Model-free Climate Sensitivity is still dangerous

On Twitter I bump into Bishop Hill, a well-known climate contrarian blogger. 

He readily agrees that a doubling of CO2 would increase the global temperature by 1.1*C per se, and puts Climate Sensitivity (CS) at 1 - 2.5*C.

To summarise, any increase of temperature over 2*C is regarded as dangerous, which is why the contrarians have made many efforts to prove that sensitivity is below 2*C.

Fair play to the Bishop. This is the first time I have had a clear and unequivocal answer on CS from a contrarian. 

This moves the game on from interminable equivocations about the vast field of climatology to a narrow discussion focused on the question of how much the planet's energy system will change in response to our greenhouse gases.

The Bishop makes his case on his blog here.

He starts with a well-known IPCC figure summarising the many lines of inquiry that converge on a CS of ~3*C. The colours represent different studies, the x-axis represents probability, and the y-axis represents the expected temperature change.

source IPCC

The Bishop then deletes all that are in any way tainted by computer models. This leaves only the green line, a  study by Forster & Gregory 2006  , which used measurements by the ERBE satellite and surface temperatures.  It originally returned CS values of 1 to 4.1 with a mean of 1.2*C, but the contrarians have reduced it to 1-2.1 because the authors had used a statistical operation (of which they disapproved) to stretch the upper limit. 

The Bish says: "the paper by Forster and Gregory, which is the only wholly empirical study in the corpus".

So I searched for empirical (that is, based on observation, not primarily computer modelling) papers on climate sensitivity.

Empirical estimation is a relatively simple thing to do (if you are a climatologist): you find some known cause of change in the earth heat budget, measure the temperature change that results from that change, and do some calculations.

The list below is what I found. It is probably not complete. I have listed them by name, year, with the lowest and highest values for CS, and a brief reference to the methodology.

1 Claude Lorius et al.                      3 - 4

Ice core records

2 Hoffert & Covey 1992         1.4 - 3.2

"Deriving Global Climate Sensitivity from Palaeoclimate Reconstructions," Nature Vol. 360, pp. 573-576. 

3 Hansen 1993                               2 - 4

Ice age data

4 Chylek et al. 2007                       1.2 - 2.0

Aerosols

5 Tung 2007                                 2.4 - 4

Solar variation

6 Bender 2010                               1.7 - 4.1

Pinatubo volcano

7 Hansen et al. 2008                       3 - 6

Paleoclimate  

8 Gregory 2009                              1.6 - 6.1

changes in ocean heat.

9 Schwartz 2010                             0.6 - 1.6                        

Ocean heat

Schwartz has been sharply criticised here.

10 Forster & Gregory  2006      1 - 2.1

Satellite observations (originally given as 1 - 4, see above)

So there we have it. Ten papers on climate sensitivity derived by empirical  means. 

Bishop complains that some of these use computer models within their empirical method. I believe that is taking the objection against models too far. It is one thing to question their projections of the future, but quite another to object to the use of models per se.

The average boils down to  1.8 - 3.5 which comfortably contains the danger threshold of 2*C, and the canonical figure of 3*C. And that's including Schwartz.

Without Schwartz  and with FG2006 at the original value, it moves to 1.9 - 3.7*C.

Therefore, if contrarians choose totally to  reject the validity of climate models to derive sensitivity and insist that only empirical evidence is admissible in the vital matter of climate sensitivity, it really makes no difference. The danger of severe perturbation of the climate of our planet is still there. 

The hypothesis that man-made CO2 will not adversely affect the planet is still without empirical evidence.

More on this topic, including - why is sensitivity slightly lower for empirical studies?

(This post has been updated)

More posts about modelling on this blog:
"Data" does not necessarily prove the models wrong
Climate models - are they scientific and reliable?
Climare model projections - what impact do ocean currents and solar variation have?