Sensitivity Of The Hydrologic Cycle To Surface Flux Enhancement By Convection

The super-parameterized version of the Community Atmosphere Model (SP-CAM) uses a simplified cloud-resolving model (CRM) to represent atmospheric processes that occur on scales finer than the CAM’s grid. A copy of the CRM is embedded in each column of the CAM’s much coarser grid. The physical processes computed on the CRM’s fine grid include cumulus convection, stratiform cloud formation, and radiative transfer. Until recently, however, all versions of the SP-CAM used surface fluxes of sensible and latent heat computed on the CAM’s coarse grid. With this approach, all CRM grid columns in a given CAM grid column received exactly the same surface and sensible heat fluxes. 

With help from software engineers at the National Center for Atmospheric Research, we have created a new version of SP-CAM in which the surface sensible and latent heat fluxes are separately computed for each CRM grid column. This allows the surface fluxes to respond to small-scale thermodynamic and wind-speed fluctuations in the boundary layer, including fluctuations associated with cumulus convection. As a result, the interactions between the surface fluxes and cumulus convection become more realistic.

This project gives an analysis of the sensitivity of the hydrologic cycle to enhanced surface evaporation by convection. With the newer, surface-flux-enhanced version of the model, we expect to see more occurrences of stronger latent heat flux because winds aren’t averaged first; thus, we can expect an increase in gusty cool dry winds that are associated with thunderstorms, which should make for an increase in the evaporation rate. Over land, the fluxes can also differ because of cloud shadows, which create cooler ground and thus a lower latent heat flux. On the other hand, wet patches of soil caused by rain the day before will tend to increase the latent heat flux. 

We have focused on particular regions over both land and ocean. Our results show interesting differences between the two versions of the model in terms of  the mean precipitation, and also in the joint probability distributions of precipitation and the surface latent heat flux.