Bishop Hill +++Government abandons temperature records+++
Feb 4, 2014 Climate: Parliament Climate: Statistics Climate: Surface Doug Keenan has just pointed me to a very interesting parliamentary question and answer. As ever the protagonists are our old friends Lord Donoughue and Baroness Verma, and once again the subject is statistical models. I've inserted some clarification suggested by Doug in brackets. This has no bearing upon the answer.
To ask Her Majesty’s Government, further to the Written Answer by Baroness Verma on 25 May 2013 (WA 44–5) which stated a linear trend model with first-order autoregressive noise [is very unlikely to be an appropriate model] in representing the evolution of global annual average surface temperature anomalies, and in the light of the Working Group I Contribution to the Intergovernmental Panel on Climate Change Fifth Assessment Report which states that statistical analyses of climatic time series “have to assume some kind of model, or restricted class of models”, what models they rely upon for statistical analyses of global temperature series; and why they chose those models.[HL4497]
The Parliamentary Under-Secretary of State, Department of Energy and Climate Change (Baroness Verma) (Con): Her Majesty’s Government does not rely upon any specific statistical model for the statistical analysis of global temperature time series.
Global temperatures, along with many other aspects of the climate system, are analysed using physically-based mathematical models, rather than purely statistical models.
Read that again. The government no longer performs statistical analyses of the temperature records. They rely instead of comparisons of the temperature records to GCM outputs. Having admitted the inadequacy of the AR1 model normally used to make claims of significant changes in the temperature record, and having also agreed that an alternative model - suggested by Doug and under which these temperature changes can only be seen as natural variation - is much more likely to be correct, this approach seems to be the only one open to them.
Carbon dioxide is still a greenhouse gas of course, but it is startling to realise that the government has effectively abandoned the temperature records as evidence for global warming. Everything comes down to the climate models.
Reader Comments (52)
Government policy which relies on output from poorly written software models by incompetent academics, is somewhat rather frightening.
Here is a corner and here is your paint, get to work.
What if medical researchers were to abandon the use of statistical analysis of the results of drug trials and use whatever criteria seemed to support their theories about the actions of the drugs? Such behaviour would be regarded as completely unethical and would provoke uproar.
Why should climate science be different?
There's the rub and therein lies the fiction; all political myths [man made global warming being the greatest lie] thirst and crave for justification, of scientific legitimacy and the modellers did them proud - until they got found out.
Then someone said, "the emperor - he's got no clothes on!"
It would seem impossible, to this non-statistician, to analyse time-dependent global temperatures data without using some statistical tool.
Phillip
Yes, in some ways they are ducking the point aren't they? In truth they should be saying that they have analysed the temperatures (using model ABC, for reasons XYZ) and have found that there is no significant change in the temperature series. Just saying "we don't do that stuff" doesn't really cut the mustard.
The climate sensitivity of CO2 is near zero because the atmosphere self-adapts.
For Phillip Bratby, here is a paper that relies on sophisticated statistics but is reasonably accessible to non-statisticians like you and me:
An Israeli group concluded, "We have shown that anthropogenic forcings do not polynomially cointegrate with global temperature and solar irradiance. Therefore, data for 1880–2007 do not support the anthropogenic interpretation of global warming during this period."
Reference: Beenstock, Reingewertz, and Paldor Polynomial cointegration tests of anthropogenic impact on global warming, Earth Syst. Dynam. Discuss., 3, 561–596, 2012.
URL: http://www.earth-syst-dynam-discuss.net/3/561/2012/esdd-3-561-2012.html
I think it's a bit of a leap from "does not rely upon any specific statistical model" to "no longer performs statistical analyses".
I read this more as "will no longer answer questions about statistical models".
..The government no longer performs statistical analyses of the temperature records...
Er... I can't see where they say that. They say they "do not rely upon any specific statistical model for the statistical analysis of global temperature time series...." - which stresses the word 'RELY'.
If I had been the civil servant responding to this question I would have thought thus:
1 - the question points out that there are mathematical difficulties with a particular type of calculation
2 - it goes on to ask which one we 'rely' on
3 - if I give an answer I will have implicitly said I 'rely' on an imperfect source of data
4 - so I had better say "We do lots of things, but do not 'RELY' on any particular one of them...
@ Fred Colbourne | Feb 5, 9:48 AM
The work of Michael Beenstock and colleagues is founded on the assumption that the relevant climatic processes are essentially linear (and Gaussian). Such assumptions are unjustified and almost certainly unjustifiable.
I sent Beenstock a copy of my critique of AR5 statistics (discussed in the BH post “Keenan does AR5”). The critique implicitly points out that Beenstock’s analyses are ill-founded. I did not receive a reply.
More generally, in order to use statistics to draw inferences from data, we have to make some assumptions. In experimental physical sciences, it is usually clear what assumptions to make. In climate science, it is not clear at all. Until someone can find some way to justify a choice of assumptions, it will not be possible to defensibly draw inferences using “purely statistical models”—as the Answer from Under Secretary Verma puts it. For more on this, see my critique.
Doug Keenan
It's nice to see someone talking about the proper way to use any model. That any results must be also presented with the underlying assumptions rather than hiding them. Or for anyone who missed it: this is the correct scientific method.
The use of temperature anomalies is a bit chicken and egg though. The actual measurents most likely have a +\- 1 degree error which is fine for absolute temperatures. But because AGW predicts changes of less than 0.1 degrees per year scientists feel compelled to create an anomaly chart. And do so without showing errors and assumptions which would show that it is a futile exercise, reserved only for theoretical musings and not something that has been driving policy. Or that justifies the $1B spending per day (if this is a good estimate)
It's bad practice to present anomalies as they are and feels more agenda driven.
How did they come up with the theories the models are based on if it wasn't from statistical analysis of the data they had available at the time? But now statistical analysis isn't any good?
They do realise these answers are kept for posterity... and enquiries?
All of which only goes further to prove DP's definition (in "What to do with a hot model" thread) of CAGW: Computer Aided Global Warming.
That should have read inquiries... I blame my cold.
The Global Temperature is now totally immaterial, they are no longer interested in Global Warming as it no longer exists.
It is now all about Climate Change and Extreme Weather.
Wasn't it Myles Allen that insisted in the hearing recently, that the output from the models is data?
There you have it, no need for old fashioned instrument readings.
John Silver:
No, it was Brian Hoskins.
Without a statistical model surely they are unable to predict (project) the future; for anything.
Transcript of that hearing is now published.
But we still can't get the mainstream climate academics to admit that their 'mountain of evidence' is just a molehill of invalid models with unjustified assumptions because turkeys don't vote for xmas. Without that the politicos will continue to defer to their 'expertise'. It's frightening just how easy you get from baseless conjecture to irrefutable fact. All this unnecessary expense based on a perfectly normal 0.6K per century!
@Dodgy Geezer
You are correct in your translation. Unfortunately, there is a reducing number of people able to parse offical 'civil servant speak' especially the particular dialect that is used for answers to Parliamentary Questions. The art is in framing a 'reply' that actually says nothing and does not commit the minister replying to anything. This was one of those answers - immediately misread by the non-cognoscenti.
@Roy
"What if medical researchers were to abandon the use of statistical analysis of the results of (something) and use whatever criteria seemed to support their theories.... "
That is more or less what was done at MAFF (now DEFRA) in the era of Foot And Mouth Disease to produce a culling policy based on theoretical mathematical modelling and unvalidated predictive models.
I refer my honourable colleagues to a critique of that policy.
See http://www.oie.int/doc/ged/D3278.PDF
"The UK experience provides a salutary warning of how models can be abused in the interests of scientific opportunism."
The same goes for Bird Flu, Bovine TB, etc
"Everything comes down to the climate models."
Inherently incapable of being validated and therefore fundamentally useless.
[Unless you accept the Met Office's ludicrous claim that the capability of models to reproduce past climate validates their ability to predict future climate. Reproducing the data used to 'parameterise' and tune a model by no means confirms that it correctly models the physical system.]
Marginally off-topic, but in the context of the 'men go mad in hordes, and come to their senses individually' truism...
Just reading Bill Bryson's excellent 'One Summer America 1927', and have come to the chapter on 'negative eugenics' - which was the astonishing movement which encouraged sterilisation (and worse) of those thought to be 'inferior' (subsequently embraced of course by the Nazis)..
My point is that CAGW could be compared with such movements, in that, like eugenics, it was started by a group of acedemics and scientists (you see, its all happened before) - and got taken up enthusiastically by the politicians (none more so than a certain Mr Hoover) and in fact got written into law (now, where have we come across THAT lately..?)...
Not sure the foregoing is much comfort - except to say that, eventually, common sense MUST prevail...
@ Fred Colbourne.
Might I suggest you use the term "complex" rather than "sophisticated".
From my 1925 Pocket OED
Sophisticated: to spoil the simplicity or purity or naturalness of, corrupt or adulterate or tamper with.
If they have, then kudos is due to Lord Donoughue and Doug Keenan for having brought about a small step in the abandonment of propaganda presented in the guise of science and statistical analysis.
@ Dodgy Geezer (10:13 AM), Ian W (12:50 PM)
See the second paragraph of the Answer, which states that global temperatures, and many other aspects of the climate system, are not analyzed using purely statistical models.
[Snip - Please refrain from namecalling]
@DJK
See the second paragraph of the Answer, which states that global temperatures, and many other aspects of the climate system, are not analyzed using purely statistical models...
Indeed. That still does NOT say that "The government no longer performs statistical analyses of the temperature records...". All it says is that " we don't use stats which are not rooted in reality." To make the point more clearly:
1 - "global temperatures ... are not analyzed using statistical models" means they don't use statistical models.
2 - global temperatures ... are not analyzed using PURELY statistical models means anything you want it to mean. In this case it implies that the models have a real physical basis, but it could be bent in any other way.
For my sins, I have spent far too long working in HM Cabinet Office than has been good for me. But for PQs and ministerial statements you should recall this quote from 'Yes, Minister' (The Right to Know, 1980):
James Hacker: The department prepared this. "No loss of amenity" it says.
Sir Humphrey Appleby: No, Minister, not no loss of amenity; no *significant* loss of amenity.
James Hacker: Well, same thing.
Sir Humphrey Appleby: On the contrary, there is all the difference in the world. Almost anything can be attacked as a loss of amenity, and almost anything can be defended as not a significant loss of amenity, which seems to signify that one should appreciate the significance of significant...
Re: Feb 5, 2014 at 2:56 PM | Dodgy Geezer
"Sir Humphrey Appleby: On the contrary, there is all the difference in the world. Almost anything can be attacked as a loss of amenity, and almost anything can be defended as not a significant loss of amenity, which seems to signify that one should appreciate the significance of significant... "
Ah, so true, and what an excellent lesson in Political 'speak' that is, indeed what an excellent series 'Yes, Minister' proved to be, they just don't make 'em like they used to......
It reminds me too of that phrase so favoured by politicians ie 'No evidence was found......' which usually means they've done their best not to find any!!! Crops up frequently in all the whitewashes, sorry, enquiries!!!
What a spectacular FAIL thread.
Read that again:
Is not the same as
L2R.
An example of a purely statistical model, assumption of AR(1) for noise.
An example of a physics based model: Application of the Navier-Stokes equation. Oooh that's SO creepy.
>.<
All they are just saying they are using physics-based models to analyze the temperature data, instead of purely statistical ones. That's bad?
• This doesn't mean statistical models aren't applied.
• This doesn't mean they are using full blown GCMs either, which they obviously aren't.
• This doesn't mean they are no longer using temperature series.
• It has bloody nothing to do with CO2.
If you want an example similar to what is being described, see the NCDC temperature field reconstruction method.
Doug Keenan, I though I saw a post of yours at WUWT where you agreed with B&R, namely that temperatures have a unit root. Isn't that the basis for saying the statistical model being used is flawed?
Or are you saying you disagree with the conclusion that temperature is driven by a change in CO2?
Feb 5, 2014 at 12:56 PM | Unregistered Commenter Keith Macdonald
What an excellent paper. It's astonishing (or perhaps not) that computer models are being similarly misused to drive policy a decade later. And in an area of science where there is poorer understanding, poorer data and a much greater adverse impact on the economy. This paper should be required reading for anyone involved in climate modelling and related policy setting.
This is what a linear model looks like.
http://tamino.wordpress.com/2014/01/30/global-temperature-the-post-1998-surprise/
I new all those statisticians in govmint could be fired
All because of "the pause": get a model that has some 'pause' in it, and you don't need to show warming for the past 16 years.
Neat trick! Essentially, curve-match whatever model gives you up to today and whatever you want for 2100.
They are torturing models, however. When 2015 rolls around and global/CET temps DROP, what then? Different model?
As said many times, an unfalsifiable proposition.
Very neat trick, indeed.
Bishop Hill
Her Majesty’s Government does not rely upon any SPECIFIC statistical model for the statistical analysis of global temperature time series.
Global temperatures, along with many other aspects of the climate system, are analysed using physically-based mathematical models, rather than PURELY statistical models.
(My emphasis)
How do you get from "the government does not rely ONLY on statistical models" to "The government no longer performs statistical analyses of the temperature records."?
One of your more absurd bits of propaganda bullshit.
From someone who knew what they were talking about, the claimed distinction between "physically-based mathematical models" and "purely statistical models" might be worthy of note.
Given that Baroness Venal (Con) doesn't, it isn't.
Entropic Man. Read Wegman's Report on Mann's and the palaeo-climate community use of statistics.
Read that again. The government no longer performs statistical analyses of the temperature records.
Except that's not what she said. She actually said "Her Majesty’s Government does not rely upon any specific statistical model for the statistical analysis of global temperature time series." Sure, weasel words, but if we are to rightly criticise her pathetic response, at least we should be accurate, no?
So to be more accurate, she's saying they DO "perform statistical analyses," it's just that they don't "rely" on any "specific" model for that statistical analysis, i.e. the ones that give them the wrong answer.
chippy -
As the response was patently written by a scientist in the Met Office and not the Baroness herself, I don't understand your objection.
Bishop -
I don't understand the claim that "there is no significant change in the temperature series." Are you suggesting that there has been no anthropogenic influence on temperature in the last century?
Like others here, I just don't get this. The objection seems to be that:
"Everything comes down to the climate models."
Are you suggesting that AR(1), or Doug Keenan's proposed alternatives are not, in the relevant sense, climate models?
Global temperatures are analyzed using physically-based mathematical models.
I wonder how confidential these models are. Can we see a documentation, source code, etc?
In AR5, and global-warming research generally, evidence for warming falls into three main classes: statistical analyses of instrumental data; statistical analyses of proxy data; simulations of the climate system. The simulations are usually done with GCMs; occasionally they are done with simple energy-balance models, box models, etc. The simulations are always based on the underlying physics.
Evidence in the first two classes is virtually always obtained with purely statistical analysis. Indeed, the most common type of analysis, by far, is to fit a straight line with AR(1) residuals. What the second paragraph of the Answer says is that they are not doing such analyses anymore.
I have some familiarity with the tricks of Sir Humphrey. I think it would be difficult to apply those tricks here.
The Answer is a proper reply, for reasons described in my critique of AR5 statistical analyses, and also described in the paper that Chief Scientist Slingo wrote. In brief, we do not know enough to choose a statistical model—and some plausible models have the increase in temperatures being significant and other plausible models do not.
So the inevitable conclusion is that her majesty's government relies on simulations only to decide if it is warming
But there are plausible simulations that indicate cooling. In fact there is an infinity of simulations, out there.
What does her majesty's government do to cherrypick "the better simulation", in order to declare it is warming?
They are funny, aren't they. At high pay as well.
@DJK
... What the second paragraph of the Answer says is that they are not doing such analyses anymore.
I have some familiarity with the tricks of Sir Humphrey. I think it would be difficult to apply those tricks here....
Perhaps you could enlighten me as to the precise position where this statement is made. My (broad) interpretation goes:
Her Majesty’s Government does not rely upon any specific statistical model for the statistical analysis of global temperature time series. equals " We do not rely on any particular method, but may use several, depending on how we feel"
Global temperatures, along with many other aspects of the climate system, are analysed using physically-based mathematical models, rather than purely statistical models. equals "Global temperature work is done with maths and some real-world data. Generally." Note that this does not say that 'the Government' does any kind of analysis at all - it almost certainly does not. The Government takes advice from duly appointed people. These people, generally, do maths on real-world data. They do lots of other things as well. I can't see that the second sentence says anything whatsoever about what particular analyses are performed and what the Government relies on.
They are using the 'question of loyalty' technique. Oblig. Sir Hump Quote:
......
Betty Oldham: But the Minister advises us to ask you.
Sir Humphrey Appleby: And I am advising you to ask the Minister.
Allen Hughes: When does this end?
Sir Humphrey Appleby: As soon as you like..
......
Sir Humphrey Appleby: It was the government's policy to test all the proposals for fuel saving.
Betty Oldham: At this fantastic waste of taxpayer's money? You agree this money was wasted?
Sir Humphrey Appleby: It is not for me to comment on government policy. You must ask the Minister.
Betty Oldham: Look, Sir Humphrey, whatever we ask the Minister, he says is an administrative question for you, and whatever we ask you, you say is a policy question for the Minister. How do you suggest we find out what is going on?
Sir Humphrey Appleby: Yes, yes, yes, I do see that there is a real dilemma here. In that, while it has been government policy to regard policy as a responsibility of Ministers and administration as a responsibility of Officials, the questions of administrative policy can cause confusion between the policy of administration and the administration of policy, especially when responsibility for the administration of the policy of administration conflicts, or overlaps with, responsibility for the policy of the administration of policy.
Betty Oldham: Well, that's a load of meaningless drivel. Isn't it?
Sir Humphrey Appleby: It is not for me to comment on government policy. You must ask the Minister.
In this case the Government are going to say that the advice given is a scientific responsibility, while the Met office are going to say that the advice accepted is a government responsibility....
In this case the Government are going to say that the advice given is a scientific responsibility, while the Met office are going to say that the advice accepted is a government responsibility....
Feb 5, 2014 at 11:58 PM | Unregistered CommenterDodgy Geezer
The AR5 committee transcript included two clear reference to this, both by John Robertson in discussion with Professor Lindzen.
"scientists in general are concerned about climate change, and they are right to be concerned. If we take that as a premise of where we are going then we take the advice, gentlemen and ladies, that you give us, and at the end of the day politicians have to weigh up the pros and cons of what they are doing,"
"Your problem is the science. Our problem is the policy. I do not get involved in the science and perhaps probably you should not get involved in the politics."
Robertson makes a clear distinction. Scientific advice, which, in the Uk is the job of the Met Office, and policy, which is the job of the politicians.
forget warming folks -
http://www.mikehulme.org/2014/02/global-warming-is-dead-long-live-global-heating/
they will confuse us into submission.
Roy
"What if medical researchers were to abandon the use of statistical analysis of the results of drug trials and use whatever criteria seemed to support their theories about the actions of the drugs?"
That is exactly what they do and worse.