Revision of the boundary layer mixing scheme in the NCEP global forecast system

While the nonlocal boundary layer vertical mixing scheme in the NCEP Global Forecast System (GFS), so called MRF (Medium-range Forecast model) PBL (Planetary Boundary Layer) scheme, provides a realistic development of a well-mixed layer, it has been recognized that the scheme tends to overestimate the daytime PBL growth. This deficiency has been found to be mainly due to too strong entrainment flux at the inversion layer especially for a PBL with strong shear, which is implicitly determined depending on the critical bulk Richardson number specified. In the revision, the entrainment flux for the convective boundary layer is explicitly determined as a function of surface buoyancy flux and friction velocity. The PBL top is taken to be the height at which a parcel near the surface with the scaled buoyancy excess is neutrally buoyant relative to the environment when it lifted upward. For the stable boundary layer, a local closure scheme is used.

A stratocumulus cloud driven vertical diffusion scheme simplified after Lock et al. (2000) is also incorporated into the model. In the simplification, the buoyancy reversal term due to evaporative cooling at the cloud top is neglected and only radiative cooling term is taken into account. For the stratocumulus-topped boundary layer, the PBL top entrainment is obtained from the surface and cloud top driven components. Within the PBL, the total eddy diffusivity is specified as the surface driven diffusivity plus cloud top driven diffusivity. In conjunction with shallow cumulus convection, a criterion to distinguish between the stratocumulus regime, which represents a well-mixed PBL up to cloud top, and the shallow cumulus or decoupled regime, is also included in the revision. For the stratocumulus regime, the shallow cumulus convection scheme is turned off.

The revised scheme appears to display more realistic mixed layer growth with substantial reduction of the PBL height compared to the MRF PBL scheme. The model predictions with the revised PBL scheme are evaluated especially in terms of precipitation and cloudiness.