Climate change accelerates the hydrologic cycle through a differentiated intensification of moderate and heavy rain rates in response to warming. Enhancement of extreme rain is commonly understood through changes in atmospheric thermodynamics, to first order, while the role of atmospheric dynamics is unclear. Complications appear from differences in modeling assumptions and analysis frameworks, and from intrinsic model limitations such as the scale separation between convection and large-scale atmospheric flow. The present work revisits how changes in the global statistics of rain are driven by atmospheric physics in a hierarchy of climate models, and underscores the modeling uncertainty that arises from the unresolved multi-scale interactions between cloud processes and the atmospheric circulation.

We use a formula which approximates heavy precipitation rates based on atmospheric dynamics and thermodynamics in order to investigate changes in extreme rainfall intensities in a global climate model (CESM), in a superparameterized climate model (SPCAM) and in an idealized cloud-resolving model (SAM) in radiative-convective equilibrium. Detailed comparison of convective dynamics in these three modeling frameworks led to the following conclusions: (1) in all models, increases in extreme rainfall closely follow the 6-7%/K thermodynamic increase in humidity dictated by the Clausius-Clapeyron formula, while changes in convective instability and in the large-scale circulation have a negligible impact, and (2) an additional acceleration of 1-2%/K could arise from the reinforcement of mesoscale circulations associated with convective organization, with nonlinear dependence on surface temperature. These circulations might be crucial ingredients that connect the large-scale atmospheric flow to local convective processes on various scales, and their omission from current convective parameterizations may be a structural source of bias when modeling changes in the hydrologic cycle.