Due to constraints upon computational power, large-scale numerical atmospheric models cannot presently resolve all pertinent scales of motion. Until such time that computational resources allow global numerical models of an arbitrarily high resolution, it will be necessary to parameterize the effects of subgrid-scale atmospheric processes upon the resolved dynamics. For studying such small-scale processes, and developing parameterizations for them, the use of high resolution numerical models is often a key element. This research applies such small-scale models to a number of different problems involving surface fluxes of heat and moisture.

The University of Utah Cloud Resolving Model is used to characterize the effects of cumulus and mesoscale circulations on the calculation of the surface turbulent fluxes of sensible and latent heat over a tropical ocean. Enhancements of over 25% were common when compared with flux calculations that neglected these effects. Parameterizations that include these enhancements were constructed using large-scale modeled quantities, including cloud mass fluxes, vertical kinetic energy, and surface precipitation rates. In general, these parameterizations were shown to perform better than the more simple methods typically used.

Leads, quasi-linear openings that form within the interior of a sea ice pack, significantly enhance local surface fluxes. Despite their typically small size, unresolvable by large-scale models, leads can also have a major impact upon the large-scale surface heat budget of the Arctic. The University of Utah Cloud Resolving Model is used to examine the convective plumes that form in the vicinity of leads, and determine their impact upon the large-scale energy budget. Under typical wintertime conditions, simulated convective plumes penetrated to heights of hundreds of meters and impacted surface fluxes over 50 km downwind.

In addition, the development of a new three-dimensional Large-Eddy Simulation (LES) model is described. This model incorporates a nonhydrostatic quasi-compressible framework, monotone scalar advection, and efficient time-split integration of the elastic modes. In comparisons with similar LES models, the new model is as good or better at reproducing observational and theoretical results for convective boundary layer and radiatively cooled smoke cloud simulations.