Development of a Scale-Adaptive TKE-Based Subgrid Mixing Scheme in the WRF Model

As the resolution of numerical weather prediction models steadily increases, such that grid spacing becomes comparable to the typical size of the largest energy-containing eddies in the convective boundary layer (CBL), the development of a more general treatment of horizontal and vertical sub-grid turbulent mixing is required to overcome the energetic inconsistency of the conventional parameterization approach. To this end, a coherent parameterization scheme based on the full 3-dimensional (3-D) prognostics equation of turbulent kinetic energy (TKE), which is scale-adaptive, has been developed in the Advanced Research Weather Research and Forecasting (ARW).

This presentation highlights major results from a series of dry CBL simulations using the ARW model that are carried out and compared with the ARW large-eddy simulation (LES) dataset for the purpose of comparing and evaluating a newly-developed general and energetically consistent TKE-based parameterization of sub-grid turbulent mixing together with conventional 1-D TKE-based schemes for parameterizing vertical sub-grid turbulent mixing in the ARW model. The same surface layer scheme is used in the sensitivity experiments so that only the sensitivity of the ARW model to different parameterization schemes of the sub-grid turbulent mixing above the surface layer is examined. The effect of horizontal grid resolution is also examined by running the same case with progressively larger grid sizes of 250 m, 500 m, 1 km and 3 km. We will first compare the characteristics of the ARW-simulated CBL using the two types of schemes. We will then illustrate the importance of including the non-local component in the vertical buoyancy specification in the newly-developed general TKE-based scheme. Finally, comparing the results from the simulations against the WRF LES dataset, we will show the feasibility of replacing conventional planetary boundary layer parameterization schemes with the new scheme in the ARW model.