A Stochastic Eddy-Diffusivity/Mass-Flux Boundary Layer and Shallow Convection Parameterization in 1-D NCEP GFS: Implementation and Initial Results

In this study a stochastic multi-plume model for turbulence and convection is developed and implemented in a one-dimensional (1-D) version of the National Center for Environmental Prediction (NCEP) Global Forecast System (GFS) model. The parameterization is based on Sušelj and Teixeira (2016) but includes substantial modifications to adapt it to the GFS's environment. Key elements of the scheme are (a) multiple updrafts that are integrated independently in vertical, (b) updrafts originate at the surface with varying thermodynamic properties based on surface conditions, (c) stochastic updraft entrainment, and (c) environmental eddy-diffusivity flux based on the turbulent kinetic energy. By design, the scheme is weakly sensitive to the prescribed initial updraft area fraction. Initial evaluation of the scheme is presented, which is based on several idealized experiments, such as dry convection, shallow cumulus convection, stratocumulus topped boundary layer, and a stratocumulus-to-cumulus transition case. Large eddy simulations (LES) serve as a reference for parameterization evaluation. The new scheme outperforms the current operational set of parameterizations in GFS, showing very good agreement with LES. Our results further substantiate advantages of the stochastic multi-plume model introduced by Sušelj and Teixeira (2016). Next steps will include evaluation in global 3D GFS simulations.