Simultaneous Characterization of the Dynamics of Tropical Precipitation Extremes on Mesoscales and Convective Scales

Substantial uncertainty persists on the fate of extreme rainfall intensities in the warming Trop- ics. The main causes are: (1) misrepresentation of mesoscale (≈100km) mass fluxes in convective parameterizations of general circulation models and (2) the misunderstood effects of large-scale dynamic changes on convective-scale (≈1km) processes simulated in cloud-resolving models. Here we compute fractional increases in extreme daily tropical precipitation intensities on both scales, and therefore defining two types of extreme events, using the multiscale modeling framework of SPCAM.

For both types, these changes are decomposed into two dynamic and one thermodynamic contributions: respectively, the increase in vertically-integrated mass flux (D1), the shift of the mass flux profile towards higher altitudes (D2), and the increase in saturation specific humidity (T). On mesoscales, SPCAM shows more realistic extremes than the standard rainfall parameterization of CAM5 and a good agreement with the Clausius-Clapeyron increase in boundary layer specific humidity (≈+7%/K). Both types of extremes differ in their locations and geometries, with larger rainy areas involved in mesoscale extremes than convective-scale extremes, implying that they may respond differently to climate change. Both agree in sign and relative magnitude on the three contributions: T (dominant, ≈+5-6%K) and D1 (small, positive) tend to enhance while D2 (small, negative) offsets the increase in rain intensity. D2 seems indeed to be controlled by the vertical displacement of the temperature profile resulting from global warming.

The distinction between both kinds of extremes manifests itself quantitatively for D1, D2 and T, and through changes in rainy surface areas of opposite sign. In particular, convective-scale extreme events show an intensification of both their strongest updrafts and strongest downdrafts, leaving the mean mass flux effectively unchanged for convective-scale extremes in a warmer climate. We conclude by commenting on the importance of the convective mass flux definition and on the need to better understand the dynamical interactions between scales (ie, what constrains D1) and their effects on extreme rainfall intensity.