On Recent Criticisms of My Research
By Dr. Roy Spencer
One of the downsides of going against the supposed “consensus of scientists” on global warming — other than great difficulty in getting your research funded and published — is that you get attacked in the media. In the modern blogging era, this is now easier to do than ever.
I have received many requests recently to respond to an extended blog critique by Barry Bickmore of my book, The Great Global Warming Blunder. The primary theme of my book was to present evidence that scientists have mixed up cause and effect when diagnosing feedbacks in the climate system, and as a result could have greatly overestimated how sensitive the climate system is to our addition of carbon dioxide to the atmosphere from fossil fuel burning.
For those interested, here is our most extensive peer reviewed and published evidence for my claim.
But for now, instead of responding to blog posts, I am devoting all the time I can spare to responding to peer-reviewed and published criticism of my work. The main one is Andy Dessler’s paper in Science from last fall, which claimed to find positive cloud feedback in the same 10 years of NASA satellite radiative energy balance (CERES) data we have been analyzing.
In his paper, Dessler dismissed all of the evidence we presented with a single claim: that since (1) the global temperature variations which occurred during the satellite record (2000-2010) were mostly caused by El Nino and La Nina, and (2) no one has ever demonstrated that “clouds cause El Nino”, then there could not be a clouds-causing-temperature-change contamination of his cloud feedback estimate.
But we now have clear evidence that El Nino and La Nina temperature variations are indeed caused in large measure by changes in clouds, with the cloud changes coming months in advance of the temperature changes.
And without going into detail, I will say it now appears that this is not the only major problem with Dessler’s diagnosis of positive cloud feedback from the data he presented. Since we will also be submitting this evidence to Science, and they are very picky about the newsworthiness of their articles, I cannot provide any details.
Of course, if Science refuses to publish it, that is another matter. Dick Lindzen has recently told me Science has been sitting on his critique of Dessler’s paper for months. Science has demonstrated an editorial bias against ’skeptical’ climate papers in recent years, something I hope they will correct.
In the meantime, I will not be wasting much time addressing blog criticisms of my work. The peer-reviewed literature is where I must focus my attention.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
Art – you are very confused. When I stated, and you agreed, that A0 = T(t0)-Te, then your equation was wrong. When you now stipulate that A0 = T(0)-Te-β Q(0)/c h, your equation works, but Q(0) in your notation, which is poor because it causes just this sort of confusion, is the convolution integral from minus-infinity to zero.
Thus, “implying A0 = 0” is non-sequitur. There is no information available which compels that T(0)-Te = β Q(0)/c h.
Let us focus on MY equation for a moment.
T(t) = Te + (T(t0)-Te)*exp(-(t-t0)/tau) + convolution-integral-from-t0-to-t
As you see, this eliminates confusion, because the start time and the current time are made explicit in all the important variables. Now, suppose I want to know what is happening in the past. Is T(-t) just
T(-t) = Te + (T(t0)-Te)*exp(-(-t-t0)/tau)
which diverges exponentially? No, it is not. It is
T(-t) = Te + (T(t0)-Te)*exp(-(-t-t0)/tau) + convolution-integral-from-t0-to-(-t)
The convolution integral is not zero, just because the upper limit is less than the lower limit. It is merely the negative integral from -t to t0. It is also unstable. So, to get a reliable estimate of the past, you have to have perfect information for the integrand, so that the instability of the exp(-(-t-t0)/tau) term and the convolution integral cancel out precisely.
Thus, your “six-million degrees” is farcical.
So, Bart, I am curious. What is your opinion of Spencer’s model?
– Do you believe that it actually provides a constraint on climate sensitivity and how well does it constrain it?
– Do you believe it demonstrates that the global temperature variations are well-explained by the PDO?
Arthur Smith says:
April 6, 2011 at 11:38 am
“Either way, Spencer’s model claimed to be explaining 20th centuries with PDO, but it does not.”
I do not know if it does or does not. You do not know if it does or does not. Your analysis is flawed for reasons I have presented. But, I do know that it matches up the amplitudes and phases for apparently nonsecular components fairly well, and that is good guide for whether he is on the right track in ferreting out a significant (at least) contributor to temperature variations in the 20th century.
You and Barry have lent your energies to those who would snuff the flame aborning. Instead of offering constructive criticism, you choose grandstanding and chest-thumping. That is not so surprising. Every great advance throughout history was resisted by vested interests and their toadies. But, I would not want to be one of the latter.
Joel Shore says:
April 6, 2011 at 12:14 pm
To tell the truth, Joel, I don’t really care. It is a sideshow for me. It or another mechanism to emerge will explain why the climate is not behaving as predicted. My main beef is when people with insufficient knowledge or experience to know what they are talking about, like Barry and Art, get up on a high horse and start calling other people dumb.
I care about whether the cloud feedback renders the overall water vapor feedback system negative. As you know, all the GCMs, either implicitly or explicitly, assume that the feedback is positive, and without this assumption, they do not predict a catastrophic temperature rise. My reading of Spencer’s evidence points to negative feedback.
Bart says:
But…You haven’t shown that Arthur’s different starting point for the integration is relevant once one considers that one can fold some of these differences into the parameters [given that you don’t have F(t) going back in time]. So, it seems like you are just sniping at things that may well be totally irrelevant.
And, it is strange that you are so bothered by this but apparently not by the fact that Spencer claimed his fit determined a value for the climate sensitivity when in fact it didn’t independently constrain it at all! Doesn’t that irk you just a little bit?
Finally, I find it weird that you think Spencer is on the right track…and believe his evidence that he claims points to the negative feedback and yet you don’t believe the absolutely overwhelming evidence that the current rise in CO2 is anthropogenic. That shows judgment that I think almost all scientists would find quite bizarre! You have to admit that your point-of-view is rather unique in the scientific community, meaning that either you are one of the smartest people alive…or maybe that your judgement isn’t quite as good as you think it is!
Bart,
Actually, it was 6 TRILLION degrees, and Arthur meant it to be farcical.
I’ll let Arthur argue with you about the integrated form of the model, since he did that work. In any case, I still don’t understand what any of your criticisms have to do with anything I said in my original critique, since I was using the non-integrated form and numerically integrated it. (And I checked my results against Spencer’s to make sure I was getting the same answers.)
Let’s step back from your point about the integrated form, and focus on the main issues I brought up.
1. Do you disagree with my assertion that there are an infinite number of “best-fit” solutions for alpha and beta? If this assertion is true, then the particular “best-fit” values Spencer obtained are meaningless, since he could have picked any number of other combinations, and gotten exactly the same results. And yet, Spencer assigned great meaning to the parameter values he obtained.
2. Do you disagree with my assertion that not only the starting temperature anomaly, but the equilibrium temperature, were effectively fitting parameters for Spencer? As Arthur pointed out, modelers always have to run time-series models for a while before they begin “production,” to iron out any kinks caused by arbitrary choices in starting conditions. This is true of climate models, molecular dynamics models… whatever.
3. Do you disagree with Arthur’s and my assertion that about the first half of Spencer’s simulation is dominated by his choice of starting anomaly and equilibrium temperature? If the assertion is true, the overall trend of his simulation in the first half of the 20th century doesn’t have much to do with the PDO.
4. Do you disagree with the assertion that Spencer’s model doesn’t fit the temperature data in the last half of the 20th century very well? I.e., it significantly under-predicts the temperature. If it’s true, then the first half of his simulation fits the data ok-ish, but the model trend doesn’t have much to do with the PDO, while the last half of his simulation is dominated by the PDO, but it doesn’t fit the data very well. And so Spencer’s model doesn’t do anything much to show that the PDO significantly affects global climate.
5. Do you disagree with the assertion that Spencer’s choice of ocean mixed layer depth is unrealistic? If he had chosen a more reasonable depth, he could have fit the data exactly as well, but his alpha value would have indicated a VERY sensitive climate system, and his beta value would not have matched well with what he estimated from satellite data.
You keep rehashing your point about the integrated form, but even if you’re right, I fail to see how it even puts a dent in the main criticisms of Spencer’s model.
Bart – sorry I agreed to your saying “A0 = T(t0)-Te,” – it had been several weeks since I looked at the equations in question and I assumed we were talking about the same thing. That is correct if you start the integration at t0. It’s also correct if PDO = 0 before t0 – either way Q(0) is zero. Which is exactly the way I calculated it – if you download the R code from my page you will see it. If you think I did something wrong, point out where the R code went wrong. As Barry noted, I reproduced Spencer’s model identically in the 20th century with these equations.
You are also exactly correct that projecting backwards gives:
T(t) = Te + (T(t0)-Te)*exp(-(t-t0)/tau) + convolution-integral-from-t0-to-t
When I do that with the proxy-PDO data (MacDonald as cited on my page) I got about -6 trillion degrees for T(1000). In order to *not* get such a ridiculously low early temperature, you need the convolution integral to be strongly positive – and since it’s over a negative time period, you need PDO to be strongly negative. That’s what I’ve been saying all along. I’m sure you’re glad that we agree on this.
So what do you think are the “other inputs” that allowed Spencer to set his year-1900 temperature so low, as it doesn’t seem to have been the PDO?
(by the way, it was 6 trillion, not 6 million degrees!)
“…either way Q(0) is zero”
It is the weighted integral of the PDO from the infinitely remote past to time zero. You do not know the PDO from the infinitely remote past to time zero. You don’t even know how long this model is approximately valid into the remote past. You have no basis for this statement.
“When I do that with the proxy-PDO data…”
What part of “you have to have perfect information for the integrand, so that the instabilit[ies] … cancel out precisely” did you not understand? ANY imperfection grows exponentially. All you have demonstrated is that exponential growth is fast. Who’da thunk it?
Barry Bickmore says:
April 6, 2011 at 2:31 pm
“1. Do you disagree with my assertion that there are an infinite number of “best-fit” solutions for alpha and beta?”
If the trade space is not degenerate, then there is a unique least squares or other normed solution. But, it does not matter. Spencer’s model is plausible. You guys have not ruled it out, even if you think you have.
“2. Do you disagree with my assertion that not only the starting temperature anomaly, but the equilibrium temperature, were effectively fitting parameters for Spencer?”
I do not see any point to this question, and in any case, I have answered it previously.
“3. Do you disagree with Arthur’s and my assertion that about the first half of Spencer’s simulation is dominated by his choice of starting anomaly and equilibrium temperature?”
That is a loaded question “. The simulation shows how the model behaves when it is fit to the data.
“4. Do you disagree with the assertion that Spencer’s model doesn’t fit the temperature data in the last half of the 20th century very well?”
Define “very well”. In what manner does it deviate? How well does it correlate? Based on how well it correlates, could you use this information to investigate changes to the model which could make it fit better?
“5. Do you disagree with the assertion that Spencer’s choice of ocean mixed layer depth is unrealistic?”
As I stated before, it appeared to me it was YOU who broke that variable out of a particular lumped parameter. That parameter may have other interpretations or legitimate modifications. And, pay careful attention to this comment.
“I fail to see how it even puts a dent in the main criticisms of Spencer’s model.”
Exactly.
Bart, you don’t have to integrate backwards to the year 1000, just integrate backwards to 1850 or so with an ensemble of different plausible values for the PDO from 1850 to 1900, and report back which PDO values allow you to have a reasonable 1850 temperature, and which do not. A simple choice would be to pick a constant PDO at some level, and see what it does. I assume we are in agreement that global temperatures in 1850 were not 3 or more degrees C below temperatures now.
You have the equations, you can download my R code. You do the math and see what values of PDO Spencer’s choice of 1900 temperature requires over that time period in order for temperatures not to blow up.
Yes, it’s true we don’t know precisely what they were. But presumably, as an oscillation, PDO stays pretty much up and down around where we’ve seen it.
I assert that, aside from the other problems Barry B. pointed out, Spencer’s PDO model does not realistically describe temperature from 1900 to 1950 or so – it requires an implausible starting value. And if you take that part of the temperature curve out, Spencer’s PDO fit to the last half of the 20th century doesn’t look very good at all.
Bart,
You said,
“If the trade space is not degenerate, then there is a unique least squares or other normed solution. But, it does not matter. Spencer’s model is plausible. You guys have not ruled it out, even if you think you have.”
No, there isn’t a unique solution. I demonstrated that alpha and beta are perfectly covariant. Arthur demonstrated that they have to be, given the form of the model.
And most of his chosen parameter values are unphysical, it is clear that his model is NOT “plausible”. I am at a loss to understand why you think it doesn’t matter if his curve fitting produces parameter values that we KNOW are way off. The model is supposed to have some connection with the physical world, after all.
“That is a loaded question. The simulation shows how the model behaves when it is fit to the data. “
But Spencer claimed that his modeling effort showed that the PDO can explain most of the warming over the 20th century. It didn’t show this. And it’s not my job to fix his model, since I have no reason to believe the PDO is controlling everything. Why don’t you try it, if you think it can be done?
“Define ‘very well’. In what manner does it deviate? How well does it correlate? Based on how well it correlates, could you use this information to investigate changes to the model which could make it fit better?”
The PDO-forced model fluctuates around equilibrium, while the real temperature keeps going up significantly above the chosen equilibrium. In other words, the model curve is significantly below the data curve OVER THE PERIOD WHERE THE PDO DOMINATES MODEL RESPONSE.
I did show how one could get the model to fit the data better, but that involved such high positive feedback that the system was unstable. (Roy would really hate that outcome.) THAT’S what you have to do to make it fit better.
“As I stated before, it appeared to me it was YOU who broke that variable out of a particular lumped parameter. That parameter may have other interpretations or legitimate modifications.”
You’re wrong. Spencer is the one who came up with a 700 m mixed layer. The Cp parameter, as Spencer defined it, is just the heat capacity of a column of water 1 m^2 on top and a certain depth. He used the depth as a fitting parameter, as he explained in his book.
“…you don’t have to integrate backwards to the year 1000…”
I take that as an admission that you realize now that your extrapolated claim of 6 trillion degrees was absurd, not for Dr. Spencer, but for the claim you were making. The punters are keeping score, and I intend to claim my beers, so I need your capitulation in writing.
But, you still don’t get it. I’m not going to bother accepting your challenge because it is a waste of my valuable time. Whether I could rise to it or not using a particular parameterization of this model, it means nothing. This model is a deterministic approximation/linearization of a complex nonlinear stochastic system. It provides a reasonable replication of the output temperature when driven by the PDO over a particular time interval. Any breakdown of that relationship at some point in time, plus or minus, likely enough means nothing more than that the equations need to be relinearized about a different state.
Or, the state could have been affected at some time by a one-off, transient event, such as the explosion of the island of Krakatoa in 1883, for which it is said:
Stop trying to prove the unprovable, or get serious about it and join one of those fringe groups trying to prove the non-existence of God. At least that concerns something important, even if it is as ultimately futile.
Barry – “And it’s not my job to fix his model, since I have no reason to believe the PDO is controlling everything.”
And, it is not my job to fix your model, since I have no reason to believe you know what you are doing.
“But Spencer claimed that his modeling effort showed that the PDO can explain most of the warming over the 20th century.”
Uh-uh. The exact quote is:
Strawman diversion. Do you ever present an argument which does not contain at least one logical fallacy?
Erratum: Part of the quote from Dr. Spencer should have read “…sufficiently strong to justify devoting…
Bart, you say Spencer’s model “provides a reasonable replication of the output temperature when driven by the PDO over a particular time interval”. What time interval do you believe it provides a reasonable replication of temperature for, and where does it break down? See Barry Bickmore’s Figure 4 for instance. Do you believe it should work from 1900-2000 (the entire displayed range)? Do you believe it fails at 1900, and at 2000? Why would it work for an entire century (or whatever period you believe) but break down there? What has changed?
Science believes in finding the causes of things – an explanation isn’t complete unless it matches physical reality in some way. What is the physical reality that changed in 1900, or in 2000, that requires abandoning the fitted parameters of Spencer’s model that worked for that particular limited range of time?
In particular, if you believe it didn’t break down in 2000 but may continue to be valid, then you can use that model going forward, to predict global surface temperatures from 2000 to 2010, and even 2011 (since the PDO dependence is smoothed over 30 years into the past, you don’t even need 2011 data to make a rough guess – but this site has it through February 2011 anyway).
If you think Spencer’s model is really giving a “reasonable replication” of temperatures, tell us how it’s doing for the past decade, and what does it predict for 2011? Or explain what happened in 2000 (or earlier?) that made it break down.
I’ve recently done precisely this prediction exercise for 2011, 2012, and 2013 based on a not much more complex statistical fit to ENSO, sunspots, and stratospheric volcanos – *plus an underlying upward trend* (and you’ll notice from the graphs, it does way better than Spencer’s model in matching the temperatures). I’m willing to put my numbers out there for the world to see, and we’ll see how I do. Are you? Is Roy Spencer?
I have tried to post this several times – it seems to be failing for some reason. I’m removing links in case that’s the problem.
Bart, you say Spencer’s model “provides a reasonable replication of the output temperature when driven by the PDO over a particular time interval”. What time interval do you believe it provides a reasonable replication of temperature for, and where does it break down? See Barry Bickmore’s Figure 4 for instance. Do you believe it should work from 1900-2000 (the entire displayed range)? Do you believe it fails at 1900, and at 2000? Why would it work for an entire century (or whatever period you believe) but break down there? What has changed?
Science believes in finding the causes of things – an explanation isn’t complete unless it matches physical reality in some way. What is the physical reality that changed in 1900, or in 2000, that requires abandoning the fitted parameters of Spencer’s model that worked for that particular limited range of time?
In particular, if you believe it didn’t break down in 2000 but may continue to be valid, then you can use that model going forward, to predict global surface temperatures from 2000 to 2010, and even 2011 (since the PDO dependence is smoothed over 30 years into the past, you don’t even need 2011 data to make a rough guess – but it is available through February 2011 anyway).
If you think Spencer’s model is really giving a “reasonable replication” of temperatures, tell us how it’s doing for the past decade, and what does it predict for 2011?
By the way, I’ve recently done precisely this prediction exercise for 2011, 2012, and 2013 based on a not much more complex statistical fit here:
http://arthur.shumwaysmith.com/life/content/predicting_future_temperatures
using ENSO, sunspots, and stratospheric volcanos – *plus an underlying upward trend* (and you’ll notice from the graphs, it does way better than Spencer’s model in matching the temperatures). I’m willing to put my numbers out there for the world to see, and we’ll see how I do. Are you? Is Roy Spencer?
Bart,
Whether Spencer said his model shows the PDO “can explain” or “can potentially explain” the 20th century temperature record is irrelevant. His model didn’t show either one, since he used physically implausible parameters and non-unique solutions.
And by the way, Bart, I never asked you to fix my model… since I don’t have a model. I only explored the one Spencer proposed.
Arthur Smith,
Interesting prediction of 0.58 for 2011. We’ll see how you do with it over the next 8 months. Also, what do you think of this chart by Bill Illis?
You guys are amazing. Tweedledumb and Tweedledumber.
Art – I’m not taking any more questions from you until you admit your “6 trillion degrees” taunt was mistaken. In fact, at this point, I’m going to need a little groveling.
Barry – it’s irrelevant, except that it changes the entire meaning of what he was doing, what he was trying to show, and the relevance of your criticism. Not only do you guys have no clue as to the system you are criticizing, you’re arguing against an opponent which doesn’t even exist.
BTW, Barry, about your 700 m vs. 100 m concern: Spencer addresses that very point in the chapter and cites the paper “Warming of the World Ocean: 1955-2003” in Geophysical Research Letters, by Levitus, Antonov, and Boyer, published 2005. If you have a problem with it, maybe you should consult them. And, next time you set out to slander a respected scientist, try reading what he wrote first.
Smokey – short term temperature fluctuates a lot. Bill’s graphic isn’t surprising given the recent La Nina. It’ll be headed up now that the La Nina is waning. And satellite-measured temperatures are far more responsive to ENSO variation for some reason.
Bart – there was no mistake about 6 trillion degrees. That is what you get when you plug in Spencer’s model with the proxy PDO data and what he did for 20th century; if you think I did the calculations at all wrong, please show where. Yes, it’s unrealistic to extrapolate back that far – you can’t extrapolate the fit I just posted about more than a decade or so because the underlying trend function isn’t right (it’s just a polynomial, which has no physical basis). But I have a good explanation for that. Do you have a good explanation for why Spencer’s model doesn’t work extrapolated forward or backward?
When you have a model based on physical phenomena, like El Nino, PDO, sunspots, etc, then extrapolation should be fine – there’s nothing special about the end-points of a fit, and it should have predictive power forwards and backwards, or if not then you should have a good explanation for what changed. When you have a model based on some mathematical function like a polynomial or an exponential, then of course it becomes invalid when you go much beyond the endpoints of the fit. My polynomial is no better than Spencer’s exponential at long-term extrapolations.
The problem is – Spencer never acknowledged that he was not just fitting to the PDO, he was actually fitting temperatures to a linear combination of (convoluted) PDO and an exponential function. The PDO part of that fit should be fine well beyond the endpoints if it is based on any physical relationship between the two measures. But the exponential function is clearly not – it needs explaining.
And Bart, making pre-conditions before responding to somebody is a pretty strong sign of a weak case. Making accusations of slander (actually, it would be libel since this is a written medium) is even more hilarious given your particular remarks in this forum.
Bart,
If you had actually read my critique, you would have seen that I did address Spencer’s rationale for the 700 m mixed layer. Here’s what I said:
First, if a 700 m mixed layer is a physically reasonable value, then maybe my objections are moot. Here’s what Spencer says about it.
“By coincidence, this figure actually matches the approximate depth over which warming has been observed to occur in the last fifty years, which is something the model did not know beforehand. (p. 116)”
Even if the water temperature has been measurably heating down that deep, however, Spencer’s model assumes that the temperature is uniform throughout that entire 700 m, which is demonstrably false. The thermocline (i.e., the boundary between the warmer, well-mixed layer at the surface of the ocean and the colder deep ocean water is typically in the range of 50-100 m deep (Baker and Roe, 2009). In a simple model like Spencer’s that doesn’t account for upwelling and diffusion of heat into the deep ocean, one needs to fudge that figure a little higher. Murphy and Forster (2010) discussed previous work on this question, and it appears mixed-layer depths of 100-200 m (probably closer to 100 m) are reasonable for models such as Spencer’s. The irony, of course, is that Murphy and Forster were criticizing Spencer and Braswell (2008) for using only a 50 m mixed layer, which skewed their results. (Spencer provides the spreadsheet he used for the 2008 study here. Go ahead and plug in a 700 m mixed layer, and see what kind of nonsense comes out. You can compare it to what you’re supposed to get here.)
I’m sure you’ll tell me why detectable heating as low as 700 m is the same thing as a 700 m mixed layer, however. I’m on the edge of my seat.
Art – that’s not groveling. You know you were totally pwned. Man up and admit it.
And, Art, review the exchange with Barry. Read Spencer’s own words for the first time. Then, explain your argument again, and why it has any meaning. What, exactly, are you trying to prove?
“Yes, it’s unrealistic to extrapolate back that far…”
So, it was basically a useless exercise, is what you are saying. I agree. Indeed, I had to explain it to you. Yet, you claimed it debunked Spencer’s entire model. Now, you say it was just a lark.
This isn’t groveling. I now require more than a little. You are in a hole. I advise you stop digging.
Bart, honestly, nothing you have stated here is not already covered in mine and Barry’s posts, but perhaps we didn’t explain clearly enough there. I have tried carefully to uncover any actual remaining problem here in dialogue with you, but so far everything that you have claimed that is of any quantitative nature is in agreement with what we both originally posted. It’s a little hard to tell if anything is getting through, or whether this dialogue is completely pointless.
Rather than telling me what to do, can you try to explain what you actually think about Spencer’s model, fitting temperatures to the sum of an exponential and the PDO? You stated up above that it “provides a reasonable replication of the output temperature when driven by the PDO over a particular time interval”. Do you still believe that? What time interval? Can you state any conclusion about the physical system of Earth’s climate that you might conclude from this “reasonable replication”?