As the resolution in numerical weather prediction models increases,turbulent motion can be partially resolved by grid scale dynamics.However,a substantial part of turbulent motion still needs to be parameterized,which results in the so-called "grey zone" problem.Assessment of the suitability of traditional PBL (Planetary Boundary Layer) schemes at "grey zone" resolutions would provide foundation for future improvement of PBL schemes,making them more adaptive to the variation of resolutions.In this study,the authors assess the performance of four PBL schemes that are commonly used in WRF,i.e.YSU (Yonsei University),MYJ (Mellor-Yamada-Janjic),MYNN (Mellor-Yamada-Nakanishi-Niino Level 2.5) and MYNN3 in the simulation of convective PBL at "grey-zone" resolutions using LES (Large-Eddy Simulation).Results indicate that at higher resolutions,domain-averaged potential temperature increases in the mixing layer with local PBL schemes such as MYJ and MYNN2.5,while subgrid scale flux decreases and falls below the reference value constructed from LES.The nonlocal terms in YSU hardly change with grid size and do not show distinct scale-sensitivity.In contrast,the counter-gradient term in MYNN3 decreases with reduced grid-size and exhibits a certain degree of scale-sensitivity.The strong nonlocal subgrid mixing in YSU results in a weak,stable stratification in the mixing layer and suppresses resolved-scale turbulence mixing,thus impeding the onset of convectively induced secondary circulations in the model.The suitability of PBL schemes at "grey zone" resolutions varies with certain resolutions.For example,at the resolution of 500 m,none of the four PBL schemes performs well to accurately simulate both the thermal structure and turbulent statistics in the PBL.