Where's the hockey stick? The 'Marcott 9' show no warming past 1950

More on the Marcott et al “hockey stick”. All of the ‘Marcott 9’ had altered dates.

marcott-A-1000[1]

The Marcott hockey stick compared to the Mann hockey stick – it seems that redating and other tricks made the uptick, not the data itself.

Guest post by John Kehr

While it took me a while to get the time together to write an article about the Marcott paper, that does not mean I have not been looking at it and discussing it from nearly the day it was released.  There has been volumes of discussion within The Right Climate Stuff group that I have been involved with.  The ones that lean towards CO2 as something to be concerned about were initially rather excited about this paper, but that has taken a course correction as it has become clear how poor the science is in the Marcott paper.

Many skeptics are calling this the newest hockey stick and there is certainly some accuracy to that, but what I initially found interesting was the Holocene cooling that he shows.  In one respect his paper is different because it shows the cooling that has been taking place for thousands of years.  That also makes the stick at the end more extreme, but it is something most will not show.

For those who missed the details of the Marcott paper I will provide a brief summary.  The paper was published on March 8th in the ultimate of peer-reviewed journals, Science Magazine.  The paper was loudly broadcast by the media as further proof of global warming.  The paper basically says that the most modern period of the Holocene (the current interglacial which the Marcott paper states as 11,300 years) has been warmer than ~75% of the Holocene.  The paper states that this is especially significant as the Holocene has shown steady cooling for the past few thousand years, but that has now completely reversed.  The conclusion is that mankind has drastically altered the natural climate of the Earth. 

The paper itself is a composite of 73 different temperature proxies.  These proxies were used to reconstruct the Earth’s climate over the past 11,300 years.  The 73 proxies were not uniformly distributed around the world.  The following is a summary of the spatial distribution.

Tropics:              33 proxies

NH Polar:             12 proxies

NH Mid:                20 proxies

NH Tropics:          16 proxies

SH Polar:              4 proxies

SH Mid:                 9 proxies

SH Tropics:           12 proxies

NH Total:              48 proxies

SH Total:             25 proxies

The NH is over represented by 3x in the polar region and 2x in the mid-latitudes.  This of course can be dealt with easily enough, but the real resolution in the NH is better than the SH.  None of this is directly critical to the paper, but it is something worth noting.

Far more troublesome to the conclusion of the paper is the dating of the proxies.  Other sites have some excellent write-ups on the re-dating in the paper itself and I will touch on it, but my more immediate concern is how recent most of the proxies are based on the published data of the proxies.

Here is the breakdown of the last date in the proxies he used.

1950+                9 proxies (1960, 1970, 1991, 3x 1995, 1998, 1999, 2000)

1900-1950:        16 proxies

1800-1899:        11 proxies

1600-1799:        7 proxies

1000-1599:        14 proxies

< 1000:              15 proxies

This leaves one proxy (GeoB 3313-1) with the last known data of ~1750 AD.  There are 3 data points for temp after that, but no dates associated.

Since the modern period by almost everyone is considered to be post 1950, only 9 of the 73 proxies contain any data that can be relevant to the global warming issue.  Right away that concerned me, but when I looked at the data for those 9 proxies something very interesting became apparent.

I will be referring to those 9 proxies as the Marcott 9.  They are perhaps the most interesting proxies that he used and those proxies disprove the conclusion of his paper.  In order of the most modern data, the Marcott 9 are:

Lake 850, most recent data is from 2000.

Flarken Lake, most recent data is from 1999.

Lake Nujulla, most recent data is from 1998.

Tsuolbmajavri Lake, most recent data is from 1995.

Homestead Scarp, most recent data is from 1995

Mount Honey, most recent data is from 1995

Composite MD01-2421…, most recent data is from 1991

Moose Lake, most recent data is from 1970

Agassiz & Renland, most recent data is from 1960

What is most interesting about all of these proxies is that none of them show the warming result the paper ended up with.  Not a single one.

Without further ado, here are charts for the Marcott 9.

clip_image002

clip_image004

clip_image006

These nine proxies are the only ones of the 73 that Marcott used that have data past 1950.  The only one that shows any kind of warming is the last one which is the Agassiz-Renland ice core and the warm point was not the most recent, but the proxy from 1940.  The last point which is 1960 shows as cooler than the data from 1940.  The ice core certainly does indicate that the warming in the 1930-1940 period was impressive, but few claim that mankind caused that warming.

There is far more to discuss about this paper.  I have an idea where I am going to go with my research, but others may beat me to it which will alter the path I take.  Based on what information is being found by others, primarily by Steve McIntyre over at Climate Audit is that Marcott re-dated data that didn’t fit into the hockey-stick result.

This agrees with what I have found as well.  All of the Marcott 9 had altered dates associated with the last date with the Moose Lake data changing the least at 20 years.  Flarken and Tsuolbmajarvi Lake were moved back into the 1800’s and the MD01 Composite removed the last 3 data points.  The end-point strategy for this paper was full of shenanigans.  Since it is only the end-point data that matters to the conclusion of the paper, well, let the Marcott 9 speak for themselves.

=============================================================

John Kehr is a Chemical Engineer by schooling and Research and Development Process Engineer by profession.  He has more than a decade of experience at the cutting edge of technology for a large semiconductor company.  That experience was critical for him while wading through the often contradictory information that exists about global warming.  He was generally neutral about the subject of global warming until he met and married a wonderful woman who challenged him to make a choice.

There are few things more dangerous than challenging an engineer to make a choice like that on a scientific topic.  While occasionally taking a break from research to breathe and go on getaways with his beautiful wife, he spent many months deeply involved in his research.  When he finally reached his conclusion, the only logical thing was to put all that research into a book, The Inconvenient Skeptic

Advertisements

  Subscribe  
newest oldest most voted
Notify of

Of course the Marcott study doesn’t tell you about warming past 1950. It was a study of the Holocene period. It used very long term proxies which do not have fine dating resolution.
For warming of the last century plus we have an extensive thermometer record. No need to look to proxies.

Peter Miller

Any sign of the Marcott FAQ yet?
This post strongly suggests the FAQs are breeding at a rate faster than the authors can respond to.

JohnWho

Another decline that must be hidden?

ChootemLiz

No wonder Mann Ipulate was touting this as more evidence of hockey stick global warming, it was simply made to measure.

RockyRoad

It’s bad enough when somebody cries “Wolf!” and none appears, but instead, a herd of Ice-Age Mastodons proceeds to trample your world. Marcott should be labeled as an empty wolf-crier.

NZ Willy

Marcott’s “perturbations” are the 2nd key to his final output. After the final data were prepared, the last step was to permutate each datum 1000 times, using the age-uncertainty of that datum as bounds. Thus a datum dated 1700AD with an age-uncertainty of 100 years would be permutated into the range 1600AD-1800AD — thus its temperature value is distributed into those years. However, for the final 1940AD bin, Marcott set its age-uncertainty to zero. Therefore all those data were permutated only within the 1940AD bin. The result is that the 1940AD bin did not share its temperature values with its neighbors, and so was not flattened as were all other bins. It is just a mathematical trick which Marcotte used to guarantee that the 1940AD bin would preserve the same value after the perturbations, that it had before.
I have made this point repeatedly on other forums, but it seems to be outside of peoples’ understanding. In my field we work with large data all the time, and this method of perturbing is a substitute for doing excruciating statistical analysis. It smooths nicely, but to assign a governing zero-uncertainty to the end point of the distribution is an absolute abuse of the method. And this is the last time that I am posting on this topic.

Adam

Is it fraud to alter the dates to obtain a different conclusion to that which would be obtained had the dates not been ordered? Can a legal prosecution arise from this? It might be interesting to get a FOIA email dump from these guys to see what was discussed.

eco-geek

I think Marcott and Mann had orthogonal approaches to the hockey stick development.
Marcott adjusted the ‘x’ axis.
Mann adjusted the ‘y’ axis.
Well somebody had to say it!

Two suggestions: plot all proxies with the same Y-scale: Lake 850 only has a range of 2.5 deg C.
Lake Nujella squeezes 6 deg into the same space.
Create another set that zooms into the most recent 2063 years (YBP 2000 to today) and again uses a common Y-axis scale (cm per degree) would also expose the “shenanigans” (no more polite word for it).
Frankly, this “Years Before Present” while scientifically defensible, is an obfuscation to camouflage his shenanigans. If Marcott had shown the hockey stick at a year labeled “1940” instead of YBP 10, he would have been laughed out of the room. Instead, using YBP he attempts to fly under the radar to deliver his payload to the IPCC.

Gene Selkov

While I find the analysis of this paper by Steve McIntyre, John Kerr and others enlightening and instructional, I can’t help but note an amusing similarity between such analyses and a crash investigation of a kamikaze flight.

“All of the Marcott 9 had altered dates associated with the last date”
What are the odds of that being accidental and getting past the authors?

Jimbo

Changing the data like this is either sloppiness, or deliberate fraud. Either way, it’s GIGO.

“Since it is only the end-point data that matters to the conclusion of the paper”
The end point is irrelevant to the conclusions of the paper.
REPLY: OK then, write to the journal, tell them to retract the paper, have Marcott remove the end point spike that has all the idiot journos in a tizzy, then republish it. Simple. – Anthony

John Trigge (in Oz)

Why would anyone use proxies in favour of actual temperature measurements when available?
There are numerous, supposedly rigorously adjusted & homogenised, temperature series from thermometers since the 1950s – why not use them, provided it is explained that this has been done (rather than hide any decline)?

Clay Marley

Thanks for the summary. I’d like to see the same 9 after the re-dating. Do they show the kind of spike that shows in the final report? I wager there’s more to it, i.e. they also did something when setting up the parameters for the Monte-Carlo runs, or they re-dated some of the others to add data to the modern era.

peter_dtm

the link to the book The Inconvenient Skeptic is broken
REPLY: Fixed thanks, wordpress can’t handle a missing http: – Anthony

Louis

“All of the Marcott 9 had altered dates associated with the last date with the Moose Lake data changing the least at 20 years.”
Does anyone have a plausible reason to offer for why the Marcott paper might legitimately do such wholesale altering of dates on the most recent data? Or did they think they could fool the world?

cui bono

Given the controversial nature of the paper, and that others like it (Mann) have had – um – a chequered history, while others (Gergis) have had to be retracted, how come good folks like Steve McIntyre and John Kehr still have to spend time with these forensic dissections rather than it being the job of thesis advisors, peer reviewers or magazine editors?
Kudos to Steve, John et al; megaFAIL for the latter lot.

John Trigge (in Oz) says:
March 26, 2013 at 5:19 pm
Why would anyone use proxies in favour of actual temperature measurements when available?
There are numerous, supposedly rigorously adjusted & homogenised, temperature series from thermometers since the 1950s – why not use them, provided it is explained that this has been done (rather than hide any decline)
=======================================================================
Because we don’t have thermometer readings from 11,000 years ago and you should not splice thermometer records on to proxy records.

jeanparisot

Not one dating error in favor of a non- AGW scenario? I’m sure the SI has a detailed justification an methodology for those date changes, when will Science release that information.

Bill Illis

NZ Willy says:
March 26, 2013 at 4:42 pm
—————
Thanks, I’ve played around with many different smoothing methods to replicate Marcott’s end point and smoothed Holocene lines and couldn’t even come close so gave up.
Your explanation is the only thing that makes sense now, so thanks a bunch.

Jantar

“John Trigge (in Oz) says:
Why would anyone use proxies in favour of actual temperature measurements when available?
There are numerous, supposedly rigorously adjusted & homogenised, temperature series from thermometers since the 1950s – why not use them, provided it is explained that this has been done (rather than hide any decline)?”

The instrumental record should be shown alongside the proxy data. However this may show that the proxy data does not align with the instrumental record and therefore indicate that one or more of these records is incorrect.

Nick Stokes

Stephen Rasey says: March 26, 2013 at 4:45 pm
“Create another set that zooms into the most recent 2063 years (YBP 2000 to today) and again uses a common Y-axis scale (cm per degree)”

Clay Marley says: March 26, 2013 at 5:25 pm
“Thanks for the summary. I’d like to see the same 9 after the re-dating. Do they show the kind of spike that shows in the final report?”

You can see them with re-dating here. You can set up a blink test to show the dating effect. There’s an option for last 2000 years, with common Y-axis.
Louis says: March 26, 2013 at 5:35 pm
‘“All of the Marcott 9 had altered dates associated with the last date with the Moose Lake data changing the least at 20 years.”
Does anyone have a plausible reason to offer for why the Marcott paper might legitimately do such wholesale altering of dates on the most recent data?’

Yes. All dates are inferred, by authors or others, usually from carbon dating. Marcott et al used the most recent programs to do this, systematically. The program they used (eg Marine09) won’t date beyond about 1954, because of atomic tests.

I found this article simply written, not too long and very useful. I was wondering what the fuss was about, and now I can see clearly. Thank you

Thrasher

Are some people really having trouble understanding that a low resolution proxy cannot have thermometer data spliced onto the end of it? For those who keep posting about it, the proxies are average temperature spanning decades but sometimes much longer on the scale of centuries. They will tend to smooth out more extreme temperature fluctuations over shorter periods that instrumental data can detect.

coalsoffire

Nick Stokes in the grand tradition of the black knight, fights on….

Nick
Are you saying none of the underlying works calibrated dates correctly. And since you note the atomic tests, which show up as a hard and definitive marker in the cores – that you do not need any dating calibration for, how do you address Steve McIntyre’s noting finding a bomb spike in one of the cores Marcott substantially re-dated?
Have you been able to reproduce the hockey stick portion of the Marcott graph, and if so can you tell us how they managed to get that large hockey stick when none of their reconstructions contained anything close to that amount of warming?
The authors admitted at least the appearance of issues, and that they would issue a FAQ to address them. It would seem this should be a pretty simple task, yet weeks are passing and nothing has been forthcoming. How long is fair in you opinion to wait for this ‘clarification’ from the authors.
The author himself noted the recent reconstructions and conclusions therefrom were “not robust” and that the majority of the hockey stick came from a single record – Agassiz-Renland. Yet the authors and their institution have promoted the hockey stick as the featured and most important finding, when in reality they admit it is not “robust” and really shouldn’t be relied upon..
Seems we’ve heard that recently – where Lewandowsky pimped and promoted the sensationalized by minor finding of the paper, which like here was supported by the thinnest threads of data, which vanished altogether once the clearly scammed handful of responses were removed.
I wonder if we’ll see a similar ‘Recursive Fury in the Blogoshere over Marcott” follow on paper.;-)

Oh … and to all readers here – make sure you visit Nick’s site and try out his truly EXCELLENT Marcott visualization app.
You can look at a graph of all, and any, of the Marcott proxy’s data in a absolutely cool interactive app.

With the way this is shaping up, can these people be charged with fraud? I don’t understand why we allow such behaviour. This deserves much more than a slap on the wrist. If we don’t start making these “climate scientists” responsible, we give no incentive for the rest of them to “honest-up”.

NZ Willy says:

It smooths nicely, but to assign a governing zero-uncertainty to the end point of the distribution is an absolute abuse of the method. And this is the last time that I am posting on this topic.

I am glad you posted–I am one of those that don’t understand the details, but i get the intent and I am very grateful for folks like you who bother to break it down and give it to us in more understandable terms.

markx

Nick Stokes says: March 26, 2013 at 4:19 pm
For warming of the last century plus we have an extensive thermometer record. No need to look to proxies.
Niiiiick….yes, and the proxies are smoothed, by their nature (each point covering many years/decades) and by statistics ….. and the temperature record is not … (let’s call it spiky!)
So the appearance of an ‘end spike’ is simply an ‘illusion’ … if the rest of the record was equally spiky it would not even be noticeable.

Mervyn Sullivan

Anthony Watts, I have sent via your ‘contact section’ a message on 25 March 2013 and 27 March 2013 … an urgent request of your permission to use some of your site content in a book I have written. I know this is not the place for this but it is important. Please contact me urgently. Thanks. Other readers, please excuse me for this.
[Reply: Suggest posting this comment under Tips & Notes, where Anthony is much more likely to read it. — mod.]

philincalifornia

Nick’s hoping that no one can see the elephant in his room, which is that Marcott et al. is either a pile of shit, or scientific fraud – or both of course, since they’re not mutually exclusive.
A 12-year old could see that from the get-go. Fight on Black Knight. Bite someone’s legs off. Wanker.

A. Scott says: March 26, 2013 at 6:56 pm
“Have you been able to reproduce the hockey stick portion of the Marcott graph, and if so can you tell us how they managed to get that large hockey stick when none of their reconstructions contained anything close to that amount of warming?”

Yes, I described the process here. I get much the same spike with Marcott re-dating, and nothing much without. I think it’s pretty much as Steve Mc (and Tamino) says – the end result is very dependent on which proxies drop out when, and so very affected by re-dating. Personally, I think they just shouldn’t plot their proxy recons in the modern period. There’s too much room for end artefacts, and not enough data. And it’s not needed – the thermometer record is much more reliable.
On the bomb spike etc, I think what Marcott et al have done is to simply use the established programs (eg INTCAL09, MARINE09, based ob Calib 6) systematically. I think that’s advisable. There’s a separate issue of inferring a coretop condition, and a bomb spike might help there, if it can be precisely located.

I’ve noticed Mann seems to be avoiding any talk of Marcott et al on Twitter timeline since the flaws have been pointed out.
I’ve been using his favorite hashtags #extendedhockeystick & HSCW to attract attention to his audience.

Theo Goodwin

Good article. Thanks.

ferdberple

Nick Stokes says:
March 26, 2013 at 4:19 pm
Of course the Marcott study doesn’t tell you about warming past 1950.
=============
You are splitting hairs in a less than honest fashion. Marcott adjusted the end dates of the proxies to 1950. Your spagetti graph of the proxies shows no pattern of modern warming, thus either debunking the modern thermometer record or Marcott. Time to call it like it is.
What the data shows is the temperatures have been falling of thousands of years and no significant modern warming within the resolution of the proxies. It is the adjusted the data post thesis that created the hockey stick.
It is hard to see how this is an error. Rather it appears to be a purposeful distortion with the earmarks of scientific fraud.

Catcracking

Anthony,
I apologize since this is slightly off topic, but it would be “cute” to capture the current humorous Weather Channel guy explaining why this cold March is actually caused by Global Warming.
I laughed out loud when he explained that since the US is only 2% of the area of the earth, the fact that March has been so cold in the US has no bearing on the fact that the globe is warming. Of course this is not the consistent with the story of last year when they repeatedly mentioned that the contiguous US set warm records as being due to global warming without mention of the 2% figure.
Then he went on to explain how the cold March was likely due to global warming being caused by ice melting and blocking highs causing the cold jet stream to dip further South.
Then finally he went on the point out that the warmest 10 out of 12 years (?) have occurred in recent years without mentioning that there has been no warming in the last 15+ years.
The Weather Channel has no integrity what so ever.
I laughed out loud, this would be great to capture, but I don’t have the technology to do so.
Delete if you feel appropriate.

Leo Geiger

Was this post supposed to contribute something that wasn’t said in the other dozen or so already done on the same subject?
Perhaps all can agree that
(a) Marcott said the final decades of the reconstruction are “not robust”, so maybe that means they are not robust.
(b) Thermometers have measured the warming in this century, which is generally what people are referring to when they talk about the recent increase in temperatures.
(c) The authors understand that their conclusions are limited by the lower resolution of the proxies: “Strategies to better resolve the full range of global temperature variability during the Holocene, particularly with regard to decadal to centennial time scales, will require better chronologic constraints…”

ckb42

If the data from the proxies since 1950 do not match the observed temperature record, how useful can these proxies really be??? Isn’t this good reason to throw them out?

Tilo Reber

Nick: “For warming of the last century plus we have an extensive thermometer record. No need to look to proxies.”
Oh, good lord Nick, not you again!
A. The thermometer record disagrees with the proxy record.
B. If it disagrees now, then why would it not also disagree in the past.
C. How can we say that the current climate situation is different from the past climate situation when we don’t have thermometers in the past and when the proxy data shows that the current climate situation is not different from the past.

In the science community, do they really thing we’re going to buy the racial divide and conker strategy? I’ve been seeing a lot of this kind of news lately. The TV MSM doesn’t seem to want to touch it though.
Within Mainstream Environmentalist Groups, Diversity is Lacking
““We essentially have a racially segregated environmental movement,” said Van Jones, co-founder of the nonprofit Rebuild the Dream and a former adviser on green jobs to the Obama administration. “We’re too polite to say that. Instead, we say we have an environmental justice movement and a mainstream movement.””
http://www.washingtonpost.com/national/health-science/within-mainstream-environmentalist-groups-diversity-is-lacking/2013/03/24/c42664dc-9235-11e2-9cfd-36d6c9b5d7ad_story_1.html

Tilo Reber

Nick: “I get much the same spike with Marcott re-dating, and nothing much without.”
Good boy Nick. You are half way there.
“Personally, I think they just shouldn’t plot their proxy recons in the modern period. There’s too much room for end artefacts, and not enough data.”
And it’s particularly embarassing that the “end artefacts” for all the different kinds of proxies seem to all disagree with the highly “adjusted” temperature records. There seem to be no proxies at all that show the same kind of temperature spike as the instrumentation record. As far as “not enough data” goes, I assume that all of the physical processes that were producing proxies 11,000 years ago are still doing so today.
Now, Nick, I guess that you will have to go to work to find rationalizations for why every one of those proxies has an “end artefact” that is always less than the instrument record.

Karl

Catcracking: He said “ice melting?” Where? Certainly not in the arctic. Does this guy at TWC feel his audience is that stupid?

Tilo Reber

Leo: (a) Marcott said the final decades of the reconstruction are “not robust”, so maybe that means they are not robust.
Without a good explanation, maybe it means that they were never robust.
You could claim that they were robust in the past because they agree with the other proxies of the past and that they are not robust today because they disagree with the instrument record today; but the problem is that the other proxies that you are comparing them to in the past also don’t agree with the instrument records today. So obviously you can’t say anything about robustness in the past. Or, alternately, you could claim that all the proxies support each other and are, therefore, robust while the instrument record is not.

ferd berple says:March 26, 2013 at 8:07 pm
“You are splitting hairs in a less than honest fashion. Marcott adjusted the end dates of the proxies to 1950. Your spagetti graph of the proxies shows no pattern of modern warming, thus either debunking the modern thermometer record or Marcott.”

Well, they are certainly not debunking the thermometer record. Nor are they debunking Marcott – the proxies aren’t his.
The fact is, they are low resolution. That generally means they are not perfectly layered. There has been mixing between layers – apart from the issue of determining the date, at the same layer there is a mixture of date material. And as this post says, on author dating only 9 proxies extend beyond 1950 (on Marcott dating, none). On that basis they just couldn’t detect modern warming.
Marcott was doing a study of the Holocene. He chose proxies that were stable over long periods and could be independently calibrated (ie did not require overlap with historic thermometer measurements). He could not have that plus the resolution needed to detect modern warming.

@cbk: If the proxies don’t match the observe temperaure record then, there is a) noise in the proxies, b) noise in the observed anomally temperature record, or c) noise in both.
Stick a 50 year moving average filter on the observed temperature record to make it appear closer to the resolution of the highest resolution proxy. Then debate how much of that smoothed signal is microsite and UHI increase over the past 100 years – from Watts et al 2012 it might be 1.5 deg C / century. Take what’s left and extend by slope only the Marcott trend from the time period where he is down to 20 proxies. It will be less a hockey stick and more of a spoon.

“Of course the Marcott study doesn’t tell you about warming past 1950. It was a study of the Holocene period. It used very long term proxies which do not have fine dating resolution.
For warming of the last century plus we have an extensive thermometer record. No need to look to proxies.”
###################################
yes we have a good thermometer record. So, lets see how good the proxies are?
after all if they cant get the present correct, what good are they for the past

AlexS

“For warming of the last century plus we have an extensive thermometer record.”
****************************************************************************
Hahaha! what a joke. Moving thermometers,More 2/3 world without moving thermometers. in case you didn’t noticed it is called sea. Maybe you can tell me the temperature of last 200 years outside cities of Europe and some in Asia and US… I am not even going into cloud cover, wind changes and temperature variations in altitude. You know there is a third dimension.
All of this to claim 0.x degrees of change. It is a social construct and you Nick Stokes only have that opinion because you value your social life more. Because many professions are increasingly more social and less scientific.