The Good, The Bad, and The Ugly: My Initial Comments on the New Dessler 2011 Study

September 7th, 2011 by Roy W. Spencer, Ph. D.

UPDATE: I have been contacted by Andy Dessler, who is now examining my calculations, and we are working to resolve a remaining difference there. Also, apparently his paper has not been officially published, and so he says he will change the galley proofs as a result of my blog post; here is his message:

“I’m happy to change the introductory paragraph of my paper when I get the galley proofs to better represent your views. My apologies for any misunderstanding. Also, I’ll be changing the sentence “over the decades or centuries relevant for long-term climate change, on the other hand, clouds can indeed cause significant warming” to make it clear that I’m talking about cloud feedbacks doing the action here, not cloud forcing.”

Update #2 (Sept. 8, 2011): I have made several updates as a result of correspondence with Dessler, which will appear underlined, below. I will leave it to the reader to decide whether it was our Remote Sensing paper that should not have passed peer review (as Trenberth has alleged), or Dessler’s paper meant to refute our paper.

NOTE: This post is important, so I’m going to sticky it at the top for quite a while.
While we have had only one day to examine Andy Dessler’s new paper in GRL, I do have some initial reaction and calculations to share. At this point, it looks quite likely we will be responding to it with our own journal submission… although I doubt we will get the fast-track, red carpet treatment he got.

There are a few positive things in this new paper which make me feel like we are at least beginning to talk the same language in this debate (part of The Good). But, I believe I can already demonstrate some of The Bad, for example, showing Dessler is off by about a factor of 10 in one of his central calculations.

Finally, Dessler must be called out on The Ugly things he put in the paper (which he has now agreed to change).


Estimating the Errors in Climate Feedback Diagnosis from Satellite Data

We are pleased that Dessler now accepts that there is at least the *potential* of a problem in diagnosing radiative feedbacks in the climate system *if* non-feedback cloud variations were to cause temperature variations. It looks like he understands the simple-forcing-feedback equation we used to address the issue (some quibbles over the equation terms aside), as well as the ratio we introduced to estimate the level of contamination of feedback estimates. This is indeed progress.

He adds a new way to estimate that ratio, and gets a number which — if accurate — would indeed suggest little contamination of feedback estimates from satellite data. This is very useful, because we can now talk about numbers and how good various estimates are, rather than responding to hand waving arguments over whether “clouds cause El Nino” or other red herrings.

I have what I believe to be good evidence that his calculation, though, is off by a factor of 10 or so. More on that under THE BAD, below.

Comparisons of Satellite Measurements to Climate Models

Figure 2 in his paper, we believe, helps make our point for us: there is a substantial difference between the satellite measurements and the climate models. He tries to minimize the discrepancy by putting 2-sigma error bounds on the plots and claiming the satellite data are not necessarily inconsistent with the models.

But this is NOT the same as saying the satellite data SUPPORT the models. After all, the IPCC’s best estimate projections of future warming from a doubling of CO2 (3 deg. C) is almost exactly the average of all of the models sensitivities! So, when the satellite observations do depart substantially from the average behavior of the models, this raises an obvious red flag.

Massive changes in the global economy based upon energy policy are not going to happen, if the best the modelers can do is claim that our observations of the climate system are not necessarily inconsistent with the models.

(BTW, a plot of all of the models, which so many people have been clamoring for, will be provided in The Ugly, below.)


The Energy Budget Estimate of How Much Clouds Cause Temperature Change

While I believe he gets a “bad” number, this is the most interesting and most useful part of Dessler’s paper. He basically uses the terms in the forcing-feedback equation we use (which is based upon basic energy budget considerations) to claim that the energy required to cause the observed changes in the global-average ocean mixed layer temperature are far too large to be caused by satellite-observed variations in the radiative input into the ocean brought about by cloud variations (my wording).

He gets a ratio of about 20:1 for non-radiatively forced (i.e. non-cloud) temperature changes versus radiatively (mostly cloud) forced variations. If that 20:1 number is indeed good, then we would have to agree this is strong evidence against our view that a significant part of temperature variations are radiatively forced. (It looks like Andy will be revising this downward, although it’s not clear by how much because his paper is ambiguous about how he computed and then combined the radiative terms in the equation, below.)

But the numbers he uses to do this, however, are quite suspect. Dessler uses NONE of the 3 most direct estimates that most researchers would use for the various terms. (A clarification on this appears below). Why? I know we won’t be so crass as to claim in our next peer-reviewed publication (as he did in his, see The Ugly, below) that he picked certain datasets because they best supported his hypothesis.

The following graphic shows the relevant equation, and the numbers he should have used since they are the best and most direct observational estimates we have of the pertinent quantities. I invite the more technically inclined to examine this. For those geeks with calculators following along at home, you can run the numbers yourself:

Here I went ahead and used Dessler’s assumed 100 meter depth for the ocean mixed layer, rather than the 25 meter depth we used in our last paper. (It now appears that Dessler will be using a 700 m depth, a number which was not mentioned in his preprint. I invite you to read his preprint and decide whether he is now changing from 100 m to 700 m as a result of issues I have raised here. It really is not obvious from his paper what he used).

Using the above equation, if I assumed a feedback parameter λ=3 Watts per sq. meter per degree, that 20:1 ratio Dessler gets becomes 2.2:1. If I use a feedback parameter of λ=6, then the ratio becomes 1.7:1. This is basically an order of magnitude difference from his calculation.

Again I ask: why did Dessler choose to NOT use the 3 most obvious and best sources of data to evaluate the terms in the above equation?:
(1) Levitus for observed changes in the ocean mixed layer temperature; (it now appears he will be using a number consistent with the Levitus 0-700 m layer).

(2) CERES Net radiative flux for the total of the 2 radiative terms in the above equation, (this looks like it could be a minor source of difference, except it appears he put all of his Rcld variability in the radiative forcing term, which he claims helps our position, but running the numbers will reveal the opposite is true since his Rcld actually contains both forcing and feedback components which partially offset each other.)

(3): HadSST for sea surface temperature variations. (this will likely be the smallest source of difference)

The Use of AMIP Models to Claim our Lag Correlations Were Spurious

I will admit, this was pretty clever…but at this early stage I believe it is a red herring.

Dessler’s Fig. 1 shows lag correlation coefficients that, I admit, do look kind of like the ones we got from satellite (and CMIP climate model) data. The claim is that since the AMIP model runs do not allow clouds to cause surface temperature changes, this means the lag correlation structures we published are not evidence of clouds causing temperature change.

Following are the first two objections which immediately come to my mind:

1) Imagine (I’m again talking mostly to you geeks out there) a time series of temperature represented by a sine wave, and then a lagged feedback response represented by another sine wave. If you then calculate regression coefficients between those 2 time series at different time leads and lags (try this in Excel if you want), you will indeed get a lag correlation structure we see in the satellite data.

But look at what Dessler has done: he has used models which DO NOT ALLOW cloud changes to affect temperature, in order to support his case that cloud changes do not affect temperature! While I will have to think about this some more, it smacks of circular reasoning. He could have more easily demonstrated it with my 2 sine waves example.

Assuming there is causation in only one direction to produce evidence there is causation in only one direction seems, at best, a little weak.

2) In the process, though, what does his Fig. 1 show that is significant to feedback diagnosis, if we accept that all of the radiative variations are, as Dessler claims, feedback-induced? Exactly what the new paper by Lindzen and Choi (2011) explores: that there is some evidence of a lagged response of radiative feedback to a temperature change.

And, if this is the case, then why isn’t Dr. Dessler doing his regression-based estimates of feedback at the time lag of maximum response, as Lindzen now advocates?

Steve McIntyre, who I have provided the data to for him to explore, is also examining this as one of several statistical issues. So, Dessler’s Fig. 1 actually raises a critical issue in feedback diagnosis he has yet to address.



The new paper contains a few statements which the reviewers should not have allowed to be published because they either completely misrepresent our position, or accuse us of cherry picking (which is easy to disprove).

Misrepresentation of Our Position

Quoting Dessler’s paper, from the Introduction:

The usual way to think about clouds in the climate system is that they are a feedback… …In recent papers, Lindzen and Choi [2011] and Spencer and Braswell [2011] have argued that reality is reversed: clouds are the cause of, and not a feedback on, changes in surface temperature. If this claim is correct, then significant revisions to climate science may be required.”

But we have never claimed anything like “clouds are the cause of, and not a feedback on, changes in surface temperature”! We claim causation works in BOTH directions, not just one direction (feedback) as he claims. Dr. Dessler knows this very well, and I would like to know:

1) what he was trying to accomplish by such a blatant misrepresentation of our position, and

2) how did all of the peer reviewers of the paper, who (if they are competent) should be familiar with our work, allow such a statement to stand?

Cherry picking of the Climate Models We Used for Comparison

This claim has been floating around the blogosphere ever since our paper was published. To quote Dessler:

“SB11 analyzed 14 models, but they plotted only six models and the particular observational data set that provided maximum support for their hypothesis. “

How is picking the 3 most sensitive models AND the 3 least sensitive models going to “provide maximum support for (our) hypothesis”? If I had picked ONLY the 3 most sensitive, or ONLY the 3 least sensitive, that might be cherry picking…depending upon what was being demonstrated.

And where is the evidence those 6 models produce the best support for our hypothesis? I would have had to run hundreds of combinations of the 14 models to accomplish that. Is that what Dr. Dessler is accusing us of?

Instead, the point of using the 3 most sensitive and 3 least sensitive models was to emphasize that not only are the most sensitive climate models inconsistent with the observations, so are the least sensitive models.

Remember, the IPCC’s best estimate of 3 deg. C warming is almost exactly the warming produced by averaging the full range of its models’ sensitivities together. The satellite data depart substantially from that. I think inspection of Dessler’s Fig. 2 supports my point.

But, since so many people are wondering about the 8 models I left out, here are all 14 of the models’ separate results, in their full, individual glory:

I STILL claim there is a large discrepancy between the satellite observations and the behavior of the models.


These are my comments and views after having only 1 day since we received the new paper. It will take weeks, at a minimum, to further explore all of the issues raised by Dessler (2011).

Based upon the evidence above, I would say we are indeed going to respond with a journal submission to answer Dessler’s claims. I hope that GRL will offer us as rapid a turnaround as Dessler got in the peer review process. Feel free to take bets on that. 🙂

And, to end on a little lighter note, we were quite surprised to see this statement in Dessler’s paper in the Conclusions (italics are mine):

“These calculations show that clouds did not cause significant climate change over the last decade (over the decades or centuries relevant for long-term climate change, on the other hand, clouds can indeed cause significant warming).”

Long term climate change can be caused by clouds??! Well, maybe Andy is finally seeing the light! 😉 (Nope. It turns out he meant ” *RADIATIVE FEEDBACK DUE TO* clouds can indeed cause significant warming”. An obvious, minor typo. My bad.)

205 Responses to “The Good, The Bad, and The Ugly: My Initial Comments on the New Dessler 2011 Study”

Toggle Trackbacks

  1. Bart says:

    I would like to submit my comments at Steve McIntyre’s blog as perhaps being of some interest to you. More information can be gotten out of the phase plane charts.

  2. Jacob C says:

    Yes! Progress!
    As you said, he at least worked around the idea that cloud forcing is possible, even if he came to a rather small number for its magnitude. That was the only issue I had ever had with SB2011 or any of its predicessors…even if the bias is there, is it really meaningful and important to the sensitivity debates? We finally seem to be working on that idea specifically. Absolutely wonderful. Keep it up Dr. Spencer. Your perspective is needed in this debate badly.

  3. Arthur Smith says:

    Dr. Spencer – why do you “assume” a value for lambda? That sure sounds like circular reasoning, if you are trying to come up with an argument for why sensitivity is different from the IPCC estimate. Dessler’s estimate was based on observational numbers, with no “assumptions” about lambda, as far as I’m aware. Also, it would be helpful if you and Dessler could agree on common notation and terminology for your energy balance equation, it’s a little confusing comparing your comments with his when one person says “Delta F_ocean” and the other says “S” – are these supposed to mean the same thing?

    • 1) You have to assume a value for lambda in order to get an answer…Dessler did exactly the same thing.

      2) I am abiding by the original notation we introduced in 2008…Dessler is the one who is departing from that notation.

      • Roy,

        Arthur’s point was that the assumed value was based on observational data, rather than pulled out of a hat.

        • Pay attention, Barry. Dessler examined a range of lambdas, from 1 to 6, because no one really knows what lambda is…that’s the point of the whole exercise. We have two publications now (not the Remote Sensing one) which show evidence for lambda=6. Lindzen and Choi (2011) gets even bigger values. I used the examples of 3 and 6.

          This is not an issue worthy of debate. There are far more worthy ones, so let’s not waste time.

        • Jack says:


          When are you going to challenge Dr Spencer’s work in a peer reviewed journal. Or are you just going to hang around this blog like a gadfly?

          • When Roy actually publishes something, I usually let the specialists hash it out. I originally just set out to critique his book, which reports some stuff he couldn’t get published in the peer-reviewed literature, but then he kept making similar mistakes on his blog posts, so I’ve kept at it.

  4. I think he should have said cloud changes could cause warming or cooling. His first of many mistakes.

    Secondly how could anyone be so bold to know the actual effects of clouds on the climate, or if temperature changes cause cloud changes (which I doubt) or vice versa, when clouds have these following variables which probably influence there impact or lack of impact on temperatures, those being the following:







    Those are a few I can think of ,off the top of my head.

    Dessler’s argument is during a 10 year period of time temperature variations were caused mostly by El Nino and La Nina, and no one has shown clouds cause El Nino ,and then there could not be a clouds causing temp. change.

    I say how could one jump from the fact if clouds cause or do not cause La Nina’s/El Nino’s , that they might or might not influence temperatures.

    I say ENSO ,when in it’s warm phase,the temperatures go up, and less clouds are present , does not prove that the increase in temperature caused less clouds.

    Say for a minute that the temperature increase/ less clouds, was a positive feedback, if that were the case why then does that not keep feeding upon itself. Why does the ENSO/CLOUDS/TEMPERATURE- relationship, more or less stay in check,if a postitive feedback were present.

    This leads to two conclusions, it is not a positive feedback, or some other negative feedback must come into play, to stop the positive feedback. If someother big negative feedback was out there , would we not have a hint of what that might be. The fact we don’t means it probably does not exist, therefore the positve feedback between the temperature increasing ,causing less clouds is probably false. It is probably a negative feeback, like Dr. Spencer says.

  5. MikeN says:

    So 3 of the models are not too far off looking at your chart.

    There are 2002 ways to choose 6 out of 14 models.

    • Thanks…I was wondering what that number was. I figured it had to be at least in the hundreds.

      • Dudley Dobinson says:

        Dr Roy

        A bit off topic but to get any 6 from 14 models
        I calculate the number to be 3003 (not 2002) possible combinations of 6 from 14 as follows (14*13*12*11*10*9) / (6*5*4*3*2*1) I learnt this formula as an accountant working for a bookmaker. Very important when calculating betting odds. 8 from 14 would give you the same result. If you wanted to put the resulting 6 in any order then the answer is 2,162,160 Calculated 3003 * 720 (6*5*4*3*2*1)
        Now you know just one way a bookmaker makes a profit.

  6. In the long term clouds can cause significant climate warming? How about significant climate cooling instead? Don’t clouds increase the Earth’s albedo? Yes, they do. This is clearly seen in Earthshine data.

  7. Dessler and Gore and the whole crew will say and do anything to try to show how theri soon to be defunct theory is still valid.

    Dessler, a 10 year time frame is to small a time frame, and the conclusions you reach isolate earth’s climatic system , not allowing for all the other many random, chaotic factors and other items that control earth’s climate to come into play.

    It is that magic bullet approach everyone wants to have for why the climate changes. Foolish to say the least.

    Some want cosmic rays more clouds to be the cause, some want co2 to be the cause, some want us to believe in false positive feedbacks ,just between one or two items in earth’s climatic system igoring everything else ,or at the very least trying to isolate these items from the everything else which one CAN’T do. It is all interconnected. End of story.

    The mainstream climate community is on the wrong path,in trying to find out how earth’s climatic system works and until anyone of them can EXPLAIN all the past abrupt climatic changes the earth has had and why, none of them have any business in trying to portray they now have the answers to earth’s climatic system.

    The phase in theory incorporates everything, and uses past history and is the best explanation for past abrupt climatic changes.

  8. Dessler ,will say and do anything to promote the false global man made co2 induced warming, despite all the evidence that is showing this is not so. That is why they keep changing their story on their climate predictions,why they keep trying to change past data(MANN’S HOCKEY STICK) ,and keep trying to use recent data to try to make it fit their theory(AEROSOLS RETARDNG WARMING THE LATEST BS). They will do and say anything to keep this alive.

    Now Dessler ,reaches this conclusion that clouds do not exert an influence on temperatures. What will they change next? They are trying to change everything we do know about climate ,to make it fit their ridiculous theory.

    They will learn this decade ,how foolish they have been.

  9. Dessler, for the record , more clouds probably cause colder conditions and less clouds warmer conditions.

  10. Scott Allan says:

    During the daytime more clouds tend to reduce temperature, but at night, at least in winter, more clouds tend to increase temps … at least at the surface

  11. P.Solar says:

    What Dessler (and many others in the field) fails to address is that you cannot just do ols regression on a signal with that kind of noise.

    It is a mathematical certainly that the ols *estimator* (it should not be called the slope) will be less than the true slope. This is known as regression attenuation. How severe it is depends upon the magnitude of the noise w.r.t. the linear signal. If one can estimate the variance of the noise and that of the true x values one can scale the ols estimator to recover a better estimate of the slope. Normally this is rarely possible since we don’t know the true x values , only observations. It may be possible that comparison with your simple model may help get such an estimate.

    The other condition is that the noise is not correlated to the signal. That may well be approximately the case at zero lag. I doubt it will be valid for the lag where there is a max correlation, so care must be taken since the feedback term although decorrelated will still have some correlation with the other part of the signal.

    That’s all I can fit into a blog post but I think there is a means to recover a better estimate and to at least partially correct the invalid slope derived by simple ols at lag zero.


    • Yes, this is an issue that is seldom if ever addressed.

      I probably spent a good 2 weeks looking into it, by using model data sets where I knew what the true slope was, then add noise to both variables, to see how well different regression methods work (which assume “errors” in both variables) versus OLS regression.

      The results were, rather surprisingly, mixed. It depends upon the source of the noise and its characteristics.

      For example, if the (non-feedback) cloud variations cause even HUGE temperature variations, but the cloud variations are truly random in time, with zero autocorrelation, you get a pretty accurate slope estimate with OLS even with a correlation of (say) 0.1.

      There are SO many issues with this whole business of estimating feedbacks (climate sensitivity) which have either been swept under the rug, or people are too lazy to investigate, that it’s difficult sometimes to know where to begin.

      We have WAY more stuff on this than we have time to publish (or, I should say, fight with the inevitable single hostile peer reviewer over).

  12. “These calculations show that clouds did not cause significant climate change over the last decade (over the decades or centuries relevant for long-term climate change, on the other hand, clouds can indeed cause significant warming).”

    Could Dessler not have said the following as well with regards to CO2:
    “These calculations show that CO2 did not cause significant climate change over the last decade (over the decades or centuries relevant for long-term climate change, on the other hand, CO2 can indeed cause significant warming).”

    This would be a marked departure from earlier claims that nothing is close to CO2.

    I believe the SB paper scored a home run, but at least in Dessler’s mind, it seems to have scored a double. However I wonder what Trenberth is saying in his private emails. Let me guess: “We cannot counter the brilliant science in the SB paper and it is a travesty that we can’t.”

    • No, actually, I’m pretty sure they really do think I’m an idiot.

      • Jon says:

        “No, actually, I’m pretty sure they really do think I’m an idiot.”

        This belief doesn’t, on its face, appear to be compatible with your often stated belief that your research is being suppressed for political reasons. If your opponents honestly believe you’re an idiot and that your output reflects that, how are their actions improper? You aren’t exactly shy about saying so when you think someone is talking rubbish, Dr. Spencer. Why should they act any differently?

  13. Arthur Smith says:

    Ok, looking at both your sets of numbers, Dessler claims that the mixed layer standard deviation monthly anomaly is 9 W/m^2, based on 2 sources (MERRA for temp change + LC11 for heat capacity, and ARGO/Douglass and Knox for heat content) while you claim based on Levitus (what is the citation? The reference I found did not provide monthly data nor 3-month standard deviations, and was not for just a 100-meter mixed layer) it is just 2.3 W/m^2. That is a factor of almost 4, and explains most of the factor of 10 between your two ratios.

    The other discrepancy is the CERES radiative measurements on your end (0.56 W/m^2) vs Dessler’s own 2010 calculations that arrive at 0.5 W/m^2. This doesn’t seem a big discrepancy – except that you have to add other causes of radiative forcing change besides clouds to the CERES analysis. Your calculation here seems to have been to simply add the magnitudes of the variation together, but that’s not a valid statistical analysis – the magnitude of the standard deviation of the sum of two independent variables is not the same as the sum of the two separate standard deviations! And it is *certainly* not the difference of the two, if subtraction is involved!!!

    That is, if cloud variation causes 0.5 W/m^2 of standard deviation, and lambda = 3 that means the temperature response contributes 0.23 W/m^2 standard deviation, so the total expected standard deviation of the radiative term CERES would see is the square root of the sum of the squares of the two, i.e. 0.55 W/m^2 in this case (assuming the two terms are independently varying).

    So Dessler’s 0.5 W/m^2 is perfectly compatible with the CERES value. The only real discrepancy seems to be the ocean heat content term.

    • Yes, I agree, it seems I the big difference is in the mixed layer heat content. And, yes, using standard deviations is not meant to be exact…Dessler is trying to estimate the relative sizes of terms in the equation.

      One fact of interest is that the standard deviation of the 100 m layer temperature variations are a little more than 50% of the surface temperature variations.

      One thing he might have done wrong is that, to be physically consistent with what is going on, you don’t use the standard deviation of the mixed layer temperature itself, but of its time rate of change. At each three month time step, a new set of energy imbalances changes the mixed layer temperature from its previous value, to a new value. This makes a significant difference to the results.

      There is a standard Levitus 3-month dataset of ocean temperatures, which during the 2000-2010 period would be mostly ARGO floats. It’s a 4-dimensional dataset of temperature anomalies you must download and do your own averaging from.

      I’m considering putting all of the 3-month anomalies for all the data sources in a spread sheet and putting it on my website for people to play with.

  14. Elsabio says:

    A fact of life in the sun, and a simple question for all you scientific geniuses.

    Where I live, we currently enjoy mostly clear skies and balmy, 30ºC+ daytime temperatures. As pleasant a situation as this is, any sort of physical labour in the full glare of a midday sun is pretty much out of the question, unless, of course, a big, friendly cloud happens to drift overhead. During any such event, though the ambient temperature doesn’t immediately change (it takes hours for concrete slabs and/or walls to cool), the direct heat from the sun drops significantly and we put down whatever cool beverage we happen to be sipping and dash outside to attend to wilting vegetables or parched pot-plants – it’s a hard life but someone has to live it…. Now, my simple question: what happens to that blistering blast of heat as the cloud passes? Is it retained by the passing cloud (to be released into the atmosphere when the cloud eventually dissipates), or is it reflected back toward its source (back into space)?

    I thank you.

    • Barry, from the title of your blog post, I must ask:

      Do you really believe what you are writing?

      Did you even understand the post you are criticizing?

      • I think so, but I guess I’ll never know if you don’t break down and point out the errors.

      • I do know, however, that on some points you have really dodged and weaved to avoid the real point of Dessler’s (and Trenberth and Fasullo’s) arguments. E.g., you don’t address the point that the deviation of the models from the data doesn’t seem to have much to do with climate sensitivity. It seems to have more to do with how well they mimic El Niño. Since ENSO oscillates over the short term, a model might do very poorly at that, and still be right on the money with respect to climate sensitivity. Acknowledging that this is the real argument would clear away about half of what you said above.

  15. Arthur Smith says:

    Not having access to the Levitus data right now, it seems to me aside from your explanation the discrepancy may possibly be due to either:

    (1) The temperature variations being analyzed in the Levitus case is for a significantly different ocean depth (700 meters vs 100 meters?) to what Dessler was looking at

    (2) Three-month variations are much smaller than 1-month variations for the mixed layer.

    On the standard deviation business – you agree, right, that std. dev(N) should be *LESS* than std. dev(CERES) in your analysis, since the variations in CERES include both N and other presumably independent sources of variation? Yet you seem to have gotten std. dev(N) larger than std. dev(CERES) by almost a factor of 2 – how?

  16. P. Solar says:

    re. OLS.

    Yes error in the dependant variable is a whole other issue that is ubiquitously ignored but what I was referring to here is regression attenuation by significant noise in the dependant variable. This is primarily what gives the lower correlation, that’s why it’s also referred to as dilution. Since , for mean zero data, the ols estimator is just a scaled version of the correlation , if the data is decorrelated ols regression no longer gives an indication of the slope. That is precisely what is occurring in all these naive, inappropriate attempts to get CS from a simple ols.

    It’s not just that its technically sloppy, it *will* give a gravely false result (in the direction of zero slope) . Most people seem to think “yeah , well it’s near enough” without informing themselves on how it does deviate.

    The simple model suggests the signal mix is about 30%/70%, that is backed up by the form of the empirical data.

    I found that correcting the ols estimator using those magnitudes as the variance of the signal and the noise as I indicated above gave a reasonably good estimate of the original f/b in the model:
    ols * (rad_coeff/nonrad_coeff+1)

    I got 4.5 to 5.5 with a f/b of 6.0 , not the “right” value but a descent estimate. Better then 1.5 to 2.5 , say.

    This was quite robust to different ocean depths and f/b parameter values.

    Doing a similar operation with the axes inverted would at least bound the min and max range for the slope estimator. One could then discuss if things like the bisector of the two had any merit but at least we would have a fairly close range to constrain the location of the Holy Grail. 😉

    The key fact to recognise here is that these two signals are orthogonal, since one if f(T) the other is g(dT/dt) . That, in principal allows a means of differentiating between the two and isolating the linear signal.

    Obviously the real data is not that simple , but the closeness in form of your lag plot to that of the model gives good reason to expect a robust result.


  17. Christopher Game says:

    This debate continues to cripple itself by using an inadequate formalism. The questions cannot be safely resolved until a second dynamically determined internal state variable is introduced and some information is added by use of an external driver. Christopher Game

  18. Christopher Game says:

    It remains the case as before, that Drs Spencer and Braswell have adequately and reliably demonstrated a large discrepancy between the observed measurements and the behaviour of the IPCC AOGCMs. The IPCC AOGCMs must be wrong in a big way. The exact way in which this arises is a secondary question, and it is this further question the answer to which calls for a second dynamically determined internal state variable. It is a pity that this secondary question is clouding appreciation of the far more important primary point that Drs Spencer and Braswell have demonstrated a large discrepancy between the observed measurements and the IPCC AOGCMs. Christopher Game

  19. Stephen Wilde says:

    External driver – The variations in solar activity somehow (probably chemical means involving ozone but maybe cosmic ray quantities too) altering the vertical temperature profile of the atmosphere.

    Another internal state variable – The speed of the water cycle via latitudinal shifting of the surface air pressure diatribution.

  20. Christopher Game says:

    Response to the post of Stephen Wilde of September 7, 2011 at 4:07 PM.

    Stephen Wilde proposes, quite reasonably, variations in solar activity as an external driver.

    There is a very simple way to get external driver information for the periodic component with period one year. It is routine to deliberately discard this simple information in this game, on the grounds that one is looking for changes on a climate time scale. Then one goes ahead and looks at a residual process and looks at time lags on the order of a few months. This is not consistent with the stated aim of looking at time lags on the climate time scale. But the time scale of a one-year period is quite near that of a few months which is being examined. There is no reason that I know of that there should be a massive jump in the response spectrum between a few months and the annual time scale. Christopher Game

  21. TomRude says:

    @ Barry Bickmore
    “It seems to have more to do with how well they mimic El Niño. Since ENSO oscillates over the short term, a model might do very poorly at that, and still be right on the money with respect to climate sensitivity. ”

    Do you know what El Nino represents as far as the general circulation is concerned? Are you suggesting that a model that fails to describe a significant dilatation of the northern hemisphere circulation would still be worthy of describing climatic evolution?

    • Why don’t you ask Roy, Tom? He uses a zero-dimensional (sometimes 1-D) model to draw conclusions about climatic evolution. And that might be ok for some purposes. In any case, all models gloss over some details. The trick is not to gloss over details that are important for the particular question you want to answer. If you want to answer questions about what goes on a multi-decadal timescale, all the El Niño stuff would average out, anyway. If you want to compare models to a single decade of data, where El Niño is very important, maybe a zero-D model isn’t going to cut it, and many more complicated models wouldn’t, either. It depends on how good they are at simulating El Niño.

  22. Steve Short says:

    As a relatively unskilled bystander (PhD geochemistry) the following looks like possible (reasonable?) evidence for the role of clouds in warming/cooling over decadal timescales to me:

    According to Pinker et al., 2005, surface solar irradiance increased by an average 0.16 W/m^2/year over the 19 year period 1983 – 2001 inclusive or some 3.0 W/m^2 over the entire period.

    This change in surface solar irradiance over 1983 – 2001 is about 1.3% of the mean total surface solar irradiance of the more recent 2000 – 2004 CERES period of 239.6 W/m^2 for which the mean Bond albedo has been claimed to be 0.298 and mean surface albedo to be 0.067 (Trenberth, Fasullo and Kiehl, 2009).

    Inspection of the ISCCP/GISS/NASA record for satellite-based cloud cover determinations suggests a mean global cloud cover over the 2000 – 2004 CERES period of about 65.4% and over the entire 1983 – 2008 26-year period the stated mean is about 66.4±1.5% (±1 sigma).

    ISCCP/FD and Earthshine albedo data for the 2000 – 2004 period enables estimation of the relationship between albedo and total cloud cover and it is best described by the simple relationship:

    Bond albedo (A) ~ 0.353C + 0.067 where C = cloud cover. The 0.067 term represents the surface SW reflection (albedo) as quoted by TFK09. For example, for all of 2000 – 2004; A (TFK09) = 0.298 = 0.353 x 0.654 + 0.067

    According to ISCCP/GISS/NASA, mean global cloud cover declined from about 0.677 (67.7%) in 1983 to about 0.649 (64.9%) in 2001 or a decline of 0.028 (2.8%).

    This means that in 1983; A ~ 0.353 x 0.677 + 0.067 = 0.305

    and in 2001; A ~ 0.353 x 0.649 + 0.067 = 0.296

    Thus in 1983; 1 – A ~ 1 – 0.305 = 0.695

    and in 2001; 1 – A ~ 1 – 0.296 = 0.704

    Therefore, between 1983 and 2001, the known reduction in the Earth’s albedo A as measured by ISCCP/GISS/NASA should presumably have increased total surface solar irradiance by ~200 x [(0.704 – 0.695)/(0.704 + 0.695)]% = 200 x (0.009/1.399)% = 1.3%

    This estimate of ~1.3% increase in solar irradiance from cloud cover reduction over the 19 year period 1983 – 2001 is obviously ‘close’ to the ~1.3% increase in solar irradiance measured by Pinker et al (2005) for the same period. Coincidental?

    We know the period 1983 – 2001 was a period of claimed significant global (surface) warming.

    However, within the likely precision of the available data for the above exercise (maybe of the order of say ±0.5% at ± 2 sigma?), it may also be concluded that it is easily possible that the finding of Pinker et al (2005) regarding the increase in surface solar irradiance over that period was due to an almost exactly equivalent decrease in Earth’s Bond albedo resulting from mean global cloud cover reduction.

  23. Stephen Wilde says:

    It seems to be a little more complicated than clouds simply being a negative or positive feedback because latitudinal cloud distribution is also very important and the oceanic response to cloudiness changes confounds the initial expectation.

    What I think happens is that for whatever reason the atmosphere expands when the sun is active and contracts when it is inactive.

    In the process the temperature of the stratosphere and mesosphere changes oppositely to the sign of the temperature change in thermosphere and troposphere.

    The effect is to draw the tropopause upward when the sun is active and push it down when the sun is less active. Globally averaged of course.

    The outcome is latitudinal shifting of all the components of the surface air pressure distribution which changes the sizes and positions of the climate zones.

    That changes the energy budget via the speed of the water cycle AND cloud quantities because that process changes the length of the air mass boundaries which is where mixing occurs to produce clouds.

    So an active sun tries to COOL the system by changing the structure of the atmosphere to let energy OUT of the system FASTER via the higher tropopause but in the process cloud bands are drawn poleward to let more energy into the oceans in the tropics which offsets the faster energy loss to space.

    So the cloud changes provide an indirect negative (warming) response to counter the direct solar cooling effect via the coolr stratosphere and mesosphere.

    The position regarding bottom up effects from periodically faster energy release from the oceans or more energy in the air from more GHGs is different. In that case the extra warmth at lower levels pushes the tropopause up as before and in that case the increased energy into the oceans is a positive feedback. However the poleward shift of the surface air pressure systems accelerates the speed of energy transfer to space which is a negative response sufficient to cancel out both the extra energy from the oceans or GHGs AND the extra solar energy into the system.

    Thus whatever changes the vertical temperature profile of the atmosphere from above or below will cause cloudiness changes that then exert a negative response either by adjusting energy flow into the oceans or by adjusting energy flow out to space as necessary to maintain equilibrium and what we then experience is shifting climate zones as the speed of energy flow through the system varies.

    That is a neat solution to the problem.

    It is the vertical temperature profile of the atmosphere that is key whether caused by top down solar effects or bottom up oceanic or GHG effects because that then causes the cloudiness changes.

    It sounds complex and it is but it is no more complex than it needs to be to fit observations.

  24. Stephen Wilde says:

    “Christopher Game says:

    September 7, 2011 at 4:16 PM

    Response to the post of Stephen Wilde of September 7, 2011 at 4:07 PM.

    Stephen Wilde proposes, quite reasonably, variations in solar activity as an external driver.”

    Just to clarify. The solar changes needed to alter the vertical temperature profile of the atmosphere appear to operate over centuries such as from 1600 to date and most likely from MWP to LIA.

    There is a lot of variability along the way and the pattern is often overridden by internal system variability.

    The best evidence we currently have is the fact that the stratosphere stopped cooling in the late 90s as the sun started to come down from the activity peak of cycle 23.

  25. Ivan says:

    Barrry says

    “Since ENSO oscillates over the short term, a model might do very poorly at that, and still be right on the money with respect to climate sensitivity.”

    No, it couldn’t. The problem is that the periods of warming are characterized by more frequent and stronger El Ninos, while the cooling periods have stronger and more frequent La Ninas. Hence, the models that do not represent reasonably well these El Nino and La Nina processes in the short run are useless for analyzing global warming and cooling in the long run.

    • Windchaser says:


      That’s a non-sequitur. Warming periods may have stronger El Nino cycles, but that doesn’t mean that stronger El Nino cycles *drive* the warming. If ENSO doesn’t drive climate change, but is just itself affected by warming/cooling, then including ENSO in long-term models is not necessary for the purpose of reliable temperature estimates.

  26. Christopher Game says:

    Dr Spencer writes: “at least beginning to talk the same language in this debate.”

    Besides the formalism problem, the debate continues to cripple itself by talking carelessly and ambiguously. It talks about how “clouds cause climate temperature change” and how “climate change causes cloud change”. Willy nilly, this has an effect of setting up an unnecessarily polemical debate about ‘which one is it: A->B or B->A?’ when we all know that both apply. Better to use more focused language such as ‘A has an effect on B and B has an effect on A’, and other such locutions that avoid the provocative and ambiguous wording ‘A causes B’ when in context both A and B are internal state variables which have dynamic effects on each other. Christopher Game

  27. This is one of those issues that is NEVER going to go away because one can make it come out the way he/she wants it to come out ,as is being shown on the board.

    The way I see it is, the greater amounts of low clouds ,which tend to increase during low solar activity periods, due to a more -AO, an increase of volcanic activity and perhaps an increase in cosmic rays, the lower the temperatures will be.

    Steve is right in that when the sun is less active the stratosphere/mesosphere warm while the the troposhere/thermosphere cool , and the atmosphere as a whole contracts, pushing the jet stream further south ,which is part of the overall cooling associated with a less active sun.

    The clouds in this situation over the lower latitudes would tend to INCREASE, which cool the oceans (NEGATIVE FEEDBACK),helping to enhance a more meridional atmospheric circulation.

    The more meridional circulation being the result of the ozone distribution/concentration change in the stratopshere , due to the low solar activity,which causes the warming of the stratosphere as a whole, but probably warming the stratosphere to a greter degree in the higher latitudes ,as oppossed to the lower latitudes, enhancing a more -AO.

    This has been the case in this current solar minimum ,and in the Dalton Minimum and the Maunder Minimum.

    Dessler, is clueless when it comes to this part of what effects earth’s climatic system, therefore all of his arguments are based on incomplete ,subjective data.

    He can use all the fancy math/models he wants, to try to justify his position, but past history certainly does not support what he is saying, in addition to him not being able to prove how the positive feedback between a temperature increase/decrease and clouds work. It is the same problem these guys have with the positive feedback between CO2 and water vapor ,they can’t show it or prove it.

    This decade due to the prolong minimum solar activity and all the items it will impact that REALLY control earth’s climate, will be the decade of temperature decline,climatic extremes, and increased geological activity, which will play havoc with earth’s climatic system. Mark my words on that last point.


  28. MikeN says:

    Yes it’s 3003, not 2002. I used 11x13x7=1001 but slipped up and took the residual to be 14/7=2 instead of 9/3=3.

  29. Dale says:

    Dr Spencer, thank you for keeping this up. I have no background in climate (I’m in IT) but reading Dessler’11 it didn’t sound right to me.

    In my really simplified thoughts on the relationship between clouds and temperature, if it’s a sunny day you feel the sun’s warmth. When a cloud passes in front of the sun you no longer feel that warmth. Where did it go? Reflected back out.

    Thinking of it large scale over time, if the surface warms more clouds are created (evaporation). These clouds block more of the sun’s heat which then cools the surface down. Obviously it isn’t going to be instantaneous as the acquired heat in the surface needs time to dissipate. But the surface is losing heat without the sun replenishing it. Overall net effect is that more clouds reduce the surface temp.

    So Dessler’s claim that clouds can’t influence temperature just doesn’t sound right. And to say that clouds can actually cause global warming in the long time just sounds ludicrous to me.

    Though it is quite possible I’m completely wrong and just a complete fool. 🙂

  30. WheelsOC says:

    Having read through SB2011, but not being a scientist myself, I’ve got some things that need clearing up before I understand them.

    1) What prompted the use sensitivity to transient CO2 concentration as a criterion for selecting which models to present, if the same analysis was performed on 14 models? I thought the issue under examination in the study was a comparison between the roles of ocean heat flux (non-radiative) and the effect of clouds (radiative) as forcing. Indeed, towards the end it’s said that the temperature trend over the period studied was dominated by ENSO and sensitivity to CO2 isn’t an obvious factor then. Unless this paper was supposed to indicate a problem with the accepted sensitivity to CO2 forcing (and I don’t see that anywhere in it), it just seems irrelevant.

    2) If I’m reading the paper right and the same analysis was performed for all 14 climate models, and now I see that some of them perform much better at reproducing the real-world data than the six originally graphed, aren’t those the models that should be used to compare against the simple forcing-feedback model? Since they have different assumptions about what role radiative forcing by clouds can play, wouldn’t that be more demonstrative?

  31. KAP says:

    You mentioned that the temperature variations at the 100-meter level are about half of the surface. Which value did you use, or did you split the difference?

    I can see how the 100-meter temperature variations can be less than the surface, but then you run into another problem, which is in that case, the mixing layer isn’t really 100 (or 25) meters deep as assumed. That’s because the definition of the mixing layer is the depth at which the temperature is 0.5° C < SST. That in turn constrains the bottom of the mixing layer to move (temperature-wise) in lockstep with SST. So if the SST is has, let's say, a diurnal variation that's stronger than at 100m, that implies that the mixing layer depth shares a diurnal variation too.

    In an ideal world, we should model the change in mixing layer depth, but that seems unnecessarily complex: it's much easier (and not wrong) to just take a mean depth for purposes of computing Cp. But for purposes of computing dT/dt, going with the 100-meter sigma would lead to an underestimate, since the *real* mixing layer must share dT/dt with the surface values.

    Regarding your speculation that Dessler might have taken the standard deviation of the temp rather than of the time change of the temp, this seems not to be the case; he says, "The time derivative is estimated by subtracting each month's global average ocean surface temperature from the previous month's value." This is similar to what you're doing, but you're using 3-month timesteps and he's using 1-month timesteps. And this, I think, is one of the major differences between the values obtained. Although the difference in temperature is certainly likely to be greater after 3 months than after 1 month, it's not likely to be three times greater, and since time is a factor in (the denominator of) Cp, going out three months would lower the LHS by a third if the temp changes were the same.

    As for the remainder of the difference, I'm rather skeptical of the mixing layer depth results you obtained by regression of the 5-month means (why 5 months?) in dT/dt vs. radiation, which seem low by a factor of 4-ish, both from models and from data. My gut feeling is that there's a systematic error in there somewhere, but not having the details of the computation, I can't say anything for sure. In any case, you and Dessler seem to be on the same page now with 100 meters, which is a step in the right direction.

  32. ” (over the decades or centuries relevant for long-term climate change, on the other hand, clouds can indeed cause significant warming).”

    Long term climate change can be caused by clouds??! Well, maybe Andy is finally seeing the light!”

    I wonder if it occurred to Andy Dessler that you don’t have to attend too many logic classes to realise that if clouds can cause significant warming, they can also cause significant cooling.

    ISCCP data shows a drop in tropical low cloud cover 1980-1998 and an increase after. This empirical fact is contemporaneous with warming 1980-1998 and a flattening of temperature variation after a short lag, plus flat to falling ocean heat content from 2005 (lagging the cloud increase and solar slowdown).

    When is the IPCC science team going to start regaining a little credibility by discussing empirical facts rather than constantly reiterating tired talking points around models and co2?

  33. Ask why is it so? says:

    Dr Roy, first I’d like to say that this is all getting a bit nasty. Regardless of whether your paper is right or not the scientific world seems to take pleasure in finding errors that may or may not be there. I am disappointed and disillusioned by these current events. My advice, you need to go back to the beginning. Any idiot knows that its cooler standing under a tree than it is in the sun. The starting point of every model is the current mean temperature of the earth and because this figure is unknown creating climate models to predict future temperature warming or cooling is just a waste of time. Looking through old papers written in the 70’s and 80’s I find the mean temperature then was 15oC, the same as today. In science everything is given to interpretation and bias. It’s also obvious that the sun is the driver of our temperature, the maximum temperature achieved per day and the atmosphere and the surface of the earth simply allocate it. There is nothing on the surface of the planet or in our atmosphere that can absorb the suns energy and increase its power to produce a higher temperature. Conversion of solar radiation by varying the absorption ability of the planet will alter the temperature. What we are all trying to do is find out why the earth is getting warmer, look out your window at what was once a green field and it’s now a high rise. IF we change the landscape it must affect the temperature. I could go on and I know you will all say I’ve got no idea what I’m talking about but like all of you I’m entitled to theorize (that’s science talk for opinion).

  34. The website ,has a great paper by Bob Tisdale, which shows HOW FAR OFF ,the models have been in predicting the ocean heat content going forward.

    I think this latest error will have to be added on the ICECAP.COM WEBSITE, UNDER GREENHOUSE SCORECARD.

    That will bring the model total BLUNDERS up to 28 out, of 32 predictions. With just one prediction correct, 33 neutral.

    Really good predicting.

    In the meantime I will be waiting (probably in vain) to an alternative theory, to the phase in theory ,to explain ALL of the abrupt climatic changes, earth has had in the past.

    Come on ,know it all global warmers ,give us your explanations.

    The phase in theory is beyond a doubt, the best explanation out there for abrupt climatic changes, and it has NOTHING to do with CO2.

  35. correction 3 predictions are neutral ,not 33.

  36. Arthur Smith says:

    I downloaded the heat content data for the top 700 meters of the ocean from here:

    and for the period 2005-2011 (and some other periods) looked at the RMS variation from one 3-month period to the next in global ocean heat content they find. This came to about 9×10^21 J; given that earth’s surface area times 3 months is about 4×10^21 m^2 s, that gives a magnitude that agrees with Dr. Spencer’s number of about 2.3 W/m^2. So I’m guessing this is the data source for the numbers stated above?

    However, note this is *not* the mixed layer heat transfer rate, this is for the entire top 700 meters of the ocean. I had thought the “S” term in the equation, Dessler’s delta F_ocean, refers to transfer from the lower ocean into the mixed layer. If you sum up the total heat content of both the mixed layer and lower layers, then you would get a much smaller number. So maybe we need some clarification on what this “S” or “delta F_ocean” is supposed to mean? Is this clearly explained anywhere?

    [By the way, whenever I try to Reply to a comment here the “Submit Comment” button does not work in my browser (Firefox 6 on a Mac). Only the main “Submit Comment” box at the bottom of the page works. Sorry for continuing to create new threads this way.]

  37. OHC has leveled off. Bob Tisdale, gives an excellent explanation as to why it rose and why the OHC has leveled off.

  38. Kasuha says:

    I’m afraid I’m too low on the list of comments to get a reply but anyway…
    I believe the argument of many of the critics of “which three models should be selected for comparison” is that you probably might have tried the three models which match observations best. Honestly I am also curious how such graph would end up.
    But the question is, what exactly is such comparison good for. In interpretations of IPCC results, mean of all models is usually used regardless whether they simulate ENSO or not because results of individual models go all around the place and the mean is the only thing which seems to behave. So comparison of these three models’ results with other aspects of real world data might give us some really interesting results too.

  39. Kasuha, your thoughts are as good as anyone else on this board.

    Just say what you honestly think. That is what I do, and time will tell, if you are correct or not.

  40. Svend Ferdinandsen says:

    It is more a suggestion than a remark.
    Could you once make an article about those climate models.

    I wonder why 14 models should be nescessary, especially when they seems to be used randomly.
    Are some models better to special parts of the climate than others or special parts of the globe?

    Or is the meaning that you pick the model that best show your point?

    By the way, the habit to use ensemble runs with different models and different starting conditions is to me a sign that the models do not do it right. I can see the point in local weatherforecasts for some weeks, but for Earths climate several years ahead, they should all stabilize to the same if they worked.

  41. boballab says:


    Steve McIntyre has a new post out showing what you get if Dr. Dessler had used Ceres Clear Sky instead of ERA and compared it to HadCrut using Dr. Dessler’s techniques from the Dessler 2010 paper:

    He got a surprising result: Using Dessler’s technique and those datasets the slope reverses from 0.54 +- 0.94 w/m2/K to -0.96 +- 0.98 w/m2/K.

  42. Just another example, of why one must not take what Dessler, has to say seriously.

  43. “UPDATE: I have been contacted by Andy Dessler, who is now examining my calculations, and we are working to resolve a remaining difference there.”

    So Andy Dessler is happy to have Roy on hand to wipe the bottoms of the reviewers who let him down so badly. That’s mighty grand of him. 🙂

  44. Andy Dessler says:
    I’ll be changing the sentence “over the decades or centuries relevant for long-term climate change, on the other hand, clouds can indeed cause significant warming” to make it clear that I’m talking about cloud feedbacks doing the action here, not cloud forcing.”

    Lol. Cloud feedbacks during centuries which are feeding back to what forcing? Not co2, which allegedly bimbled along at a nice steady 270ppm.

    So, if the feedback was causing warming, can’t have been volcanos… must be solar then? But if that’s the case, then why would clouds have stopped feeding back to solar variation when co2 started to increase? Hmmm? 🙂

    Maybe Andy Dessler should read Nir Shaviv’s JGR paper ‘Using the oceans as a calorimeter’.

  45. Steve says:

    Could it be Dessler sees writing on the wall and is switching sides… hmm….

  46. glacierman says:

    Update – looks like the backtracking has begun. Will there be any apology from others like Trenberth. He is only used to getting personal apologies I guess.

  47. Jack says:

    Good to see Dr(s) Spencer and Dessler communicating politely and discussing there papers. Lets hope it’s a sign of things to come. Maybe the resident Gadfly (Bickmore) could learn from this discourse.

    • MarkB says:

      Gadflies are good – they keep the discussion lively. Nothing more boring than to read posts where everybody agrees. Besides, I prefer Barry’s posts to Obscurity’s any day. 🙂

    • Yes I agree. I think Roy’s humorous and cheerful response to Dessler’s rebuttal paper has broken some ice here. If any progress is to be made, the era of talking past each other has to end. Testing ideas with several datasets and accepting the successes and failures openly has to be the way forward.

  48. TomRude says:

    @ Dr. Roy Spencer
    So your update suggests that you are actually helping Dessler with his paper! How considerate of you! How convenient that Dessler’s tone won’t then be in the record…
    I wonder if the Team will ever extend the same courtesy especially after other members went flat out discrediting your work and attacking your character!

    Here is Tuco’s advice: “when you have to shoot, shoot, don’t talk!”

  49. Maybe there will be a great outcome here for Dr Roy and the sequel post will be ‘A Fistful of Dollars’.

    Round ’em up Roy 🙂

  50. Pehr Bjornbom says:

    Dr Spencer,

    I tried a simple calculation for the period 2007 – 2009 including La Nina.

    From figure 2 in your Remote Sensing paper I calculated the radiative forcing by taking the (- Net) radiation flux according to b: and subtracting the feedback calculatet by ??Tsfc (temperature from a.).

    With ? = 6 the radiative forcing total variation was 3.5 W/m2 during this period.

    With ? = 1 the variation was 2 W/m2.

    Hence the effect on temperature by this radiative forcing is not possible to neglect during this period with a La Nina.

    A likely reason for those changes in radiative forcing could be changes in cloud cover and cloud albedo due to the changes in the state of the climate system during a La Nina event.

    Furthermore the regression method used by Dessler in Science 2010 cannot give correct result for the feedback parameter since such a method is only valid when changes in radiative forcing may be neglected.

  51. In order to reach a definitive conclusion one way or the other one needs complete data, with no unknowns. Those two factors will not happen, therefore only time will prove who is right(Dr. Spencer) and who is wrong (Andy Dessler)

  52. Pehr Bjornbom says:

    Salvatore del prete,

    I have also plotted the phase plane curve for this period.

    In the phase plane there are points with the same temperature with a distance of 2 W/m2 in the y-coordinate. This means that the difference in radiative forcing between those two points is 2 W/m2.

    Hence the minimum variation in radiative forcing is 2 W/m2 but the variation could be 3,5 W/m2 if the feedback parameter ? = 6.

  53. Arthur Smith says:

    Dr. Spencer,

    I see you have updated your text in several ways, but you have *not* corrected the silly error you made when you stated:

    “I assumed a feedback parameter ?=3 Watts per sq. meter per degree, that 20:1 ratio Dessler gets becomes 2.2:1. If I use a feedback parameter of ?=6, then the ratio becomes 1.7:1. ”

    Given your numbers of 2.3 W/m^2 for ocean heat content variation and 0.56 W/m^2 for CERES radiative forcing changes, the ratio in question cannot be less than 2.3/0.56, i.e. 4.1. That is because the standard deviation of cloud radiative forcing must be *less* than the standard deviation of all radiative forcing, no matter the source.

    So you must have made a mistake in calculating 2.2 or 1.7 for the ratio. You really need to take a look at this, it’s rather central to your claims here; even if the ratio is not 20:1, 4:1 or 5:1 is enough to cast significant doubt on the capacity of the cloud changes to be causative of ocean heat content changes.

  54. John Keye says:

    The frenetic pursuit of filthy lucre and metaphysical satisfaction by the idolatrous obsession with rising atmospheric CO2 as the certain cause of global warming i.e. algorism, is fading fast.
    Can’t we now focus on the New Theory of Climate Change just confirmed by CERN’s CLOUD experiment (Svensmark’s research).
    The temptation to explain cyclical warming and cooling via solar minimums/maximums and galactic ray ionization of H2So4 + water + NH3 to form cloud droplets is compelling and needs to be incorporated into the complex mix of oceanic current fluctuations, temperatures at varying depths, volcanic atmospheric debri, etc. At least this seems important to this retired-no-nothing-octagenerian pathologist who got his sea legs in 1939 as cabin boy on the schooner E. W. Scripp’s 3 month oceanographic expedition in Baja California led by Roger Revelle. John Keye, Reddin, CA

  55. pochas says:

    Dr Spencer,

    I believe you have indeed shown there is a negative feedback from something in the air (water vapor, clouds, aerosols, red balloons, etc.). You might possibly be able to take the analysis further, even calculating a feedback ‘sensitivity’ for the ghostly vapor by calculating an amplitude ratio and going from there to a sensitivity.

    We recently went through this over at Niche Modeling

    The math for this is discussed in Process Control Texts. Look for “Open Loop First Order Systems with Sinusoidal Input.” The analytical solution for your differential equation will be there.

    Anyway, you would get the amplitude ratio by converting outgoing flux to a temperature by dividing by 3, finding its absolute amplitude, doing the same for surface temperature, then comparing the two as an amplitude ratio.

  56. JJ says:

    Dr. Spencer:

    Next time, consider not responding to the preprint. Let the arrogant @#$% and his dishonest, disrespectful teammates publish their rush to poor judgement, then let them take their lumps in the literature.

    Pissants who whine about you ‘forcing others to correct your serial mistakes’ do not deserve your patient assistance in keeping their agenda driven incompetance and shoddiness off the record.


    • Dessler is not Trenberth, and the preprint, the sceptical part of the blogosphere’s reaction to it, and the backpedalling are firmly on the record.

      Time to move forward. Quickly.

  57. fido says:

    Dessler states he will change sentences. But he is not allowed at this stage, without the paper going back to reviewers…

  58. Bart says:

    Is there anyone with a voice in these matters who could just do a cross spectral estimation of the two series and read off the phase relationship directly?

    If you have an input x(t) and an output y(t), the cross spectrum is the expected value of X'(f)*Y(f), where X(f) and Y(f) are the Fourier transforms and “‘” denotes the complex conjugate. Since Y(f) = X(f)*H(f), where H(f) is the transfer function, the cross spectrum is C(f) = P(f)*H(f), where P(f) is the power spectral density of x(t). Since P(f) is a zero phase, positive real function, the phase response of C(f) is identical to that of H(f). If it is near 180 degrees at low frequency, then you’ve got negative feedback. If near zero, it is positive.

    The dispersion in the phase plane makes it very difficult to diagnose the sign of the feedback – the nonlinear phase characteristic of H(f) ensures that you will get a messy cloud unless you focus on a particular frequency range and filter everything else out. For sure, the cross spectrum will be polluted with other processes and noise, too, but I expect it should be a pretty good indicator of the dominant process occurring, if there is one.

    Caveat: estimation of the cross spectrum is not as easy as multiplying the FFTs of the two data streams. The outcome of that will be highly erratic and of questionable value. A good method is to do that, then inverse transform to get the impulse response. Window that impulse response with a good window (e.g., Hamming or other) at a truncated limit before it visibly starts breaking up, then transform the windowed impulse response function back to get the estimate of C(f). There are, of course, other reasonable methods. A mathematical analysis tool like MATLAB has its own built-in functions for doing it.

  59. Conrad Dunkerson says:

    Dr. Spencer, you vehemently object to the statement in the Dessler draft that you used, “the particular observational data set that provided maximum support for their hypothesis”… but then use just that same single data set, the satellite observations shown in green in the graph above, to ‘refute’ this.

    I believe the point he was making is that other observational data sets show a closer correlation to the model runs than the one you chose. Indeed, some of the model runs you did not graph correlate very closely with observational data sets you did not include. This can be seen in the graph at:

    Thus, it can be seen that an examination of a larger pool of models and observational data shows less divergence (overall) than your original presentation. Whether this constitutes ‘cherry picking’ or ‘incomplete information’ is largely a subjective evaluation and not relevant to the scientific conclusions. Given that looking at certain observational data and model runs shows very close correlation while looking at others shows wide divergence it seems clear that conclusions should not be drawn unless strong reasons for identifying a particular subset as ‘most accurate’ can be shown. Logically, that would be unlikely to include either the high and low outliers of the model runs or the high outlier of the observational data sets.


    This I believe is the best theory out there to explain climatic change and abrupt climatic changes. It incorporates all the factors, and does not try to have a silver bullet, like so many of the other theories have.

    PHASE IN THEROY INCLUDES THESE FACTORS- which also vary within themselves as to the degree of magnitude they phase into a cold/warm mode, and duration of time they phase into that mode. If degree of magnitude is STRONG enough ,LONG enough, certain THRESHOLDS may be reached, result an abrupt climatic change.













    How anyone can argue against this type of theory ,or come up with a better theory for climate changes, is beyond me.

    I say this is it, and past history supports this ,and I feel this decade wiLl suPport this, if the sun can stay in this prolong minimum state. It takes time, due to the OHC for the temperatures to respond, although the extremes in climate are much quicker to respond.

    This is the correct path, why/how the mainstream can’t see this is beyond me, it makes so much SENSE.

  61. Richard Saumarez says:

    I was surprised to see on Climate Audit that a rerun of Dessler’s analysis usinh Hadcrutt data gave results very close to that of Spenser’s.

    Is this cherry picking data sets or is there a good reason for using one set over anthor?

    If there are such substantial differences between different data sets, what does this tell us about their reliability as model inputs?

  62. Exactly, one can get any result they want when they concentrate on ,isolated items of earth’s climatic system. You can make it come out the way you want, NOT the way it is.

    That is the whole problem, and why I am so disgusted with the mainstream climatic research. They try to have the silve bullet, and then try to make the data make the silver bullet appear to be correct.


    There is no silver bullet, it is a combination of many of the items I mentioned above, that no climatic model can have incorporated into it’s data bank,in order to try to come up with a result, for the climate.

    It can’t be done in a complete ,accurate way.

    That is my whole argument for past history, and the phase in theory ,and why the climatic models,climate projections will forever be USELESS.

  63. Massimo PORZIO says:

    S Basinger said September 8, 2011 at 12:26 PM
    “I guess if you’re always indoors the fact that clouds can cause cooling. Some climate scientists clearly need to get out more.”

    S Basinger said September 8, 2011 at 12:29 PM
    “Wow, my post is brutal.

    That should read: “.. fact that clouds can cause cooling isn’t obvious.””

    Imagine yourself on the Moon looking to the Earth enlighten by the Sun during a starry night. The Earth will shine brighter on cloudy days than in clear sky days because of the change of albedo. For the law of conservation of energy if more energy is reflected from the top of the clouds then less energy reaches the ground cooling the atmosphere below them.
    So the fact that clouds can cause cooling is obvious indeed.
    Some climate scientists clearly need to get out in the outer space more 🙂

  64. La Nina is back, as I have been predicting it would be, in contrast to the global warmers who were calling for El Nino, another blunder to add to their list.

    Also we have a k index of 5, TAMBORA VOLCANO ,raised to level 3 out of a possible 4. The last time it had a major eruption was during the Dalton Minimum in 1815, with an explosive index of 7, making it 100x more powerful then the Mt. Saint Helens eruption of 1980 with an explosive Index of 5.

    I am not saying it will erupt but just the fact that now all of a sudden after the prolong solar minimum started in late 2005, now we have so many volcanos displaying unrest or erupting. Katla, is another one, along with Mt Etna, just to name some.

    Prior to 2005, no big volcanos were displaying this action or this degree of unrest, which was a period of very active solar activity.

    Again if one goes back to 1600 and plots all major earthquake/volcanic activity, one will find 85% of them are associated with solar minimum periods. That is what past history shows ,for what it is worth. Then again mainstream climatolgy can give a hoot about past history.

    This fits in with my phase in theory, so well with the sun setting the tables.


  66. Scott Allan says:

    Conrad Dunkerson – do you guys even bother to read before you post?

    If you had read Dr. Spencer’s comments above you would have seen (a.) his entirely reasonable explanation on why he use the 3 best and 3 worst models originally, and also (b.) that he presented a graph showing ALL the models above:

  67. MarcH says:

    GRL state about papers in press:

    “Papers in Press is a service for subscribers that allows immediate citation and access to accepted manuscripts prior to copyediting and formatting according to AGU style. Manuscripts are removed from this list upon publication.”

    The AGU Authors Guide states: “Once the figures pass technical requirements, your final figures and text will be combined into
    a PDF file that is placed on the journal’s Papers in Press page. Papers in Press is a service for subscribers that
    allows immediate citation and access to accepted manuscripts prior to copyediting and formatting according to
    AGU style.”

    The Publishing Guidelines state:
    “An author should make no changes to a paper after it has been accepted. If there is a compelling reason to make changes, the author is obligated to inform the editor directly of the nature of the desired change. Only the editor has the final authority to approve any such requested changes.”

    As the changes suggested by Dessler appear greater than “copyediting and formatting” it seems the paper must be withdrawn and a new version submitted and reviewed. Any comment?

  68. D. Williams says:

    This exercise, climate science, seems to better characterized as “climate art”, the rigor required to reach significant understanding of a highly complex system (global and regional) seems to be lost on everyone. My primary argument is that there is NO RELIALBE FIRST ORDER DATA WITH RESPECT TO IRRADIANCE–PERIOD. If the very object (the sun) is responsible for the great majority of planetary heat gains, then I would suggest that an accurate measure of total absorbed versus total irradiated energy is necessary–at least from a historical perspective. There are no accurate records of the total irradiated energy that is absorbed by the exosphere, magnetosphere, stratosphere, ionosphere, or troposphere…period. For rhetorical purposes; answer the question of “Was the total amount of irradiated energy from the sun from the years 1600 to 1700 and how different was the ‘AVERAGE’ temperature as compared to the years 1100 to 1200?”

    • KAP says:

      D. Williams,

      First-order total solar irradiance has been measured by satellite since 1979, and that data is publicly available at PMOD.

      Going farther back than that, I would suggest that the SATIRE method of TSI reconstruction has been remarkably skillful in hindcasting TSI during the satellite era, and if I had to bet, I would go with that method (see Krivova 2009) which gets you back to 1610, albeit with increasingly uncertainty as you go back in time.

      Beyond that, you need to use proxies.

  69. Myrrh says:

    salvatore del prete says:
    September 8, 2011 at 5:19 PM
    “In order to reach a definitive conclusion one way or the other one needs complete data, with no unknowns. Those two factors will not happen, therefore only time will prove who is right(Dr. Spencer) and who is wrong (Andy Dessler)”

    Not quite. There are ground rules for whether or not a model even has the ability to forecast, see here for example audit:


    In order to audit the forecasting processes described in Chapter 8 of the IPCC’s report, we each read it prior to any discussion. The chapter was, in our judgment, poorly
    written. The writing showed little concern for the target readership. It provided extensive detail on items that are of little interest in judging the merits of the forecasting process, provided references without describing what readers might find, and imposed an incredible burden on readers by providing 788 references.
    Of the 89 forecasting principles that we were able to rate, the Chapter violated 72. Of these, we agreed that there were clear violations of 60 principles.
    Some principles are so important that any forecasting process that does not adhere to them cannot produce valid forecasts. We address four such principles, all of which
    are based on strong empirical evidence. All four of these key principles were violated by the forecasting procedures described in IPCC Chapter 8.

    It is difficult to understand how scientific forecasting could be conducted without reference to the research literature on how to make forecasts. One would expect to see
    empirical justification for the forecasting methods that were used. We concluded that climate forecasts are informed by the modelers’ experience and by their models—but that they are unaided by the application of forecasting principles.

  70. KAP says:

    Since the LHS of the equation measures the *change* in heat content over time, and since the deep ocean is remarkably stable over these timescales, going deeper (i.e., integrating down to 700m) shouldn’t change the result unless the shallower result has been contaminated by mixing from below. In other words, going deeper just insures that the result you’ve got is rock-solid correct. So one can’t fault Dessler for going deeper, since the deeper layers shouldn’t (theoretically) contribute much if anything to the change in heat content over short timescales.

    That was one thing that bothered me about Dr. Spencer’s paper, in that he used a shallow mixing layer (25 m) combined with a rather long timescale (3 months). Normally the mixing layer (however deep it is) should turn over within a few weeks time, so it makes sense to use a few weeks as the timescale, i.e., monthly data. The risk of using quarterly data, especially if you’re using a shallow mixing layer depth for the whole ocean, is that some of the heat that’s moving in and out of a 25 meter layer might be coming and going from/to below rather than from/to above, which would seriously compromise your results.

    Regarding the issue of having Dessler’s paper re-reviewed with the 700m change, I’ll take a pass. It would depend mostly on how much the value of the LHS changes as a result. But whether it changed much or not, integrating over the entire 700 meters will improve the robustness of the paper and should be done regardless.

  71. MikeN says:

    Arthur Smith, you did several blog posts on Mann and Tiljander. Are you ever going to try and get Mann to fix his basic error?

  72. Ron Cram says:

    Obviously the clouds issue is a big one. I noticed you cited a 2005 paper by GL Stephens, but you did not cite his 2010 paper in GEWEX found at

    Of course, the second paper may not be peer-reviewed and so you may not have seen it and would not be criticized for not citing it. However, I think it is possible this paper makes a contribution to our understanding. Stephens is basically claiming that clouds with rain or drizzle decrease albedo – certainly a reasonable guess and worth investigating. Do satellites have the capability to categorize cloud cover based on a whiteness scale? Does that kind of dataset exist now?

    Thanks for responding!

  73. So far 2011 ,is the second coldest year this century.

    Further OHC, is going to slow down the temperature response to all the times which are now in a cold mode phase, but the decline will be more obvious ,as this decade proceeds, IF they continue to phase as they are presently.

    I claim, Weak Magnetic Fields ,correspond to weather systems moving slower and weather patterns being more persistent. Not to metion the greater geological activity ,when spurts of activity occur, in an otherwise quiet sun.

    Year 2011, could not be a better example of what I am talking about. It is going like I thought it would, from the climatic extremes to the high geological activity.Not to mention a stop in the temperature increase, which actually started way back around 2002.

    I say so far so good, for the PHASE IN THEORY.

  74. Time will tell ,who is gong to be the most correct. This is going to be a very interesting, and educational decade for earth’s climatic system.

    This is the test decade, in my opinion, and INSTEAD of having crazy inadequate models doing the testing, the earth natural phase in items that control the climate/solar relationships,are going to give us the test/opportunity ,to see, what is what.

    This is the first time since the Dalton Minimum, that we have an opportunity like this.

    We have on the one hand the CO2 increase which should mean warming according to the global warmers, versus the phasing in of items that control the climate into a cold mode ,which should mean colder going forward. We will find out which one will be correct, in the not to distant future.

  75. Tilo Reber says:

    Bickmore: “If you want to answer questions about what goes on a multi-decadal timescale, all the El Niño stuff would average out, anyway.”

    This is not true. One or the other part of the ENSO cycle can dominate for decades at a time. Look at this chart, for example.

    You can see that La Nina dominated from 1950 to 1977 and El Nino dominated from 1977 to 1998. So given the period of the satellite temperature record, there has been far more El Nino effect than La Nina effect. In think that in order to extract a valid climate sensitivity number the models will have to do ENSO and PDO and likely the Svensmark effect correctly.

  76. Tilo Reber says:

    Richard Saumaurez: “If there are such substantial differences between different data sets, what does this tell us about their reliability as model inputs?”

    I can tell you that the GISS data set is wrong. A couple of years ago Hansen and RealClimate made an attempt to explain the difference between HadCrut3 and GISS. It turned out that if you remove the poles, the results for the two data sets were nearly identical. It seemed strange to me that such a small part of the planet could cause such a large divergence in the data sets. So I took a closer look at what was going on. As you know, Hansen get’s his polar coverage by extrapolating up to 1000 Km from shore stations around the Arctic. In other words, what he sees happening at those shore stations is really what you get for Arctic coverage. But that extrapolation doesn’t work at the Arctic. The shore stations are heavily influenced by melting shore ice. When the shore ice retreats (and it retreats from the shore first), the circulating water acts like a heater for those shore stations. Their reported temperatures take a huge leap. But the problem is, that temperature is only correct where the ice has melted. But Hansen extends that change across vast areas of the Arctic that are still under ice. So the values assigned for those areas are much hotter than what they are in reality.

    • KAP says:

      Since land those Arctic coastal stations are included in HADCRUT, it seems that what you’re complaining about is that the (legitimately) rapidly increasing temps on the coast are (illegitimately) being extrapolated across the Arctic Ocean.

      But that extrapolation would only be illegitimate if the Arctic Ocean were not melting too, like the coastal areas are.

      Since the Arctic Ocean is in fact melting, the GISS procedure seems perfectly fine to me.

  77. Andrew says:

    Can someone explain to me what exactly is meant by “mixed layer depth” and why people just casually use wildly varying values for it all the time?

    • Socratic says:

      The depth of the mixed layer is the depth at which the temperature is 0.5 C less than the surface. This depth changes with latitude and season, with warmer seasons/latitudes having a shallower mixed layer and cooler seasons/latitudes having a deeper mixed layer.

  78. Socratic says:

    Regarding the different values obtained on the left-hand side of the main equation (2.3 for Spencer vs. 9 for Dessler) I did some quick work myself. Spencer recommends Levitus for a data source. This appears to the the World Ocean Atlas, an updated version now available online from NOAA at

    Downloading quarterly data for temperature (analyzed means) allows you to compute a weighted mean SST for the globe by using cosine(latitude) as the weight. The four weighted means thus computed were: JFM=18.293; AMJ=18.1672; JAS=18.128;OND=17.935, with a global annual mean of 18.130 C.
    The four differences between quarters are then ,358, -.131, -.034, and -.193, with a standard deviation of those values being .247 C.

    To compute heat capacity (and change thereof) I used the mean temp of 18.130 as a “before” value and a changed temp of 18.130+.247=18.377 as an “after” value. I computed density and heat capacity for both using a salinity of 35 g/kg and the equations of Sharqawy 2010. For a 1m x 1m x 25m column, I get mass=25634.58 kg (before), 25633.14 kg (after); HC=29854059 kJ (before), 29878463 kJ (after) for a quarterly change of 24404 kJ. The rate of change per quarter is therefore 2440384 J / 7889400 seconds = 3.1 Wm^-2. This is a bit higher than the 2.3 value given by Dr. Spencer. Note also that this value may be wrong; one could argue that we should be operating on a constant mass of water rather than a constant volume. Computing on that basis, the result would be 3.3 Wm^-2.

    Frankly, I’m new at all these equations and perhaps I’ve made a mistake somewhere. If so, I hope Dr. Spencer will correct me. In a future comment, I will try the same computations using Dessler’s assumptions, i.e., monthly data and a 100m mixing layer.

    Dr. Dessler uses different assumptions and different data.

  79. SOCRATIC, it sounds so smart so wonderful(that I am sincere about, don’t take it the wrong way ) but after you take into account all the uncertainties and differences to come up wih your answer, what does it mean?

    Is your result, and use of data correct ,or is someone more correct ,less correct?

    Again don’t take this the wrong way , I am just trying to point out the FLAW in this type of approach to climatic research.

  80. TILO REBER, you said it right. Excellent, and very true post.

  81. Bart says:

    I would like to alert you to a new analysis here which looks like it could be fairly convincing on the topic of negative cloud feedback.

  82. Andrew says:

    Socratic, thanks.

    So then, there is no true single, global value valid for all times and seasons. Hm…that causes one to wonder how appropriate it really is to just use one value…

    Although none of this makes it clear why different people use different global average values.

  83. Bart says:

    Rog Tallbloke says:
    September 10, 2011 at 2:58 PM

    TB – what I have found is significantly greater evidence that the feedback relationship is, indeed, negative, contrary to Dessler. The phase shift at low frequencies, which determines the sign of the feedback, is very clearly near 180 degrees. Saying an input is 180 degrees from the output is a long winded way of saying that the one is the negative of the other.

  84. Tilo Reber says:

    KAP: “But that extrapolation would only be illegitimate if the Arctic Ocean were not melting too, like the coastal areas are.”

    Yes, HadCrut3 also uses the costal stations. But it doesn’t extrapolate them across the ice. HadCrut3 uses SSTs for as much area as is available before the ice. Hansen’s problem isn’t that he is extrapolating across open Ocean, his problem is that the shore stations are exposed to open ocean early in the melt season while the rest of the Arctic is still frozen. Then he extrapolates from a small amount of open ocean near the shore all the way across the Arctic, most of which is still frozen at that time of year. So the extrapolation is from an area with it’s temperature moderated by open water to an area where there is no open water and where it is actually much colder.

    If the Arctic is melting or not is irrelevant. The point is that for a large part of the year those costal stations are still sitting next to a frozen Arctic. And when that shore ice begins to melt, the temperature in that immediate vicinity does not apply across the remainder of the Arctic that is still frozen.

  85. Socratic says:

    Following up on my previous post, I have results from using monthly (rather than quarterly) WOA data.

    Global weighted-average SSTs by month: 18.176, 18.347, 18.357, 18.282, 18.147, 18.057, 18.134, 18.173, 18.078, 17.950, 17.871, 17.984 giving the same 18.13 average as we saw in quarterly data.
    This gives Delta-Ts of: .192, .171, .010, -.075, -.135, -.090, .077, .038, -.094, -.128, -.078, .113, and the standard deviation of these is .117°C. Already we notice a major difference: if the SST is changing by typically .117 C in a month, we might expect it to change by .117 x 3 = .35 C per quarter. But the actual quarterly change is .25 (see previous post), which means that monthly data is more variable than quarterly data. Not a surprise, but here it is quantified.

    In this run I added a slight improvement: I also downloaded and used salinity data from WAO, which is a little less than the 35 I had been using (mean=34.586). Using T=18.130 as the “before” temp and 18.130+.117=18.246 as “after”, I find a “before” density of 1025.0639, and for a column 1x1x25 meters a mass of 25626.6 kg, specific heat of 4.001219 KJ/kg/K and heat capacity of 29867131 KJ. “After” density is 1025.0369, specific heat is 4.001262, and heat capacity is 29878629 KJ. The change over time is therefore 11498 KJ in 2629800 seconds, for 4.4 Wm^-2. This compares with 3.1 Wm^-2 for quarterly data, all of which is due to the greater variability of the monthly data.

    However, Dr. Dessler used 100 m as the depth of the mixing layer in his paper, as as expected this quadruples the value obtained. For various mixing layer depths, the values I get are: 50m = 8.7 Wm^-2; 75m=13.1 Wm^-2; 100m=17.5 Wm^-2.

    If I’m not making a mistake somewhere (and I might be), it seems to me that Drs. Spencer and Dessler are both too low in the LHS of the main equation.

    Andrew wondered if there was a way to reduce the differences, and I believe there are. As Einstein said, things should be as simple as possible but no simpler. I suspect using a mixing layer is a case of making things too simple. What we really want to get out of all this is how much the heat content of the ocean changes, and both Drs. Spencer and Dessler are assuming that the temp changes ONLY in the mixing layer over these timescales. Someone else brought up the issue of losing or gaining heat out of the bottom of the mixing layer and that’s potentially a big issue. You can eliminate all the depth issues (which is the major area of numerical disagreement) by looking at the temp profile all the way down to where it actually stops changing, which is typically 700 meters or so.

    In his recent additions and comments above, Dr. Spencer indicates that Dr. Dessler is planning to do this, presumably for a subsequent paper. That would eliminate most of the disagreement (at least in the left-hand side of the equation) because ALL energy change would be captured regardless.

    The other big issue is timescale, because monthly data changes faster than quarterly data. What we’re trying to measure is the effect of clouds, and clouds (and weather systems generally) form and dissipate over rather short timescales. So it seems to me that the shorter the timescale, the better we can capture that. In other words, I’m going to cast my lot with the 1-month data (and I would go with 1-week data were it available).

    For the same reason, I have to look skeptically at Bart’s analysis, finding a 5-year lag between cloud and temperature; that’s gotta be a statistical artifact, IMHO.

    Just my 2 cents.

  86. Bart says:

    Socratic says:
    September 10, 2011 at 8:19 PM

    “For the same reason, I have to look skeptically at Bart’s analysis, finding a 5-year lag between cloud and temperature; that’s gotta be a statistical artifact, IMHO.”

    I had trouble believing it, too. But, I have looked at it thoroughly, and generated artificial data streams and analyzed them in the same way to confirm that the analysis holds up. It’s not a statistical artifact. As to any other type of artifact, I cannot speak to how the data are prepared or treated prior to them coming into my hands.

  87. Bart says:

    “…clouds (and weather systems generally) form and dissipate over rather short timescales.”

    Over a particular area. But, statistically averaged over the entire earth? What is the timescale for that?

  88. P. Solar says:

    [By the way, whenever I try to Reply to a comment here the “Submit Comment” button does not work in my browser (Firefox 6 on a Mac). Only the main “Submit Comment” box at the bottom of the page works. Sorry for continuing to create new threads this way.]

    yes, same problem here: linux opera or firefox.

    I guess WordPress is windows only tested. 🙄

  89. Dallas says:

    Bart said, “I had trouble believing it, too.” Why? It is a fair match for the ENSO cycle/peudocycle. Wouldn’t the trick be find what leads what?

  90. The lag can be explained by this comment I made in response to a similar question on Bishop Hill:

    “Bart’s analysis identifies a 5 year lag between cloud changes and temperature changes, and agrees this is surprising. If true, then how should this be interpreted relative to the inverse proportionality between sensitivity and response time?”

    Between El Nino events, heat builds up in the pacific warm pool, most of the energy is therefore hidden from the surface temperature record. The approximate period of the ENSO cycle is around 5 years.

    The last 3 solar cycles were all around 10 years long. It is noticeable that the big El Nino’s occur just after solar minimum. I have a theory why….

  91. The trap everyone is falling into is so many are trying to come up with definitive causes and effects, as if earth’s climatic system is based on two or three items ,that change only in a certain manner, and for a certain duration.


    It is going to be different each and everytime,although general trends as I said can be established.

    An article just came out titled 800,000 years of ABRUPT CLIMATE VARIABILITY.

    I will quote the first part of the article ,which hits home to what I am, and have been saying.

    Drill cores taken from Greenland’s vast ice sheets provided the first clue that Earth’s climate is capable of very rapid transitions . I will end it there.

    The PHASE IN THEORY I subscribe to is the best explanation for this ,and all other climatic changes, rather then what so many seem to be doing now, which is to take a few climatic items in ISOLATION, and then try to give those items some definitve cause and effect versus time lag. IT IS NOT, going to work.

    I know no one likes at all what I am saying, but I always say what I believe, like it or not.

    At the end of the EEMAIN PERIOD, the climate descended from a time of warmth ,warmer then today ,to full blown glaciation in less then 20 years. That is what past history shows.

    Someone explain it, explain that, and how what is being talked about now , and studied now in climatalogy can explain that.

    My phase in theory does,with certain thresholds being reached due to items controlling the climate reaching acetain MAGNITUDE AND DURATION OF TIME. It is the only explanation.

    I post much, so I can be on record as much as possible,and will be promoting this theory, until proven otherwise.

  93. The ABRUPT CLIMATE issue which everyone wants to ignore , is where it is at.

    That needs to be solved, before a real understanding of earth’s climatic system can be reached.

    Of course perhaps one can say it does not exist ,like the global warmers always do, when the data goes against them, which is just about on a constant basis.

  94. When the sun is in a prolong solar minimum state,climatic records and cycles , along with geological activity ,during typical 11 year sunspot activity, can be thrown out of the window.

    It is a new ballgame, when the sun enters a prolong minimum state.

    DESSLER’S claim of a positive cloud feedback is beyond RIDICULOUS. Then again look at the source ,a member of the AL GORE TEAM. He probably wants to undermine the cosmic ray/cloud/temperature relationship theory, which has some traction,although that is not the answerr either, only a part of the answer.

    DR. SPENCER ,at least has more of an open mind then most ,although I wish it were more, but he is still the best high visibility guy out there, to repell the global warmers, and their utter NONSENSE.

  95. Bart says:

    Rog Tallbloke says:
    September 11, 2011 at 11:28 AM

    “The approximate period of the ENSO cycle is around 5 years.”

    I’m not talking about a cycle time, though. I am talking about the time it takes for clouds to react to temperature changes. That is, if you increase global temperature by 1 degC (and could hold it there), within 4.88 years, you will be 1-exp(-1) = 63% of the way to creating an opposing 9.5 to 10 W/m^2 reduction in insolation. The unit step response is plotted here.

    • Hi Bart,

      Yes, I appreciate that. What I’m thinking is that this reaction time is the reason why the ENSO cycle has the periodicity it does have, and the fact it fits the half period of the recent solar cycle is icing on the cake for me, as I’ve been hypothesising the link between ENSO and the solar cycle for the last three years. 🙂

      Think damped oscillator. Alternate El Nino’s are stronger/weaker, due to the varying solar input.

  96. SVENSMARK, should call out DESSLER.

    Until DESSLER, retracts his claim, and says at least there is a possibility negative feedbacks can be associated with clouds ,and more clouds could cause cooling, nothing in his position has really changed.

    Until he says that, nothing has really changed.

  97. Bart, how could you make such a statement. You are acting as though clouds and temperature are operating in a vacum.

    Further temperatures react to clouds,not clouds reacting to temperature.

    But even if clouds did react to temperature changes, which they don’t, it would not be the same each and everytime due to other factors, such as the atmospheric circulation, the earth’s magnetic field strength, the solar magnetic field strength(cosmic rays), so2 particle amounts,soi index, pdo/amo cycle, and on and on and on.

    Bart, has fallen into the classic trap to try to isolate something in earth’s climatic system and have a cause and effect that wil forever and ever remain the same. Right.

    So you are saying really is just not so, it is not reality.

  98. Bart says not only do clouds react to temperatures but it take percisely 4.88years ,if one increases the temperature by exactly 1.0 degree c and holds it there. Then if one does that ,expect exactly a 9.5 to 10.0 W/ ^m 2 reduction in insolation.

    Next theory please.

  99. He said within 4.88 years,not exactly 4.88 years, a little lead way. Wow.

  100. D. Williams says:

    Again, with direct historic measurements from the sun there is no method for establishing exothermic effects from an atmospheric perspective–do we know how interface (exosphere, magnetosphere, stratosphere, ionosphere, or troposphere) reacts to variations in external radiated sources of time (geological)? I would love to see established records for measuring what I term “atmospheric dielectrics” for any energy type (photons, plasma, stellar gases, x and gamma rays, etc.) as this energy transfers through the atmospheres…an experiment:

    1.) Measure total external sources (photos, phonons, plasma, gamma and x rays) and measure the field (assuming the same field of view) for total source energy at every atmospheric interface (exo, strato, magneto, ion, and troposphere).
    2.) Make a differential calculation at each layer and measure losses at each layer (total absorbed energy versus total irradiated energy).

  101. Socratic says:

    Mixing Layer Depth

    Hopefully my final post on this topic. While looking at what it would take to compute the mixing layer from WAO data, I found that it has already been computed on a global grid, and available from NOAA, here:

    There are three criteria in use: 1) (most common definition) depth at which temp is .5 C lower than the surface; 2) depth at which the density is .125 standard deviations greater than the surface; and 3) depth at which the density is equal to what the density would be with a .5 C change. These three definitions give rather different results.

    After downloading all the data and running global weighted averages (see my first post), the global average mixing layer depth for each definition was:
    1. 71.5 meters
    2. 57.2 meters
    3. 45.9 meters

    For quarterly data, using these mixing layer depths gives for the LHS of the equation energy change rates of 8.9, 7.1, and 5.7 Wm^-2 respectively.

    For monthly data, these mixing layer depths give energy changes rates of 12.5, 10.0, and 8.0 Wm^-2 respectively.

    Dr. Dessler’s use of 9 Wm^-2 with monthly data therefore seems about right (though his mixing layer is too deep); Dr. Spencer’s use of 2.3 Wm^-2 with quarterly data seems too small (in large part because his mixing layer is too shallow).

  102. Bart says:

    salvatore del prete says:
    September 11, 2011 at 1:19 PM

    “Further temperatures react to clouds,not clouds reacting to temperature.”

    The phase relationships are pretty clear – clouds primarily react to temperatures. But, it is a feedback loop, so clouds react to temperatures, which then react to clouds.

    “…it would not be the same each and everytime due to other factors…”

    On a globally averaged basis, the reaction is the same over the decade long span of data analyzed.

    “Bart, has fallen into the classic trap to try to isolate something in earth’s climatic system and have a cause and effect that wil forever and ever remain the same.”

    I have recognized a steady, well-defined, classic 2nd order, underdamped response in the data during the interval in which it was collected. The odds of that occurring randomly without it being a real response are astronomical.

    “He said within 4.88 years,not exactly 4.88 years, a little lead way.”

    He (I) said the time constant was 4.88 years, meaning the step response is 63% of the way to the final value in that time. Convergence is exponential in the standard, usual way. You do not appear to know a great deal about such standard, ordinary, and ubiquitously observed behaviors which commonly arise in every engineering discipline on Earth.

  103. Bart says:

    salvatore del prete says:
    September 11, 2011 at 1:03 PM

    “Until DESSLER, retracts his claim, and says at least there is a possibility negative feedbacks…”

    What a very odd thing to say, and then turn around and attack me for showing and quantifying just such a negative feedback.

  104. Tilo Reber says:

    Bart: “That is, if you increase global temperature by 1 degC (and could hold it there), within 4.88 years, you will be 1-exp(-1) = 63% of the way to creating an opposing 9.5 to 10 W/m^2 reduction in insolation.”

    Love the work that you are trying to do here, Bart. But count me a skeptic as well on that lag time. I can’t even come close to imagining – physically – why the lag time would be so large. I would expect days, maybe weeks, but not years.

  105. “Tilo Reber says:
    September 10, 2011 at 10:16 AM

    I can tell you that the GISS data set is wrong.”

    I agree completely! Below is what I have calculated.

    Is GISS more accurate?

    I have read that GISS is the only record that is accurate since it adequately considers what happens in the polar regions, unlike other data sets. I have done some “back of the envelope calculations” to see if this is a valid assumption. I challenge any GISS supporter to challenge my assumptions and/or calculations and show that I am way out to lunch. If you cannot do this, I will assume it is the GISS calculations that are out to lunch.

    Here are my assumptions and/or calculations: (I will generally work to 2 significant digits.)
    1. The surface area of Earth is 5.1 x 10^8 km squared.
    2. The RSS data is only good to 82.5 degrees. (I will assume this applies to HADCRUT3 as well.)
    3. It is almost exclusively the northern Arctic that is presumably way warmer and not Antarctica. For example, we always read about the northern ice melting and not what the southern areas are gaining in ice.
    4. The circumference of Earth is 40,000 km.
    5. I will assume the area between 82.5 degrees and 90 degrees can be assumed to be a flat circle so spherical trigonometry is not needed.
    6. The area of a circle is pi r squared.
    7. The distance between 82.5 degrees and 90.0 degrees is 40,000 x 7.5/360 = 830 km
    8. The area in the north polar region above 82.5 degrees is 2.2 x 10^6 km squared.
    9. The ratio of the area between the whole earth and the north polar region above 82.5 degrees is 5.1 x 10^8 km squared/2.2 x 10^6 km squared = 230.
    10. Let us compare GISS and HADCRUT3 for 2010 and 1998.
    11. According to GISS, the difference in anomaly was 0.07 degrees C higher for 2010 versus 1998.
    12. According to HADCRUT3, it was 0.07 degrees C higher for 1998 versus 2010.
    13. The net difference between 1998 and 2010 between HADCRUT3 and GISS is 0.14 degrees C.
    14. If we are to assume the only difference between these is due to GISS accurately accounting for what happens above 82.5 degrees, then this area had to be 230 x 0.14 = 32 degrees warmer in 2010 than 1998.
    15. If we assume the arctic site can be trusted for temperatures above 80 degrees north, we see very little difference between 1998 and 2010. The 2010 seems slightly warmer, but nothing remotely close to 32 degrees warmer as an average for the whole year. (I omitted the site since moderation may take a while with a URL, but I will assume you know which one it is.)

    Readers may disagree with some assumptions I used, but whatever issue anyone may have, does it affect the final conclusion about the lack of superiority of GISS data to any real extent?

    • Tom Curtis says:

      If you are going to test out a comparison of GIStemp and HadCRUt3, actually do so, rather than comparing versions of your own concoction. Where you to do so you would see that HadCRUt3 does not have data for:

      a) The Arctic north of 80 degrees North;

      b) Much of Greenland;

      c) Much of the Sahara;

      d) Much of Central Africa;

      e) Much of the Amazon basin;

      f) A significant portion of the Canadian Arctic;

      g) Most of the Antarctic south of 60 degrees South.

      Adjusting your figure for the Arctic to 80 degrees North and allowing for spherical geometry (which cannot be ignored) nearly doubles the relevant area to 3.87 million km^2. The area south of 60 degrees South adds an extra 34 million km^2 to the calculation, with the other areas adding perhaps another 10 million km^2.

      Further, if you are going to compare particular years, compare them. Doing so would show that GIStemp shows a 1.2 degree anomaly in 2010 relative to 1998 over much of the Antarctic. That Antarctic temperatures are often cool does not mean you can ignore the actually (relatively) warm temperatures in 2010 relative to 1998.

  106. Bart says:
    …the time constant was 4.88 years, meaning the step response is 63% of the way to the final value in that time. Convergence is exponential in the standard, usual way.

    The feedback effect will kick in and take the temperature back the other way at some point. Is there a mathematical way that point can be determined via the data used here?


    • Bearing in mind we are working with averages. The real situation will of course be strongly affected by the seasonality of insolation and differences between northern and southern hemisphere geography.

  107. Bart- I respect your point of view, but your premise is wrong in my opinion. Yes you have the NEGATIVE feedback, which is good, but I don’t think temperature change leads clouds, I think clouds lead temperature change, and hence the negative feedback.

    If it were true that temperture change leads clouds ,much of what is thought in climatalogy would have to be changed.

    I will try to present my argument for what it is worth.

    ALBEDO -low clouds have been shown to increase earth’s albedo, which in turn will result in lower temperatures. Clouds leading the temperature change. Svensmark cosmic ray theory supports this conclusion, not to mention the recent EARTHSHINE PROJECT, which showed during the recent warming cloud cover was lower, as was the albedo, hence the warming.

    I say the cloud/albedo relationship ,is a postive feedback.

    I say it stands to reason if clouds change the overall albedo of the earth ,they must lead the temperature change.

    To take this further data shows that when the atmosphere is less transparent , whether that be from an increase in particulate matter from say volcanic eruptions, a so called nuclear winter, dust , as some examples, the amount of solar radiation coming into the earth via the atmosphere to begin with ,is reduced due to increased reflection of incoming solar radiation back into space ,by those items before it ever once hits the earth’s surface.

    I know the reflection/ transfer of energy down to the earth’s surface is not a clear cut item with clouds with counter effects, but I say on balence due to clouds having the effect of increasing earth’s overall albedo ,that temperatures will respond to this change in albedo, started by the clouds, and hence temperatures will follow the clouds.

    The thing about clouds is not all clouds are created equal when it comes to their impacts on incoming solar radiation ,which I feel makes it very hard to reach percise conclusions on clouds effects, or lack of effects on temperature, although I think the overall effect is more clouds, lower temperatures ,less clouds higher temperatures.

    Bart , everything in earth’s climatic system is random/chaotic ,that is why it is going to be so hard to ever forecast future climate. It is based on a zillion different items, coming together in a zillion different combinations, which some times phase to create a dramatic climatic change ,while at other times create very little change.


    That is why for you to come up with your conclusions just probably is not so, among other items I mention in this post.


    Some of the ones that come to mind.








    Based on the above I think conclusions such as Bart has reached (nothing personal) are impossible to reach or attain.

  108. I have to say the following,look at what we have.

    1. temperature leads clouds a positive feedback

    2. temperature leads clouds a negative feedback

    3. clouds lead temperature a negative feedback

    4. I don’t think anyone thinks, clouds lead temperature a positive feedback.



    I say this in a sincere manner. Until this can be solved ,we will never be able to forecast future climatic change. If one can’t understand this ,which past data shows to be the case in a convincing manner ,we will not be able to project the future climate.

    I say at least the PHASE IN THEORY , shows how the abrupt climatic change can come about by certain thresholds being reached due to a phasing in of various items that control earth’s climate ,with the sun setting the stage. At least it shows a way. I say if not this, what????


  110. Bart, is doing great work ,although I disagree with him. You can still appreciate the work and effort someone puts into something ,even if you don’t agree. I do respect ,and appreciate the efforts Bart, has made.

    Don’t get me wrong on that score.


    If Bart’s feedback theory is correct, clouds would essentially over the long run be a non factor , in earth’s climatic system.

    Bart, is trying to say, I think, that clouds react to temperature and temperature reacts to clouds, as if they happen in conjunction with one another. I say it has to be one or the other, NOT both. I don’t think you can have it both ways. My opinion.

  112. I don’t mean to go on an on, but 10 years worth of time is way to short a period of time to reach any SERIOUS conclusions, on a subject so deeply complex.

    I am done.

  113. Paul says:

    Dressler has shown in his first statements his bias and desperation to paint Dr. Spencer and the other “deniers” as wackos. He states in his introduction that… “The usual way to think about clouds in the climate system is that they are a feedback… …In recent papers, Lindzen and Choi [2011] and Spencer and Braswell [2011] have argued that reality is reversed:…” He first states that the “usual way to think about” something is “this way”, implying there are other ways to represent some behavior. Then he states Lindzen, Choi, Spencer, and Braswell are claiming that “reality” is reversed. In essense Dressler is claiming that the “usual way”, or their way, is in fact “reality”. Not very scientific and puts the rest of his arguments in a bad light.

  114. D. Williams says:


    Again, without direct historic (geological time) measurements from the sun there is no method for establishing exothermic effects from an atmospheric perspective–do we know how atmospheric interfaces (exosphere, magnetosphere, stratosphere, ionosphere, or troposphere) react to variations in external radiated sources over (geological) time? I would love to see established records for measuring what I term “atmospheric dielectrics” for any energy type (photons, plasma, stellar gases, x and gamma rays, etc.) as this energy transfers through the atmospheres…a proposed experiment:

    1.) Measure total external sources (photons, phonons, plasma, gamma and x rays) and measure the field (assuming the same field of view) for total source energy at every atmospheric interface (exo, strato, magneto, ion, and troposphere).
    2.) Make a differential calculation at each layer and measure losses at each layer (total absorbed energy versus total irradiated energy). Rinse and repeat.

  115. Tilo Reber says:

    Tom Curtis: “Where you to do so you would see that HadCRUt3 does not have data for:”

    GISS doesn’t have data for those areas either. The difference is that GISS extrapolates to where it doesn’t have data. But there is no reason to think that yields a more accurate result. And in the Arctic the extrapolation definitely fails and yields incorrect numbers.

    By the way, concerning the rest of the planet, RealClimate and Hansen have shown that when the polar data is removed from GISS, there is no divergence between HadCrut3 and GISS.

    • Tom Curtis says:

      Tilo Reber, GIStemp and HadCRUt do not use the same data bases, although there is considerable overlap. Consequently your assumption that missing data for Hadley is missing data for GISS is incorrect.

      There is significant empirical justification of GIStemp’s extrapolations, which are based on studies of correlation between the anomalies of distant weather stations. What is more, if you do not like the extrapolation, GIStemp provides a product limited to 250 km extrapolations, which is more conservative than the HadleyCRUT methodology. For the record, the GIStemp 250 km product shows 2010 as being 0.03 degrees warmer than 1998.

      The GIStemp extrapolation in the Arctic in particular has been confirmed by the nearly equivalent trend of GIStemp and the DMI reanalysis product in the Arctic. The DMI reanalysis is based on a large number of shipbourne, airbourne, bouy and surface station data giving it a very high resolution relative to GIStemp.

      Finally, I require a citation for your claims about what Hansen and Realclimate has shown, as I believe you have misinterpreted their claim. In particular, do they claim to have shown the close similarity of trends (different from individual yearly values)? And if so, for what specific regions? Did they just exclude the high Arctic? All of the Arctic north of 60 degrees? What portions of the Antarctic? Did they exclude other areas not covered by HadCRUt? These distinctions make crucial differences to what was claimed by Hansen and Realclimate.

  116. Bart says:

    Tilo Reber says:
    September 11, 2011 at 8:43 PM

    “I can’t even come close to imagining – physically – why the lag time would be so large. I would expect days, maybe weeks, but not years.”

    Clouds require moisture. Consider how long it might take for that moisture to be produced and settle out in the atmosphere.

    salvatore del prete says:
    September 12, 2011 at 7:41 AM

    ” I think clouds lead temperature change, and hence the negative feedback.”

    This is a bit of a circular argument (pun intended) since what we are talking about is a feedback LOOP.

    salvatore del prete says:
    September 12, 2011 at 8:32 AM

    “I say it has to be one or the other, NOT both. I don’t think you can have it both ways.”

    But, you can. This is the very definition of a feedback LOOP. It looks like this. As you can see, clouds both drive and are driven by temperature. The only break in the symmetry is where the radiation forcing comes in.

  117. Bart says:

    It looks like this.

  118. oxonmoron says:

    salvatore del prete says:
    September 12, 2011 at 8:32 AM

    “I say it has to be one or the other, NOT both. I don’t think you can have it both ways.”

    Relax SDP, if you think about it the way Bart has it figured the clouds sort of act as a temperature stabilising system – sort of a govenor as Willis puts it. That’s good news. Now if you insist there are other ways than temperature increase that causes increased cloud cover and I’m sure you’re right then those extra clouds will maybe join in the merry-go-round that Bart ilustrates so clearly and drive the temperature down further. All good stuff then?

  119. Bart, let me ask you this, have you concluded when all is said and done that more clouds will equate to lower temperatures ,and less clouds to higher temperatures, regardless of which one leads or does not lead the other, to begin with?

    I think you do.

  120. D. Willams post is where most of the climatic research should be directed, not this idiotic CO2 temperature relationship, which does NOT exist.

  121. oxonmoron makes good sense.

  122. One of the other factors which I think can impact cloud cover is the AO. My thinking is the more the AO is negative, the more clouds, due to a more meridional circulation, which brings polar air far south and tropical air far north, the result greater contrast and contacts between different air masses, more intense storms more clouds, then if the atmospheric circulation were more zonal (+AO).

    The above applying to the mid and high latitude cloud cover, and not the tropics.

  123. Conrad Dunkerson says:

    Scott Allan, I note that, like Dr. Spencer, you have ignored Dessler’s actual objection about the lack of other observational data sets and instead stridently complained that the model choices are ‘correct’.

    Other observational data closely matches some of the model runs. It is the exclusion of the data and model runs showing that close correlation which Dessler objected to. Adding in the model runs, but continuing to exclude the observational data does nothing to refute that objection.

  124. BART, Here is another thought that just came to me. Perhaps your feedback loop would work in the tropics, but I don’t think for the middle and higher latitudes as well, where I am thinking the Arctic Oscillation, would have a greater impact. In that latter case, a -AO ,is not going to only equate to more low clouds, but colder temperatures overall,or at least a colder distribution of temperatures in the middle, high latitudes, in contrast to a +AO which would equate to colder polar temperatures and warmer mid/high latitude temperatures, with probably an overall increase in temperatures, in addition to less cloud cover overall, from the middle to polar latitudes.

    Maybe your feedback loop is a good fit for the tropics and not as good a fit for the middle and high latitudes ,where other factors could come more into play, in determinatin of cloud cover amounts. A thought.

    Does that make any sense to you?

  125. Bart says:

    salvatore del prete says:
    September 12, 2011 at 11:48 AM

    “Bart, let me ask you this, have you concluded when all is said and done that more clouds will equate to lower temperatures ,and less clouds to higher temperatures, regardless of which one leads or does not lead the other, to begin with?”

    That seems a reasonable assumption to me. The idea of the feedback here, I think, is that rising temperature will lead to an increase in clouds, reducing the radiation back to space but, more importantly, reflecting radiation incident from the Sun, thus ameliorating the temperature rise in an active thermostat loop.

    But, I have to say straight out that I have not given much thought to these matters and am only marginally qualified to do so. I am merely a systems analyst, and I have reported the result I have found in my analysis. Since Dr. Spencer has been intent on proving a negative gain, and Dr. Dessler a positive, I assume I am coming out rather firmly on the side of Dr. Spencer.

  126. Bart says:

    “…I assume I am coming out rather firmly on the side of Dr. Spencer.”

    Or, rather, the evidence is coming out rather firmly on the side of Dr. Spencer.

  127. Bart says:

    It is kind of funny. People are all of a sudden asking me questions like I am some kind of all-knowing climate guru. I don’t know an awful lot about the overall climate system. But, I do know general systems theory very well. Given the data, I can generally derive what the relationships are. Drawing conclusions about those relationships in regard to the climate system is a job for experts in that field.

  128. Thank you for your response “Tom Curtis says:
    September 12, 2011 at 9:36 AM”
    In the URL that Tilo alluded to at “Tilo Reber says:
    September 10, 2011 at 10:16 AM,” the following statement is found:
    “I think that it is safe to say that the divergence of the GISS record is due to the interpolation and extrapolations at the poles.”

    I had read this somewhere a while back so it seemed reasonable to assume that the major difference was the north polar region since all of the talk is about the north polar ice. I originally did the calculations comparing RSS to GISS. (And RSS ‘sees’ to 82.5 degrees north.) However UAH and GISS also show a similar discrepancy. Now I realize that the satellite numbers measure different things so it is legitimately possible to get differences between satellite measures and ground measures over a given 12 month period. But as far as I know, the only blind spots for satellites are the polar regions. And both RSS and UAH agree that 1998 was hotter than 2010, although just barely in the case of UAH. It could be possible that 2010 really was warmer than 1998 on the ground, however it seems a bit odd to me that the gaps in the data did not even out but that there were apparently many more cooler gaps in 1998 and many more warmer gaps in 2010 according to GISS.

    • Tom Curtis says:

      Werner, one feature of the UAH and RSS Lower Troposphere temperature indexes is that they are much more susceptible to perturbation than are the surface temperature indexes. That is, they become warmer relative to the mean during Le Nino events than do the surface records, and cooler relative to the mean during La Nina events. Given this, and because 1997/98 was such a strong El Nino event, it is not surprising that UAH (and RSS) show 1998 as warmer than 2010, while GISS shows 2010 as warmer.

  129. Tom Curtis says:

    I have found the Realclimate discussion of the HadCRU/GIStemp differences. The comparison is made between HaCRUT and GIStemp masked to cover only those areas covered by HadCRUT, and modified to have uniform temperatures within each HadCRUT cell area (as does HadCRUT).

    In other words, the demonstration was made not by removing polar data, as claimed by Tilo Reber, but by removing all data fields not included in the HadCRUT analysis.

  130. fido says:

    Dear Dr. Spencer, Dessler states he will change sentences. But he is not allowed to do that at this stage, without the paper going back to reviewers…It seems to me this fact has not been stressed enough: what exceptions are allowed for this paper?

  131. Bart: if my own hypothesis is correct, insolation into the ocean is much more important than levels of longwave radiative flux in the atmosphere, because there is no mechanism for rapidly entraining significant heat from the atmospheric radiation into the ocean mixed layer, and the ocean drives the atmospheric temperature due to its vastly higher heat capacity.

    This is evidenced by the ~3 month lag of air temperature behind sea surface temperature. It is also evidenced by the relationship between insolation, albedo, and outgoing longwave radiation you have been working around.

    So I think you are exactly correct when you say that reflection of incoming solar is the most important cloud effect.

    Your determination of a -9.5W/m^2 cloud feedback plus the empirically measured drop in low cloud over the tropics from 1980-1998 pretty much nails the global warming issue in my opinion. Co2 is along for the ride.

  132. Ask why is it so? says:

    Does all this mean that with an increase in the surface temperature of the earth this increases the amount of water vapour in the atmosphere and this makes more clouds and this causes a reduction of solar radiation reaching the surface but increases the amount of long wave returned to the surface but the reduction of solar radiation is higher than the increase in long wave therefore with a reduction in solar radiation the temperature of the surface is reduced which in turn reduces the amount of clouds and all this takes a few years or so or what? Or have I completely missed it?

    • pochas says:

      You’ve got it! Now the question is, why does the earth’s temperature vary a small amount despite this powerful stabilizing mechanism? Is there some way to conjure up demons and dragons which will justify massive taxation and One World Government and get rid of those nasty American capitalists once and for all? We’ll all be soooooo much better off.

  133. G. H. Larouche says:

    I think you got a good point there, the way I see it.

  134. Stephen Wilde says:

    “but the reduction of solar radiation is higher than the increase in long wave.”

    More precisely the reduction in solar shortwave to the oceans is energy that never gets into the system at all, it is lost forever.

    In contrast the increase in downward longwave just represents a very slight delay in the exit of energy that did get into the system.

    If energy is lost to the system altogether then that causes a reduction in total energy content and in itself reduces downward longwave after a while.

    What matters most above all else is how much energy enters the system in the first place.

    And whatever happens the climate response is negative so as to maintain stability over 4.5 billion years despite the power output of the sun having increased over the period by some 30% or so.

  135. Stephen Wilde says:

    “The idea of the feedback here, I think, is that rising temperature will lead to an increase in clouds, reducing the radiation back to space but, more importantly, reflecting radiation incident from the Sun, thus ameliorating the temperature rise in an active thermostat loop.”

    Observations tell us that it is slightly more complicated.

    When the sun was more active the temperature rose but cloud amounts fell and in the process the air circulation systems shifted poleward and/or became more zonal.

    Meanwhile the stratosphere cooled and the tropopause rose to accelerate energy flow out to space.

    So the negative feedback in that situation (solar forcing) is not more clouds but a faster water cycle.In a solar warming situation the active sun PULLS energy out of the system as fast as the reduced cloudiness allows extra energy into the oceans.

    In an oceanic warming situation the negative feedback is again a rising tropopause because the warmer surface increases the surface/stratosphere temperature differential but again the system response is negative via a faster water cycle. In an oceanic warming situation the warmer sea surfaces PUSH energy out of the system faster.

    Furthermore both solar and oceanic forcings can go into reverse and may or may not offset/supplement one another at any given time.

    So in a warming world the extra energy into the oceans is completely negated by faster energy transfer to space despite the reduction in cloudiness. The technical reason for that is related to atmospheric pressure which sets the energy values for the phase changes of water.I’ve gone into that in a lot more detail elsewhere.The price paid in climate terms is a shift in the climate zones at the surface.

    In a cooling world the reduced energy into the oceans is only partly negated by a slower energy transfer to space because solar energy that never gets into the oceans is lost to the system forever.

    That is why we have had liquid oceans for 4.5 billion years despite an increase in solar power output over the period of some 30%.

  136. At least everyone has gave an opinion on clouds which is good.

    My biggest obstacle for my PHASE IN THEORY , and all that comes about with the sun setting the tables, is will the solar activity stay LOW enough, LONG enough ,to see if things will happen the way I think they might.

    Since SEPT.01 , the solar flux a good measurement of solar activity has been running around 115 ,now at 131, if that should last ,solar activity will be to high.

    In my opinion solar activity has to stay close to the typical 11 year solar minimums, of a flux reading around 75 (with some spurts of activity), in order to get significant solar effects on the items, controlling the climate. The further the solar activity rises above this level, and the longer it stays above that level ,the less efffectiveness the sun will have on the climate, as far as causing items to phase into the COLD mode.

    Then again GEOFFRY SHARP, from the LAYMAN SUNSPOT SITE, thinks solar cycle 24 might be in the process of maxing out ,which would be very early, but that is his thinking. If solar cycle 24 does max out with a solar flux in the 130 area ,then I think, that would keep everything intact, as the flux readings would go down from there, and we would then get to see just what effects a sun in a constant prolong solar minimum state, will have on earth’s climatic system.

    It is frustrating, because the SUN will do what it is going to do, and no one knows for sure what that might be. Past history suggest a GRAND MINIMUM , lasting untill 2034 or so. We will see.

    As long as the PROLONG SOLAR MINIMUM remains intact, which started in late 2005,evidence will be provided one way or the other, which will support the PHASE IN THEORY with the sun setting the tables, or NOT support this theory.

    I would like to know one way or the other. Past history suggest it may be right ,but I much rather have it happening before my eyes.

  137. Bart says:

    Stephen Wilde says:
    September 13, 2011 at 6:58 AM

    “When the sun was more active the temperature rose but cloud amounts fell and in the process the air circulation systems shifted poleward and/or became more zonal.”

    I think you need to be careful about ascribing cause and effect here. Were these processes caused by, or merely coincident with the temperature rise? Moreover, the feedback process I have found has a time constant of almost 5 years associated with it, so you need to look ahead in time a little to see it manifested.

  138. The hardest part of the PHASE IN THEORY, is to know what degree of magnitude and what duration of time, is needed for all of the various items it includes, to get the climatic change.

    For example, how many volcanic eruptions ,how strong? Do you need two major ones or one major one?

    The sun, does the solar flux have to be in the 70’s or could it be in the 90’s?

    I don’t know the answer to these questions ,only the apparent trend, which results as these conditions phase into a more extreme magnitude or duration. Whether that be a warm mode or cold mode or a neutral mode.

    Let’s take the AO. Negative 1 versus negative 4, would not have the same impact.

    One can go on and on with each item that influences earth’s climatic system.

    Take the earth’s magnetic field off by about 70% from its high but what is the point that really makes an impact? Is it 70%off, 80 % off? I think you can all get my drift.

    Just endless combinations of items, that will give countless, endless ,climatic results.

  139. Jaison Jacob says:

    Sorry to repost my comments again. However, a news article has been published today under the heading “Dangerous bacteria spreading in warm oceans”, again blaming global warming for the increased incidence of cholera infections around the world.

    Warm oceans can increase the incidence of vibrio infections, however, researchers have gone overboard in this case and exaggerated their data. There is only anecdotal evidence for this. In fact, I had written a blog on this subject questioning many of their assumptions. There are 13 blogposts on this subject starting at

    To me, there are many reasons for the spread of cholera worldwide, the most important being global transport and immigration and pollution.

  140. Stephen Wilde says:

    Bart,I think I have been careful about cause and effect.

    Throughout the period of more active sun the polar vortices became deeper and stronger resulting in a more positive polar oscillation which became smaller at the surface thus pulling the jets poleward and allowing a larger set of sub tropical high pressure cells which in turn allowed more solar energy into the oceans.

    When the sun started to come down from the peak of cycle 23 all those features went into reverse.

    Now much the same must have happened in the MWP because there were more poleward jets and a more active sun then too as evidenced by the warmth in Greenland at the time of the Viking settlements.

    It could be mere coincidence but I don’t think so because too many other features of climate changes fit into that scenario.

  141. Bart says:

    Steve – I certainly defer to your greater knowledge of the system. Is there anything in all that which might lead to a second order type response with a ~5 year time constant? Two processes (energy storage states), perhaps, which could interact with one another on that kind of a timeline?

    • pochas says:

      Yes, Bart, there is (sorry for butting in, Steven). The Hale Cycle. The standard quarter cycle lag for the Hale cycle would be about 5 years. The mechanism would be that during half of the Hale cycle the sun’s magnetic polarity is oriented opposite to the earth’s, allowing for a different set of earth/sun magnetic and particle interactions. This seems to produce observable hydrological interactions as shown by Alexander (and dismissed by warmists).

  142. Stephen Wilde says:


    I’m sure a range of second order responses are possible and I’m open minded as to what timescale of such response might dominate.

    A 5 year response seems reasonable given the inertia of the oceans. Remember too that each ocean basin reacts differently and they are generally out of phase with one another so the net global oceanic response is very complex. That gives multiple energy storage ‘pools’ constantly interacting.

    I think the solar effect dominates on long timescales such as MWP to LIA to date but that the oceans dominate on shorter timescales.

    It seems reasonable to suppose that if a solar change is affecting clouds and albedo to alter shortwave input to the oceans then it might take 5 years or so for a significant oceanic response to build up so that it becomes apparent over and above the usual ENSO variability.

    Ultimately the process gives ‘stepping’ of temperature trends from one Pacific Multidecadal Oscillation to the next such as happened twice in an upward direction during the 20th century.

    I suspect downward stepping from MWP to LIA and upward stepping from LIA to date.


    When one examines the typical solar flux values during a typical solar minimuM, during the typical 11 year sunspot cycle, one finds a reading of 75-80 to be common.

    I use that as a base, to say I feel that in order for a PROLONG SOLAR MINIMUM ,to accomplish, what I call for in THE PHASE IN THEORY, that the level of solar activity has to be such that the solar flux avErage througout the PROLONG SOLAR MINIMUM ,has to stay at 90 or lower.

    That is my best estimate. So what I am saying, is if the solar flux stays at 90 or below, then I would expect all the items the PHASE IN THEORY , calls for with the sun setting the tables ,to materialize, if they did not, I would admit to being wrong.


    That is my key number , as far as I can tell.




    ORAEFAJOKULL VOLCANO ICELAND showing unrest. This volcano hardly ever shows any activity at all.

    TAMBORA, KATLA/MT ETNA to name a few more showing unrest.

    It is amazing how the spurts of activity in this still prolong solar minimum, keep corresponding to an increase in geological activity.

    I am expecting the solar flux readins to fall off from here.

    Also what is very curious about the solar activity, is that when it does come about, it is almost entirely only in the NORTHERN HEMISPHERE ,of the sun ,with the SOUTHERN HEMISPHERE ,being just about as blank as can be. Not normal behavior.

  145. I have to say us people on this board as a whole, are probably making more head way into what makes the climate system of earth work ,then anyone else.

    Again mainstream is completely on the wrong path, they will never ever get anywhere.

    They have no feeling for the astronomical side of climatalogy or the geologic side of it.

    To really master climatolgy ,one must know his /her astronomy and geology ,not mention oceanagraphy and chemistry, those perhaps not being so obvious among everythng else, but vital to this field.

  146. Thank you for your comments Tom Curtis.

    I guess the one thing that is constant between ALL (five) data sets is that there has been essentially no change in temperature over the last 10 years as per:

    Does this fact support Spencer or Dessler? Or do we need more time to be sure? Or is this fact not relevant to this discussion? Thank you in advance.

  147. ” it might take 5 years or so for a significant oceanic response to build up so that it becomes apparent over and above the usual ENSO variability.”

    Stephen, the period of El Nino followed by La Nina often *is* around five years. It *is* the dominant ocean response to the waxing and waning of solar cycles. The Pacific is after all the biggest of the ocean basins.


    I doubt if Dessler, knows a thing about astronomy or geology, and OBVIOUSLY how they may apply to earth’s climatic system.

    Remember he is part of the AL GORE club,which believe climate begins, and ends with CO2.

    The reality of course being CO2 ,is a non factor, and past history shows this over and over again.

  149. ROG, what happens during a prolong solar minimum ? That is the question you should be asking yourself.

    Reason being the items that effect climate will not respond in the same way ,as to when the sun is in it’s quote, normal 11 year sunspot cycle, as oppossed to a prolong solr minimum state.

    When the sun is in it’s more or less normakl, 11 year sunspot cycle ,climatic effects will be mimimal, everything gets cancelled out.

    No, the question that needs to be examined further (becasue past hsitory shows this to be so) is what happens during prolong minimums. I have stated many times what I think has happened, and will happen.

    The decade 2000-2009 is a bad decade to try to draw conclusions about about earth’s climate system. A bad decade to use in my opinion.

    This decade in contrast should be a great decade to use in order to draw conclusions about earth’s climatic system.

  150. ad further geological activity, as most everyone is oblivious to the connection. What will it take ?

    Since Sep 11, shortly after the latest spurt of solar acitivy , we have to add 3, earthquakes of magnitiude 6.0 or better to what I sent in the above post.

    The latest just happening in the ALEUTIANS.

    This can’t be by chance ,because it happens everytime, following a burst of activity on the sun(when the sun is otherwise quiet) resulting in a K value of 5 or more here on earth. The lag time being usually from 1 to 25 days or so.


    All one has to do is look at spurts of solar activity this year and then see what kind of geological activity folowed in thenext 1to 25 days , you wil have your answer.

    This latest being no exception.

  151. make it 4, I forgot the 6.4 earthquake at Vancouver. SEP 09

  152. RobB says:

    SDP – I am slightly reluctant to encourage you, but what is the physical mechanism that links solar and geological activity?

  153. Bart says:

    Rog – “…the period of El Nino followed by La Nina often *is* around five years.”

    If you are looking for possible related cyclic activity, the proper timeframe to be looking for is not that of the time constant, but of the inverse of the bandwidth. In order to be a part of a frequency-specific, randomly excited oscillation of a lightly damped mode, the frequency response needs to be able to pass through that frequency. The natural frequency of my model is about 0.0725 years^-1, so roughly speaking it allows periods of greater than about 1/0.0725 or 13.8 years (and maybe a little shorter, depending on the gains in the rest of the system) to get through.

  154. Bart says:

    salvatore del prete says:
    September 14, 2011 at 12:20 PM


    Dessler is not relevant because of this. A linear regression on a phase plane plot of dispersively lagged signal components gives an essentially random result.

  155. D. Williams says:


    Would you purchase an automobile if you couldn’t know how many tires or wheels are attached or mounted to the vehicle?

    Well that is exactly what climate science, with global warming, wishes us all to do. What I find interesting is that there are experiments (as of today 14 Sept) to determine basic principles (particulate interchange in the atmosphere) in the UK (see the telegraph’s website).

    Why do we need to consider any more experiments or fund anymore satellite projects when “the science is in”. Just from a rhetorical perspective this whole climate science and political gerrymandering is a huge waste of time (and dare I say energy). Either the climate scientist are correct as they contend (thus we can end their funding) or they can admit that their scientific work product is shoddy and go back to the drawing board.

    No reliable claim can be made to anthropogenic warming (specifically based on only CO2 effects) based on the absurdly apparent lack of causal relationships to and of a complex global climate system–call Vanna White because you all need to buy a vowel (or a least a clue).


  156. BART, thanks for the info.

    ROB B. -First of all past history supports what I am saying. There are a some papers on it, if you google solar/volcanic relationships.

    One theory goes something like this. When the magnetic fields in general are weak ,as is the case with the sun and the earth presently, it provides the background, in that any sudden jump in magnetic activity will have a much greater effect in giving the earth a jolt, due to the sudden increase/change, of magnetic activity hitting the earth. This in turn effects earth’s magnetic field ,due to causing a shift of the molton iron in earth’s inner core. This shift ,however slight, is enough to cause the plates which are unstable to begin with ,to become just a little more unstable ,hence more geological activity.

    In addition these jolts seem to change the rotational rates of the various layers of the earth ,from the crust though, the mantle to the core ,at slightly different rates ,which again result, in imparting more instablity to the already unstable plates.

    We are talking fractions of a second ,but nevertheless enough to set things in motion ,along with the iron core probably shifting, due to the jolt to the earth’s magnetic field being greater then normal when the sun is in a queit phase ,but has spikes in activity from time to time.

    Look at this way if your in a car going 100 m.p.h (active sun) and it picks up to 140 m.p.h you will feel a jolt, but not as great as if the car were at 0 m.p.h (quiet sun) and suddenly speeds up to 40 m.p.h.

  157. D. Williams is right on.

    Piers Corbyn , has a good explanation, as good as any, for the solar /volcanic relationships. On the climaterealist website a few months back.

  158. In closing I don’t know if it for real, but the correlation seems to be there.

    If one plots all major geological activity from 1600-present ,you will see what I mean.

  159. “The natural frequency of my model is about 0.0725 years^-1, so roughly speaking it allows periods of greater than about 1/0.0725 or 13.8 years (and maybe a little shorter, depending on the gains in the rest of the system) to get through.”

    Hi Bart and thanks for that. How I wish I could get my head round what your model is. The ENSO cycle is clearly strongly related to clouds and ocean heat content, and it often is around 5 years as I said. So I don’t understand what you mean about “allows periods greater than 13.8 years… to get through.”

    ENSO gets through, so is it an oscillation that somehow occurs despite the negative feedback in your model because it is being driven by “gains in the rest of the system”?

    I have trouble with that, because the Sun and the albedo, and the ocean mixed layer are the big boys on the block. What other “gains in the system” could account for it?

    Salvatore: Long solar minima are associated with seismic activity as you say. They also allow heat sequestered in the ocean during runs of higher than average solar activity to escape. This is why temperatures at the surface are maintained when the Sun is weak. Look at the run of el nino events evident in the sst’s at the end of the 1800’s. The phasing and amplitude of the solar cycles then is similar to the last decade.

    Then comes the cold.

  160. Rog, thanks for confirming the seismic activity with solar minimums.

    The problem is why? I gave it my best shot, I don’t know if that is the correct path or not, but that is the best explanation I can come up with . In any event the correlation is there.

    Two more earthquakes today, magnitude 6.0 or higher. Just reinforcing the uptick in geological activity, that follows a spurt of activity on the sun, when the sun is in a prolong minimum state.

  161. What we need is for the solar flux to stay at 90 or less, and this will prove either the PHASE IN THEORY ,is correct or not correct.

    Although thus far I am quite pleased with the evidence that shows support for this theory, from our current climate/geological situation, thus far. Since the sun started it’s prolong minimum phase back in late 2005, geological activity is on the rise,the AO ,has tended more negative, the temperature increase has stopped, and climatic extremes have picked up. In addition to the PDO turning cold, LA NINA’S coming on, and OCEAN HEAT CONTENT ,at a stand still or even declining. All these items the PHASE IN THEORY , has called for with the sun setting the tables for this to take place. So far ,so good.

    I am quite sure if this PROLONG SOLAR MINIMUM continues, a further phasing of the items which control earth’s climate into a cold mode , will continue.

  162. let’s add yet another major earthquake 7.3 fiji islands

  163. A. C. Osborn says:

    Dr Spencer have you seen this analysis at Niche Modeling, does it add anything to your original analysis?

  164. P. Solar says:

    re. Rog Tallbloke September 14, 2011 at 5:00 PM

    Sorry to be late to the party on this but I thought I should just post this for the record. Using Dr Spencer’s simple model with random rad and non-rad terms I attempted to get the best fit to his satellite data show shown in his comparison with IPCC referenced models.

    This was just ad hoc fiddling with parameters to see what I could get. I don’t regard it as a scientific method.

    Here is an overly of the two.

    Note I had to shift about 3mth in lag to match the two. This was a persistant offset and not realated to model parameters. I have not understood what is represents yet.

    What is interesting is that the model is fairly consistent to changes in feedback and ocean depth as long as the ratio stays the same. So I’m not suggesting 9.2 means much but the ration was robust. If you look at what this represents it is the same time constant Bart has derived by a completely different method. So look what you get from the run shown in my graph:

    45/9.2 = 4.891304 years.

    Now that is so close to Bart’s figure I’m astounded.

Leave a Reply