Stochastic Backscatter for Cloud-Resolving Models. Part I: Implementation and Testing in a Dry Convective Boundary Layer

Abstract

In simulations of deep convection with cloud-resolving models the turbulence is often rather poorly resolved, and the influence of the subfilter-scale parameterization used in such circumstances is probably greater than in better-resolved simulations. Therefore a study to investigate the influence of stochastic backscatter was performed and presented in two papers. This first paper focuses on a description of the stochastic backscatter model and its effect on a neutral and a dry convective boundary layer. The second paper then deals with two cases of deep convection. The dry convective boundary layer is typical of the subcloud layer in deep convection and this study allows for influences on this layer to be investigated separately. As a simple case of convection it also allows for general effects to be identified. The implementation of stochastic backscatter was improved to ensure an appropriate scale of backscatter that is independent of any mesh refinements and always spatially isotropic. It can also be applied in a deep atmosphere with use of the anelastic approximation. Generally the backscatter is found to affect the velocity and scalar variances as well as the spatial structure and time development of the simulations. An additional issue is the marked influence of both varying resolution and backscatter on the near-surface temperature differences. This is an element of the complex issue of surface transfers under free convection conditions. Overall, the investigations show that the backscatter gives some beneficial changes to the simulations, which tend to keep results in less well- resolved cases closer to those in better-resolved simulations.