Reevaluation of Energy Balance Framework Against Observations: A Focus on Cloud Feedback

Abstract

Climate sensitivity, the amount of surface warming in response to increasing carbon dioxide, is a fundamental indicator of the severity of climate change. However, the climate sensitivity estimated by global climate models exhibits large intermodel spread. Satellite observations since 2000 provide a reliable tool to better understand the Earth’s energy balance framework and serve as an observational constraint to cloud feedback inferred from global climate models. This dissertation focuses on climate feedbacks in response to interannual variability and finds systematic disagreement between model-simulated and observed climate feedbacks. However, there is a wide range in the performance of individual models in reproducing the observed feedbacks. Internal variability plays a crucial role in modulating the magnitude of feedbacks on short-term time scale due to the unforced pattern effect, resulting from dependence of climate feedbacks on the surface warming pattern. The modified energy balance framework relates changes in radiative response to changes in tropical atmospheric temperature, yielding more robust comparisons between models and observations as it is less influenced by the unforced pattern effect. Cloud feedback is the major contributor to the pattern effect. During 2000-2020, the difference in observed cloud feedbacks between two consecutive 10-year periods is 1.6 ± 1.1 W/m^2/K, indicating a large unforced pattern effect. Climate models produce the unforced pattern effects of similar magnitude, and models can reproduce the main features of spatial pattern, especially over the East Pacific. This dissertation further evaluates the individual cloud feedbacks in CMIP6 AMIP models by comparing them with the observations. Cloud feedback is decomposed into changes in cloud amount, altitude, and optical depth components for low and non-low clouds. The results suggest tropical marine low cloud, high-cloud altitude, and extratropical high-cloud optical depth feedbacks are the three major components that lead to the discrepancies between observations and models. The comparison provides insights into the cloud feedback components that need improvement in models and motivates future study.