The nub of the AGW hypothesis is not simply that human actions, especially in burning fossil fuels, increase the concentration of carbon dioxide in the atmosphere and cause global warming, but also that the effect of those actions is multiplied by something called ‘climate sensitivity’. It is widely agreed, on all sides of the climate debate, that doubling carbon dioxide in the atmosphere is likely to lead, all other things being equal, to an increase in global mean surface temperature of about 1.1 degrees Celsius.
By itself such an increase is not all that alarming. Other things are rarely equal, and it takes quite a time to double the proportion of CO2 in the air — it was estimated to be about 280 ppm in 1780, and is now about 400 ppm. So an increase of about a degree in global temperature is no reason to press the panic button. The daily range in temperatures for many parts of the world is twenty or more degrees. The panic button comes from the secondary ‘sensitivity’ hypothesis: an increase in temperature will, it is argued, lead to an increase in water vapour and cloud, which will in turn retain more heat in the air, which will lead to more clouds and therefore more heat… The notion of a runaway climate disaster, during which we all fry, and our countryside becomes desert or is flooded, flows from this proposition.
It has to be said that while it is plausible, climate sensitivity as a concept has proved very difficult to pin down observationally, and it is by no means obvious that its sign must be positive — that is, that warming must increase. It is based on several assumptions, none of which has yet been shown to be true. But it is there, in two forms, in the IPCC reports. The first form is ‘equilibrium climate sensitivity’ (ECS), which you can think of as the change in global surface mean temperature caused by a doubling of atmospheric CO2 concentration.
‘Equilibrium’ here is another assumption, and is related to the notion that somewhere, or at some time, our planet had a ‘stable’ climate. It is argued, by the orthodoxy, that by adding CO2 to the atmosphere we human beings have ‘forced’ the climate in an unusual way — jolted it out of equilibrium, as it were. It will take some time (maybe a very long time) for the climate to adjust, the argument goes on, but when it does, its average temperature will go up to a new level. More on that in a moment.
The second form is referred to as ‘transient climate response’ (TCR), and that is said to be the average temperature response over a twenty-year period, with CO2 increasing at 1% per annum. TCR is smaller than ECS. Among the assumptions that are involved here, apart from ECS and TCR themselves, is the notion that CO2 stays in the atmosphere for a very long time, that there really is something called global mean surface temperature that we measure accurately, that sensitivity actually means something, and that clouds and water vapour really do magnify the 1.1 degree C increase that would come from simply doubling the proportion of CO2 in the air.
No matter, the IPCC has no doubts, and indeed its whole case for an urgent and substantial lowering of greenhouse gas emissions falls unless climate sensitivity is real, and works, and is pretty high. The IPCC’s AR4 in 2007 provided a range of 2.0 to 4.5 degrees C, gave a best estimate of 3 degrees C, said it was very unlikely to be less than 1.5 degrees C, and warned that high values couldn’t be excluded. Last year’s AR5 said the range was 1.5 degrees C to 4.5, but this time it did not offer a best estimate. The reasons for the range is that observationally you can’t actually measure ECS precisely, and most of the IPCC case rests on model simulations. The best anyone can do is to estimate and argue, and dozens have.
Increasingly, however, recent estimates have become lower and lower. A new paper by Nick Lewis and Judith Curry (yes, that Dr Curry) published in Climate Dynamics took the line of accepting the IPCC’s own estimates of uncertainty, forcing and heat uptake, and then using matched periods and observational data, rather than model simulations. The desired periods would have the same amount of volcanic activity, and ocean heat that was much the same in all datasets. The results can be seen in the table below.
The authors’ preferred period comparison, because it is the longest, and thus allows the largest change in forcing and also the lowest uncertainty, is shown in the first row of the table. Their best estimate for ECS is 1.64 degrees C, and for TCR 1.33 degrees C. You will see that the best estimate is much closer to the lower edge of the range than to the higher, which suggests (to me at least) that the range has a weighted lower bound — that is, the reality is much more likely to be low than high. The same applies to all the other period comparisons.
Why is this paper important? Because the authors have said, ‘OK. We’ll use the IPCCs own definitions and estimates, and then use observations, not model runs, to see what we get’. There is a fascinating debate in the Comments section of the Climate etc website on the rationale for doing this. So many (and I’m one) have become so tired of the IPCC’s reliance on untested and unverified models that we tend to suspect anything that depends on them.
But Lewis and Curry have accepted the IPCC base as their starting point, and that makes their conclusions really powerful. For if climate sensitivity is as low as 1.64, what is all the anxiety about? This paper is not necessarily a game-changer, but it demands a response from the orthodoxy. You can see an attempt over at Skeptical Science, but I’m not sure I understood what it was about.[Update: There is a better critique on Lewis+Curry at RealClimate: http://www.realclimate.org/index.php?p=17582]