Was alerted to this recently (Lindzen, R. S., and Y.-S. Choi (2009), On the determination of climate feedbacks from ERBE data, Geophys. Res. Lett., doi:10.1029/2009GL039628, in press, and no doubt available at sceptic blogs everywhere).
The paper claims that the feedback of the real climate system, as indicated by analysis of ERBE data and SST, differ substantially from the feedback of all models, as indicated by the model radiation budget and SST. Details of radiation and feedbacks are not really my thing, so I may be barking up the wrong tree here, but one thing really jumped out at me (once I had noticed it).
The real climate data is, of course, derived from a fully coupled system where the SST and atmosphere evolve simultaneously.
However, the model simulations they chose to use were those from the AMIP project, which is a comparison of atmospheric models forced by historical SSTs.
Now, I don't know exactly what effect this will have, but it is easy to see that it might be different. In the model case, a change in SST is being directly imposed as a forcing at the lower boundary. However, in the natural case, the change in SST is part of the natural variability of the system (and may even be considered a response to a spot of natural variability in the atmosphere).
So, the two systems are not at all equivalent. Can anyone explain why Lindzen would not want to look at a dynamically equivalent model system (ie a fully coupled AOGCM) to see how well his diagnosis works? Like this paper did, for example. The data for these model results are freely available.
See also this nice pair of papers which discusses how forcing a system may change the dynamical behaviour.
The paper claims that the feedback of the real climate system, as indicated by analysis of ERBE data and SST, differ substantially from the feedback of all models, as indicated by the model radiation budget and SST. Details of radiation and feedbacks are not really my thing, so I may be barking up the wrong tree here, but one thing really jumped out at me (once I had noticed it).
The real climate data is, of course, derived from a fully coupled system where the SST and atmosphere evolve simultaneously.
However, the model simulations they chose to use were those from the AMIP project, which is a comparison of atmospheric models forced by historical SSTs.
Now, I don't know exactly what effect this will have, but it is easy to see that it might be different. In the model case, a change in SST is being directly imposed as a forcing at the lower boundary. However, in the natural case, the change in SST is part of the natural variability of the system (and may even be considered a response to a spot of natural variability in the atmosphere).
So, the two systems are not at all equivalent. Can anyone explain why Lindzen would not want to look at a dynamically equivalent model system (ie a fully coupled AOGCM) to see how well his diagnosis works? Like this paper did, for example. The data for these model results are freely available.
See also this nice pair of papers which discusses how forcing a system may change the dynamical behaviour.
26 comments:
James
In absurdly simplified layman's terms, how do you go about calculating the climate's sensitivity to CO2?
And, of course, the complete lack of any error analysis concering the actual satellite data. There are several ERBE data sets and it just happens that Lindzen picked the "right" one.
http://chriscolose.wordpress.com/2009/03/31/lindzen-on-climate-feedback/
I think your point, James, could get down the entire paper. It's worth checking on the coupled GCMs.
If it was absurdly easy, there wouldn't be any disagreement :-)
But anyway...
One approach is to use models that simulate the physical changes arising from an increase in CO2, due to changes in radiative transfer (this component alone is well understood and accurately measurable) and consequential changes in atmospheric behaviour (this is much less well understood). This approach gives quite a range of answers, which depend on fine details of model construction.
We can also look at historical data from when CO2 (and other forcing factors) were different. To a reasonable approximation, models suggest that different forcing factors can be considered additive and equivalent, so we can generate a unified "sensitivity" parameter. But the data in historical times are limited and again there's a fair uncertainty range on how well we understand any specific epoc.
However, when you put all the evidence together, it is hard to support a number that isn't quite close to 3C - a much higher or lower number would require various independent lines of evidence to all happen to be biased in the same direction, which is a priori unlikely (are all the clocks in your house too slow, or are some slow and some fast?). Personally, I suspect a slightly lower value of 2.5 might be better (and this was the IPCC's official best guess for a fair bit of its history).
Georg,
Thanks, I wasn't aware of the data issue. I spotted that Gavin has independently flagged the issue of AMIP vs CMIP in this comment on RC. I wonder if anyone is going to bother crunching the numbers to see what difference this makes (they are not readily to hand for me).
Would it be rude to wonder out loud whether Lindzen and co-author appreciate the difference between weather and climate?
Well, the Lord is on tour, funded by he-doesn't-know-or-care-who, to tell the world about it:
http://deepclimate.org/2009/09/22/friends-of-science-behind-moncktons-magical-mystery-tour/#comment-409
Correction:
http://deepclimate.org/2009/09/22/friends-of-science-behind-moncktons-magical-mystery-tour/#comment-407
I looked at the Lindzen and Choi paper in detail. I'm not a climate expert, so I may be wrong here, but this is what I found :
Lindzen did a correlation between changes in outbound radiation against natural changes in sea-surface temperature. He found that radiation goes up at about 4 W/m^2 per increase in sea surface temperature, almost exclusively caused by an increase in long-wave (IR) radiation.
Now, Stephan Boltzmann's law says this is exactly what you would expect from a planet radiating at around 290 K, as long as there is no feedback mechanism in place.
So I'd say that Lindzen showed that there is no feedback mechanism measurable over the short (months) periods that he did his data analysis for.
Still, somehow Lindzen claims that this finding implies a strong negative feedback, and even claims that the 'models' predict a negative slope (a decrease in radiation if sea surface temperatures go up), which is essentially impossible.
I think the cause of this error is that he misrepresents the radiative "forcing" (such as from CO2) with changes in sea temperatures. That confusion leads to an incorrect figure 3 in his paper. In that figure, the SW (short-wave) graph is off-set by 4 W/m^2. All models, and the right scale (feedback factor) should move up by 4 W/m^2.
Of course, after correcting this error, the conclusions of his paper would need to be adjusted as well. Not only is the ERBE data essentially is in line with the model predictions, but also the ERBE data shows that there is NO feedback at all (feedback factor 0) for short-term sea surface temperature changes.
In summary : The Lindzen and Choi paper shows only that there is no significant feedback on the short term, and that this is in line with model predictions.
Not sure how Lindzen himself came to completely other conclusions...
Rob
I took a simplistic view of the Lindzen/Choi result. Can you please help me see how this logic is off base? (Please reply to me as a layman physiologist, not a climatologist). When earth temperature goes up, radiation measured by satellite goes up instead of down. Therefore, water vapor is not causing a greenhouse effect. If water doesn't, neither do other gases. Atmospheric PCO2 < 10% of PH2O, and the CO2 absorption in the IR is small compared to water, so it is not likely to capture any significant amount of radiated heat.
"So I'd say that Lindzen showed that there is no feedback mechanism measurable over the short (months) periods that he did his data analysis for."
Wasn't the ERBE/CERES data set 26 points from a 10 year period?
"Still, somehow Lindzen claims that this finding implies a strong negative feedback, and even claims that the 'models' predict a negative slope (a decrease in radiation if sea surface temperatures go up), which is essentially impossible."
I thought the negative slope from the models was an indication of greenhouse effect: the warmer the water, the more the water vapor in the air, the less radiation can escape. Lindzen's "negative feedback" is the opposite of the greenhouse effect, the warmer the water gets, the more heat the earth gives up, which prevents overheating...
"When earth temperature goes up, radiation measured by satellite goes up instead of down. Therefore, water vapor is not causing a greenhouse effect."
I have no idea how you are getting from the first sentence to the second. It is absurd to suggest that water vapour does not act as a greenhouse gas.
Thank you James for your reply. I was trying to describe the ERBE/CERES data analyzed by Lindzen and Choi, 2009.
By the way, an earlier post here described the data set as covering only a few months of history, but my understanding is that it covers the time period from 1984-2008 or 2009. Am I mistaken?
Hello James,
I am sure you are aware of the recent developments around Lindzen and Choi 2009, including the response by Motl :
and by Dr. Spencer :
Still, there seems to be much confusion in the blogosphere as to what is really being claimed and what is really refuted by now.
In regards to this, could you please comment on the following issues :
(1) Did Dr. Lindzen indeed make a mistake in the formula that calculates the feedback parameters. Also, does it look like this mistake was deliberately inserted. Moreover, if this mistake is corrected, what remains of his conclusions regarding negative feedback and low climate sensitivity ?
(2) The scatterplots with AMIP models are shown in every blog and even Fox News presentation about this paper. What are these AMIP models, and how did Lindzen managed to let them show an instable climate system ?
(3) What is correct and what is incorrect about Figure 3 (a,b and c), which is a representation of the ERBE data and model results combined.
(4) If there were really as many mistakes made in this paper as the discussions suggest, how come this paper was accepted by Physical Research Letters ?
I'd appreciate some comments from you, since so far it seems that serious climate scientists are extremely silent on this paper.
Regards
Rob
Sorry. I forgot to insert the links to responses about Lindzen and Choi :
Motl :
http://motls.blogspot.com/2009/11/spencer-on-lindzen-choi.html
Spencer : http://www.drroyspencer.com/2009/11/some-comments-on-the-lindzen-and-choi-2009-feedback-study/
Rob, Thank you for the links to the motls posting by lumo and the analysis by Roy Spencer.
From lumo, I realize that I should not ask just whether the plot of outgoing radiation vs. sea surface temperature is positive or negative, but rather: How much is it less positive than the black body radiation slope, 4 W/m^2/degree? If it is less, one would attribute that to some feedback effect (e.g. a greenhouse effect due to increased water evaporation resulting from the heating). (Rob, you said this too above, but now I understand what you meant).
Even water evaporation, though, would probably not produce a negative slope, Roy argues. If you actually had a negative slope, like the AMIP models predict, the earth's temperature would be very unstable. Roy confirms the point in James's original post on this thread that the AMIP models are probably not the right point of comparison. Roy explains this further, saying:
"Now for my main concern. Lindzen and Choi examined the AMIP (Atmospheric Model Intercomparison Project) climate model runs, where the sea surface temperatures (SSTs) were specified, and the model atmosphere was then allowed to respond to the specified surface temperature changes. Energy is not conserved in such model experiments since any atmospheric radiative feedback which develops (e.g. a change in vapor or clouds) is not allowed to then feed-back upon the surface temperature, which is what happens in the real world.
Now, this seems like it might actually be a GOOD thing for estimating feedbacks, since (as just mentioned) most feedbacks are the atmospheric response to surface forcing, not the surface response to atmospheric forcing. But the results I have been getting from the fully coupled ocean-atmosphere (CMIP) model runs that the IPCC depends upon for their global warming predictions do NOT show what Lindzen and Choi found in the AMIP model runs. While the authors found decreases in radiation loss with short-term temperature increases, I find that the CMIP models exhibit an INCREASE in radiative loss with short term warming."
Roy goes on to provide his own evaluation of the ERBE data in which he takes the orbit time of the satellite and latitude precessions into account for his window averaging (Lindzen and Choi used longer averages) and, for this reason or others, gets a less positive slope in radiation vs. SST.
If Roy Spencer's calculations are right, at least one can say that the slope is less than the black body radiation expectation (which I still read to mean that there is a water vapor evaporation greenhouse effect, but I'm not sure this is the explanation - I'm still confused about the comments on time scale). Lindzen and Choi's slope is steeper than the black body prediction: i.e. ~6 W/m^2/degree, which would seem to be physically unrealistic, or at least would call for a feedback mechanism whereby radiation is increased by warming rather than inhibited.
Perhaps I am starting to understand the time scale issue a little bit. Short term responses would preceed extensive changes in atmospheric water vapor content and should be compared to AMIP predictions, whereas responses over a long time scale might better be compared to CMIP predictions. Does that seem correct?
Roy Spencer states in his posting that Lindzen and Choi show radiation responses up to 6 W/m^2K, which may be true (there is scatter in the points) but the line they draw through all of their data points has a slope of ~4.1 W/m^2K, as expected for blackbody radiation.
For a short term response, would one ever expect any slope other than the blackbody slope?
If not, the point of the Lindzen and Choi result boils down to: Why would the AMIP models predict a response function different than blackbody, in fact with a catastrophic negative slope?
Surely there's a lot more to understand here that I am not seeing yet. Might someone help me understand what the data points are on the other plots (which are compared to the AMIP prediction lines) in the Lindzen-Choi montage?
David,
I do not know enough about AMIP or other models to give you any reasonable answer. I just know that the scatter plots in Lindzen's paper predict an out of control climate system. This should have triggered at least some question marks with Lindzen that maybe he was not using the models the right way.
But I do know about feedback systems.
In Lindzen’s formula’s, the factor to get from radiation change to temperature change is G.
So, if you want to insert a feedback loop from temperature change back to radiation change, the zero-feedback response is always 1/G.
It should thus not matter if G is determined by the Stefan Boltzman equation, by erratic/spatial temporal distribution of temperatures in (say) the lower stratosphere, or by little green men dancing in the sky.
If Lindzen fiels that these simple mathematics do not apply in his formula, then I would at least expect that he explains why that is so. He does not do that at all in the GRL paper.
In fact, he does add 1/G to to FOLW correctly, but in the next sentence he subtracts it again, from FSW (without explanation), for a net resulting zero-feedback parameter 0.
And with that trick, he eliminated the zero-feedback response altogether.
James, PLEASE tell me that I am wrong, because if I am right, this paper violates every standard for scientific ethics that I ever learned.
Hmmmm....I'm reluctant to dig into this again, for the simple reason that since Spencer has already shown that CMIP and AMIP are substantially different, the results are basically worthless regardless of what LC actually did. There is an ambiguity in these sorts of studies, since although on the one hand a surface temperature anomaly should give rise to a change in radiative balance, on the other hand any change in radiative balance will give rise to a temperature anomaly and thus there is a chicken and egg question as to which is the forcing and which the feedback...
James, thank you very much for your response.
I understand your reluctance to look at this paper again. After all, it is useless to kick a dead horse.
However, LC draws two separate conclusions which are also treated separately in discussions about this paper.
The first conclusion, about the appropriate models to use, has been addressed by Dr. Spencer, albeit not yet in an officially published article.
The second conclusion, that Earth's climate systems show a negative feedback, has only been addressed by Dr. Spencer to the extend that he could not reproduce the negative feedback that LC reported.
Since Dr. Spencer's data analysis is different from LC's, this difference is still used by skeptics as proof that Earth's climate shows negative feedback, contrary to the models.
As for your second sentence, I understand that when Earth's climate systems are observed, that we only measure the "black-box" response of an extremely complex system. To tackle that complexity, we separate primary forcing from feedback in scientific literature as well as in climate models. Since we humans created that separation and we developed a feedback system formula for it, and LC are using that formula, that they should at least use it correctly.
This said, you would provide a significant service to all those people that are still discussing climate science on internet with a clear statement of your opinion on how Lindzen and Choi obtained negative feedback numbers.
To make it easier for you, the mistake (subtraction of 1/G without expalnation) is in the formula for FSW, in paragraph 13 of the paper :
http://www.drroyspencer.com/Lindzen-and-Choi-GRL-2009.pdf
I hope you can spare a few moments of your time to read that paragraph and tell us what you think or LC's reasoning on this issue.
Thank you very much !
OK, you twisted my arm....and I think you are right. There is potential for confusion in the definitions of feedback factor/parameter, but I don't see how that excuses Lindzen subtracting 4 off the SW response. Mind you this still leaves the disagreement with the LW response (although IMO this is pretty worthless given the CMIP/AMIP thing, and Spencer also points to issues with the data processing).
Thank you for your response.
I hope that this should do it if Lindzen and Choi 2009 ever sticks up its ugly head again.
I'm still wonder what Lindzen was thinking when he wrote this paper, and how this got through the GRL review process.
Just a note from a layperson on this issue- I am interested in staying up to date on this issue, but it is difficult when you guys write these posts without references to what the jargon is. Perhaps you can be sure to use the practice of always giving the full name of some piece of jargon before you use the abreviation, so I don't have to search all over the interenet to get the definition of ERBE or ERBS, etc...
Sorry, it's always too easy to drop into the jargon. Here ERBE = earth radiation budget experiment, which refers to data from a satellite.
and to add to that, GRL stands for Geophysical Research Letters, the scientific journal where Lindzen and Choi 2009 was reviewed, approved and then published.
I made some comments on Lindzen & Choi (2009) (some of them are just repeats from here).
How biased can the IPCC be if Christy and Lindzen have written sections for their reports?
Post a Comment