A new constraint on a key metric that is used to gauge Earth's response to increasing carbon dioxide - the equilibrium climate sensitivity (ECS) - is presented in a paper published this week in Nature. The analysis suggests that extremely high estimates of this sensitivity can be ruled out.

ECS is defined as the global mean warming that would occur if the atmospheric carbon dioxide concentration doubled. It is a key tool for discussing and comparing climate models and an important point of policy discussions, including international climate change agreements. However, a range of ECS estimates have been calculated, which have been hard to reconcile.

Many attempts to constrain ECS have used either the historical warming record or reconstructions of past climates. Methods based on historical warming are affected by multiple factors - including uncertainties in ocean heat uptake and the contribution of aerosols to net radiative forcing - which can affect the accuracy of the estimate. Peter Cox and colleagues now present a probability distribution of ECS, based on the observed historical variability in temperature, rather than the warming trend itself. The authors determine a likely range for ECS of 2.2 - 3.4 degrees Celsius, which is a 60% reduction in uncertainty compared to the estimate from the Intergovernmental Panel on Climate Change of 1.5 - 4.5 degrees Celsius. In addition, this approach allows the authors to almost exclude ECS estimates above 4.5 degrees Celsius or below 1.5 degrees Celsius.