Item

Abstract

Equilibrium Climate Sensitivity (ECS) – a measure of the sensitivity of global temperature change to a doubling of globally averaged atmospheric carbon dioxide concentration - is a critically important property of the climate system (IPCC, 2022). Unfortunately, the magnitude of ECS is not well constrained, its likely range currently sitting between 2.6 and 4.1°C. Narrowing this range is critical to reducing spread in projected climate change and has the potential to save trillions of dollars from better targeted adaptation and mitigation planning. Previous studies have suggested that this uncertainty might be reduced by exploiting relationships between model estimates of ECS and of observable quantities, including the magnitude of variability in globally averaged surface air temperature (GT). Here we estimate GT variability by the standard deviation in annual mean running GT trends and examine its links to ECS in CMIP5 and CMIP6 models. We show that model to model differences (M2MDs) in ECS are statistically significantly linked to M2MDs in the scaled magnitude of internally generated climate variability, with 0.4 correlation in both sets of models. Scaling is obtained by normalising the variability of GT by the variability in monthly Top-of-Atmosphere radiative imbalance. The resulting relationship, together with observations of scaled GT, is used to estimate real world ECS. In addition, we find strong evidence of a robust link between the statistical properties of GT and ECS. This strongly suggests that some of the physical processes responsible for internally generated climate variability are also partly responsible for setting ECS.