No matter how well we understand the climate system, the possibility of extreme temperature increases is unlikely be ruled out entirely

Past attempts at narrowing down the range of predicted climate sensitivity — the long-term change in global temperatures with doubling atmospheric carbon dioxide concentrations — have not been met with much success. A new theoretical study1 shows that even a much better understanding of the Earth's responses to increasing greenhouse gases is not likely to make a difference to this problem.

Gerard Roe and Marcia Baker from the University of Washington in Seattle use mathematical arguments to investigate the reasons for the elusiveness of climate sensitivity. They find that the uncertainties in individual feedbacks such as sea ice extent, clouds, water vapour or deforestation — which mutually reinforce changes in temperatures — all add up. So as long as our understanding of this multitude of feedbacks operating in the Earth system is not perfect — which it probably never will be — we are unlikely to narrow down the range of possible climate sensitivities much more than we already have done.

Extremely high temperature changes are most affected by these uncertainties, making it very hard to rule out temperature changes in excess of 4.5 °C as CO2 concentrations double. As we get closer to this level of greenhouse gas, the likely temperature trajectory will, however, become clearer.