Historical Pattern Effects and Climate Sensitivity Revisited with Novel Constraints on Past Warming Patterns

Vincent T Cooper , Kyle C. Armour, Jonathan M. Gregory, Timothy Andrews, Moritz Guenther, Chongxing Fan

The historical record (1850–present) can constrain equilibrium climate sensitivity (ECS) only if we know the patterns of sea-surface temperature (SST) and sea ice concentration (SIC) that shape radiative feedbacks. Feedbacks depend on spatial patterns of SST and SIC (“pattern effects”), yet the impact of historical SST/SIC uncertainty on feedback estimates has not been systematically quantified. Here we use a new coupled reanalysis of SST/SIC to quantify feedbacks over 1850–2023. We prescribe 3–14 ensemble members of SST/SIC in six atmospheric models with constant preindustrial forcings. Compared to previous estimates, our ensemble-mean feedbacks vary less over time and are consistent with coupled models’ abrupt-4xCO₂ feedbacks over 1901–1960. With this context, the strongly stabilizing feedback over 1975–2014 is even more anomalous than previously identified. Importantly, feedbacks have become less stabilizing since 2014 in simulations and observations, but this development has little impact on feedbacks over the full record. From anomalies over 2006–2023 relative to 1850–1900, we estimate a pattern effect (i.e., the long-term 2xCO₂ feedback minus the historical feedback) of 0.62 W m⁻² K⁻¹ (−0.05 to 1.03, 5–95%). SST/SIC tends to be a larger source of pattern-effect uncertainty than intermodel spread. Revising historical constraints on ECS through 2023 (including increases in global warming, Earth’s energy imbalance, and radiative forcing) yields a maximum likelihood of 3.4 K from the historical record alone, and combining lines of evidence yields a median of 2.8 K (2.1–3.9 K, 5–95%; approximate robust upper bound at 4.4 K).