Climate Sensitivity and Cloud Feedback through Three Generations of Parameterized Physics in Aquaplanet Configurations

Climate models produce a wide range of climate sensitivities that has not appreciably changed since the inception of modern atmosphere-ocean coupled climate models. There is widespread belief that a fundamental source of the spread is structural uncertainty associated with representing physical processes that occur on scales smaller than model grid cells, especially the representation of clouds. In this work, we directly examine the dependence of climate response on these parameterized processes using the Community Atmosphere Model (CAM) in idealized configurations that use the same dynamical core and horizontal resolution but different physics packages. Three generations of the model's physics are used (called CAM4, CAM5, and CAM6); the model has evolved dramatically over the past decade, with each successive generation improving the simulated climate in various ways. As the model has changed, the equilibrium climate sensitivity in the fully-coupled configuration has also changed, largely associated with different cloud feedbacks. To isolate and understand these feedbacks, we present a suite of aquaplanet simulations that expands upon the aquaplanet experiments requested for the Cloud Feedback Model Intercomparison Project. In particular we include configurations with a slab-ocean model, allowing richer air-sea interactions and the potential for low-frequency variability. Experiments that increase atmospheric carbon dioxide provide estimates of equilibrium climate sensitivity. We show that cloud feedbacks are generally similar between fixed-SST and slab-ocean experiments, yet the aquaplanet climate sensitivity sometimes differs from more realistic configurations. We explore the different feedbacks among the physics packages as well as the differences with realistic configurations. We show some additional experiments that separately focus on tropical cloud responses and zonal asymmetries as potential causes for the differing feedbacks. Even when differing from the realistic configurations, such experiments provide a fundamental view of the role of parameterized physics and air-sea interaction in climate feedbacks.