Guest essay by Douglas J. Keenan
Temperatures on Earth’s surface—i.e. where people live—are widely believed to provide evidence for global warming. Demonstrating that those temperatures actually provide evidence, though, requires doing statistical analysis.
All such statistical analyses of the temperatures that have been done so far are fatally flawed. Astoundingly, those flaws are effectively acknowledged in the IPCC Fifth Assessment Report (AR5).
The flaws imply that there is no demonstrated observational evidence that global temperatures have significantly increased (i.e. increased more than would be expected from natural climatic variation alone).
Despite this, one of the main conclusions of AR5 is that global temperatures have increased very significantly. That conclusion is based on analysis that AR5 itself acknowledges is fatally flawed. The correct conclusion is that there is no demonstrated observational evidence for global warming.
Full critique at http://www.informath.org/AR5stat.pdf
===========================================================
He closes with these quotes:
The reason for so much bad science is not that talent is rare, not at all; what is rare is character. People are not honest, they don’t admit their ignorance, and that is why they write such nonsense. — Sigmund Freud
Half the harm that is done in this world is due to people who want to feel important. They don’t mean to do harm — but the harm does not interest them. Or they do not see it, or they justify it because they are absorbed in the endless struggle to think well of themselves. — T.S. Eliot
You should, in science, believe logic and arguments,carefully drawn, and not authorities.—Richard Feynman
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
I just completed an analysis of a representative sample of the ~5,500 surface air temperature records used in the construction of the HadCRUT4 global time series, which is why I’m late in commenting.
I reviewed a total of 638 individual records with at least 50 years of data located between the high Arctic and the South Pole and found that 531 of them showed an overall warming trend. Coin-toss statistics show that the chances of getting 531 heads in 638 tosses are on the order of a trillion to one against, so obviously the coin is very heavily weighted.
Does this result confirm that the warming is statistically valid? I’ve been trying to think of a reason why it doesn’t but so far have had no success. Maybe someone else can come up with one.
D.J. Keenan:
Dr.K’s work may be the only example that you know from reading “climate science” literature that appeals to a physical principle (max. entropy) in choice of statistical model, but that is certainly not the only such case in geophysics. Entirely different models are used as a matter of course in wave problems than, say, in those of turbulent diffusion. The crucial difference is that each of those models has been verified and refined by extensive field experiments. By contrast, the H-K model, which has found some verification in hydrology, is patently inadequate for understanding the long-term behavior of terrestrial temperatures.
It should be apparent from the spectral graph I posted that quasi-millenial and trans-centennial oscillations dominate the GISP2 record, with multidecadal oscillations accounting for only ~1/3 of the total variance. The former thus introduce what are perceived as “trends” in records that are only ~100yrs long. No such oscillations can be accounted for by any H-K model, which produces the “red noise” model spectrum shown in the graph. Enough said.
Dr. K:
The discussion you reference is quite revealing, in that you express belief that there is no intrinsic difference between mathematics and physics. Having made the (often painful) transition from the former field to the latter many decades ago, I assure you that the difference is categorical. If I had time, I could supply quotations from great physicists to that effect. For reasons sketched above, the H-K formalism, which may be useful in physical processes governed entirely by first-order DEs, will provide misleading results when second-order, oscillation producing processes are at work. Although the drivers for these oscillations are not clear, their presence in empirical records is unmistakable. This is especially true if one stays away from manufactured or corrupted indices that de-emphasize such oscillations at the expense of LTP-increasing “trends.”
Dear 1sky1,
I can tell from reading and comparing both the discussion thread linked by Dr. Koutsoyiannis and your comment at 5:27pm today that either you misread what Koutsoyiannis wrote in the linked thread, or you are deliberately mischaracterizing what he said there. I cite as evidence for your error another commenter’s (Bart Verheggen, who was arguing vigorously with Koutsoyiannis in that thread) response to the math-physics dichotomy issue Koutsoyiannis discussed in that thread:
B. Verheggen.
Before I can take seriously anything you write from now on (and I WANT to — you are a bright, articulate, obviously well-educated person), 1sky1, your credibility will need some rehabilitation. I would LOVE to hear from you that you simply misread what Koutsoyiannis wrote.
With high hopes that all is truly well,
Janice
Dear Demetris,
You are so very welcome. And, thank you for the privilege of using your first name. Hang in there.
I hope you spend a LOT more time with your grandchildren than with people on threads like the one you linked above and this one. Colleagues are great, but family is what makes life worth living.
God bless you,
Janice
1sky1,
What you can assure me is about your own feelings, painful or not, about mathematics and physics. These do not necessarily represent an objective truth.
Mathematics has helped to try to be accurate in what I am saying. In fact you misquote me as I did not say “there is no intrinsic difference between mathematics and physics”. Rather, I said “I think the language of physics is (or at least includes) mathematics”. Since you are invoking quotations, here is one relevant: “[The] great book which ever lies before our eyes — I mean the universe — […] is written in the mathematical language (Galileo).” By the way, when you mention first-order DEs and second-order, oscillation producing processes, aren’t these DEs mathematical objects?
For combining oscillating behaviours with HK see “Climatic variability over time scales spanning nine orders of magnitude: Connecting Milankovitch cycles with Hurst–Kolmogorov dynamics”, http://itia.ntua.gr/ 1297/
Otherwise, I am afraid I have been off topic and distracted the discussion, so I stop here. So, feel free to misquote me again—I don’t think I’ll reply 🙂
Corrected link: http://itia.ntua.gr/1297/
Janice, thanks for your support. Also, thanks for your suggestion–indeed sometimes I forget the importance of family and focus on minor things,
Demetris
Roger Andrews: “Does this result confirm that the warming is statistically valid?”
in the absence of human interference is given by 
, where 
is a zero-mean sequence of independent identically distributed random values of a certain (e.g., unity-standard-deviation Gaussian) distribution and 
and 
are functions you specify. If you had done that and then asked whether the chances of seeing a trend as high as you saw exceeded one in twenty, I would have understood the question. But I don’t understand “significantly valid.”
What do you mean by “statistically valid”? (And I could just as well have asked Mr. Keenan–or any of the other adepts who’ve made similar pronouncements over the years–what he meant by “significant.” Being a statistics naif, I have heretofore kept silent in these discussions, but my frustration at what seems a total lack of clarity has finally reached the breaking point.)
Now, you might instead have postulated that temperature values are generated by a random process such that the (temperature) output
Thank you for the thread and your paper regarding AR5 Mr Keenan. It must be extremely frustrating to have to work so hard just to get people to see things correctly when you have demonstrated the facts. I certainly hope you and Lord Donoughue succeed in getting Dr. Slingo and others to respond to your questions because if you do it will remove one of the piles the AGW edifice has been constructed on i.e. the climate has and still continues to warm in an unnatural way.
@Joe Kirklin Born . . my understanding of significance in statistics is that the null hypothesis has been breached. The null hypothesis being a trend line, coefficient, that is at or very near zero.
Keitho:
At November 1, 2013 at 10:38 am you say
Oh dear, sorry, but no.
I will explain the matter again.
The Null Hypothesis says it must be assumed a system has not experienced a change unless there is evidence of a change.
The Null Hypothesis is a fundamental scientific principle and forms the basis of all scientific understanding, investigation and interpretation. Indeed, it is the basic principle of experimental procedure where an input to a system is altered to discern a change: if the system is not observed to respond to the alteration then it has to be assumed the system did not respond to the alteration.
In the case of climate science there is a hypothesis that increased greenhouse gases (GHGs, notably CO2) in the air will increase global temperature. There are good reasons to suppose this hypothesis may be true, but the Null Hypothesis says it must be assumed the GHG changes have no effect unless and until increased GHGs are observed to increase global temperature. That is what the scientific method decrees. It does not matter how certain some people may be that the hypothesis is right because observation of reality (i.e. empiricism) trumps all opinions.
Please note that the Null Hypothesis is a hypothesis which exists to be refuted by empirical observation. It is a rejection of the scientific method to assert that one can “choose” any subjective Null Hypothesis one likes. There is only one Null Hypothesis: i.e. it has to be assumed a system has not changed unless it is observed that the system has changed.
However, deciding a method which would discern a change may require a detailed statistical specification.
In the case of global climate no unprecedented climate behaviours are observed so the Null Hypothesis decrees that the climate system has not changed.
Importantly, an effect may be real but not overcome the Null Hypothesis because it is too trivial for the effect to be observable. Human activities have some effect on global temperature for several reasons. An example of an anthropogenic effect on global temperature is the urban heat island (UHI). Cities are warmer than the land around them, so cities cause some warming. But the temperature rise from cities is too small to be detected when averaged over the entire surface of the planet, although this global warming from cities can be estimated by measuring the warming of all cities and their areas.
Clearly, the Null Hypothesis decrees that UHI is not affecting global temperature although there are good reasons to think UHI has some effect. Similarly, it is very probable that AGW from GHG emissions are too trivial to have observable effects.
The feedbacks in the climate system are negative and, therefore, any effect of increased CO2 will be probably too small to discern because natural climate variability is much, much larger. This concurs with the empirically determined values of low climate sensitivity.
Empirical – n.b. not model-derived – determinations indicate climate sensitivity is less than 1.0°C for a doubling of atmospheric CO2 equivalent. This is indicated by the studies of
Idso from surface measurements
http://www.warwickhughes.com/papers/Idso_CR_1998.pdf
and Lindzen & Choi from ERBE satellite data
http://www.drroyspencer.com/Lindzen-and-Choi-GRL-2009.pdf
and Gregory from balloon radiosonde data
http://www.friendsofscience.org/assets/documents/OLR&NGF_June2011.pdf
Indeed, because climate sensitivity is less than 1.0°C for a doubling of CO2 equivalent, it is physically impossible for the man-made global warming to be large enough to be detected (just as the global warming from UHI is too small to be detected). If something exists but is too small to be detected then it only has an abstract existence; it does not have a discernible existence that has effects (observation of the effects would be its detection).
To date there are no discernible effects of AGW. Hence, the Null Hypothesis decrees that AGW does not affect global climate to a discernible degree. That is the ONLY scientific conclusion possible at present.
Richard
Thank you for that full and detailed explanation Richard, I am much obliged.
I was simply referencing the meaning in a purely statistical sense.
Joe Born:
Okay, let me rephrase the statement in a form that a “statistics naif” – which you clearly aren’t – can understand.
“Does this result confirm that the warming is REAL?”
As I said, I’m having difficulty in coming up with a “no it doesn’t” answer.
Thanks to Keitho and richardscourtney for the attempts to respond to my inquiry. Particularly since such inquiries from us plodders are usually ignored, I’m reluctant to exhibit such ill grace as to admit explicitly that those attempts left me none the wiser. Still . . .
It seems to me as a matter of logic that unless the data’s probability given the null hypothesis is zero the data do not disprove that hypothesis. In the unlikely event that my logic is sound, “significance” would at best be a measure of how close to zero the data place that probability. So I’m having trouble warming to Keitho’s proposition that significance means that the null hypothesis has been breached–at least if “breached” means disproved.
In any event, I suppose my real difficulty in most such discussions is getting a solid purchase on what the “significance” user considers the null hypothesis actually to be. I am thankful in this regard for the benefit of richardscourtney’s thoughts–and not eager to question them since on so many matters of definition he seems more certain than I am on any. Indeed, his identification of the null hypothesis here as that man has not affected global average temperature does indeed seem tacitly to be implied by many such users.
But if there is indeed some probability threshold that they use to decide whether their data are significant, I have been unable to infer their mathematical basis for calculating the probabilities against which the threshold is compared and what connection that basis has to the hypothesis that man has caused no warming.
I suppose that this is what Mr. Keenan is driving at, and I join others in commending him for his efforts. Unfortunately, though, this returns me to Roger Andrews’ question that gave rise to my previous comment. Mr. Andrews noted his difficulty in avoiding the conclusion that global average temperature has exhibited a positive trend over some (unless I missed it, unspecified) time interval. Modulo the irregularities in the temperature records’ maintenance, I share that difficulty, But I suspect that Mr. Andrews’ question arose from comments such as this from Mr. Keenan: “Simply put, no one has yet presented valid statistical analysis of any observational data to show global warming is real.”
Now, perhaps Mr. Keenan did not mean that literally. Indeed, Ms. Moore and Mr. Marler seem to imply so, and there is some basis for that conclusion, including the excerpts in Ms. Moore’s second suggestion and Mr. Keenan’s response to that suggestion.
But Mr. Keenan said it more than once, and he has no Jay Carney-type exegete to place a more-congenial gloss on his writings. So I remain unsure of just what he means.
Joe Born:
Both Keitho and I tried to explain the null hypothesis to you but – at November 1, 2013 at 12:57 pm – you say you remain “none the wiser”. This leaves me at a loss because I do not know how to state the Null Hypothesis more clearly than
The Null Hypothesis says it must be assumed a system has not experienced a change unless there is evidence of a change.
If you can provide evidence of AGW having made any change to the behaviour of the climate system then I assure you that you will obtain at least one Nobel Prize. This is because 3 decades of research conducted worldwide at a cost of more than $5 billion per year has failed to find any such evidence. Ben Santer claimed to have found such evidence but that was soon shown to be false (he deliberately cherry picked a part of a data series).
You quote Douglas J. Keenan as saying, “Simply put, no one has yet presented valid statistical analysis of any observational data to show global warming is real”. In his above essay he actually says
He is simply correct when he says
“The correct conclusion is that there is no demonstrated observational evidence for global warming.”
because
the IPCC itself says the statistical analyses it uses are fatally flawed.
The clearest demonstration of that Douglas Keenan is right is that the IPCC says it is 95% certain the major cause of the recent warming is anthropogenic. If the IPCC had evidence – statistical or otherwise – for AGW then they would not say they are 95% certain of it.
People say gravity exists.
People say DNA contains genetic information.
People say an action has an equal and opposite reaction.
The knowledge of those things is provided by observational evidence, so nobody says they are X% certain of those things: people say them.
Simply, there is no empirical evidence that AGW exists; none, zilch, nada.
There may be effects of AGW but – if so – they are so trivial that they cannot be distinguished from previously experienced natural climate variability. So, as I explained to you
To date there are no discernible effects of AGW. Hence, the Null Hypothesis decrees that AGW does not affect global climate to a discernible degree. That is the ONLY scientific conclusion possible at present.
Richard
Roger Andrews:
re your post at November 1, 2013 at 11:11 am.
Nobody disputes that there has been warming from the Little Ice AGE (LIA). The warming from the LIA started centuries before there could have been any significant AGW. At issue is whether AGW has contributed to that warming, and there is no evidence of any kind that AGW has contributed to it.
Warming is not evidence of the cause of warming.
Richard
Richard:
You would be perfectly correct if Keenan had confined himself to AGW, but he doesn’t. He states that we have no proof that there’s been ANY warming. The relevant quote reads “Simply put, no one has yet presented valid statistical analysis of any observational data to show global warming is real. Moreover, that applies to any warming—whether attributable to humans or to nature”. He more or less confirms this by adding the Figure 5 time series plots, which he claims show that all of the observed warming could be simply a random effect.
Well, it can’t. To obtain representative results Keenan should have combined thousands of random runs, one for each of the individual surface station records that contribute to the global time series. Had he done this the chances that any one of his “models” would have replicated the observed warming would have been at least a trillion to one against according to my results.
Roger Andrews:
I take your point in your post at November 1, 2013 at 3:29 pm, and I thank you for your correction of my misunderstanding.
However, your correction of me raises two issues.
Firstly, as I now understand that you and I agree, there has been warming over e.g. the twentieth century but we make no comment on the cause(s) of it. However, the evidence for that is various (e.g. your analysis of individual station records, predominantly retreat of glaciers, etc.).
Secondly, that does not alter the validity of Keenan’s point about the time series analyses of global temperature not providing valid statistically significant indication of that warming. Indeed, it raises the importance of his point.
Again, thankyou for correcting my misunderstanding.
Richard
RIchard: No problems with your first issue. I agree there has indeed been warming over the 20th century and as far as this discussion is concerned we needn’t specify the cause of it.
But I’m a little confused on your second issue, where you seem to imply that while there has indeed been warming it isn’t statistically significant. Is this in fact what you are saying, or have I misunderstood you?
Roger Andrews:
I apologise for my lack of clarity. It was not intentional: I always try to be clear often to the point of bluntness. But at November 1, 2013 at 4:26 pm you say to me
No, actually I was expanding on a point which you made in the post I was answering.
You wrote
Perhaps, but Keenan’s paper is not discussing that. His paper discusses the compilations of global temperature time series provided by i.e. HadCRU, GISS, etc. and concludes that it is not possible to validly determine warming by statistical analyses of those compilations.
As I understand your comment which I here quote, you agree with him.
You are saying that “individual surface station records” (n.b. NOT the compilations of global temperature time series provided by i.e. HadCRU, GISS, etc.) should be used as the basis of an analysis to determine warming. So, I wrote my paragraph which you query; i.e.
As I see it, your argument “raises the importance of his point” for two reasons; i.e.
(a) as you seem to be saying, the time series analyses of GLOBAL temperature are not the appropriate data for determining a statistically valid significant indication of global warming,
and
(b) the the time series analyses of GLOBAL temperature have little – if any – practical value when they cannot provide a statistically valid significant indication of the global warming which you and I agree has happened.
I hope my meaning is now clear. And I assure that my lack of clarity was not intentional.
Richard
Dr. K:
Pray tell, where have I ever directly quoted you? In the context of my
ongoing critique about your lack of adequate empirical verification for “HK
dynamics,” I simply abstracted your ultra-academic view that “there is no
dichotomy between physics and statistics” and that “statistics is physics”
[stated on May 1 at 8:07pm in your linked discussion]. Nor is my
parenthetical remark about the “often painful” transition from mathematics
to physics an expression of personal “emotion,” as you would have it.
It merely ecapsulates Einstein’s humble recognition that a single
experiment can prove an entire theory wrong. Feynman put this issue much
more pungently: “Physics is to mathematics, as sex is to masturbation.”
The H-K formalism is plainly a kinematic, rather than dynamic, methodology.
It expresses no universal law. Its applicabilty to any physical problem
needs to be verified empirically on a case-by-case basis, instead of being
assumed as a given. Your novel “climacogram” effectively wrings most
information of possible physical interest out of a motley lot of proxy
time-series in reducing everything to a single variance metric. And then
the results are plotted on a log-scale, which always visually diminishes any
discrepancies. The results may impress novices, but it scarcely constitutes
geophysical verification.
That you may not respond further here is not unexpected. Only crickets were
heard when, a few years ago at Climate Audit, I numerically posted the acf
of the aggregate average of a geographically representative set of USA
stations, which contradicted all H-K expectations of functional form.
richardscourtney
Thank you for your response.. My reading of it is that you have conflated global warming with anthropogenic global warming–and that is my problem with Mr. Keenan’s paper. But your answer to Mr. Andrews suggests that you may have recognized that error. Although I was and still am thankful for Mr. Keenan’s efforts, which have clarified some things for me, his paper does not make it clear that he is distinguishing the two concepts.
As to your definition of what a null hypothesis, I had not intended to ask for elucidation. Truly, I appreciate your setting it in bold face so that we less gifted observers would be more likely to grasp it. However, I had already known what you thought the definition was; I am just unable to bring myself to accept it. I have identified no reason in your remarks not to consider a null hypothesis simply to be any assumption upon which one could base a calculation of observed data’s probability.
Again, though, thank you for the attention.
Dear Janice Moore:
I regret your impressions. But there are nuances in a highly technical discussion that are often missed by those unfamiliar with the science. Please read my response to Dr. K above. Let’s all enjoy the weekend.
Joe Born:
In your post at November 1, 2013 at 5:06 pm you say to me
I find that strange because in my “remarks” at November 1, 2013 at 10:54 am I wrote
In other words, the reason one cannot “consider a null hypothesis simply to be any assumption upon which one could base a calculation of observed data’s probability” is because that would be a rejection of the scientific method.
Richard
@ur momisugly 1sky1 (re: your 5:10 sneer) — With dismay, I read your response to Dr. Koutsoyiannis (at 5:03pm this evening); sadly, you only confirmed my bad impression of your character for veracity. And your attempt to prevaricate by quibbling over whether you directly quoted Koutsoyiannis or not is disgusting. Since I won’t know whether what you say is accurate or not, I will be ignoring anything you have to say from now on — what would be the point? Why should ANY of us here now take you seriously?
Indeed, “there are nuances…,” your mischaracterization of what Koutsoyiannis said, however, was blatantly obvious.
No WONDER you don’t reveal your real name, here.
Please know, 1sky1, that you are not disgusting; your behavior is. You are better than that.
You are loved. And you are worth redeeming.
I, even now, have hope for you.
Even if you snarl at me in response to this,
there is always — hope.
Janice