How would you improve this critique of Roy Spencer's recently published study?

Question

This is an argument from an online debate I'm currently participating in. I plan on posting it tomorrow, so I would appreciate any quick suggestions or recommendations that you may have.


"Dr. Roy Spencer argues that previous studies have overestimated the Earth’s climate sensitivity. He presents two main pieces of evidence to support this argument:

1) Lag regression analysis of satellite data shows that net radiative gain precedes, and radiative loss follows temperature maxima. A simple forcing-feedback model appears to show that this is the behavior expected from radiatively forced temperature changes. However, only in the case of non-radiative forcing can an accurate diagnosis of the feedback parameter be made.

2) “There is a rather large discrepancy in the time-lagged regression coefficients between the radiative signatures displayed by the real climate system in satellite data versus the climate models.”

I will demonstrate that both of these arguments are likely incorrect, and Spencer’s resulting conclusions are physically unrealistic. 


Regarding Spencer’s first argument:


I agree that Spencer’s lag regression analysis “supports the interpretation that net radiative gain precedes, and radiative loss follows temperature maxima”. However, this does not necessarily imply that the temperature variations during the relative time period were radiatively forced. 

The simple forcing-feedback model Spencer used to draw this conclusion was much too simple. For instance, his model did not have the ability to simulate ENSO or the hydrologic cycle. 

In addition, the model was likely tuned to give the result it gave. Spencer used an ocean mixed-layer depth of only 25 meters. However, it appears that mixed-layer depths of 100–200 meters are more appropriate for simple climate models. Spencer also used a lambda value of 3, much greater than the consensus value of about 1.4. 

Notably, Spencer made no attempt to find out how sensitive his model fits were to these different parameter values. In fact, he never presented any error bars or uncertainties in his study. 

As Dr. Andrew Dessler stated, “The argument made in these papers...is extremely weak. What they do is show some data, then they show a very simple model with some free parameters that they tweak until they fit the data. They then conclude that their model is right. However, if the underlying model is wrong, then the agreement between the model and data proves nothing.”


Regarding Spencer’s second argument:


Spencer claims to have demonstrated that, “There is a rather large discrepancy in the time-lagged regression coefficients between the radiative signatures displayed by the real climate system in satellite data versus the climate models.”

However, this argument does not withstand the slightest amount of scrutiny. As other scientists have pointed out, Spencer failed to account for decadal variability in the model simulations. 

Clearly, climate model simulations for one decade will never precisely match model simulations for another decade. This is simply due to factors such as random fluctuations in ocean circulation within the climate models.

When scientists account for this decadal variability, they find that the satellite observations fall well within the range of simulations for most models. Thus, there is no fundamental discrepancy between the IPCC climate models and the satellite data. 


The implications of Spencer’s paper are physically unrealistic:


As Spencer stated himself, “Since much of the temperature variability during 2000–2010 was due to ENSO, we conclude that ENSO-related temperature variations are partly radiatively forced. We hypothesize that changes in the coupled ocean-atmosphere circulation during the El Niño and La Niña phases of ENSO cause differing changes in cloud cover.”

However, energy budgets of the surface show that ENSO related temperature variations are not radiatively forced. Therefore, Spencer’s conclusions cannot be correct. 

Moreover, Spencer’s conclusions imply that climate sensitivity is considerably lower than estimates from current climate models. However, numerous paleoclimate reconstructions do not support this. Therefore, past climate changes couldn’t have occurred if climate sensitivity were as low as Dr. Roy Spencer alleges."

A Modest Proposal · Accepted Answer

>>>When scientists account for this decadal variability, they find that the satellite observations fall well within the range of simulations for most models. Thus, there is no fundamental discrepancy between the IPCC climate models and the satellite data.

I assume you derive this discussion from the post at RealClimate:
http://www.realclimate.org/index.php/archives/2011/07/misdiagnosis-of-surface-temperature-feedback/

but I think it could be phrased differently.  I don't think "When scientists account for this decadal variability" correctly nor fully conveys the message that was displayed in the post, which was that when you take into account the full range of model outputs, both from all models and also all models running each decade, the observations fit within that range.  Individual models do or do not match observations well - as was shown in the post, those that replicate ENSO well better match the observations of the past decade, I would presume because of the large role ENSO played in shaping the temperature trends of the past decade.

So "taking into account decadal variability," specifically by comparing against the full range of output values for individual decades (as opposed to Spencer's detrended century comparison), gives part of the picture, but the ability to replicate ENSO completes it.

I would question whether "Thus, there is no fundamental discrepancy between the IPCC climate models and the satellite data" is necessarily true, as they say in the post that there indeed are some models that replicate this decade's data poorly.  I think it's too broad, similar to Spencer's claim.  Spencer's claim, though, that the IPCC models *don't* replicate the data has weak support, one could say no support as his paper does not even properly address the situation.

>>>However, energy budgets of the surface show that ENSO related temperature variations are not radiatively forced. Therefore, Spencer’s conclusions cannot be correct.

Didn't Spencer's results also imply that ENSO was driven by changes in cloud cover?  Is that related to this specific snippet?

jyush - Dessler has already stated he is working on a response paper, my guess to be published.  He said as much in the same email he sent to Romm where Mike paraphrases his statements that the paper is not meant for other climate scientists.

Anonymous · Answer

I'm with Mike on this one. If Spencer's study is flawed, someone should review his study and publish the results. Personal attacks (ad hominem attacks) on Roy Spencer only lead credence to his study. Your own points about ENSO related temperature variations not being radiatively forced and his low estimate of climate sensitivity not being supported by paleoclimate reconstruction is a good start to a possible debunking of his study.

Actually I do have a certain respect for Roy Spencer. If he is proven to be correct, he will win the Nobel Prize in physics. I would not bet the farm on it or but a 1% chance of a Nobel Prizes is thousands of times higher probability that most people, including most scientists have of winning it.

Modest Proposal

Thanks for letting me know.

Ottawa Mike · Answer

I was actually interested in this question until I read this:  "In addition, the model was likely tuned to give the result it gave."

But I gave you the benefit of doubt until I read this:  "...much greater than the consensus value of about 1.4. "

Why don't you just save yourself a lot of effort and conclude that Spencer manipulated his models to get a desired result and his results are therefore incorrect since they are based on assumptions contrary to the consensus?  Or you could just pull a "Trenberth" and claim this study should never have been published.  Or you could pull a "Dressler" and claim that there is no point in even reading this study because "real scientists" don't listen to Spencer any more.

In the greatest of ironies, the more the warmer side seems upset by this study, the more I think it might have merit.

Tim · Answer

I will demonstrate that both of these arguments are likely incorrect, and Spencer’s resulting conclusions are physically unrealistic. 

Sounds like your mind is made up. Your going to prove him wrong.

flossie · Answer

Agree entirely?

How would you improve this critique of Roy Spencer's recently published study?

Related Questions

Trending Questions