Can feedback analysis be used to uncover the physical origin of climate sensitivity and efficacy differences?

Different strengths and types of radiative forcings cause variations in the climate sensitivities and efficacies. To relate these changes to their physical origin, this study tests whether a feedback analysis is a suitable approach. For this end, we apply the partial radiative perturbation method. Combining the forward and backward calculation turns out to be indispensable to ensure the additivity of feedbacks and to yield a closed forcing-feedback-balance at top of the atmosphere. For a set of CO 2 -forced simulations, the climate sensitivity changes with increasing forcing. The albedo, cloud and combined water vapour and lapse rate feedback are found to be responsible for the variations in the climate sensitivity. An O 3 -forced simulation (induced by enhanced NO x and CO surface emissions) causes a smaller efficacy than a CO 2 -forced simulation with a similar magnitude of forcing. We find that the Planck, albedo and most likely the cloud feedback are responsible for this effect. Reducing the radiative forcing impedes the statistical separability of feedbacks. We additionally discuss formal inconsistencies between the common ways of comparing climate sensitivities and feedbacks. Moreover, methodical recommendations for future work are given.