Climate models underestimate global decreases in high-cloud amount with warming

Cloud feedback has prevailed as a leading source of uncertainty in climate model projections under increasing atmospheric carbon dioxide. Cloud-controlling factor (CCF) analysis is an approach used to observationally constrain cloud feedback, and subsequently the climate sensitivity. Although high clouds contribute significantly toward uncertainty, they have received comparatively little attention in CCF and other observational analyses. Here we use CCF analysis for the first time to constrain the high-cloud radiative feedback, focusing on the cloud amount component owing to its dominant contribution to uncertainty in high-cloud feedback. Globally, observations indicate larger decreases in high cloudiness than state-of-the-art climate models suggest. In fact, half of the 16 models considered here predict radiative feedbacks inconsistent with observations, likely due to misrepresenting the stability iris mechanism. Despite the suggested strong high-cloud amount decreases with warming, observations point toward a near-neutral net high-cloud amount radiative feedback, owing to almost canceling longwave and shortwave contributions.