A new Method for Constraining Effective Climate Sensitivity over Multiple Timescales from Observational Records

It is now well established that Effective Climate Sensitivity (EffCS) evolves over multiple response timescales, due to the different timescales of climate feedback processes in the Earth system. For example, complex climate models reveal the EffCS evolves over many decades as the warming pattern of sea surface temperatures adjusts to anthropogenic forcing. However, observation-constrained estimates generally consider only the Effective Climate Sensitivity value applicable to a single response timescale captured by the historical record, and so may not be applicable to the future.

Here, we use a Bayesian approach to constrain the Effective Climate Sensitivity over multiple response timescales, by combining historical observational data with a novel model of independent fast and slow climate feedbacks. We find, using historical observational reconstructions of surface warming and ocean heat content, that EffCS is best constrained over a 20-year timescale, with significantly greater uncertainty in the upper end of EffCS over a 100-year timescale. We also demonstrate that the different statistical methods for generating global mean reconstructions of surface temperature and ocean heat content have a significant impact in the uncertainty in climate sensitivity, again affecting the long tail of high possible EffCS values.