Abstract: Turbulent mixing in the atmospheric planetary boundary layer (PBL) is one of the most important process to transport energy to maintain atmospheric movement. As the model grid spacing is close to the length scale of energetic turbulent eddy, turbulence can only be partly resolved. This is called the "grey-zone" problem. Traditional PBL schemes are not suitable for the description of turbulent transport in numerical models. In order to improve the capability of PBL scheme in multi-scale models, including the "grey-zone" simulation, the Mellor-Yamada-Nakanishi-Niino (MYNN) scheme was improved by introducing non-local turbulent parameterization of heat and moisture turbulent flux and mesh-scale dependent turbulent length scale for self-adapting arrangement in a wide range of model resolutions based on the Reynolds average numerical simulation. The self-adapting MYNN scheme is then used to simulate a sea fog case in the Huanghai region that occurred on 26 February 2014 using the one-way nesting Weather Research and Forecast (WRF) model with horizontal resolutions of 3 km and 1 km, and 1.5 km and 0.5 km, respectively. Results are compared with simulations using the original MYNN scheme. The new MYNN scheme demonstrates capability for reasonable simulation of vertical turbulent transport of heat and moisture in a kilometer-resolution model. Compared with satellite images, the integrated low-level cloud water shows a similar horizontal coverage of sea fog in the simulations with different resolutions using the self-adapting MYNN scheme. Vertical profiles of temperature and humidity also illustrate structural distributions that agree better with the ERA-interim reanalysis data compared with results simulated with the original scheme. Preliminary results show that the self-adapting MYNN scheme can be applied in multi-scale models.