Bishop Hill reports that Doug Keenan’s article about statistical significance in the temperature records seems to have had a response from the Met Office.
WUWT readers may recall our story here: Uh oh, the Met Office has set the cat amongst the pigeons:
===========================================
The Parliamentary Question that started this was put by Lord Donoughue on 8 November 2012. The Question is as follows.
To ask Her Majesty’s Government … whether they consider a rise in global temperature of 0.8 degrees Celsius since 1880 to be significant. [HL3050]
The Answer claimed that “the temperature rise since about 1880 is statistically significant”. This means that the temperature rise could not be reasonably attributed to natural random variation — i.e. global warming is real.
…
The issue here is the claim that “the temperature rise since about 1880 is statistically significant”, which was made by the Met Office in response to the original Question (HL3050). The basis for that claim has now been effectively acknowledged to be untenable. Possibly there is some other basis for the claim, but that seems extremely implausible: the claim does not seem to have any valid basis.
=============================================
The Met office website text is here and there is a blog post here.
I think this would depend on the parameters of the model, the size of the step in the random walk is dependent on the forcing for a given year being greater than the previous. Because of the cubic factor in emission, it would seem that the hotter it gets, the lower the probability of a positive step. This physical constraint should keep the result bounded.
Even accepting the deficiencies, the fact remains that the MET has clearly failed to justify it’s choice of model. The MET in their haste even suggests there are models which have a better fit probability than the one offered – the blog post fails to say why they are not using THAT model either. Foot, meet mouth
Finally the MET say their view is consistent with the “Physics and chemistry” when the positive feedback loop gain they are assuming is of the order of 0.95 and has NO time component (is scalar) which is very non-physical – fairy tale lala land stuff to us Engineers. Feedbacks have lag, lag causes instability, climate science postulates huge feedbacks without instability – Nutty!
FOI Question
Please supply all documentation regarding the relative fit of statistical models to global warming series data.
Margaret Hardman says:
“Null hypothesis: [statistics] A statement that essentially outlines an expected outcome when there is no pattern, no relationship, and/or no systematic cause or process at work; any observed differences are the result of random chance alone.”
==============================================
That definition is wrong, Margaret. “Random chance” is not involved in the Null Hypothesis. The Null Hypothesis is the statistical hypothesis that states that there are no differences between observed and expected data.
Thus, what is “expected” is that past climate parameters will not be exceeded. And as a matter of fact, past climate parameters have not been exceeded: the current climate is well within past parameters; there has been no acceleration of global warming, and the planet has been both warmer and colder than in the past.
Which leads to Occam’s Razor: “One should not increase, beyond what is necessary, the number of entities required to explain anything.” [William of Ockham, 1285-1349]
The “carbon” entity is not necessary to explain the current climate. CO2 is an unecessary distraction, which only serves to confuse the issue. As we now know, CO2 has no effect on global temperatures. The demonization of ‘carbon’ is a false alarm.
“Re gas etc – these processes don’t have time boundaries. Argon atoms have been following a random walk for billions of years.”
Including before the big bang, and after the heat death of the universe?
Infinity is bigger than you think.
dbstealey says: May 31, 2013 at 6:54 pm
“Many times Smokey has put Nick Stokes in his place. But some folks are impervious to reason.”
Smokey and dbstealey together might be able to do it 🙂
When I arrived at my US Air Force duty station in England in January of 1970, the Labour government of Harold Wilson had almost achieved third-world nation status for the UK through their socialist economic policies. However, discovery and exploitation of North Sea gas and oil saved the UK from their economy from going under, thanks to their clunky nationalized industries, long enough to survive until Margaret Thatcher turned things around. When I returned for long visits in 1988, I couldn’t believe the vibrancy of the British economy. Tony Blair was “Margaret Thatcher – Lite” and didn’t bugger things up.
However, this Reuters article “UK climate act limits energy choices: Gerard Wynn” shows that the British really didn’t lose their knack for snatching defeat from the jaws of victory, in this case by hamstringing their energy future. It was just hibernating, waiting for the moment when feckless leadership allowed it to rise and muck everything up again.
http://www.reuters.com/article/2013/05/31/column-wynn-uk-energy-idUSL5N0EA2XE20130531
The UK has significant fossil fuel prospects through fracking, just as they once had from North Sea gas and oil. With their genius for mismanagement, I’m sure they will muck this up too.
On a recent thread, someone stated that temperature time series look like 1/f noise (which is naturally occurring in electronic circuits). Just wondering if there is a statistical model for 1/f noise.
“Just wondering if there is a statistical model for 1/f noise.”
I think ARFIMA(1, 1/2, 1) is an example of 1/f noise, although not the only such process.
Whenever these climate loons are making completely ludicrous predictions, I always respond with this:
As previous claims are no longer matching the observed reality, the edifice is collapsing. I like to think of this as the theme song.
dbstealey says:
May 31, 2013 at 7:11 pm
Margaret Hardman says:
—————————–
At least she’s trying to get there.
Margaret, what you need to do is to take the CONCLUSION that CO2 is a terrible monster and causes global warming, climate change, extreme weather weirding and all that crap and flush it from your brain.
Start again from scratch, from the supporting DATA (of which there is none, as you will find).
We’ll go to stage two when you think you’ve found some.
Nick Stokes says:
May 31, 2013 at 1:32 pm
Stephen Rasey says: May 31, 2013 at 12:34 pm
‘“First off, Keenan’s model isn’t wrong because it leads to physically impossible scenarios and violates boundary conditions.”
Yes it is. A random walk is unbounded and has no fixed reference point. If it were to apply, at some stage (not too far away), all life would be extinguished (with probability 1). If it applied in the past, we would not be here. It is physically impossible because there is not the energy available for that unbounded behaviour.”
No statistic describes the world or any part of reality. When statisticians refer to “bell curves” they do not believe that bell curves exist in the world. A statistic is nothing more nor less than a line drawn through points on a graph. It may be formulated as a lovely mathematical equation but it no more describes reality than any other purely mathematical equation, an equation containing no predicates that are descriptive such as “___is spinning.”
When a statistician describes a statistic as a random walk statistic, he does not imply that there is some actual physical process that is the random walk. Nor does he imply that reality has a structure that could accommodate some physical process that you imagine to be the random walk.
Your understanding of statistics, at least what you expressed above, goes beyond the naive and approaches the childish.
Climate scientists must learn that science is a description of the world. Science uses mathematics to assign measures to the world. A statistic is a measure but not a description of the world
If one treats the physical analogies that are used to explain statistics as describing the world then one is making the pure metaphysician’s fundamental error; that is, he is making inferences from the characteristics of his system of representation to the characteristics of reality. That is the formula for metaphysics and anti-empiricism.
From Nick Stokes on May 31, 2013 at 3:16 pm:
“As I said above, if you want to say that it’s just natural variation and nothing has changed, then you have to propose something that would work without changing.”
Nick Stokes’ brain, at least the part that gathers information and makes inferences. But it does have a range of behaviors and various, interesting higher and lower points in that range..
Theo Goodwin says: May 31, 2013 at 9:13 pm
“A statistic is a measure but not a description of the world”
Theo, this is topsy-turvy. This is exactly my argument that you are expressing. The Met calculated statistics of the time series – trend and its uncertainty. Trend is just a statistic – a scaled first moment. Their calculation was an appropriate response to the question asked.
Yet all the argument here is about fitting. About the only number out of Keenan’s analysis is that his ARI(3,1,0) model fits better than trend+AR(1) (AIC). But that’s irrelevant. And yet on that pointless observation (as you’ve expressed well) it is said that the Met has admitted, well, something.
However, if you really want to argue about fitting, then it’s pointless to fit a non-physical model.
dbstealey
May 31, 2013 at 7:11 pm
“The “carbon” entity is not necessary to explain the current climate. CO2 is an unecessary distraction, which only serves to confuse the issue. As we now know, CO2 has no effect on global temperatures. The demonization of ‘carbon’ is a false alarm.”
When I click your link I go to a cherry picked graph. I would welcome a link to a well sourced, peer reviewed and published in a high impact scientific journal rather than to a blog.
As for the definition of null hypothesis. I lifted it, with acknowledgement. I didn’t write it myself.
Stephen Mosher says May 31, 2013 11:20pm
“Well, before we even start we know the model is physically wrong.How do we know that? Well, at some time in the future your model will predict negative ice area.”
Alternately a model that predicts a positive trend will increase infinitely. This is the other side of the coin of your statement above.
Your argument about this would probably be that since we can never reach infinite amounts of anything then positive trends can never be shown to be physically wrong.
To be amusing — in the real world, funding for the negative trend model would reach a point where it must logically terminate. For the model that predicts a positive trend the funding never ends and so the amount of money funding it stretches to infinity.
Interestingly I can conceive of a type of mathematics that deals only in positive numbers. Then infinity and zero would be perfect opposites of each other and both unreachable. The concept of negative numbers is a mathematical concept — an artifact of mathematics. Negative amounts have no physical existence — only a conceptual one (or more correctly only a mathematical one).
(I don’t really want to go here but things in the physical world have a physical existence or they don’t exist. Things in the mathematical world have a mathematical existence or they don’t existence. “Concepts” are an artifact of language and can existence without either a physical or mathematical basis. You may not recognize it but that last statement is really just very old Greek philosophy.)
(Sigh, I can’t help myself here and must make a funny. Applying the above, it is because humans use language, a system neither physically or mathematically based, that all humans are crazy. Conceptual thinking is crazy thinking. It has no basis in any type of “reality”.
Wait, wait, have i not just shown that everything i have said above must be crazy thinking? Oh, damn, what i have i just done to myself. Perhaps i better shut my jaw.
Eugene WR Gallun
Margaret Hardman
The effect of CO2 is not very evident in this graph with temperatures at the start and 2000 being similar end a sharp falling off at the end as co2 rises
http://wattsupwiththat.com/2013/05/08/the-curious-case-of-rising-co2-and-falling-temperatures/
tonyb
Compare the situation with the following example. You have won the capital prize in a lottery and someone says that this is unfair because the lottery organization is corrupt. The prize did not result from a random (or natural) process. This is possible and he has the duty to prove it. The question is of whether this can be done with statistical means. He could demonstrate for example that too much winners are friends of the lottery organization. However, we are dealing here with just one unique period in history. We have got the prize for one century with rising temperatures. Other centuries got Warm Periods or Little Ice Ages and we may suppose that these were not the result of human intervention. What about our century?
For statistical models as proposed, we should look for logical fallacies at a level beyond its content. It compares with analyzing perpetuum-mobile engines, which cannot do what they promise because we cannot obtain energy from nothing. We cannot obtain decisive statistical information from one example. The whole issue of statistical significance is a fallacy. The example Keenan used in his article (see Bishop Hill) is telling. He took the observation of ten heads in a row while tossing a coin (you certainly would get a prize for that). Does that suggest human intervention or unfair play? He forgot to tell us that, on the assumption of independence and fifty-fifty chance, the probability of every possible sequence of ten tosses equals 1/2 ^ 10, which is much more significant than a magic five percent. There is nothing special about ten heads in a row or one thousand heads in a row besides our fascination with certain regularities. From a statistical point of view and with sufficiently many measurements each temperature development of the past century would be trivially significant whatever null hypothesis we choose or whatever models we employ for comparison.
If you have won the capital prize in a lottery, and nothing is wrong with the winners so far, is it possible to demonstrate with statistical means that your prize is the result of human intervention? My answer is no.
tonyb says:
May 31, 2013 at 11:41 pm
///////////////////////////////////
Tony
You are being more than kind.
No mathematician would say there was any correlation between CO2 and temperature in the global temperature records. This follows from the twin fact that:
(i) the rate of warming during the 1970s/1990s warming period is no greater than the rate of warming during the 1920/1940 warming period, although during the 1970s/1990s there is a significant rise in CO2, hence supposedly significant driving forcing, whereas during the 1920s/1940s there was no significant rise in CO2 and hence no significant driving force. This shows the rate of change is the same with or without CO2.
(ii) there is anti-correlation. The temperatures cooled duringb the 1949s/early 1970s just when CO2 emissions began to rise significantly. Whilst correlation does not mean causation, anti-correlation is almost invarably fatal to a claim of causation.
These are not the only examples of a lack of correlation between CO2 and temperature, but are the most stark. Further examples included (iii) that there is no first order correlation in the satellite data sets. These contain 33 years of data and there is no first order correlation (flat temperatures between say 1979 and about 1997, flat beteen about 1999 to date, merely with a one off step change in and around the 1998 super El Nino, and unless that event was caused by rising CO2, there is simply no first order correlation in that data set), and (iv) the present temperature stasis of between about 15 to 22 years depending upon which data set is used.
The fact that CET shows a temperature fall this century of about 0.5degC, just over half of the warming seen in the last century, and the fact that CET shows a fall in winter temperatures since 2000 of almost 1.5degC is further evidence that a correlation between upward rising CO2 levels driving temperature is not evident in the data sets.
The same is so of the paleo record. Whilst there is some similarities between CO2 levels and temperature, there are many examples of anti-correlation. Further, it appears that CO2 lags temperature, not drives it. . :
Why would anyone expect that, after a step function input of large magnitude(the end of the last ice age) that the surface temperature of the planet would be constant?
I would expect oscillations in temperature. The trick is to figure out the periods and magnitudes of these. It may not be possible with our present understanding of the system.
Margaret Hardman says:
May 31, 2013 at 10:44 pm
//////////////////////////
Margaret.
You fall into an often seen trap.
One can not cherry pick data to prove a theory. At best, any such chery picked data is consistent with the theory, and no more than that.
However, it is quite legitimate to cherry pick data to suggest that there is a problem with a theory. If a theory is sound, it should be able to explain the cherry picked data sample, and if it cannot, then there is a problem with the theory. This may be just slight, requiring further refinement to the theory, or it may be fatal to the theory.
Most theories are falsified by cherry picked scenarios which the theory is unable to explain. For example Newtonian mechanics is ‘fine’ in most scenarios and is still being used today in most scenarios. however, we know in the extremes (if you like, the cherry picked scenario that it cannot properly and adequately explain the extreme scenario).
The theory of CO2 induced global warming must be able to explain (i) why there were ice ages when there were very high levels of CO2, (ii) why in the paleo record there are periods of rising temperatures when CO2 was falling, (why in the paleo record there are warm periods when CO2 levels were low, (iv) why in the paleo record there are periods of falling temperatures as CO2 levels rise (v) why in the paleo record CO2 levels lag temperature change if CO2 is the driver of temperature changes as opposed to the consequence of a change in temperature (vi) more recently, the Holocene optimum, the Minoan warm period, the Roman warm period, the MWP, the LIA, (vii) the 1860s to 1880s warming, the 1880s to 1910 cooling, the 1920 to 1940s warming, the 1940s to early 1970s cooling, the temperature stasis of the past 14 or so years, (viii) the reason why the rate of warming during the 1970s to 1990s warming is no greater than the rate of the 1920s to 1940s warming when the rise in CO2 emissions is significantly higher during the later warming period than it was during the former warming period, in short why is the rate of change not far greater during the later warming period/
These are not an exhaustive list (for example one could ask why there is no first order correlation between CO2 levels and temperatuire change in the 33 years worth of satellite data), but they are examples of issues that the AGW theory must adequately explain, failing which it is falsified .
@richard Verney
I said it was a cherry picked graph, not cherry picked data. You might argue that your post covers that point. I would argue that choosing how to present the data and what data to present is part of demonstrating not that your theory is correct or incorrect but that it leads some people to believe one over the other.
Statistically significant compared to what? What distribution was assumed?
If we need to use complex statistical models and processes to spot/define a small trend in data that varies considerably, do we need to ask the question: if this trend is so small that we need to do all of this complex work to try and find it, and if the answer we get depends upon which method is chosen from a range of contested methods, then is it valid to do it?
philincalifornia says:
Margaret Hardman says:
—————————–
At least she’s trying to get there.
With all due respect, my fellow Kalifornian [and I do respect you], Margaret isn’t trying very hard — man.☺
Margaret Hardman asserts plenty of vague criticisms, but she has no real, substantive facts that solidly support her belief system. She criticizes this chart as not being “peer reviewed” [as if that means anything in modern climatism]. But the rest of us know that the Wood For Trees databases are widely accepted by both sides as legitimate. And that chart shows clearly that CO2 has no measurable effect on temperature.
But I like commentators like Margaret. They show that the alarmist crowd lacks any testable, empirical, reproducible data to support their beliefs, which amount to simple — and incorrect — assertions. On this “Best Science” site, Margaret’s assertions are not enough. She needs to post verifiable data to support her argument; something that is missing from her posts.
==========================================
Nick Stokes says:
“…this is topsy-turvy.”
Nick is in Australia, so that is to be expected. ☺
Read the Met Office statement, VERY VERY Carefully.
I Hope everyone picks up on the absolutely monumental statement about GW warming the MO have issued here.
Think about what the Met Office haven’t said as much as what they have said…..
http://metofficenews.wordpress.com/2013/05/31/a-response-on-statistical-models-and-global-temperature/#respond
“Mr Keenan says that there is “no basis” for the claim that the increase in global temperatures since the late 1800s is too large to be reasonably attributed to natural random variation. He goes on to argue that this is because we haven’t used the right statistical model.”
The Met Office, have not actually challenged Mr Keenans’ argument, and are in effect agreeing with him.
They haven’t rebutted the theory that according to statistical modelling the change of temperatures since 1880 is anything other than random fluctuations.
They make a strawmans argument. We know it warming and its down to man because of all these other factors. This press release that all global warming since 1880 isn’t statistically significant.
They haven’t argued against his Maths or Stats. They have just said we know its warming, we have other indicators other than temperature, shut up and Trust us………
In Slingo’s paper…..
“Our calculations of whether a linear trend with first order autoregressive noise model is more
likely to emulate the global temperature timeseries than the driftless third order
autoregressive integrated model, show a range of relative likelihood values from 0.001 to
0.32 depending on which dataset is used (HadCRUT4, NASA, NOAA) and the starting date
of the timeseries (1850-1900). This means that the driftless third order autoregressive model
is more likely to represent the timeseries with these starting dates than the linear trend
model. ”
Basically the temperatures could just be random noise………