Friday, March 15, 2019

20th Anniversary of the Hockey Stick

Today is the 20th anniversary of the hockey stick paper by Michael Mann, Ray Bradley and Malcolm Hughes that reached back to 1,000 BP.

Jerry Mahlman, a climatologist at GFDL of NOAA, came up with the term "hockey stick graph." Their first graph was for the last 600 years of the northern hemisphere; today's anniversary is for the paper on the last millennium, then Mann and Phil Jones later reconstructed temperatures back to 2000 years BP.

Their work has been attacked like few others in science, but it has held up just fine. No one expected Mann to be the pugilist he turned out to be, who tirelessly fought off everyone at every turn -- auditors, congressman, lawyers and trolls. Sure, it wasn't the last word in temperature reconstruction, but first papers rarely -- for example, the Bohr Model, pre-Einstein papers on Lorentz transformations, the parton model of hadrons, and many many others in science. (I just happen to know the history of physics better than that of any other field.)

Many papers did their part in backing up the HS result by bringing in new data and improving the methodology. Here's a comparison of MBH98 to the PAGES 2k results of a couple years ago, from Stefan Rahmstorf's Facebook page:


One of the particular advancements of MBH was the inclusion of an uncertainty band around the temperatures.

I don't feel like relitigating any of the particulars of a case against the hockey stick, unless someone wants to bring something up in the comments. As I've written before, the hockey stick is required by basic physics, in particular radiative forcing's logarithmic dependence on CO2, which goes back to Arrhenius in 1896. Given that, it's a short argument to the hockey stick, one that can be condensed into a single tweet:
The hockey stick is required by basic physics:
1. temperature change is proportional to forcing change.
2. CO2 forcing change is proportional to log(CO2).
3. CO2 has been increasing exponentially.
=> hockey stick.
(209 characters.) Before the industrial era, the atmospheric CO2 concentration changed little during the Holocene, which implies the flat shaft of the HS. CO2 is increasing exponentially in the industrial era, which implies a linear increase in temperature, which is the blade of the hockey stick.


It'd be far more surprising if there wasn't a hockey stick in the data.

It's been interesting to follow all the twists and turns over the years of the effort to defeat this graph. It would have been interesting to have had social media during some of the big debates in science, such as the debates over the wave or corpuscular view of light, or the probabilistic nature of quantum mechanics, or the establishment of evolution by natural selection. (Here's a good book on some notable feuds in science.) I'd glad all that mischief is over. Science always wins. 

54 comments:

David in Cal said...

David - I do not know what you mean by the hockey stick "has held up just fine." I suppose it depends on what you mean by "hockey stick". Originally, it meant a particular model that predicted a very sharp increase in temperature. This model was cast into doubt when it was shown that a wide variety of inputs would automatically produce a hockey stick output. IMHO the model was far too complex for the relatively crude data available. I think that particular model has not held up. Perhaps you acknowledge this fact when you write, "Sure, it wasn't the last word in temperature reconstruction."

You're on more solid ground if by hockey stick holding up you mean that the temperature increase during the last 50 years has been relatively rapid.

Cheers

David Appell said...

David, *again*, the hockey stick didn't and doesn't predict anything. It reconstructs temperatures of the *past*.

We've been over this before.

Layzej said...

The various reconstructions all seem to agree. Is there a reconstruction that gives a contrary result?

Marco said...

"This model was cast into doubt when it was shown that a wide variety of inputs would automatically produce a hockey stick output."

And *that* criticism was cast into doubt when people rapidly pointed out that even with extreme choices of the type of noise, the "hockey stick" shape was still considerably smaller than the one observed.

Add the fact that using other methodology that removed this *potential* issue with the methodology used...gave the same result.

David in Cal said...

David -- I am not clear on what it means to say the hockey stick is or is not a "projection". I and most of the public understood the hockey stick to mean that temperature would rise very rapidly, absent some big changes in emissions. This threat of dangerously rapid warming made the hockey stick so frightening.

A forecast of something that's going to happen would seem be a "projection" in the ordinary use of the word. What am I missing?

Cheers


David Appell said...

David:

The hockey stick says nothing whatsoever about future temperatures. It does not project or predict anything.

It is just a determination of past temperatures. From proxies, it determines what the past average temperature was, for a certain region. (Northern hemisphere, usually.) And then only up to about 1960, for technical reasons called the divergence problems, when some proxies stop being representative of temperature.

And the hockey stick doesn't attribute any temperature change it uncovers to CO2, or the Sun, or anything. It simply says, here is a history of temperature over the last X years.

Anonymous said...

David - thanks for the clarification. Nevertheless, I think the hockey stick was understood (or misunderstood) as telling us something about the future. E.g., the hockey stick was used on the cover of the IPCC report as a way to dramatize the threat of global warming. They didn't say the hockey stick was of academic interest only because it tells us nothing about the future of global warming. In other words, despite its explicit meaning, the implicit message of the hockey stick was to emphasize the threat of global warming.

Cheers

David Appell said...

David/Anon: If you misunderstood the hockey stick, I don't think you should blame others.

The HS wasn't on the cover of the IPCC TAR (3AR) WG1, but it was on the second page of the Summary for Policymakers:

https://www.ipcc.ch/site/assets/uploads/2018/03/WGI_TAR_full_report.pdf

Here's what it says about the figure:

"(b) Additionally, the year by year (blue
curve) and 50 year average (black
curve) variations of the average surface
temperature of the Northern Hemisphere
for the past 1000 years have been
reconstructed from “proxy” data
calibrated against thermometer data (see
list of the main proxy data in the
diagram). The 95% confidence range in
the annual data is represented by the
grey region. These uncertainties increase
in more distant times and are always
much larger than in the instrumental
record due to the use of relatively sparse
proxy data. Nevertheless the rate and
duration of warming of the 20th century
has been much greater than in any of
the previous nine centuries. Similarly, it
is likely7 that the 1990s have been the
warmest decade and 1998 the warmest
year of the millennium."

It doesn't say anything about temperatures beyond the 1990s.

The HS is of more than "academic interest." It shows a lot of warming up to the 1990s. BUT IT DOESN'T ATTRIBUTE THE WARMING TO ANYTHING. Sure, a natural question after seeing the HS graph is, what's changed? But it takes other science to address that.

Layzej said...

Nevertheless, I think the hockey stick was understood (or misunderstood) as telling us something about the future.

I'd be curious to see an example where someone else had this misconception.

I've looked for examples on the web and failed to find any. During my search I did find the answer to my previous question about whether there is a reconstruction that gives a contrary result: (short answer: nope)

"Quantitative reconstructions have consistently shown earlier temperatures below the temperature levels reached in the late 20th century. This pattern as seen in Mann, Bradley & Hughes 1999 was dubbed the hockey stick graph, and as of 2010 this broad conclusion was supported by more than two dozen reconstructions, using various statistical methods and combinations of proxy records, with variations in how flat the pre-20th century "shaft" appears."

David Appell said...

Here's an article I wrote about an alternative calculation that gives the hockey stick:

"Novel Analysis Confirms Climate "Hockey Stick" Graph," Scientific American, November 2009, pp 21-22.
https://www.scientificamerican.com/article/still-hotter-than-ever/

Unknown said...

Mann's hockey stick ended in 1950, right? But the IPCC says most CO2 warming occurred after that year. That means Mann's hockey stick was caused by something else. What is it?

David Appell said...

I believe their data (for MBH98 at least) went up to 1995. They used the overlap of proxy temperatures and thermometer temperatures from 1902-1980 to correlate past (proxy) temperatures to modern (recorded) temperatures.

Here's the paper:

Mann, M.E., Bradley, R.S., Hughes, M.K.Global-Scale Temperature Patterns and Climate Forcing Over the Past Six Centuries, Nature, 392, 779-787, 1998.
http://www.meteo.psu.edu/holocene/public_html/Mann/articles/articles/mbh98.pdf

David Appell said...

Re: 1950

Here's a plot of total radiative forcing (RF) from all GHGs from 1750 onward:

https://www.esrl.noaa.gov/gmd/aggi/aggi.fig5.png

Their AGGI is the ratio of a year's RF relative to 1990:

AGGI = RF(year)/RF(1990)

By 1950, AGGI was about 0.4.
In 2017, AGGI was 1.41.

So 28% of today's RF was there by 1950.

David Appell said...

And warming is proportional to forcing. So about 28% of warming (assuming it was all due to GHGs) was there by 1950, or about 0.3 C.

Looking at the HadCRUT4.6 data, I get a warming by 1950 of 0.25 C, which is 29% of their total of (now) 0.87 C.

data:
https://www.metoffice.gov.uk/hadobs/hadcrut4/data/current/time_series/HadCRUT.4.6.0.0.monthly_ns_avg.txt

Unknown said...

If you look at his temperature reconstruction, at least half of that hockey stick occurred before 1950. So his paper showed unprecedented rapid warming before the impact of CO2. Kinda shoots down the CO2 argument, doesn't it?

David Appell said...

You have to add about 0.4 C onto the end of the MBH98 graph for the warming from 1998-present.

PS: Why are you using "Unknown" again for your name?

jonathan sawyer said...

I was using the option to use my google account. It apparently does not allow one to use a name.

But you did not answer my question. Mann's hockey stick paper shows centuries of relatively constant temperatures followed by a rapid rise. But that rise predated the impact of CO2. So what caused it? If it isn't human caused then what was the mechanism which only happens once in many centuries?

Also .4 degree C since 1998? What temperature record shows that?

jonathan sawyer said...


Sorry, I missed your follow on replies. The AGGI graph you cited shows 60% of the value at 1950 was already present by 1910 when the hockey stick took off. Yet we see no warming. In fact if we subtract that forcing from Mann's curve we would have unprecedented cooling in the century before 1910.

Marco said...

"Yet we see no warming."

Welcome to the real world where there are more forcings than just greenhouse gas forcings to consider. In particular several very large volcanic eruptions in the late 19th century + early 20th century put a lot of small particulates in the stratosphere, causing 'cooling' (i.e., masking GHG warming), followed by a prolonged quiet volcanic phase after 1910 ('faster' warming) and yet again some big volcanic eruptions after the 1960s). Add the fact that MBH98/99 is a reconstruction of the extra-tropical northern hemisphere, where there the expected cooling trend due to the phases of the Milankovitch cycles has more impact.

Layzej said...
This comment has been removed by the author.
David in Cal said...

I will just add that from a statistical POV Mann's hockey stick model did not deserve to be considered reliable. You can't just throw a bunch of complex statistical formulas together, without extensive validation, and claim that the result is valid.

What made the hockey appear significant was that it claimed to show that recent temperature rises were very large compared with the past. In a way, the handle of the hockey stick was more significant than the blade. We more or less know how large the recent rise in temperature was, because we measured it. We do not know just how stable past thousands of years' temperatures were.

A common sense hypothetical example should make my point. The hockey stick claims to show that temperature rose during the last few decades at a rate unparalleled for the last thousand years. Now, suppose that temperatures did rise by 1°C during some 70-year period long ago, for some unknown reason. How could a statistical model be expected to find that anomalous period? We might conceivably be able to find such a period using proxies, but a purely statistical model couldn't possibly do that.

cheers

Layzej said...

What is the "purely statistical model" that you are referring to?

David in Cal said...

Layzej - It was my understanding that the original Mann hockey stick was essentially a statistical model. Was I wrong?

Cheers

Marco said...

DiC, in essence any approach where you combine data from different sources will involve statistical methods. After all, you need to make choices, and preferentially as objective as possible, on how to combine those. That does not make the final result a "purely statistical model", nor would it automatically mean that a 1 degree increase over a 70-year period would be missed.

David in Cal said...

Marco, I appreciate your comment. However, the burden of proof should be on the model. We don't have to prove that the model might fail to catch a 1 degree increase. It's the job of the model maker to prove that the model would catch such an increase. I do not believe Mann met this standard.

cheers

Marco said...

DiC, you can believe whatever you want. However, already in 1995 the same mathematical methodology was capable of catching changes much subtler than the 1 degree over 70 years you are looking for:
https://www.nature.com/articles/378266a0

Layzej said...

One other point is that you probably shouldn't take any one study as 'proof', but once you have a few studies using different data and methods - all showing a coherent picture - then you can have some confidence in that finding. In this case there are more than a few:

Jones et al. (1998) , calibrated by Jones, Osborn & Briffa 2001 "The Evolution of Climate Over the Last Millennium".
Mann, Bradley & Hughes (1999)
Briffa (2000) , calibrated by Briffa, Osborn & Schweingruber 2004 "Large-scale temperature inferences from tree rings: a review".
Crowley & Lowery 2000 "How Warm Was the Medieval Warm Period?"
Briffa et al. 2001 "Low-frequency temperature variations from a northern tree ring density network".
Esper, Cook & Schweingruber 2002 "Low-Frequency Signals in Long Tree-Ring Chronologies for Reconstructing Past Temperature Variability",
recalibrated by Cook, Esper & D'Arrigo 2004 "Extra-tropical Northern Hemisphere land temperature variability over the past 1000 years".
Mann & Jones 2003 "Global surface temperatures over the past two millennia."
Pollack & Smerdon 2004 "Borehole climate reconstructions: Spatial structure and hemispheric averages".
Oerlemans 2005 "Extracting a climate signal from 169 glacier records".
Rutherford et al. 2005 "Proxy-based Northern Hemisphere surface temperature reconstructions: Sensitivity to method, predictor network, target season, and target domain".
Moberg et al. 2005 "Highly variable Northern Hemisphere temperatures reconstructed from low- and high-resolution proxy data".
D'Arrigo, Wilson & Jacoby 2006 "On the long-term context for late twentieth century warming".
Osborn & Briffa 2006 "The spatial extent of 20th-century warmth in the context of the past 1200 years".
Hegerl et al. 2006 "Climate sensitivity constrained by temperature reconstructions over the past seven centuries".
Frank, Esper & Cook (2007) "Adjustment for proxy number and coherence in a large-scale temperature reconstruction".
Hegerl et al. (2007) "Detection of human influence on a new, validated 1500–year temperature reconstruction".
Juckes et al. 2007 "Millennial temperature reconstruction intercomparison and evaluation".
Loehle & McCulloch (2008) "Correction to: A 2000-year global temperature reconstruction based on non-tree ring proxies".

Layzej said...

Mann et al. 2008 "Proxy-based reconstructions of hemispheric and global surface temperature variations over the past two millennia".
Mann et al. 2009 "Global Signatures and Dynamical Origins of the Little Ice Age and Medieval Climate Anomaly".
Ljungqvist 2010 "A New Reconstruction of Temperature Variability in the Extra-Tropical Northern Hemisphere During the Last Two Millennia".
Christiansen & Ljungqvist 2012 "The extra-tropical Northern Hemisphere temperature in the last two millennia: Reconstructions of low-frequency variability".
Leclercq & Oerlemans (2012) "Global and Hemispheric temperature reconstruction from glacier length fluctuations".
Shi et al. 2013 "Northern Hemisphere temperature reconstruction during the last millennium using multiple annual proxies".
Further reconstructions
Smith et al. 2006 "Reconstructing hemispheric-scale climates from multiple stalagmite records".
Lee, Zwiers & Tsao 2008 "Evaluation of proxy-based millennial reconstruction methods".
Huang, Pollack & Shen 2008 "A late Quaternary climate reconstruction based on borehole heat flux data, borehole temperature data, and the instrumental record"
Kaufman et al. 2009 "Recent warming reverses long-term arctic cooling".
Tingley & Huybers 2010a "A Bayesian Algorithm for Reconstructing Climate Anomalies in Space and Time".
Christiansen & Ljungqvist 2011 "Reconstruction of the Extratropical NH Mean Temperature over the Last Millennium with a Method that Preserves Low-Frequency Variability".
Ljungqvist et al. 2012 "Northern Hemisphere temperature patterns in the last 12 centuries".
Marcott et al. 2013 "A Reconstruction of Regional and Global Temperature for the Past 11,300 Years"
PAGES 2k Consortium 2013 (78 researchers, corresponding author Darrell S. Kaufman) "Continental-scale temperature variability during the past two millennia"

As noted above, there are variations in how flat the pre-20th century "shaft" appears, but these consistently shown earlier temperatures below the temperature levels reached in the late 20th century. With uncertainty, there is some chance that temperatures were warmer 8000 years ago than 2010, but as temps keep rising that uncertainty is becoming less and less relevant.

JoeT said...

"It was my understanding that the original Mann hockey stick was essentially a statistical model. Was I wrong?"

Yes

It amazes me how people can have such firm beliefs about a paper that they obviously haven't read. David even gave you a link to the paper. What Mann is showing is the DATA -- but David already explained that to you. If Mann was simply trying to show the hockey stick temperature, then he could have made the paper just about that.

The reason why you're confused is because Mann wasn't trying to show some hockey stick representation of the data; he was trying to understand the NATURAL variation in the past climate. To do this he applied a commonly used method called Principal Component Analysis (PCA) to determine the main spatio-temporal variations in the past climate. From this he could unravel the global mean temperature which showed the hockey stick, the ENSO oscillation and the North Atlantic Oscillation. These are the first, second and third eigenvectors from the PCA. It's because Mann used PCA that he got viciously attacked by people like McIntyre and McKitrick who, in their own work, simply disposed of the eigenvector that corresponded to the warming signal.

To sum up: the hockey stick isn't a statistical model --- it's in the data. The reason this gets confused is because Mann wanted to see if he could understand the natural oscillations in the climate and used PCA.

Ned W said...

"How could a statistical model be expected to find that anomalous period? We might conceivably be able to find such a period using proxies, but a purely statistical model couldn't possibly do that."

This is hilarious. I would file it under "not even wrong" because it's actually nonsense.

The word "statistical" inherently implies quantitative analysis of data. A "statistical model" is built on data. There is no such thing as a "purely statistical model" that exists without some underlying data that are being described by the model.

In Mann's case, the data that were the basis for his PCA model [cf Joe T] were from proxy measurements. There have been many such studies over the years, using different mixes of proxies [cf Layzej].

So, David in California, when you say "We might conceivably be able to find such a period using proxies, but a purely statistical model couldn't possibly do that" it is literally nonsensical. The data from proxies are the foundation for the model (or in this case, the PCA).

The best analogy I can think of would be this:

David in California sez: "We might conceivably be able to express complex thoughts using words, but pure sentences couldn't possibly do that."

The sentences are made up of words and the relationships among them.

It boggles my mind that someone with such a poor understanding of the things he's talking about would try to lecture much more informed people about this stuff. This thread is an excellent example of Dunning-Krueger in action.

David in Cal said...

Ned -- Thanks for your question. My wife and I are both statistics professional, with many dozens of published papers and some awards for research. I will explain my point in more detail, from the POV of a statistics professional.

Many -- far too many -- users of statistics think one can just plug some numbers into a common statistical formula. Unfortunately, various programs, such as Stat-Pak, make this easy for a layperson to do. Actually, each statistical formula has assumptions which must be shown to be satisfied, in advance, in order for the results to be valid. A related common error is to assume that various known distributions and formulas, like the Normal Distribution or the Pareto Distribution, have some magical significance. In other words, some people think the real world tends to behave like some distribution or method that we're familiar with.

Suppose hypothetically that the global temperature had increased by 1 degree during some long-ago 70-year long period. We might conceivably be able to detect the sharp increase if there were some reliable proxy that showed the earth's temperature at the beginning and the end of that 70-year long period. But, without such data, no statistical formula could answer the question, because there's no way to show in advance that any selected formula applies to the planet's temperature during that period.

Cheers

David in Cal said...

Ned - now that I'm up on my soapbox, I will add that there are two common mistakes in statistical analysis

1. Select the method after looking at the data. Even professionals sometimes make this mistake.

2. Give more credence to a result if it's what one is happy with that result.

IMHO the climate change community made mistake #2 with their reaction to the hockey stick. They gave it wide publicity because they liked the result.

Cheers

David Appell said...

David in Cal wrote:
IMHO the climate change community made mistake #2 with their reaction to the hockey stick. They gave it wide publicity because they liked the result.

What can possibly be wrong with giving publicity to a result which supported the leading hypothesis that the planet is warming??

Marco said...

DiC - MBH98/99 weren't that different from earlier work - their main addition was the use of other proxies than tree rings, their uncertainty analysis (largely missing from earlier work), and the almost annual resolution.

Please, do read up a little bit more on MBH98/99, instead of coming with criticism that just confirms you don't really know what it did and didn't do, and could and couldn't do.


Layzej said...

IMHO the climate change community made mistake #2 with their reaction to the hockey stick. They gave it wide publicity because they liked the result.

As noted in the OP though, those results have held up just fine. So maybe not a mistake after all...

Ned W said...

"We might conceivably be able to detect the sharp increase if there were some reliable proxy that showed the earth's temperature at the beginning and the end of that 70-year long period. But, without such data, no statistical formula could answer the question"

Without such data, there would be no statistical model, because the statistical model is based on data. If the model is specified based on something other than data -- first principles, wishful thinking, whatever -- it's not "statistical".

And all of the hemispheric or global paleoclimate reconstructions from proxies -- like the Mann paper -- are based on data. That's what the proxies are.

How can you continue to post in this thread without bothering to even look at the paper you're pontificating about?

David in Cal said...

David - yes, that model correctly supported the idea that the planet is warming.* OTOH it incorrectly showed that the global temperature had been quite stable for 1000 years. Layzej, that latter result cannot be said to have held up. In fact, we just don't know how much short-term variation there was in global temperature, going back a long time ago.

Marco - thanks for your comment. Yes, the almost annual resolution is what I am bellyaching about. I maintain that it's simply impossible for a statistical model to provide that kind of resolution. At best, a model could say that past temperatures were stable assuming that this model provides a valid depiction of what happened. But, there is no conceivable way to prove that a given model is indeed a valid depiction of what happened. At best, the result might be suggestive.

I will make you an offer, Marco. If you provide a link to the model you're referring to, I will try to find the spot where they used a model or method by simply assuming that it would apply accurately to the data they had. In particular, Joe T. informed us above that Mann applied a commonly used method called Principal Component Analysis (PCA). (Steve McIntyre asserted that Mann monkeyed with PCA improperly. But, set that aside.) I strongly doubt that Mann's paper included proof that PCA can be expected to provide almost annual resolution that accurately describes global temperatures. "Commonly used" is not an adequate standard.

*David - One could even quibble about what Mann's hockey stick paper showed about warming. It showed that warming had occurred in recent years, but, as you yourself pointed out, it didn't show that this warming would continue in the future. In that sense, the continued warming was not a validation of Mann's result. However, I don't mean this to be a serious point. It's just a quibble.

Ned W said...

"IMHO the climate change community made mistake #2 with their reaction to the hockey stick. "

You began this thread calling the hockey stick a "prediction" and a "projection" and a "forecast". Why should anyone pay attention to your opinion, when it's based on a near-total ignorance of the topic at hand?

David in Cal said...

Ned - I make the same offer to you as I did to Marco. If you provide a link to the model you're referring to, I will try to find the spot where they used a model or method by simply assuming that it would apply accurately to the data they had.

Cheers

Layzej said...

"In fact, we just don't know how much short-term variation there was in global temperature"

Do you have any evidence to support that fact?

Layzej said...

Here's the method and data for MBH98: http://www.meteo.psu.edu/holocene/public_html/shared/research/old/mbh98.html

Marco said...

DiC, I'm not going to shoot at goal again - you keep on moving the goalposts and claim no goal anyway. First the model wasn't proven, then the data wasn't good enough, now it's the model again, and we've gone from doubts about the models being able to show increases of a degree over 70 years (ignoring the blade in the hockeystick - clearly the model thus can do this), to the data not having annual resolution, which is not relevant for the original supposed issue with the 1 degree over 70 years.

I also provided you with a link to the 1995 paper of Mann (the pre-MBH paper, showing the existence of the AMO using much subtler temperature changes and over shorter periods than your requested 1 degree over 70 years).

Your last comment also strongly suggests you are telling us "I won't believe the method can do this, so don't bother trying to convince me".

David in Cal said...

Layjeh -- thanks for the link. I will read the Mann, Bradley, Hughes article and get back. I notice that the bold-faced summary makes no claim about temperature stability. I wonder whether the picture of a flat handle on the hockey stick is artistic license. That summary also uses the weakish word "suggest" for its conclusion. Anyhow, I will comment further when I've read the entire paper.

Regarding science not knowing the past short-term temperature variation: the only way for me to be sure would be to read every possible related paper. I should have said that I am not aware of any reputable paper with this conclusion. However, I may have missed or forgotten some such papers.

Marco - you are half correct. I do not believe any method can do this. However, I would love to be convinced that I'm wrong. By all means, show the method that can do this and explain why it works. I would be happy something new.

Cheers

David Appell said...

David in Cal wrote:
I notice that the bold-faced summary makes no claim about temperature stability. I wonder whether the picture of a flat handle on the hockey stick is artistic license

See figure 5b in their paper:

http://www.meteo.psu.edu/holocene/public_html/shared/articles/mbh98.pdf

The years 1400-1900 show a temperature variance (with the uncertainties) of less than 0.4 C. That's not perfectly flat, but it's certainly relatively flat.

David Appell said...

David in Cal said...
David - yes, that model correctly supported the idea that the planet is warming.* OTOH it incorrectly showed that the global temperature had been quite stable for 1000 years.

What data shows that to be incorrect?

Layzej said...

I notice that the bold-faced summary makes no claim about temperature stability. I wonder whether the picture of a flat handle on the hockey stick is artistic license.

As noted earlier, "Quantitative reconstructions have consistently shown earlier temperatures below the temperature levels reached in the late 20th century. This pattern as seen in Mann, Bradley & Hughes 1999 was dubbed the hockey stick graph, and as of 2010 this broad conclusion was supported by more than two dozen reconstructions, using various statistical methods and combinations of proxy records, with variations in how flat the pre-20th century "shaft" appears."

Regarding science not knowing the past short-term temperature variation: the only way for me to be sure would be to read every possible related paper. I should have said that I am not aware of any reputable paper with this conclusion.

I've provided about 30. Taken together, these results give insights on the limits of short term variability through history.

David Appell said...

David in Cal said...
It showed that warming had occurred in recent years, but, as you yourself pointed out, it didn't show that this warming would continue in the future. In that sense, the continued warming was not a validation of Mann's result. However, I don't mean this to be a serious point. It's just a quibble.

Whatever it is, it's not relevant -- no one claims continued warming was (or was not) a validation of Mann et al's result. Because their result said nothing about the future.

David Appell said...

David in Cal said...
Regarding science not knowing the past short-term temperature variation: the only way for me to be sure would be to read every possible related paper.

MBH's reconstruction was for annual temperatures.

David in Cal said...

MBH says, "We do, however, assume that the fundamental spatial patterns of variation which the climate has shown during the past century are similar to those by which it has varied during
past recent centuries." MBH straightforwardly says they simply assumed that there was relatively little temperature variation in past recent centuries.

Cheers

David in Cal said...

I skimmed MBH and found what I expect. That is, what I expected to be missing is indeed missing. There's no clear explanation of how we can be sure that the method and assumptions reliably model many centuries of temperature. They do say, "Our approach to climate pattern reconstruction relates closely to statistical approaches which have recently been applied to the problem of filling-in sparse early instrumental climate fields, based on calibration of the sparse sub-networks against the more widespread patterns of variability that can be resolved in shorter data sets." That's fine and honest, but doesn't prove that we should trust this particular model. They compared their method to relatively recent periods and say it worked. But, you can't get from there to hundreds of years ago without simply assuming that the long-ago past resembles the recent past.

A point I've been trying to make is that there are infinitely many possible models. Most of them don't even involve known functions or approaches. No doubt an enormous variety of results could be obtained from all these different models. Why is this model right and the others wrong? Without explaining why this model is correct, the results are merely suggestive. MBH agree with me, saying that their correlations, reconstructions, etc. "suggest" certain results.

I stand by my prior comment that the climate change community gave more prominence to a merely "suggestive" model because they liked the result. (And, I am not even addressing criticisms of how they used their principal component analysis and how they evaluated their proxies.)

Cheers

Layzej said...
This comment has been removed by the author.
Layzej said...

MBH straightforwardly says they simply assumed that there was relatively little temperature variation in past recent centuries.

No. They were referring to spatial patterns of variation. BTW, if that assumption is wrong, isn't it likely their results exaggerate GMT variability in the distant past?

Why is this model right and the others wrong

Where they all agree, maybe they are all right?

David in Cal said...

Layjeh, thanks for the correction about which type of variation they were talking about. You are right.

What would it mean if their assumption about past spacial temperature variation is wrong? I am not sure, but I think they used this assumption in order to deduce planetary temperature from proxies in specific areas. Without this assumption, I think they would be unable to start, because they'd have no estimates of worldwide temperature. But, maybe I am wrong. If so, please feel free to correct me.

Cheers

David Appell said...

David in Cal said...
There's no clear explanation of how we can be sure that the method and assumptions reliably model many centuries of temperature.

That's impossible, in principle, because there are no direct temperature measurements from that time. If there were, a reconstruction via proxies wouldn't be needed.

Why is this model right and the others wrong?

Because other, mathematically independent reconstruction models give the same results.

Because climate models give the same results.

Because basic physics reasoning strongly hints at this result.

Without explaining why this model is correct, the results are merely suggestive.

David, this is silly -- if there were a way of showing that the model is correct, the model wouldn't be needed. A reconstruction is required *precisely* because there are no direct observations of that time period.