Decoding marine low cloud changes reveals more resilient climate feedbacks

Marine low-level clouds' response to warming remains the largest uncertainty source in climate projections, limiting confidence in future warming estimates. We present a principal component-based method that integrates the physical processes associated with complex and indirect nonlinear covariations among multiple interdependent meteorological factors, providing a robust observationally constrained framework for reducing model biases. Using this framework, we find a smaller marine low-level cloud fraction reduction than direct model projections, particularly under the high CO₂ emission scenario. While models simulate a substantial future decrease in marine low-level cloud fraction during the 21st century, our results show marine low-level clouds reveal more moderate declines and even regional increases. The resulting global marine low-level clouds feedback under the high CO₂ emission scenario is 0.09 ± 0.09 Wm⁻²K⁻¹, less uncertain than idealized abrupt CO₂ increase experiments estimates, suggesting that atmosphere-ocean adjustments buffer cloud responses and the climate system may be more resilient than previously thought.