LAUSR

A climate model emulator that mimics an ensemble of state‐of‐the‐art coupled climate models has been used for probabilistic climate projections. To emulate and compare the latest and previous multimodel ensembles, this study establishes a new method to diagnose a set of parameters of effective radiative forcing, feedback, and impulse response functions by fitting a minimal emulator to time series of individual models in response to step‐ and ramp‐shaped CO2 forcing up to a quadrupling concentration level. The diagnosed CO2 forcing is scaled down to a doubling level, leading to an unbiased estimate of equilibrium climate sensitivity. The average climate sensitivity of the latest ensemble is 18% and 13% greater than that of the previous ensemble for equilibrium and transient states. Although these increases are subject to data availability, the latter smaller rate is significant and is consistent with the relationship between feedback strength and response timescales.