the subject could (and did) immediately find the whole proxy record in the peer-reviewed literature” … So all the EOS coauthors and IPCC LAs were aware of, and approved, what was being done?
“this single lapse of judgment on the part of a small number of scientists” … that seemed to occur over and over with the consent of so many team members.
“why many cannot see it now is a mystery to me” … hope this helps.
You are spot on. You write:
‘“only omitted that part of it that was provably false” … and there is no way of knowing which segments of earlier proxy records are false in the same way, yet they are all taken at face value.’
Clearly, proxy studies have not been conducted in accordance with scientific method. The only justification for using tree ring proxies that stretch into the distant past is that they have been used in the past. We must insist that proxy evidence be subject to the self criticak demands of scientific method.
Re: Ross McKitrick (Dec 1 11:34)
I completely agree with the sentiment but surely Steve began ‘exposing it on Climate Audit’ in 2005. A Strange Truncation of the Briffa MXD Series was 1 May 05. The date CA itself began is imprecise – one could choose 31 Jan 05, 19 Jan 05 or even 26 Oct 04, which is a post on Spaghetti Diagrams to an earlier blog by Steve copied to this one by John A in Jan 05. Try appending 2004/10/26/spaghetti-diagrams/ to the site name and click forward (put like that to avoid going into automatic moderation).
Ha, it’s futile to resist automatic moderation. So here are two other useful links: 26 Oct 04 on Spaghetti Diagrams and 19 Jan 05, which seems to be the real post by Steve to Climate Audit. One day historians will come across this stuff and say this guy also got it wrong. But I did my best.
The first long post by Steve on the hockey stick to the Climate Audit we know today was New research published on MBH98 on 27 Jan 05. (The context for this comment will be clearer when two others emerge from automatic moderation.)
I stand corrected. It was the 2005 post I was thinking of.
“Indeed, they did not simply “hide the decline”, their “hide the decline” was worse than we thought. Mann et al did not merely delete data after 1960, they deleted data from 1940 on…”
Is Cucinelli still investigating Mann?
Ross, thanks for clarifying some relevant points. Your reply is more useful than the crude challenges others provided.
It is the most useful for this reader too. I commend the clarity of what Ross has written in the field for many years. It will stand the test of time.
…so….they didn’t just try to paper over their doubting data (EOS) but tried to paint over it too (Storch)!!!
David Beckham as………………………FOIA (doesn’t have to say much)
Balderick (black adder) as………..Phil Jones (Head of Climatic Research Unit)
Christina Hendricks as………Christina H(Head of Climactic Research Unit)
Bill Nighy (actor)as…………………Anthony Watts
Daniel Craig as…………………………Steve McIntyre
Jimmy Swaggert as……………………..Al Gore
The application of error bars from prior studies and their coterminous disappearance upon the emergence of divergent data is equally deceptive, creating the illusion of reducing uncertainty into the modern era. Inclusion of divergent data would seriously puncture the correlation statistics, perhaps even rendering the end of the graph below the 95 percent magic certainty number. Then ugly words like “falsification” and “null hypothesis” start getting thrown around.
Craig Loehle wrote:
“It is also NOT true that no other trees show divergence.”
No disagreement there. I am aware of your disdain for treemometers, so I presume you would agree that I could find a sufficient sample of trees that would show anything I want over a 50 year period. But numerous studies from the Alps and other locations do not show divergence. To elaborate my point a bit more, suppose that trees with a strong MWP were omitted from the study in favor of those that could be massaged into a flat line. Now further suppose that the “chronology construction methods” used to create that flat line, most likely creative calibration, had the unfortunate side effect of causing the data to go crazy beyond 1940. How would we know? That is why I was hoping to convince someone (because I lack the skills myself) to cherrypick from the same Team databases to find trees that best track the recent warming (i.e., assume that the best information we have, satellite and thermometer records, provides the best calibration). My guess is that the MWP would pop right out.
“But numerous studies from the Alps and other locations do not show divergence. ”
Interesting, have you a reference ?
Having practiced experimental science for the past quarter century, it isn’t uncommon during research to find questionable data and to discard it based on information that would call into question it’s validity. For example, you’re monitoring a photoelectron spectrum. During the scan, spurious spikes due to cosmic rays or electronic glitches can occur. So the experimenter would look at the spectrum and say “I don’t really believe those spikes…let me redo the experiment and see if those spikes are there a second time”. Or it’s questionable because the spikes occur in unusual places.
However, in the temperature reconstructions, what truly baffles the mind is that they’ve discarded the recent data which I would think is really the only truly robust data in the entire series. It’s harder to trust thermometer readings from the early 18th century…if those numbers where “funny looking” I’d say you might want to be skeptical about them fitting your theory and keep them out of the model. But they’ve done the opposite…they keep the ancient data but throw out the recent stuff. It’s really hard to see how the center piece of the whole study wouldn’t be the recent stuff…the stuff collected within your lifetime, with your instruments, and your judgement. This alone creates tremendous doubt for me in their story.
And seriously…they don’t know why the tree rings don’t match the current temperature but somehow they trust they are perfect proxies for temperatures centuries ago???? Give me a break.
The problem is of course that they can’t “redo” the experiment with the same trees…
taking a core sample does not kill the tree. They could do exactly that. It would likely give the same result.
I meant the same trees starting over with a different known climate. One thing they could do is develop their calibration with the first half of the instrumental period vs second half and compare the two. But they they claim it is not enough data…
The disappearance of the end point has obvious parallels with the TAR graphic. I think that joint consideration of the circumstances of each case strengthens the “Hide the Decline” hypothesis. In the TAR with an arbitrary series cutoff of 1960 the series had to be aligned with an upward offset to get the desired endpoint effect. In this graphic, the series has the proper alignment therefore the cutoff needed to change to accomplish the same effect.
It’s not easy to find trees that contain both the modern warming and the MWP. They would have to be at least 800-1200 years old. Very few trees except bristlecone and foxtail pines reach such ages, and everybody (except climate scientists) know that those species are limited by moisture, not temperature.
The siberian scientists who provided data to CRU (name starts with an H–dang I wish I had Steve’s memory) used subfossil buried trees and crossdated to extend their record back. Their larch do not live 1000 years.
Does anyone know what the statistical correlation is between the instrumental data and the different proxies? Some of them look good and some don’t. But it’d be nice to have an actual statistical analysis. How come you never see one? Seems like some basic facts that should be part of any discussion.
Another point that I’ve never seen addressed is how utterly poor the the correlations are between the different proxies before the instrumental data. You have some periods where one proxy is rising for 100 years while another proxy is falling over the same period. At other times, the proxies disagree on the global temp by as much as 0.7 degrees (it looks like). How can you possibly rely on the proxies in the past when there is so much disagreement between the different proxies?
The official Eos version of the paper is online for free at
http://holocene.meteo.psu.edu/shared/articles/eos03.pdf . Readers might download a copy now just in case Mann decides to yank it from the PSU website.
This final version has Mann + 12 coauthors (the 11 on von Storch’s draft version plus Hughes).
What might be called “Snip the Decline” is invisible to the naked eye, and requires at least 600% magnification for me to see it.
Eos describes itself as a “Newspaper” of earth sciences, published by AGU every Tuesday. See http://www.agu.org/pubs/eos-news/ . OSU only has hard copies before about 2009, but fortunately Mann has made the above easily magnifiable PDF available to us. Note that the title Eos is the name of the Greek Earth goddess, and not an acronym like AGU, so that EOS is incorrect.
Correction — Eos is the goddess of dawn, cp Eocene.
For reasons not to do with climate, but archaeology, I was looking earlier today at http://archaeology.about.com/cs/dating techniques/a/timing_2.htm which discusses various dating techniques of sites and artifacts. In the discussion on tree rings, or dendrochronology, it says it was first developed in the American SW by an astronomer, AE Douglass, as long ago as 1901. He began investigating tree ring growth as an indicator of solar cycles. Douglass thought solar flares affected climate and hence the amount of growth a tree might gain in a given year. His research found that tree ring width varies with annual rainfall and varies regionally, reflecting wet years and dry years. If this was known a 100 years ago why do climate scientists assume tree rings measure temperature?
Carol – To me the real question is how different scientists try to use tree rings as proxies for either precip or temps. If both are factors in tree-ring growth, isolating either ones seems utterly impossible. And if they can’t then neither proxy output can be valid. Climatologists must be assuming constant precip, which is absolutely absurd. I doubt one could find even one location where precip from decade to decade is constant, much less year to year and century to century. But perhaps I am wrong, and if one could find a few dozen such places for which this holds true in the last 100 years, then such extrapolations might make some sense. But it is certain that this is not done now.
You are absolutely right–the Team never even tries to verify that precip is constant (or in surplus)over time at the sites they use. They just ASSUME it.
Re: Craig Loehle (Dec 2 09:35),
My impression is that Briffa was originally very mindful of using “temperature-limited” (i.e. not precipitation limited) microclimates; but before long the temptation of substantial data from e.g. the American Southwest won out.
Off topic, but I can’t believe Wigley,Hume and Jones were trying to get Von Storch thrown off the journal because he invites the likes of Lindzen to participate. Talk about back stabbing. I guess fighting for tha “cause” excuses any behavior!
http://di2.nu/foia/1051190249.txt
Reminds me of something I heard on “Wall Street Week” many years ago (You remember Wall Street, don’t you?–it was in the news recently): “A corporate report is like a bikini: What it reveals is interesting; what it conceals is vital.” So therefore: “A Mann et al. spaghetti graph is like a bikini…”
Looking at the magnified image, it appears that the Jones et al. 1998 curve (solid black line) has also been truncated, at around 1970.
I noticed it too. It ends at about 1977. 14 years short. But then you have the problem of how to smooth it all the way to the end. For a 30 year running average you would have to pad it with 14 years or leave the last 14 years out.
All the curves should of course have been displayed the same way regardless of how the smoothing was done.
phi,
Try these:
Szeicz and MacDonald, 1995;Jacoby et al., 1996;Biondi et al., 1999;D’Arrigo et al., 2000, 2001; Kirchhefer, 2001;Wilson and Luckman, 2002;Cook et al., 2003;Davi et al., 2003 (for density);Wilson and Topham, 2004;Bu¨ntgen et al., 2005, 2006b, 2007;Frank and Esper, 2005;Luckman and Wilson, 2005
tty: You don’t need 1000-yo trees, but you do need overlaps that can be conclusively correlated. Bristlecones in, say, Bryce National Park would not be a good choice for temperature, but foxtails at tree line in the Trinity Alps are unlikely to be affected by moisture stress.
Carol Smith: Basic dendro methods looking for temperature correlation involve only trees at the northern limit, or at treeline in the mountains. Folks have put thousands of hours into this, much has been established.
This argument leads to the conclusion that we cannot obtain a Northern hemisphere reconstruction from trees nor is there any basis to claim that MWP was limited in extent. If you are going to argue solely about the Arctic then the best proxies to use are surely the GRIP ice cores which show the MWP and other warming periods before it.
There is no physical theory that can be used to extract a temperature from a tree ring. Correlation does not establish causation. The thousands of hours put into normalizing dendro proxies against the modern temperature record has been a waste of time.
For alpine treelines:
http://www.co2science.org/data/mwp/studies/l1_piancabella.php
“Thus, we assign the MWP (at a minimum) to the period AD 1040-1280; and adjusting for warming between 1950 and the present, we calculate that the MWP was about 0.5°C warmer than the peak warmth of the CWP.”
And tree reconstructions in California:
http://uanews.org/node/30720
“What’s not so well known about the Medieval Warm Period is how warm it was in the western U.S.,” Swetnam said. “This is one line of evidence that it was very fiery on the western slopes of the Sierra Nevada – and there’s a very strong relationship between drought and fire.”
It’s funny how something the team so often insists is highly local, is still referred to by everyone else working in the field as if it was an unquestionably global phenomenon!
Of course Mann too finds an MWP in the tropics when he needs to; ie to correlate hurricane activity.
There are two problems with this approach:
1) Your data is limited to the highest trees in the world and you would have to justify using it for global temperature change.
2) You have not engaged the underlying science. You still do not know what causes growth changes in these trees. All you have done is maximize the one change that you believed to be the most important before you undertook this empirical research which is limited to this one factor.
A tree cannot move its location. How can a tree remain at the critical altitude for temperature sensitivity as the temperature fluctuates over hundreds of years? There is plenty of evidence (including photos of stumps) to show treelines higher in the last few thousand years. I cannot see how a tree can be a continuous sensitive indicator of temperature at the very limit of its envelope. As the temperature falls it might, surviving periods in a climate beyond its limit, but as temperature rises it drops back into its “comfort zone” and then it is likely that its growth response is no longer simply related to temperature but becomes a compound response to other factors including moisture availability. Lets also not forget that the idea of a tree as solely dependent on temperature throughout its life during a changing climate which includes other factors such as moisture deficit is essentially an unknowable and unsolvable problem. Too many variables, not enough data. The fact that the verification statistics, including R^2, are so low clearly shows trees are very poor thermometers and respond to multiple compound factors.
Actually there are at least three versions of this running around…
Click to access Soon.EosForum20032.pdf
Click to access eos03.pdf
Click to access EosForum2003_revised20June.pdf
DHG–
Your first and third links are just early drafts (with only Mann + 11), and so don’t count for much. Your second link (Mann + 12, including Hughes) is the official July 8 published version that matters most.
Hu,
By my recollection this paper was central to the ‘”contain” the putative MWP’ controversy so it is interesting to consider the drafts in that context. I haven’t taken the time to review what others have found in that regard…apologies if I am restating old news…
Firstly, you have written several times on this thread that the final draft was Mann +12 including Hughes. By my count all 3 versions had 13 plots (including the instrumental record) none of which are labeled Hughes. Can you help me understand that point?
As I recall this issue, the Team dropped “Briffa and Osborne” from the June draft on the von Storch website in favor “Mann and Jones with uncertainties” on the final version. Putative MWP contained.
In the June draft Briffa and Osborn is described as “a more tentative extension…based on very long Eurasian tree-ring width chronologies…” I don’t imagine Briffa appreciated that characterization. Particularly when Mann and Jones is apparently less tentative described as “…an extension back through the past 2000 years based on eight long reconstructions…”
So “tentative” was the Briffa and Osborn plot that Mann recalls the language several years later in his testimony before the PSU inquiry here http://www.cce-review.org/evidence/MannResponsesUK.pdf (see page 4 paragraph 1).
DGH, Hu is talking about the number authors of the paper, not the number of plots.
It is interesting that they replaced the long series Briffa and Osborne with Mann and Jones. Can anyone find an email where this switch is discussed? M&J was clearly preferable because the handle is less wiggly.
Of course M&J was the paper about which Bradley wrote:
” the Mann/Jones GRL paper was truly pathetic and should never have been published. “.
But this view didn’t prevent him from signing up as an author of the EOS paper that used M&J!
Steve: look at my recent post on the introduction of the Yang series into Mann et al EOS. This was done concurrently with the change in the series in the spaghetti graph – by Mann and Jones AFTER the acceptance of the article under a form of protest by Bradley which he did not press.
.
Paul,
Thanks for setting me straight on that point.
In regards to tying this to emails start with #0031 where the paper originated as an early reaction to Soon & Baliunas. http://foia2011.org/index.php?id=31
In discussing the Figure that is the subject of Steve’s post, Mann suggests that
“>I would suggest we scale the resulting PC to the CRUREDACTEDannual
>Northern Hemisphere mean instrumental record, which should overlap w/ all
>of the series, and which pre-dates the MXD decline issue…”
#285 has them discussing the details of Figure 1. http://foia2011.org/index.php?id=235
Mann instructs the others that he and Rutherford have been experimenting with various plots.
“Scott and I have experimented a lot w/ line types/thicknesses, etc. I take it none of you are partially color blind? Out finding has been that using too thin coloured lines makes them indistinguishable to many people. The thicker coloured lines are easier to make out, for people who have trouble distinguishing fine colour differences. So I’d lobby for the thicker lines, using thin lines in a few cases to draw further distinctions (with this many curves, we need to use colour, thickness, and line pattern type as much as posslbe, to distinguish).”
He continues, “One can also scale the variance (as you and I did in our submitted GRL article) and the result is basically the same…The only exception is Brifffa et al MXD, where the 1856-1940 period is used instead (because it starts to diverge downward about 1940 relative to the NH annual mean record).
We also don’t show it after 1940. I agree this has to be made very clear in the caption, and Scott should be able to help you guys make sure the caption is accurate.”
(This is discussed above and elsewhere.)
From 0595
Mann says, “I wouldn’t bother w/ a version with all thin lines–will be too difficult to tell apart the different colors (for me anyway, and I bet for lots of people). Instead, why don’t you try a scheme that uses a combination of thick and thin. What about thick-dashed for models, thick-solid for, and thin for dendro only (would include Esper).”
#0682 with emails from early June as the paper is being drafted has some interesting quotes by Mann and Tom Wigley…
Mann writes, “…but most of all we really have to do, in as simple terms as possible, is explain why the SB03 stuff is so fundmentally flawed. And, to boot, we have to do so in such a way that it seems more a casual consequence of what we say, than (as it is in fact) the central motivation of the article…”
From the email where Wigley’s son discovers that water impacts the growth of trees which was covered by WUWT yesterday, Wigley replies, “…By chance SB03 may have got some of these precip things right, but we don’t want to give them any way to claim credit…”
Finally found the first reference to swapping out the Briff and Osborn plot in #2530 (Kan mentions this above, but I find select quotes interesting.)
This email centers on the Mann Jones piece that is being written concurrently. They only two people participating in the discussion are Mann and Jones with Rutherford CC’d.
Jones writes, “…I would suggest with EOS we add this series into Fig 1, back to AD200, possibly by replacing the long Briffa series.”
Mann replies, “I like your idea of checking w/ Ellen or Judy Jacobs if we can substitute in the NH reconstruction (area and local-correlation weighted version) from the Mann and Jones (2003) paper. When I originally asked, Judy said we probably couldn’t do it, because it was not accepted/in press. Now that it is, I’m sure we can substitute it for the long Briffa series–I agree that would be better.”
Yes, it seems that the swap was agreed upon just between mike and Phil without consulting the rest … nice “Team play” 😉
see my post on Yang a couple of days ago – it’s the same topic though I didn’t connect the replacement in the proxy diagram with the concurrent replacement in the spaghetti graph.
In #2945 there is some concern about how busy Figure 1 has become…
Mann and Rutherford disagree with that concern…Rutherford writes,
“I think the current figure illustrates the range of reconstructions, the range of models and how well they agree (similar to one of our original ideas of a “cloud of reconstructions”).”
Cloudy indeed.
Briffa tells us this was a group decision in #3555,
” It was a majority decision to leave the Mann and Jones 2000-year series in the Figure 1 (as it was to remove the Briffa and Osborn tree-ring based one), and the details of the logic used to derive the Mann and Jones series is to be found in the (cited) text of their paper.”
Sorry for the continued spam on this point…these emails are the gift that keeps giving. And as I have noted, others have hit on all of these points. But attempting to put them in order presents an interesting narrative with context. I’ve ignored some of the branches the emails take, for example the emails about Inhofe, but they are also revealing.
Continuing…
Briffa wasn’t necessarily in complete agreement with the decision however…#5027…
“Now to the comments re the EOS piece. I believe you criticised the inclusion of the 2000 (Eurasian ) tree-ring series (since reiterated by Malcolm). Fair enough , though again misguided in my opinion if on the basis of “contains few data ” or ” has weak climate response” . I was perfectly happy to drop it ( I never suggested its inclusion in the first place), but I find it somewhat ironic that it should be replaced with the latest (Mann and Jones) series that contains the same three series plus a mixture of other far more dubious (not to say bad ) series – I agree with the remarks you made re some of these (particularly the Chinese series) in your recent email to someone. I consider that this new series (plus the illustration of the Western US series in the EOS) piece will “stimulate further discussion ” in the field , both between we palaeo-types and the Sceptics . I and Tim have been left to submit this and the balance of pressure seems to be to submit as is – if we remove the suspicious Chinese series we would have to delay things further (Ellen is hassling for us to submit) and , anyway, it is still contained in the Long series. I am of the opinion that the points made in the piece still stand – and by signing on , we are not individually sanctioning all the curves or data used in the illustrations ( There are genuine problems with ALL of them). We will therefore , add Malcolm’s name and submit the version we now have.”
#2530
Hughes was brought in on June 20, see 1056133160.txt (CG1).
Was there any text added from Hughes? Notice also that mike used in Fig 2 the ITRDB PC#1 (AD200 version) without “correction” (as in Mann&Jones 2003 I suppose), i.e. without the Mannkovitch bodge, despite Hughes raising the issue! So we now have a situation where it is necessary to “remove the non-climatic effect” in the AD1000 version (MBH99) but this is not necessary in AD1400 (MBH98) nor AD200 (Eos 2003, Mann&Jones 2003 (?)) versions! Amazing inconsistency.
Comparing the June 20th version that include “In Press” in the title I find very few differences in the body and the figure captions. The differences are primarily to fix typos and adjust style.
Hughes did not appear on the June 20 version and was subsequently added as you know. But it doesn’t appear that any late comments by him were incorporated after that date.
Is it true that they also don’t use any series that does not correlate to the early 20th century? To me, that sounds just as dubious as hide the decline. Correlation of a certain range does not mean that it correlates for a longer range. Hide the decline proves that point. I guess Mann et al. just though this was true only on one side of the range. JK (or am I?) However, deleting proxies that don’t correlate is pure pseudo-science. If you can’t determine the signal that is responsive to temperature, then deletion of the series that are known to be wrong isn’t going to help you one bit. It sounds attractive at first. Hey, we’ve narrowed down the samples. But no. It actually makes it worse because now your confidence in the remaining series should be ZERO until you figure out what the signal is. Otherwise, all you have is noise. Anyone that’s dealt with encryption will know what I mean. You can find patterns in ANYTHING. So if you start selecting the parts that look good and discarding the rest, you end up with word soup. In this case, it’s proxy spaghetti. It’s complete garbage.
There are 9 series shown below the Snip-the-decline graph in both Eos paper versions.
I suspect that shining the forensic spotlight on these might produce some more interesting material. Those 9 graphs look suspiciously bland and upticking around the end.
I can extract and digitise all the lines individually from the pdf that was in the last post.
If you know a site where I can drop the data I’ll get on it.
RJ Hendrickson posted a link about ocean heat content that I’d read a while ago but forgotten about. Thanks for bringing this up.
On the second page there is a graph showing heat content (without a scale or units) and a radiation budget.
What strikes me is that these “top” scientists don’t even seem to be looking at the right thing.
They seem happy that heat content rises roughly together with energy imbalance from 1993-2003 but don’t seem to notice that this quantity is measured in W/m2 (power or rate of change of energy) while heat content is in MJ (energy)
http://earthobservatory.nasa.gov/Features/OceanCooling/page2.php
If it is the energy imbalance that causes the warming they should be plotting rate of change of ocean heat content. ie the time derivative not the heat content. With or without the post 2003 drop they don’t like, this radiation budget in now way fits OHC data. It looks like Wong is wrong by about 0.5 W/m2. He finds a continued warming where there isn’t one.
Only if it was the ocean heat content , (reflected in surface temperature) causing the imbalance, would their plot have any relevance. That does not seem to be their position.
Maybe if they compared dimensionally equivalent quantities, they may get further.
I think it would be worth having a closer look but, as usual, they don’t make their data visible. Even all their papers are behind the GRL paywall.
Fergal, if you could digitise that graph, or if someone can point me to the tools to do this sort of task on Linux, it would be very helpful.
Solar,
It’s all done on Linux compatible free software, but it’s frustrating to learn by error and I have it down quick and dirty. I think.
I picked a file dump at random, hope it’s okay (I have adverts blocked). It’s a plain text file with Mann’99, Jones 1856-1980, Crowley&Lowery(?) and Briffa et al in that order separated by a line of X’s. I rarely use a spreadsheet so you’ll have to deal the 6(!) significant figures, varied decimal years and anything else I’m not used to coping with. They graph like they should look but please check my work. Doubtless many could do it better. I’ll do the rest if these are okay:
http://www.mediafire.com/?jjsfg66ye4z3puz
Fergal, in fact I was not asking for Mann’s graph but the one I linked to, (which is a lot simpler and cleaner).
If you don’t fancy digitising it , could you tell me what progs you use?
Thx
Sorry, I was reading 3 things at once.
Here’s the numbers from that ocean heat storage graph:
http://www.mediafire.com/?990k9egpzq084re
Don’t download or run third-party software unless you’re sure you know what you’re doing! Bad things will happen.
The digitizing program I use is called Engauge Digitizer, downloaded from Sourceforge. It’s much easier to use than it looks, but (using an imaging program) lines should be filled in black first and if they overlap you’ll have to do them one at a time or semi-manually. Doubling or quadrupling the size of the image helps a lot too.
Simple tutorial for Engauge here:
http://bililite.com/blog/2010/01/14/engauge-digitizer-tutorial/
If a pdf has a graph with all the vectors intact you can see them individually with a program called Inkscape – also downloaded from Sourceforge. Delete the series you don’t need to leave the one(s) you want with the axes then take a screenshot.
I probably should have added: if you’re not familiar with photoshop/GIMP-style image manipulation you’ll tear out some hair at the bottom of the learning cliff.
Many thanks , I’ll see what I can do with that data.
I’d actually installed engauge a year ago but forgot what is was. Reasonably capable on gimp. Thanks for the tip on using inkscape on pdf , I didn’t know about that.
Thanks again for obliging my request. Saved me a fair bit of time. 😉
Fergal,
I superposed your four series on the Eos figure, and they match as perfectly as I can align them.
What was your method?
Steve —
As I was comparing Fergal’s extracted curve with your figure 3, I noticed that your version of Briffa 2001 doesn’t match at the early end (1400). Perhaps an edge effect as you were applying Mann’s smoother?
Steve: yes. His end-smoothing has a quirky algorithm and he may have done something a little different in 2003 than he was doing in 2008.
confirming my earlier post. End-filling should be banned in scientific practice. It’s an (invariably) undocumented bodge that implements a subjective and undeclared extrapolation.
That Mann is using different methods at each of the same dataset seems a clear indication that he is inserting what ever subject frig best fits his intended conclusion.
More malpractice.
looking at the band of uncertainly range “Mann and Jones” plotted faintly in the background it does not seem like the data are upticking all the much.
Again this is an illusion created by the big fat red line they obscure the rest of the data with. Note how the red line is thicker than all the others.
M&J error margin seems centred about the final values of the yellow plot attributed to Cowley.
We can only see M&J from it’s shadow but the profile is one of a peak rising over the last 50y that topped out around 2000. It does not look like a continuing rise.
Does Mann & Jones’ data actually show the last 50-60 years is to a large degree, a cyclic change?
They’re just doing their best to hide the fact behind a big red line from incompatible data.
If you look at what is behind the red line it’s not too scary, even without Briffa.
For the Team , it’s all a game of smoke on mirrors, slight of hand.
They are not scientists, they are illusionists.
Here is how the Team responds:
—————————————————
Please plan to attend
Directions in Climate Change Education and Communication
The compelling nature of anthropogenic climate change is well documented in science literature. But the public and educators clearly lack understanding and appropriate knowledge of climate change due to legitimate confusion and deliberate obfuscation and distortion of science knowledge by interest groups and corporate entities. This town hall meeting will feature a short panel discussion of climate science communication and education, and a series of small group discussions that will conclude with short format verbal and written reports on future directions. Panelists include Michael Mann, Richard Somerville
and James Hoggan.
Location:Moscone West, Room 2005
Monday, 5 December, 1815h-1915h
James Byrne, Professor and Chair of Geography, University of Lethbridge
A woman can tell a lot about a man from the size of his carbon footprint
(Big Bang Theory).
—————————————————–
As the message says, plan to attend…
Worry when talk turns to energy footprint. Think of the little neutrino that could.
=======
I have not noticed any contribution from hans von storch…his comments would be interesting because he has always struck me as an honest guy
RealClimate may have a similar tool, not sure (tic)
http://co2science.org/data/mwp/mwpp.php
I always thought that if tree rings depended on temperature only, then they would all look the same. The last time I checked, they weren’t.
We have in the past seen funny business in the end years of these curves. All the curves are smoothed right up to the end, but the only way to smooth up to the end and keep the curve going up (vs backwards looking running mean, which they don’t like) is to project somehow into the future before smoothing. Mann likes to end pad by reflection and use a butterworth filter. Others like to replace the “future” of the end of the series with the instrumental before smoothing (jones likes that). In any case, both methods would tend to make the last few decades more closely match the instrumental than the raw reconstruction, which is annual, except reflection does not work if the series is headed down, like Briffa’s is. So there may be multiple monkey business going on in this graph, all in the name of proving a pre-conceived notion.
I sure wish the average Joe understood this. But, alas, he does not. Can you write an exposition that explains this for us statistically challenged folks?
I prefer the Mrs. Butterworth filter.
>> Others like to replace the “future” of the end of the series with the instrumental before smoothing (jones likes that).
>>
Wasn’t that precisely what Jones referred to as “Mikes Nature trick”?
He cropped off available scientific data that did not suit him, replaced it with temperature data to pad the smoothing window, then applied the smoothing filter. (Butterworth of Gaussian is a bit irrelevant)
This means the result is a weighted average (the usual implementation of these filters) that is an unholy mix of two different datasets reflecting physically very different quantities.
Mann seems to have limitted himself to quietly doing this to then end of Briffa to get it to turn in the right direction before cropping it off.
Jones went the whole way and graphed the two together.
Thanks Craig, very interesting summary of the smoothing. The trees that one might pick to flatten a handle would either (a) have a flat response to temperature, (b) actually be upside down with temperature (yikes!), or (c) be random with respect to temperature. Possibility (b) obviously requires little encouragement via processing to produce a divergence. Is there a way, either through creative calibration or smoothing, to go from (a) or (c) to a deterministic divergence?
I believe the common case is c: random with respect to temperature. If you take a buncy of sites where the trees have little functional relation to temperature (red noise series) and pick the ones that correlate with temperature in the last 80 yrs or so, the first part of their respective curves will be random noise whose average (by the linear combination method of Mann) is a nice straight handle. Steve showed this a couple of years ago. The Team refuse to understand it.
Craig,
You are right than randomness works all the way through the logic tree, with the divergence being random as well. But Osborn in 0237 states that the divergence is “almost certainly in part due to the methodological aspects of the way tree-ring series are produced.” He seems to be hinting that the divergence is deterministic.
Matt –
Not certain the usage of “deterministic” you mean. Why would the method used give good linearity pre-1940 and then diverge doesn’t seem to be covered by Osborne’s speculation, and he knows it.
Craig – Your description about “end=of-record” smoothing is a good one, but does it apply to the divergence? I don’t think so.
End of record smoothing does not create the divergence, but if they used end point reflection to get the future data to smooth with, any curve with a trend (up or down) will have that trend exaggerated near the end compared to a running backwards-looking mean.
If the red noise series persists beyond the calibration range as well as before the calibration range, you would not get divergence from simple averaging. So randomness would mean that the divergence is an artifact of nothing more than too short of a time series after the calibration period. I’m not sure I buy that.
If we generate a large number of series which are random and flat, overall, then pick out those that have a definite upward slope for the 100 years calibration period, it follows that somewhere in the series there is a downward trend.
You then call those that show an upward slope good thermometers of temp and throw out all other samples as being non specific.
Magically you now have a data series that shows… An upward slope!
Imagin if drug trials were done like this.
Gosh, “hide the decline” seems to be a common methodology used by climate scientists. Here’s another ‘team’ that substituted alternate data that conformed to expected results when the original ARGO buoys didn’t yield the desired rise in temperature:
http://earthobservatory.nasa.gov/Features/OceanCooling/
My, my, indeed.
How do I deceive thee? Let me count the ways:
1) Hide the decline
2) Only include favored reconstructions that support “the cause”
3) Exclude trees/sites from the recons that are not responsive
4) Thick red line to hide stuff and to emphasize alarm (red=hot)
5) End-point padding by reflection or using instrumental data before
6) Smoothing with a centered smooth
7) Rescaling the SMOOTHED series (not using the original data)
8) Choosing different anomaly base years in different graphics to get best agreement of the curves
9) Deceptive (too small) confidence intervals
10) I’m sure I missed some.
Wow, the # 8 became a smiley face.
I did leave out 11) Mining for hockey sticks using PC or other methods.
(12) using pseudo-proxy pseudo-data ie models: those dotted lines. Used to visually support the final thermometer uptick.
This may be a silly question, but have any scientists carried out experiments growing trees in environments controlled for average temperature, temperature variations – day/night, summer/winter, water supply & seasonal variations in it, nutrient supply, soil percolation/drainage characteristics, CO2 in atmosphere, tree species, tree age, daylight/sunlight hours, daylight changes (eg when you cut down trees shading a remaining tree, its growth rate increases) and any other factors that biologists might think influence tree ring width & density? I am speaking as a gardener, not a scientist, but all these things all seem likely to affect the growth rates of my trees.
Without dealing with all the variables which might determine tree ring width & density, I don’t see how one variable, temperature, can be inferred from historical tree ring data. I get the impression that the people who attempt to make historical temperature reconstructions from tree ring data don’t seem to be experimental scientists, merely data miners with poor statistical skills and little or no integrity. Am I missing something?
IIRC, look up the Idsos. Also, if you click my name, my “Climate Science Skeptics Primer” has lots of relevant links that might help you find what you want. Also, look at my Yamal pages (click “index to topics”).
News alert: researchers who discovered an error in their analysis promptly issues a press release, are doing the study over:
http://fuelfix.com/blog/2011/12/01/researchers-say-there-is-an-error-in-gas-drilling-study/
Mr. McIntyre, seems to me it would be a fine and courteous gesture on your part to re-acknowedge in this post what you acknowledge in the comments of the “Direct Action at Harvard” post, namely that FergalR was the first to notice this 1940 truncation. My view–take it or leave it.
Steve –
A few comments earlier, I had previously drawn attention of readers to the graphic in Mann et al 2003 (EOS) as a hide-the-decline issue https://climateaudit.org/2011/11/28/direct-action-at-harvard/#comment-313625 and thanked Fergal for his sharp eyes in spotting the snip in a high-res blowup. https://climateaudit.org/2011/11/28/direct-action-at-harvard/#comment-313752. In the post itself, I credited “sharp eyed CA readers” for spotting the snip (also spotted by HaroldW.) As you suggest, I’ll add their names into the post.
Steve,
A bit O/T for this thread, but if you have not read 3874, you should (from a mixture of CG1 and CG2, not sure which it came from). Mann describes exactly how MM03 went off the rails…mostly because Rutherford bollixed up the spreadsheet. There is also evidence that he genuinely believed that you asked for a spreadsheet, even though you have proven that you did not.
Steve: Mann’s story to his associates was untrue on multiple levels. He accuses us of not noticing the problems in the data set, but we had noticed them and re-did the calculation of PCs. Mann’s allegation that we hadn’t noticed the errors was absurd.
His untrue statements about the Excel spreadsheet are hard to understand, given that there was no such request. Osborn seems to have realized that Mann’s spreadsheet allegations were bogus, but, as typical, didn’t contradict Mann.
My guess is that Rutherford deserves some share of the blame. There’s an email in which Mann asked Rutherford about the date of the file (which preceded our request). The answer isn’t in the files.
Strangely, Mann re-iterated this untrue story to the Penn State Investigation.
Well … just very occasionally contributing general climate change discussion, I just want to express my disappointment on the understanding of climate cyclicity. If considered, I am very sorry to
I think my original hypothesis is proven correct by 4450. The divergence problem that led to hide-the-decline was never anything more than a statistical artifact of the way the series were standardized.
Briffa 4450:
“The divergence is only apparent (at the NH average scale) in the smoothed i.e. lower-frequency domain. You need to smooth the tree-ring records and the temperature to see it. However, the divergence is largely an artifact of using curve fitting (i.e. based on least-squares fitted regression lines or functions ) to estimate the unwanted (biological) growth trend in the tree-ring data. These fits are influenced by climate warming signals in the recent data, and this signal is inadvertently removed in the standardising process. When non-curve fitting methods are used (such as RCS) this problem is largely removed.”
I am a bit gobsmacked by the amount of work over the years that has gone into trying to find a natural cause for this, especially since Osborn seems to have suspected the cause for quite some time.
I am not sure that I can buy Briffa’s explanation.
If I understand his words correctly, he is claiming that using an adaptive methodology to account for natural growth patterns rather than a fixed functional form (with possibly estimated parameters) is the factor responsible for causing a decline in the estimation of the temperature series. I would suggest that given all other factors being constant, the most it could do is to flatten the series. There would have to be other influences to produce a decline of the observed magnitude.
If, as he says, RCS can solve the problem, then why wouldn’t he use it? I doubt that he actually tested this and it strikes me as just more arm-waving on the situation.
RomanM,
I think a biological growth standardization graphs as an upside down U-shape. If the algorithm was used to flatten the handle of the hockey stick, perhaps the data was forced into the wrong place on the U. If the algorithm treated older trees as younger trees, the modest temperature response might be swamped by the biological growth correction. I suspect this all occurs after the time interval that the team found useful. Obviously just guessing here.
Craig Loehle,
If you are still on this thread, can you deconvolute the statement about smoothing? That has to be some sort of end effect.
Briffa is only right about needing to smooth to see the divergence in the sense that with annual reconstructed data it is so noisy it is hard to see what is going on. He does not seem to be considering (in my view) the effect of smoothing on the end of the curve, and in any case his data that trend down in 1940 or 1961 start going down way before they would be influenced by the lack of data at the end and any padding going on.
If you look at Figure #3 in Briffa 01 the normalized “age bands” are shown (left side of graph in light grey). Definitely different from RCS. IIRC Briffa seems to suggest a trend artifact in the age bands due to a common growth signal, and that the effect would be eliminated by fitting to an RCS curve. To me it is at least plausible and it is a matter I have been thinking about over the last week or so after my post at tAV.
Matt, you might be interested in taking a look atthis post.
LL,
Very interesting analysis at tAV, mostly above my head but I can barely keep up. I found this other clue from 0493:
“It is well known that individual standardisation of tree-ring series does not allow reconstructions for frequency bands greater than the mean segment length of the single series. Additionally this standardization produces even end effects that might be at least partly responsible for the a divergence problem in recent years.”
This one gets no attribution because it is an anonymous review comment. At my primitive level of understanding, it sounds like if you smooth, then standardize for biological growth, then combine multiple overlapping series using curve fitting, it blows up at the end points unless you have lots of really long series.
LL,
Now I am curious whether your Figure 3 might be correlated to a biological growth standardization. Probably best thatt I take my questions to tAV at your link above. Anybody know where to find standardization curves for latewood density, mxd(?), etc?
Matt, see response at tAV.
Is tricking the Senate unlawful ?
Well, given all the criticisms of the temperature curves offered by mainstream science, I’d like to see an alternative that the OP or many of those here considers a proper or at least improved alternative, and its derivation. I know, it isn’t necessary to offer a specific alternative if you think you’ve got criticisms of other methodology, but it is more impressive. Such may have been noted here before but really, few of us have time to pore over comments etc.
I personally prefer proxies that are based on physical/chemical partitioning. Some methods related to cave deposits, ice cores etc. show promise but need more work.
Anyone know what happend circa 1460, the low point? Volcano?
Steve, I’m surprised that you missed the smoothing factor in the caption. >40 year time scales would mean that the last possible year that the Briffa curve could be shown was 1940 since Briffa tossed the unreliable post 1960 data in the original work. I they hadn’t done that you would have been screaming about “end point pinning”.
Steve: oh, puhleeze. The data after 1960 wasn’t “unreliable”. The measurements were accurately taken. Nor have I taken exception to people using smooths to the last year of the available data. I did observe some years ago that Emanuel’s foolish method of smoothing resulted in end point pinning but that was specific to Emanuel.
If you believe that a 1940 endpoint resulted from smoothing exigencies and a 1960 end, then the Mann smooth should have ended in 1960. Unless of course he used Mike’s nature trick.
How is the data unreliable?
Well Steve , perhaps you should be criticising end-filled smoothing filters. There is no way that they can ever be a correct result. They _always_ an unscientific frig.
There are all sorts of schemes, as you know, but none of them have any mathematical legitimacy. They will always be wrong and it then down to subjective choices by the data manipulator which he thinks is “best” for his data.
If it is just edge effects in filtering a photo , who cares, but you just cannot justify this sort of bodging of scientific data. Especially when all eyes are on the end of the data, it’s not peripheral.
Steve – we’ve had numerous discussions on end effects of smoothing and, yes, they raise issues. However, these are issues are not germane to enhanced hide-the-decline. The rollback to 1940 was not done because of exigencies of smoothing end-effects – or else similar effects would be seen in other series. The rollback to 1940 was a calculated decision to enhance the hide-the-decline, as Rat should acknowledge.
At the end of the day any end-filling is basically extrapolation. It’s just never declared as such. In fact in a graph it’s almost never declared that it’s even happening.
when you run out of data you stop the line, it’s that simple.
In the EOS response, Mann et al. identify 3 issues that Soon et al. 03 failed to properly take into account.
Point 1) “The existence of possible underlying dynamical relationships between temperature and hydrological variability should not be confused with the patently invalid assumption that hydrological influences can literally be equated with temperature influences
in assessing past climate.”
In email # 0682 Tom Wigley indicates SB03 maybe on to something:
“By chance SB03 may have got some of these precip [sic] things right, but we don’t want to give them any way to claim credit.” Wigley’s comment is in response to Mann discussing a lot of issues that are deemed too subtle for a response that is intended to “demolish” an improperly peer- reviewed paper.
Point 2) “It is essential to distinguish [e.g. by compositing or otherwise assimilating different proxy information in a consistent manner e.g. Jones et al. 1998; Mann et al. 1998; Briffa et al. 2001] between regional temperature anomalies and anomalies in hemispheric
mean temperature, which must represent an average of temperature estimates over a sufficiently large number of distinct regions.”
From email #0404 we see the difficulty of combining the chosen series in the EOS response:
Mann to Osborn “The position of Crowley and Lowery, in particular, is quite inconsistent between our respective comparisons. When we scale the various series to the full N. Hem instrumental annual mean CRU record 1856-1980, we get a a very different relative ordering of the different series, as shown in the attached figure from my Science perspective piece from last year”
The science Perspecitive piece can be seen here:
Click to access MannPersp2002.pdf
Interesting side note – from email #0078 we get reviewers #2 comment on a contemporaneous paper, “Changes in the Northern Hemisphere annual cycle – implications for paleoclimatology?” JGR 2003 by Jones and Briffa, indicating that by using the Mann 1998:1999 paper they might be attempting the very thing Mann EOS 03 accuse SB03 of.
Point 3) “It is, furthermore, important to identify and, where possible, quantify uncertainties; and demonstrate, using independent data, the reliability of any deconstructions.”
From email #2461 we learn in constructing their SB03 response it is, indeed, difficult to accomplish this:
“The reason for the slow delay is that I was attempting to estimate uncertainty ranges on the re-calibrated composite – but haven’t succeeded yet! The difficulty is that I cannot simply combine the published uncertainty ranges under the assumption that they are independent series – because they have common proxy data in, especially early on. On the other hand, I cannot simply use the calibration statistics of the composite to estimate uncertainty ranges, since that ignores the deterioration in reliability early on that occurs in some of the constituent reconstructions because of fewer proxy records early on.”
Ross McKitrick, while you are right about the implications of ‘getting rid of the Medieval Warm Period,’ that it adds weight to the chance that the current warming is manmade, there is a counterargument to this. Mann has said that things were warmer in North America and Europe, but not globally. His reason for this he theorizes is perhaps because of the Pacific Thermostat hypothesis producing LaNina like effects in the tropics. This would mean that if the Medieval Warm Period was not global, but as Mann says, then current warming due to CO2 would be dampened by LaNina like effects as well. When asked about this, Mann said I agree with that. I think there’s a missing negative feedback.
http://www.realclimate.org/index.php/archives/2004/12/temperaturevariations-in-past-centuries-and-the-so-called-hockey-stick/comment-page-1/#comment-666
If random numbers are put into Mann’s procedure, does a hockey stick shape result?
Response: Thanks for your question, and the opportunity it provides for clearing up yet further “Hockey Stick” disinformation. First, I’ll give you a two word short answer: Absolutely Not!
Was it Wigley who confirmed a hockey stick would result from random numbers using Mann’s procedure?
It was Rob Wilson.
He wrote the following in #4241:
——–
I first generated 1000 random time-series in Excel – I did not try and approximate the persistence structure in tree-ring data. The autocorrelation therefore of the time-series was close to zero, although it did vary between each time-series. Playing around therefore with the AR persistent structure of these time-series would make a difference. However, as these series are generally random white noise processes, I thought this would be a conservative test of any potential bias.
I then screened the time-series against NH mean annual temperatures and retained those series that correlated at the 90% C.L.
48 series passed this screening process.
Using three different methods, I developed a NH temperature reconstruction from these data:
1. simple mean of all 48 series after they had been normalised to their common period
2. Stepwise multiple regression
3. Principle component regression using a stepwise selection process.
The results are attached.
Interestingly, the averaging method produced the best results, although for each method there is a linear trend in the model residuals – perhaps an end-effect problem of over-fitting.
The reconstructions clearly show a ‘hockey-stick’ trend. I guess this is precisely the phenomenon that Macintyre has been going on about.
——–
It’s all rather late to comment here, and I don’t know whether anyone has mentioned this, but Figure 6.10b in Chapter 6 of WG1 of the 4Ar has a spaghetti graph of climate reconstructions that recapitulates the fine EOS work dissected here.
One of the reconstructions in 610b is the mislabeled sky-blue line (it should be BOS 2001), and is from Figure 4 in Briffa, K.R., et al., (2001) “Low-frequency temperature variations from a northern tree ring density network. J. Geophys. Res. (Atmospheres) 106, 2929-2941.
It’s also the same line as the truncated Briffa, et al., scaled 1856-1980, line in the first Figure in the head-post.
In IPCC Figure 6.10b, Briffa’s line is truncated at 1960, effectively hiding the same decline.
In 6.10b, it appears the IPCC didn’t use the “preferred reconstruction” in Briffa, 2001 Figure 4. Instead, they seem to have used one of the eight alternative reconstructions, specifically and likely the thin line that stands out with a slight inflection point at 1960.
Briffa, 2001 as it appears in IPCC Figure 6.10b has a slight provocative up-bend at 1960 that doesn’t seem to be in the “preferred reconstruction” of the original Figure 4. The felicitous terminal bend in 6.10b makes it seem like the line would rise if there were further data. But, of course, there are further data and it doesn’t rise.
Jonathan Overpeck was one of two Coordinating Lead Authors on AR4 CH6, and was also a co-author of the EOS article. Other cross-publishing personna are Tim Osborn and Keith Briffa, among AR4 CH6 contributing authors and lead authors, respectively.
So, institutionally, the IPCC/4AR is evidently guilty of the same lie of omission as the AGU/EOS.
13 Trackbacks
[…] He writes now in Hide-the-Decline Plus […]
[…] He writes now in Hide-the-Decline Plus […]
[…] Hide-the-Decline Plus […]
[…] big news today seems to be Steve McIntyre’s recent discovery that the Briffa series in print has been chopped back in a Mann paper even further than advertised to hide “more” decline. Amazingly, […]
[…] He writes now in Hide-the-Decline Plus […]
[…] più di un pizzico di ‘hide the decline’ (cioè nascondi il declino delle temperature), Steve McIntyre ha appena pubblicato un nuovo post in cui documenta che il taglio alle serie proxy non gradite e il famoso colpo di […]
[…] Steve McIntyre / ClimateAudit / December 1, 2011 […]
[…] in the caption ”. This data deletion never was made clear in the caption, leading to a repetition of the notorious “hide the decline” trick. 4207.txt: “ You commented that the […]
[…] https://climateaudit.org/2011/12/01/hide-the-decline-plus/ […]
[…] how scientists associated with the University of East Anglia Climate Research Unit not only manipulated research data but also intentionally destroyed their work. The CRU scientists also knowingly […]
[…] fiasco, in which University of East Anglia climate scientist Phil Jones conspired to “hide the decline” in global temperatures, the U.S. Department of Labor, Bureau of Labor Statistics has just […]
[…] the decline" explained here. That little orange line peaking out is the key part – read the link for a full explanation. […]
[…] This divergence can be seen in what Steve McIntyre exposed when he discovered that the diverging tree ring data had been deleted from the hockey stick graph [click to enlarge]: […]