Debate on Climate Sensitivity

I have been having extensive interchanges on Twitter with a couple of sceptics on Climate Sensitivity (CS). One, @TertiusIII, has posted a longer reply, the opening part of which I have pasted here.

Jim (TertiusIII) : CS is not directly measured empirically. 

It is determined through use of complex mathematical models of our earth (ocean, land and atmosphere). 

RL:  Not so. It is true that models are one line of approach, and the models are more accurate than sceptics imagine, since they are tested against past observations, but there are other lines of approach, all of which converge remarkably on the value of 3*C +/- 1.5*C. Climate Sensitivity must be greater than zero, since otherwise we would not see the natural variations in climate that some/most/all sceptics believe is the sole cause of present observed warming. 

We must always bear in mind that models usually only include fast-acting forcings such as water vapour, clouds, solar activity, lapse rates and ocean currents. Slow positive feedbacks like loss of snow cover and methane releases from tundra are not included, so the models are underestimating temperatures in decades to come.

Rahmsdorf explains (p39) how models are fed data with thousands of different data sets, and then run. Those run with a climate sensitivity of ~3*C are found to give the best match for observational data.

More here on models.

Observational basis for CS


1 Palaeoclimate.
In the Ice Ages, temperatures climbed some 6*C in the periods between ice maxima and minima. If there were no climate sensitivity, these increases would not have been possible. Using temperature, CO2, methane and dust,

2 Volcanoes
Wigley looked at the global temperature recovery times from cooling due to large volcanoes. He found that only a climate sensitivity in the range 1.5-4.5 would account for the observations.
Forster had similar results.


3 Seasonal Cycles
Temperature Variation between seasons gives more data for testing climate sensitivity models. This is a particularly challenging test due to the complexity of modelling the specific regional conditions that govern local climate.

4 Solar Cycles
Studies of solar cycles give values of 2.3-4.1.

5 Ocean temperature changes
Studies of temperature changes associated with oceans gave a lower limit of 1.6, a mode of 2.1 and a median of 6.




Jim, I'm going to pause here (I've kept back the rest of your posting for later).
I hope I have shown fairly that there is a wide observational basis for CS. It is not just generated by computer programmes. 


Of course CS is not directly measured empirically, since it is a derived value.


Do you accept that there are many lines of evidence pointing to the range of 1-4 for CS ?