摘要

We compare daytime planetary boundary-layer (PBL) characteristics under fair-weather conditions simulated using a single column version of the Weather Research and Forecasting model with different PBL parametrization schemes. The model is driven only by prescribed surface heat fluxes and horizontal pressure gradient forcing. Parametrizations for all physical processes except for turbulence and transport in the PBL are turned off in the simulations to ensure the discrepancies in the simulated PBL flow are due only to differences in the PBL schemes. A large-eddy simulation (LES) of the evolution of a daytime PBL is performed as a benchmark to examine how well a PBL parametrization scheme reproduces the LES results, and performance statistics are compared to rank those schemes. In general, hybrid local and non-local schemes such as the Yonsei University and Asymmetrical Convective Model (version 2) schemes perform better in reproducing the LES results, particularly well-mixed features, than do local schemes. Among local schemes, the University of Washington scheme produces the results closest to the LES. Local schemes, such as those of Mellor-Yamada-Janjic and Mellor-Yamada-Nakanishi-Niino, simulate too low an entrainment flux, while PBL heights diagnosed from the simulations using local schemes are lower than those from the LES results. Hybrid local and non-local schemes are more sensitive to vertical grid resolution than local schemes. With a higher vertical resolution in the PBL, the schemes using the eddy-diffusivity and mass-flux methods perform better. Differences in the values of eddy diffusivity, length scale, and turbulence kinetic energy and their vertical distributions are large.