This paper proposes an improved vertical diffusion package with a nonlocal boundary-layer vertical diffusion. Based on the study of Noh et al. (2003), and accumulated realism of the behavior of the Troen and Mahrt concept implemented by Hong and Pan (1996) for the National Centers for Environmental prediction (NCEP) Medium-range forecast model (MRF) and the National Center for Atmospheric Research (NCAR) 5th generation Mesoscale Model (MM5), a revised vertical diffusion algorithm that are suitable for weather forecasting and climate prediction models is developed after further generalization and reformulation.

The scheme has been successfully implemented onto the WRF model and MRF model by producing a realistic structure of the PBL. The sensitivity of the boundary-layer development to the tuning parameters in the revised nonlocal diffusion scheme is also investigated. The scheme is capable of resolving a problem of too strong mixing in the mixed layer, particularly, when wind is strong. Too early development of the PBL before noon is also resolved. Some systematic biases of the large-scale features including cold bias at 850 hPa after noon in the MRF PBL are improved. The scheme shows a better skill in the simulation of tropical rainfall in response to the SST anomalies over the Tropics.