A new paper by Steven Sherwood et al. published in Nature this week is predicting a higher climate sensitivity of 3°C or more per doubling of CO2. The paper is called Spread in model climate sensitivity traced to atmospheric convective mixing.
Current projections of climate sensitivity fall in the range of 1.5-4.5°C. Part of the reason there is such a large range is because of the uncertainty in how clouds will behave in a warmer world. Low cloud cover has a net cooling effect on the globe through the reflection of incoming sunlight whereas mid- and high cloud clover has a net warming effect due to the absorption of long wavelength radiation.
Sherwood et al. find that a warmer world will produce fewer low-level clouds and so reduce the cooling effect that these clouds will have on our climate. Without this added reflectivity, the warming will be greater. The physical explanation for this is as follows: As air collects water from the ocean surface it rises, sometimes by a only few kilometres before returning back to the surface and at other times by 10-15 kilometres. The rise and descent of water vapour at the lower level is termed lower-tropospheric mixing and acts to pull water vapour away from layers where clouds that cool the climate form. Hence fewer clouds form in these layers.
Climate models that pitch climate sensitivity at less than 3°C do not incorporate this shallow circulation of water vapour and so, according to Steven Sherwood, are inconsistent with observations.
Steven Sherwood explains this all rather well in the following video:
For a more detailed explanation see A bit more sensitive… at Real Climate.
Some other articles about the same paper:
The Guardian: Planet likely to warm by 4C by 2100, scientists warn.