RealClimate logo

Misrepresentation from Lindzen

Filed under: — gavin @ 6 March 2012

Richard Lindzen is a very special character in the climate debate – very smart, high profile, and with a solid background in atmospheric dynamics. He has, in times past, raised interesting critiques of the mainstream science. None of them, however, have stood the test of time – but exploring the issues was useful. More recently though, and especially in his more public outings, he spends most of his time misrepresenting the science and is a master at leading people to believe things that are not true without him ever saying them explicitly.

However, in his latest excursion at a briefing at the House of Lords Commons in the UK, among the standard Lindzen arguments was the following slide (which appears to be a new addition):

What Lindzen is purporting to do is to compare the NASA GISS temperature product from 2012 to the version in 2008 (i.e. the y-axis is the supposedly the difference between what GISS estimated the anomaly to be in 2012 relative to 2008). A rising trend would imply that temperatures in more recent years had been preferentially enhanced in the 2012 product. The claim being made is that NASA GISS has ‘manipulated’ (in a bad way) the data in order to produce an increasing trend of global mean temperature anomalies (to the tune of 0.14ºC/Century compared to the overall trend of 0.8ºC/Century) between the 2008 and 2012 versions of the data, which are apparently shown subtracted from each other in Lindzen’s figure. Apparently, this got ‘a big laugh’ at his presentation.

However, this is not in the least bit true: the data are not what he claims, the interpretation is wrong, and the insinuations are spurious.

The annotation indicates that Lindzen is using the GISTEMP Land-Ocean Temperature index (LOTI, i.e. the index that includes weather station data and sea surface temperature data to give a global anomaly index with wide spatial coverage) (“GLB.Ts+dSST.txt”). There is another GISTEMP index (the Met station index) which only uses weather station data (“GLB.Ts.txt”) which doesn’t have as much coverage and has a substantially larger trend reflecting the relative predominance of faster-warming continental data in the average.

Old versions of the data can be retrieved from the wayback machine quite readily, for instance, from February 2006, October 2008 or December 2007. The current version is here. I plot these four versions and their differences below:

As should be clear, the differences are tiny, and mostly reflect slightly more data in the earlier years in the latest data and the different homogenisation in GHCN v3 compared to GHCN v2 (which was used up to Dec 2011). This is however in clear contradiction with Lindzen – the biggest difference in trend (between 2006 and today), is a mere 0.05ºC/Century, and from 2008 to 2012 it is only 0.003ºC/Century – a factor of 40 smaller than Lindzen’s claim. What is going on?

The clue is that the transient behaviour of Lindzen’s points actually resembles the time evolution of temperature itself – not homogenisation issues, or instrumental or coverage changes. Indeed, if one plots the two GISTEMP indices and their difference (using current data), you get this:

Thus it looks very much like Lindzen has plotted the difference between the current Met Station index and an earlier version of the LOTI index. I plotted the Feb 2012 Met index data minus the Feb 2009 LOTI index, and I get something very close to Lindzen’s figure (though it isn’t exact):

This is sufficient to conclude that Lindzen did indeed make the mistake of confusing his temperature indices, though a more accurate replication would need some playing around since the exact data that Lindzen used is obscure.

Thus, instead of correctly attributing the difference to the different methods and source data, he has jumped to the conclusion that GISS is manipulating the data inappropriately. At the very minimum, this is extremely careless, and given the gravity of the insinuation, seriously irresponsible. There are indeed issues with producing climate data records going back in time, but nothing here is remotely relevant to the actual issues.

Such a cavalier attitude to analysing and presenting data probably has some lessons for how seriously one should take Lindzen’s comments. I anticipate with interest Lindzen’s corrections of this in future presentations and his apology for misleading his audience last month.

Update: Lindzen did indeed apologise (sort of) (archived) though see comments for more discussion.

539 Responses to “Misrepresentation from Lindzen”

  1. 501
    Dave says:


    I’m not trying for humour. I asked for information on “experimental tests which could have falsified the climate models which presently imply drastic long term warming i.e. X degrees over Y years or something similar whereby X and Y would have been defined at the time of the test and would correspond to dangerous long-term warming”. Perhaps the use of the word “time” confused you. It is simply used in the sense that values X and Y would have been generally agreed upon prior to the measurement taking place i.e. no redefining what “success” means if the theories fail to match generally agreed criteria. This is rather standard experimental protocol when assessing the validity of predictions.

    If you have a sensible answer to the question I’d very much like to hear it.

  2. 502
    SecularAnimist says:

    Dave wrote: “If you have a sensible answer to the question I’d very much like to hear it.”

    I humbly and respectfully suggest that you thoroughly familiarize yourself with the actual climate models that are in use in the field, as well as the actual temperature data sets.

    Once you have done that, please read the “Updates to model-data comparisons” which the moderators of this site have helpfully provided for 2009, 2010 and 2011.

    Then revisit this question.

  3. 503
    Ray Ladbury says:

    Out of curiosity, what is your research specialty? I am also a physicist.

    As to falsifying tests of models, there are a variety of pieces of evidence that if not true would call the models in question:

    1)A cessation of the warming trend over the long term (>30 years)
    2)A lack of cooling in the stratosphere as the troposphere warms.
    3)Drastically different response to impulse forcing, such as volcanic eruptions.

    However, it is unlikely that the entire theory would be falsified, because it is exceedingly unlikely that the entire theory is wrong. In particular, it is very unlikely that any new theory will not have positive feedback–you just cannot explain Earth’s climate without it.

  4. 504
    Hank Roberts says:

    Sure, Dave.

    You’re asking a trick question having fun.

    Of course, “a professor of physics at a leading
    European University” knows how to address this
    kind of question. You’ve taught stellar physics without
    building your own star, remember? Did it blow up?
    Got an idea why, if it did?

    Same situation with planetary atmospheres.

    That’s why I suggest asking a real question at John
    Baez’s site

    You’ll get a reception appropriate to your actual
    understanding, and the rest of us will enjoy watching.

    As “a professor of physics at a leading European
    University” you are no doubt well known to other physicists.
    They’ll recognize you there.

    Go ask.


  5. 505
    Rob Dekker says:

    Hi Troy,

    Thanks for your note, and the link to your page where you discuss the scientific arguments made on this subject.

    However, I am not even close to assessing the claims made by S&B on bias caused by dependent or independent noise in the modeling.

    In fact, in my first experiments, I was just checking what LC11 has done.

    Please check LC11 figure 6 (scatter plot), and notice the offset on T. That means that they found many more positive delta-T segments than negative delta-T segments.

    Now, how could they have gotten that ?

    I think they simply plotted T against TOA flux from their model run, rather than look at short-term segments as other researchers did.

    What I’ve show in my first post on this, is that IF they indeed plotted T against TOA_flux for each of the 288 months of their run, then the feedback bias they inferred can be explained by the long-term drift (in SST and flux) that their AGW forcing they applied.

    By the way, my model does correctly put Q behind the TOA flux estimate, but so far I did not even include any noise at all. I only just quantified the long-term drift (of the ‘center point’ for scatter plots) that their AGW forcing causes, which appears consistent with their T center-offset in figure 6 and their assessment of positive feedback bias in figure 7 A.

    You mention that you wrote the S&B model in M.
    Can you share that model with us ?
    Or can you run it with 0 internal noise (Q=0) and just set f to +0.4W/decade ramp ?
    And see what comes out for estimated feedback ?

  6. 506
    Alex Harvey says:

    Gavin, #483:

    I have made enough off-topic comments here so I’ll limit this response to a single point. I claimed that there is in climate science “no effort to mitigate for the confirmation bias of the scientists”. You responded:

    I do find it amusing though that you are accusing mainstream scientists of confirmation bias, while quoting from Lindzen, the erstwhile subject of this thread.

    Why would you regard my statement as an ‘accusation’? I don’t dispute that Lindzen is biased; I don’t think he does either. But why should it be an ‘accusation’ to suggest that his critics are also biased? It was perhaps first pointed out by Thomas Kuhn in The structure of scientific revolutions that scientists as humans necessarily suffer from confirmation bias. Psychologists have shown further that the bias is stronger when the topic is emotionally charged – such as the case with climate change science. Considering further the large uncertainty in both theory and observations it seems to me that climate science is perfectly set up so that the confirmation bias of scientists collectively could make a large difference to our understanding.

    Barton Paul Levenson produced an interesting graph showing climate sensitivity estimates over time. (I note that he hasn’t plotted all estimates in the literature so it would be even more interesting if it was updated.) Anyhow, when he looked at it he saw climate science estimates converging over time on around 3 K per doubling CO2. When I look at it, however, I don’t see the convergence but I do see an odd step-wise jump from low to high estimates immediately following the Charney report in 1979. Perhaps like Kuhn’s gestalt we’ll all see what we expect to see when we look at this graph.

    It has been recognised now by physicists that blind experimental design is required in order to protect the result from the bias of the experimenter. The same has long been recognised in psychology and medical research. You may be interested in Roodman, A. 2003 on Blind Analysis in Particle Physics.

    Why would such experimental design be considered necessary in particle physics if it is not necessary in climate science?

  7. 507

    #501 Dave

    First, I’m curious why you, being a well known physicist would be reticent to post your real name? Most people on this blog do post their real name.

    Second, how about the attribution models which imply that we should be relatively stable to cooling.

    Third, How about the early models from Hansen and Manabe that could have been falsified and unfortunately were not… it’s warming.

    Forth, how about Svante Arrhenius predicting adding more CO2 should warm the planet.

    We’ve all heard the falsification argument here before and those that pose it seem to think they are somehow novel, or original, or clever. Interesting.

  8. 508
    dbostrom says:

    Rob 20 Mar 2012 at 7:37 PM, what you’re doing sounds like a great topic for a collabo at Tamino’s site.

  9. 509

    We have to wonder, without Lindzen or Monckton, what would we be talking about here?

    I dare say that Lindzen does not dwell on this comment section.

    But since Lindzen is such an effective diversion and distraction – I would imagine that his handlers (funding foundations) are strategizing on precisely how he should to act.

    Of course, I do not know this for sure, but if I heavily invested in a deliberately deceptive campaign of aggressive ignorance – especially one that enabled a trillion dollar industry that was afraid of the science, then I would expect shareholders would demand that teams of professionals engage in deep planning for optimal tactics.

    Lindzen is earning his pay. He succeeds merely by drawing our attention to his actions.

  10. 510
    MARodger says:

    Dave @501
    Those who repose on prestigious chairs of physics do not always find climatology a subject they can understand, a point I feel well demonstrated here.
    Indeed, not all science is like physics with its simple mathematical relationships. Physicists get the credit for showing we don’t live on a flat earth or in a heliocentric universe but physicists should be very thankful we live in a flat universe and not some warped (or even warping) Riemann-space.
    Other sciences are not so lucky.
    If you are familiar with the concept ‘climate sensitivity,’ you would perhaps be less bold with your enquiries about X degrees and Y time and your talk of “…rather standard experimental protocol.” Sensitivity is a scientific quantity whose value is of incredible importance for humanity yet the best we can say is “very likely” 1.5-4.5 deg C with a ‘fat tail’.
    There is much that is well established within climatology although denialist commentators do a very good job muddying the waters. And there are climatologists like Lindzen involved in this muddying process. So when the likes of Lindzen breezily presents a meeting of wannabe movers & shakers that ‘climate sensitivity’ is so low it’s not worth bothering about, it is couched within propaganda and includes all the usual ingredients – cherry-picking, nonsense, conspiracy theory & flawed science.

    My response to such denialism is to try to expose the lies and hypocracy that supports it because climatology is very very important.
    Professor Richard Lindzen – a climatologists who reckons his doubts on climate change would remain unless the oceans freeze over or the tropics become uninhabitable.” These are not the words Lindzen uses but it is definitely part of the message he gave in the Westminster meeting. (Why not watch the videos linked within this thread.)
    So is Physics plagued by lunatics like Lindzen?

    Your complain @488t is that you are “more than fed up with hearing about a consensus” within climate science. From a science like physics where the likes of the LHC is built with hardily a squeak of public complaint for the cost, perhaps physicists have some expertise you could share with us in how to silence the denialists & baying crowds and get the job done.

  11. 511

    “From a science like physics where the likes of the LHC is built with hardily a squeak of public complaint for the cost..”

    And then there was the Supercollider…

  12. 512
    Ray Ladbury says:

    Alex Harvey, I must say that your assertion of a “step” to higher sensitivity estimates in 1979 requires a “creative” mind. What I see is that the estimates below about 1.5 and above about 5 are clearly outliers (some clearly motivated by ideology, others by systematic error.). What is more, excluding these outliers, there is a clear convergence–just what you would expect in a mature field like climate science.

    As to your contentions regarding confirmation bias, do you really think it is a problem that is news to the climate science community? My doctoral research in the dark and distant past was in experimental particle physics. And yes, we were also aware of the issue. That is why we had more than one individual working on any one important analysis. There are a number of measures you can take. However, probably the most important counter measures to confirmation bias are inherent to the scientific process:
    1)independent replication
    2)Sheep do not prosper in science.

    This last is important. If you are a follower, you will never shape the consensus in your field. Those who do shape the field are innovative iconoclasts who have good judgment and powerful ideas. Do you have any idea how big a prize it would be to a scientist to make the climate crisis go away? To contend that there are no such innovators in climate science is to be wilfully blind. So, yes, your contention that climate science is especially plagued by confirmation bias is not just an insult, but an ignorant one at that.

  13. 513
    SecularAnimist says:

    Alex Harvey wrote: “I claimed that there is in climate science ‘no effort to mitigate for the confirmation bias of the scientists’.”

    And that claim is blatantly false, which renders the remainder of your comment moot.

  14. 514
  15. 515

    #506 Alex Harvey

    You seem focused on the argument rather than the science. Either way, Gavin’s point is valid, you do at the least ‘infer’ an accusation in the woven fabric of your arguments.

    And who, of whom indulges in the argument over the science, doesn’t trot out Khun at times like this. Right?

    As to the general argument re physicists, there are many lines of evidence in climate science pointing to human causation.

    A hypothetical: If the only way for you to prove a theory is to wait until it kills you, how productive is that to ‘your’ continued work in the future that no longer exists for you because you would be dead from the proof?

    As to IPCC and paleo, of course they are looking at it strongly, keep reading.

    As to: “(4) no effort to mitigate for the confirmation bias of the scientists; and”

    You’re kidding, right?

    As toP “(5) an answer at the end that doesn’t actually rule out the low sensitivities anyway.”

    Since your seem to be a big believer in low sensitivity, just how to we get in an out of ice ages with such low sensitivities?

    sound of crickets chirping…

  16. 516

    For those that are dictionary challenged such as…, Alex Harvey has made statements that someone has done something wrong. This is otherwise known as an accusation.

    Alex doesn’t like Gavin’s usage of the word accusation because it makes Alex look bad.

    Alex would likely prefer Gavin to use more ambiguous language when pointing out facts, while Alex prefers to make accusations that are not supportable by the relevant evidence; largely because Alex doesn’t know enough about what he is talking about to make an argument that stands on said relevant evidence.

    Hope that helps clarify for those just following along.

  17. 517
    T. Marvell says:

    MARodger (460) is, I find, essentially correct about ENSO. Regressions with monthly data show it has an extremely strong impact on SST in the same month and the following month, but not later. A one point increase in the ENSO index leads to a 0.1 incease in SST, both in the north and the south, over these two months. There is also a strong negative relationship between CO2 levels and ENSO – when CO2 goes up, the ENSO index goes down 3 to 4 months later. The relationship is with ENSO, not temperature as MARodger argues.

    [Response: You have the causality backwards – and of course, applied to the trend in CO2 this is obviously incorrect. – gavin]

    MARodger questons my regressions finding that higher SST leads to more CO2 in the atmosphere. I thought that relationship was obvious. The effect is indeed large, with SST increases accounting for roughly 10% of CO2 increases.

    [Response: Not sure what ‘10%’ refers to, but if this is supposed to be relevant for the trend it implies a carbon cycle/climate sensitivity that is greater than that seen over the glacial cycles (~20 ppmv/deg C) – and even there it takes hundreds of years to manifest. – gavin]

    A couple of scientists (e.g. 501) argued that the climate models do not stand up to traditional scientific methods of proof. They cannot be falsified. Several people responsed to the effect that one should not criticize the models without understanding them, which seems arrogent and defensive to me, a non-climate scientist. One does not have to understand the models to ask how they can be falsified in any reasonable time period. I doubt whether they can be, a least when they predict future temperatures. Temperature changes are so erratic that if predictions are met, it might be just due to chance.

    [Response: This is certainly true in the short term (say < 10 years). But it is interesting that climate models themselves indicate that short term trends are unpredictable just from the external forcings. Thus if you want to test model predictions you need situations and/or timescales where the forced trend is detectable above the noise of the short-term variations. That will happen either for very strong forcings (big volcanic eruptions) or for longer term changes (the Holocene trends, the 20th Century, 20-30 years, the last glacial maximum, the 8.2kya event etc.) - gavin]

    I don’t see how climate models can make short or medium term predictions when temperature changes are so dominated by El_Nino, and El_Nino is so erratic (as MARodger’s graphs show) and hard to predict over the time frames in the models.

    Out of curiosity, do the climate models include the impact of SST on CO2 levels?

    [Response: Yes. This is used in the carbon cycle class of models where you don’t know the CO2 level already but you can estimate emissions and is a significant part of the new CMIP5 simulations. (Note that for the historical period where we know CO2 concentration these models are used to constrain estimates of the emissions). – gavin]

    The overarching concern here is that global warming theory has important policy considerations. Climate scientists, no matter what they think of the merit of their models, have to look outward and convince policy makers that the models are accurate. They cannot afford to be as insular as their comments suggest. If they are at all concerned with policy and the outside world, they cannot brush off critics like Lindzen as not worth the effort to respond. Policy makers are likely to be influenced by a MIT professor who is making unanswered points.

    [Response: None of Lindzen’s points are ‘unanswered’. Arguments against a low sensitivity for instance, abound in the literature and in the assessments. – gavin]

    Because policy is important, the models cannot be ignored just because they are not up to snuff scientifically. They will probably never be subject to falsification, so the climate scientists should look hard for alternate methods to judge the models in ways persuasive to policy-makers.

    [Response: You have somewhat of a Popperian view of science that doesn’t really accord with how it actually works. All models are wrong – that is trivially true – but the issue is whether they can make skillful predictions. The answer is that they can. i.e. using models gives better information than just assuming persistence or extrapolating into the future. Improving that skill and quantifying how robust predictions are, and on what spatial and temporal scale is a useful activity – and one that most of the scientists involved are focused on. – gavin]

  18. 518
    MARodger says:

    T. Marvell @517
    We agree that ENSO wobbles cause SST wobbles. We disagree about SST-CO2 linkage, of which you say you “thought that relationship was obvious.” That’s a worry.
    And @517 you propose a CO2-ENSO relationship that I find most bizarre. You say CO2 rises precede ENSO falls by 3-4 months!

    Having linked to a pair of graphs @460 plotting these three variables ENSO, SST & CO2, the second of which illustrates the point of our agreement, I did find an old graph that may assist in the other relationships & have uploaded it onto a defunct website.

    I cannot vouch for the synchronisity of the plots (the graph is old work-in-progress) but they do demonstrate quite well that the three wobbles do so pretty much in unison, big wobble matching big wobble. Thus surely puts your negative CO2-ENSO relationship in doubt.

    Regarding the third relationship SST-CO2, I said no more @463 & @487 than CO2 wobbles were in some way the result of the ENSO/SST wobble. Plotted on the graph below is the ‘de-trended’ CO2 plot. The wobble thus has no net impact on long-term atmospheric CO2 levels.

    Graph of three wobbles here

    I hope this helps.

  19. 519
    Ray Ladbury says:

    T. Marvell,
    OK. First, where are you getting this crap?

    T. Marvell: “The overarching concern here is that global warming theory has important policy considerations.”

    Allow me to introduce you to the logical fallacy called “argument from consequences,”. The consequences of a proposition have no bearing whatsoever on the truth of the proposition. Thus, if scientists have determined with high confidence (e.g. >90%) that humans are warming the fricking planet, then humans are most likely warming the fricking planet. This proposition was first posited 116 years ago and has generally been accepted for about 50 years. Anthropogenic causation is an inevitable consequence of the consensus theory of Earth’s climate. There is no rival theory that explains anywhere near as much of Earth’s history that would not also imply anthropogenic causation of the current warming epoch.

    Climate sceince is nowhere near as insular as you imply. The science has been vetted and explicitly endorsed by the National Academy of Sciences of the US and most comparable scientific bodies worldwide. There is not one dissenting voice. Likewise, nearly every one of the professional societies of scientists in related fields (physics, chemistry, meteorology, statistics…) has similarly examined and endorsed the science. Even the American Association of Petroleum Geologists–fricking petroleum geologists–was forced to drop its dissenting position. There is no scientific rationale for opposing the conclusion that we are warming the planet.

    Now it appears to me that you as a layman are faced with a choice: Side with the overwhelming majority of climate experts and the overwhelming majority of the scientific community or go directly against them and side with a bunch of loons like Lord Ha Ha of Monkton and the occasional contrarian/shill scientist who has quit doing science. May you choose wisely.

  20. 520
    Radge Havers says:

    T. Marvell @ 517

    Climate scientists, no matter what they think of the merit of their models, have to look outward and convince policy makers that the models are accurate. They cannot afford to be as insular as their comments suggest. If they are at all concerned with policy and the outside world, they cannot brush off critics like Lindzen as not worth the effort to respond. Policy makers are likely to be influenced by a MIT professor who is making unanswered points.

    Tsk! O, those silly climate scientists sitting around in their ivory towers being all silly and stuff!

    So we’re back to that. Again. Got that annoying tune stuck in our head have we?

    That you are apparently unaware of the efforts of scientists to communicate the science (including the models), and that Lindzen has been dissected in fine detail over the years here and elsewhere, doesn’t mean that it hasn’t happened. Perhaps you can’t see through all the denialist noise.
    On the remote possibility that you actually happen to be an expert in communicating to congress critters and lay audiences, then do your due diligence and then offer concrete suggestions on SPECIFIC INSTANCES of communications failure.

    Please drop the tetchy, broad brush accusations.

    And see, for starters:
    Misrepresentation from Lindzen (read the article itself)

  21. 521
    SecularAnimist says:

    T. Marvell wrote: “Several people responsed to the effect that one should not criticize the models without understanding them, which seems arrogent and defensive to me, a non-climate scientist”

    So when someone asks a question that clearly shows that they don’t understand what they are asking about, and in some cases clearly shows that they haven’t made any effort to understand what they are asking about, and they receive respectful and helpful responses encouraging them to learn more about the models, how they work, how they are developed and refined, and about the data sets against which the models are constantly compared, and then revisit their questions in the light of this understanding — you think that is “arrogant and defensive”?

    The moderators of this site, and the scientifically-knowledgeable commenters, consistently demonstrate the patience of saints in their attempts to educate visitors about climate science, to explain the complexities of the science, and to answer sincere questions — and even blatantly insincere, belligerent and hostile questions — and for this, you call them “arrogant and defensive”?

  22. 522

    #517 T. Marvell

    “A couple of scientists (e.g. 501) argued that the climate models do not stand up to traditional scientific methods of proof. They cannot be falsified.”

    I don’t understand why people think the theory is not falsifiable.

    – Arrhenius said it would warm, it did. That was falsifiable (granted the science was not mature then).
    – Hansen/Manabe said it would warm and did a model that could have been falsified if it didn’t warm.

    Just because it ‘did’ warm doesn’t mean that the predictions and models were not falsifiable. They were. It’s just that they proved to be generally correct. They would have been falsified if it ‘did not’ warm. Therefore, the model was falsifiable.

    I’ll say it again. The early predictions and models were falsifiable. Hansen/Manabe were strong enough science to meet the criteria of ‘good science’ with a reasonable amount of the bugs worked out. The work was falsifiable over decades. Just because some people don’t like the results does not mean that the models were not falsifiable.

    – Lab experiments with CO2 are falsifiable also.
    – Lindzen’s ‘Iris Hypothesis’ was falsifiable and has been falsified.

    There is falisifiability if you look. The old adage still applies that there are none so blind as those that choose not to see.

  23. 523
    SecularAnimist says:

    John P. Reisman wrote: “I don’t understand why people think the theory is not falsifiable.”

    Because Fox News or Anthony Watts told them so. And of course they unquestioningly and absolutely believe whatever Fox News and Anthony Watts say, because they are “skeptics”.

    Thus, they have no need to learn or understand anything about climate models before “criticizing” them, and to suggest that they should first learn something about climate models and temperature data sets is “arrogant and defensive”.

  24. 524

    #523 SecularAnimist

    Ah, okay. Right, I forgot. Sorry ;)

    I suppose it’s not an argument to authority, so I guess it’s argument to Dunning/Kruger now.

  25. 525
    Troy_CA says:

    Re Rob Dekker (#505, 20 Mar 2012 @7:37):

    I can definitely share the model script. You may have already found my post with the script (in R) through that page, but I’ve also updated it to make it a bit more relevant to this LC11 question: (you’ll need to install the TTR package, if you haven’t yet already).

    I think the issue is as I expected. Technically, as I said and according to the FG06 method, the regressions should not be purely TOA flux against T_s, but for TOAFlux minus external forcing (WMGHG in this case) against T_s. Otherwise, the issue crops up that you mentioned in your original comment, where you get a bias if the only thing creating your temperature changes is that external WMGHG forcing.

    However, I don’t believe this is what is creating the bias described in LC11, and shown in figure 6. The reason that is that for their model, the internal/ocean forcing used to create the monthly temperature fluctuations dwarfs the impact of the WMGHG forcing on those monthly Ts. If the LC11 model were simulating volcanic eruptions that had abrupt, external instantaneous forcings, then this external forcing would certainly need to be removed, but as it is their simple regression using TOAFlux rather than TOAFlux-external forcing is not strongly impacted by this (although I do think it would be good to correct this). You can confirm this by setting the WMGHG forcing to “0” in the model I linked to, and see that the bias is still present. Indeed, the simple model demonstration used to show this bias in the S&B11 paper does not include a GHG forcing at all.

    The amount of bias is more sensitive to the other parameters of the model. For instance, the ratio of the internal radiative forcing to the ocean forcing, or the “smoothing months”/decorrelation time for the internal radiative forcing term. The noise model itself (which I’ve used from SB11, it is likely slightly different for LC11), which uses a low-pass filter for the internal radiative forcing, is also subject to question. Similarly, the effective mixed layer heat capacity has an impact. Finally, you’ll note that if you use a long enough time period for the model, and THEN properly remove the external forcing (during which time the WMGHG wins out, depending on the sensitivity), you can get an accurate diagnosis from the simple regression model, but this is not necessarily relevant given the limited amount of flux data we currently possess.

    Anyhow, as I said, I think the 0.4 W/m^2/K GHG forcing aspect is largely irrelevant to the LC11 demonstration of bias, where the bigger point of contention for this topic is the noise model and parameters.

  26. 526
    Jim Larsen says:

    485 John said, “Just because the public thinks something is sound does not make it sound.”

    True but irrelevant. Just like Monoskeptics railing against facts won’t change next decade’s weather, stating truths about relevance and merit won’t change the fact that in the minds of the relevant parties not refuting the currently dominant Monoskeptical paper is de facto admission that it’s indisputable.

  27. 527
    Alex Harvey says:

    Ray Ladbury, #512:

    Aside from BPL’s graph, I am not able to see where you disagree with anything I actually said.

    I made a number of points and I would appreciate if you can say which if any of them you disagree with.

    If you are aware that all humans have a tendency to confirmation bias, why do you think it is an ‘accusation’ to say this? Note that I did not say “climate science is especially plagued by confirmation bias”.

    Do you agree with psychologists that confirmation bias is stronger when the topic is emotionally charged? For instance, Lindzen is evidently angry about some things (see Lindzen, 2008, ‘Climate science: is it currently designed to answer questions?’). Thus his confirmation bias is more likely to come into play (witness the avoidable mistake he made that is the subject of this thread). But surely you also must admit that climate science generally is an emotive topic, especially given that many believe we are headed to catastrophe. Do you agree?

    Kuhn showed that during paradigm shifts even the raw data themselves change. His examples, though, were from fields where the data was much more certain than in climate science. Presumably you don’t dispute that climate science data is highly uncertain compared to other fields like physics and medicine; so why would you disagree that this should lead us to expect that the bias of scientists can come into play more easily? Do you contend that Kuhn was simply wrong for instance?

    Finally, I showed that theoretical physicists lately see the need to manage the experimenter’s bias using blind experiment design. You can’t dispute this, I presume. Do you dispute that blind analysis has never been used in climate science? If not, why is it necessary in physics but not climate science? Maybe you contend that it isn’t necessary in physics and they are just being pedantic?

  28. 528
    T. Marvell says:

    MARodger (518) – in your chart, shouldn’t red be temp and yellow CO2? I don’t think 12 month running averages help when studying short term relationships. It’s better to graph changes, yearly changes if the data are seasonal.

  29. 529
    Rob Dekker says:

    Troy, thanks for your response.
    You may very well be right that the bias caused by (internal and/or external) noise creates a bias that is larger than the GHG bias that LC11 introduced.

    And I readily agree that “known” forcing changes (such as GHG forcing) should be removed from from regressions run on actual observations (and from model runs too), or else an artificial bias would be introduced.

    However, I am not sure if LC11 actually did that. In fact, all the evidence I see so far suggests that they did NON remove the GHG bias that I quantified. Otherwise, how can they get a residual (offset) in T in figure 6, and how come that the ‘bias’ they found in figure 7A happens to be of the same magnitude as the bias I found SHOULD be there if you put a sudden change ramp of GHG forcing at the start of the simulation ?

    Thanks a lot for your R script !
    I am unfamiliar with R, but will install the necessary tools and run some experiments.
    For starters, I would like to reproduce bith run you suggest (noise with GHG forcing 0), and also the run that I suggest (0 noise and GHG forcing as LC11 dictates).

    I do have one question on the core section of your model though :
    (forgive me for re-formatting your code so it’s readable on this restricted HTML post) :

    #run through each time step
    for (mon in 2:modelLength)
    forcing = S[mon] + N[mon] + WMGHG[mon]
    T[mon] = (forcing+Cp/dt*T[mon-1])/(Cp/dt + lambda)
    Feedback[mon] = lambda*T[mon]
    TOAFlux[mon] = WMGHG[mon]+N[mon]-Feedback[mon]

    Here, I see your integration of forcing into T, but I don’t see that the loop is closed.
    Shouldn’t the ‘forcing’ depend on the temperature (or ‘Feedback’ or ‘TOAFlux’ ?
    Or is there some way that R moves ‘TOAFlux’ back into variable N or so ?

  30. 530
    MARodger says:

    T. Marvell @528
    Concerning graph linked @518 –
    Yellow is recognisably HadCRUT3 so that makes Yellow temperature. And thus Red is annual CO2 changes.
    Concerning use of 12 month running averages – I would suggest that in this situation they provide a quick & understandable way of smooting data & thus appropriate to demonstrate that ENSO wobbles, temperature wobbles & CO2 annual increase wobbles all wobble in unison (with ‘small’ lags although I will not vouch for the syncronicity of the data on the graph presented) – that being the point at hand.

  31. 531
    SecularAnimist says:

    Alex Harvey wrote: ” … I did not say ‘climate science is especially plagued by confirmation bias’ …”

    Oh, really? Here’s what you said, in comment #506 (emphasis added):

    … there is in climate science “no effort to mitigate for the confirmation bias of the scientists” …

    Psychologists have shown further that the bias is stronger when the topic is emotionally charged – such as the case with climate change science. Considering further the large uncertainty in both theory and observations it seems to me that climate science is perfectly set up so that the confirmation bias of scientists collectively could make a large difference to our understanding

    You said that confirmation bias “is stronger … with climate change science”.

    You said that climate science is “perfectly set up” for confirmation bias to make “a large difference”.

    And you asserted an absolute, blatant falsehood, that “there is in climate science no effort to mitigate for” confirmation bias.

    But after claiming repeatedly that confirmation bias is “stronger” in climate science, that climate science is “perfectly set up” for confirmation bias, and that climate scientists have made “no effort to mitigate” confirmation bias, now you pretend that you “did not say” that climate science is “especially plagued by confirmation bias”.


    With all due respect, sir, what you are doing is making wild, general, sweeping, unsupported accusations of confirmation bias on the part of thousands of scientists involved in climate research over many decades — along with blatantly, insultingly false claims that those scientists have made “no effort” to mitigate for that alleged bias — which you puff up with irrelevant and nonsensical hand-waving at Kuhn.

    And what’s worse — when you are challenged on these assertions, you attempt to obfuscate what you actually said. Why? Because your original assertions are baseless and insupportable, and you know it.

  32. 532
    Susan Anderson says:

    MARodger and T Marvell:

    Clicking on MARodger reveals a wonderful compendium of visual aides to understanding. This is a terrific collection from a guy who’s used his expertise and taken infinite pains (well, not quite infinite if you are in the mood to quibble). Don’t knock it!


  33. 533
    Troy_CA says:

    Rob (#529):

    I agree that LC11 probably regressed Flux against T, rather than Flux-WMGHG against T. However, I’m not sure we’re on the same page, because the offset on T in Figure 6 is not evidence of this — for proper implementation of the FG06 regression method, you don’t remove the WMGHG forcing from its effect on T (indeed, this would be extremely complicated with timing offsets, ocean transport, etc.), so T would look the same either way in that graph. Rather, you remove the radiative anomaly due to WMGHG forcing from the TOA flux. So, evidence that this is not being removed would need to come in the form of an offset on that _Y_ axis, and over the 288 months should be around 0.5 W/m^2 offset (I can’t tell if this is the case from the graph). But it should then be clear why this offset, even if present, does not matter much: it is tiny compared to the Flux variations. Regarding why you get a similar bias to figure 7, I would suspect coincidence, as both issues would create a positive bias. Can you try plotting a graph similar to figure 6 for the simple regression? I would imagine that in your GHG only scenario, the Flux will not show the variability of the LC11 graph.

    Regarding the model, the forcing should NOT depend on T. The radiative forcing for a doubling of CO2 is ~3.7 W/m^2, regardless of the temperature. As you can see in eq. 8 of LC11, Q(t) (or the forcing at time t) does not depend on T. The change in outgoing radiation (i.e., “Feedback” in the code), certainly DOES depend on T, as can be seen on that Feedback line. Similarly, the apparent TOA flux (what we measure) is a combination of all the radiative forcing/noise + the feedback term, as is set on that next line.

  34. 534
    Troy_CA says:


    To add on to my last comment (23 Mar 2012 at 10:34 AM), regarding the “loop”, you’ll notice it is actually present within the T[mon] = line. The “lambda” (F in LC11) and “T[mon-1]” terms complete that feedback loop. If you simply replace the dSST term in LC11 eq. 8 with T[mon] – T[mon-1], and using lambda*T[mon-1] for the F*SST(t), then re-arrange to solve for T[mon], this is what you get. Obviously, this is a discrete form, but the monthly time-step is fine enough that it has little effect.

  35. 535
    Ray Ladbury says:

    Alex Harvey,
    First, I would contend that humans have a tendency toward wishful thinking and in many cases magical thinking. Science is the appropriate countermeasure to this tendency. What that means is that we must try to weave ALL THE EVIDENCE into a single tapestry that has both PREDICTIVE AND EXPLANATORY power. Climate science fits very well into that paradigm. It has done a very good job at understanding an incredibly complicated system over much of its history, and it has had some impressive–in some cases astounding–predictive successes. Progress has been steady over a period of several decades–quite unlike a field undergoing a revolution.

    The catch is that in order for all this data to make sense, there has to be positive feedback in the system. You simply cannot understand Earth’s climate without it. You can’t explain why we are not a ball of ice. You cannot explain how small changes in insolation result in glacial/interglacial cycles. You cannot explain how the climate responds to impulse forcings such as injection of aerosols by volcanic eruptions. And so on. Lindzen and the other “dissenters” offer no predictions. They offers no understanding. They offer questionable analyses of obscure datasets that even they don’t fully understand and claim they are invalidating understanding built over decades. They do nothing of the sort.

    You suggest that climate science suffers from confirmation bias. Do you really think that thousands of climate scientists are such sheep that they would forego the opportunity to revolutionize their field? Do you realize the celebrity they would attain? I assure you that they realize it.

    Look, Alex, science works. It works even when practiced by fallible humans. It capitalizes on human traits such as curiosity, ambition, stubbornness and the need to be valued by our peers to acheive reliable understanding of natural world. Climate science is no different from any other discipline other than it has stumbled into a beeshive of glibertarian assclams who wish to be left to their profession of [edit: easy] in peace.

  36. 536
    Radge Havers says:

    In case the point is lost for not being bluntly restated, the scientific process itself has evolved precisely to mitigate against confirmation bias. This is perhaps more true of the hard physical sciences, which you can see for yourself (and unlike some other areas) haven’t been polluted by postmodernism. As for “…a beeshive of glibertarian assclams…” OK. For collectors, that one’s a keeper!

  37. 537
    Brian Dodge says:

    “…climate models do not stand up to traditional scientific methods of proof. They cannot be falsified.”
    Sure they can.
    Show that the infrared absorption spectra of polyatomic gases are significantly different from current scientific consensus.
    Show that Clausius–Clapeyron is significantly different from current scientific consensus.
    Show that TSI is increasing, driving temperatures up, or GCR decreasing, driving albedo down.

    Climate models could have been falsified – they just haven’t been.

    There are lots of components to climate models; better, or different, understanding of how most of them work won’t falsify the big picture, but may change the sensitivity. (It’s arguable that the sensitivity is higher than models indicate. We know that tundra, clathrate, and ice sheet collapse have nonlinearities that aren’t well understood; the loss of summer sea ice in the Arctic is faster than models predict, the collapse of Larsen Ice Shelf was a surprise, and misunderestimated nonlinearities of clathrate and permafrost melt carry low probability high impact positive sensitivity risk. Cloud cover trends are still uncertain, but apparently negative – ~0.04% per decade decrease globally 1954-2008; so far, no surprises.

    A smart scientist like Lindzen can tell which key parts being wrong would falsify the models. I thought of two, and I’m a college dropout.

    IMHO, the fact that Lindzen’s only arguing for low sensitivity, and hasn’t proposed any fatal flaws in the models means he doesn’t think the models are fundamentally flawed. The only place left to hang his political hat is to exaggerate the model’s inaccuracy, rather than say they are wrong.

  38. 538
    SecularAnimist says:

    Ray Ladbury wrote: “Climate science is no different from any other discipline other than it has stumbled into a beeshive of glibertarian assclams …”

    If only.

    Unfortunately, climate science has “stumbled into” the wealthiest and most powerful corporations in the history of the world, who can buy entire governments with their petty cash, not to mention major media organizations, and who are determined that a little thing such as the likely destruction of the Earth’s biosphere from continued business-as-usual consumption of their products is not going to stand in the way of the trillions of dollars in profit that they expect to reap from burning every last drop of oil and every last crumb of coal that can be ripped from the Earth’s crust.

    Their “glibertarian” pseudo-ideology is just another propaganda tool, as bogus as the pseudo-science and pseudo-economics cranked out by their bought-and-paid-for “think tanks”.

  39. 539
    t marvell says:

    Anderson (532) pointed out that if you click on MARodger, you get a web site with lots of interesting graphs. That site is:

    Another interesting set of graphs is:

    As a non-climate scientist I find these facinating, and I have spend a lot of time browsing through them.

    I couldn’t find a link to them on the home page. There should be one, and links to similar sites, even though there are already lots of links.

    Again, from browsing the graphs, I think that the best way to compare trends visually is to compare yearly changes. That is, temperature should be, say, the December 2011 temperature less the December 2010 number. That takes out seasonal variations.

    In post #517 I remarked that increased CO2 precedes a reduced ENSO index. Gavin and MARodger were highly skeptical. The effect exists, however, although it is not as strong as the ENSO-to-temp and temp-to-CO2 connections, which are extremely strong (post 517). This does not mean that CO2 has a negative “cause” with respect to El Nino, but there may be some third variable involved, such as arosols, that is associated with CO2 and that affects El Nino.

    This is important because when I look at the association between CO2 and temperature, I find a lagged relationship in the “wrong” direction. More CO2 is followed by lower SST temperature. The “wrong sign” problem. But if the ENSO index is added to the analysis, that goes away.

    Also this is important because it can be a short-term forecasting tool for the El Nino, provided the relationship is firmly established and the reason for it found.

    All these relationships are based on monthly-time-series Granger analysis (Granger got the nobel prize in 2003). They can be easily replicated since climate scientists probably have both the data and regression programs readily available. The benefit of the Granger is that it provides good controls for factors not entered into the regression, although very short term relationships are not controlled.

    The El-Nino-to-temp relationship is partly for the current year (although there is a large one-year-lagged relationship), so it might be partly an identity relationship, in that temperature factors into the ENSO index.