Future changes in heavy rainfall intensities have been analyzed in simulations of present and future climate on a variety of scales and model configurations, from the 1-2◦ resolutions typical of current General Circulation Models (GCMs) down to the 100m-5km grids of Cloud-Resolving Models (CRMs). Thermodynamic increases in specific humidity systematically lead to an increase in precipitation extremes, but the sign, magnitude and relative importance of changes in the dynamics are not well understood. Dynamic contributions have multiple sources and include changes in the large-scale circulation (the apparent cause for inter-GCM spread in the increase of tropical extremes), in the mean mass flux and its vertical profile in strongest updrafts, and in the area of convective updrafts possibly associated with convective organization.

Here we quantify these dynamic effects on the intensity of daily precipitation extremes at both large and convective scales. We use the multiscale modeling framework of SP-CESM, where CRMs are embedded in the GCM’s grid cells to replace parameterizations of moist convection, and compare global simulations under pre-industrial and 2×CO2 conditions. Large- scale quantities are defined at the GCM grid resolution. Convective-scale quantities are defined within a GCM grid cell as averages over the points containing the largest 90% of the large- scale condensation. A scaling expression for extreme rainfall intensities is then used to quantify the contributions from changes in the mass flux vertical mean and profile, on both scales. In addition, the ratio of large-to-convective-scale precipitation values gives a measure of the surface fraction of convective updrafts, and is used to quantify contributions from changing updraft geometry.

We first show the benefits of explicitly resolving convection for large-scale extremes are assessed by comparing SP-CESM and CESM runs with identical forcings. We then compare the behaviors of large-scale and convective-scale extremes and their response to each of the dynamical contributions listed above. Finally, we comment on the importance of improving our understanding of the effects of large-scale convergence on the properties of convective-scale rainfall.