AbstractPlanetary boundary layer (PBL) modeling is a primary contributor to uncertainties in a numerical weather prediction (NWP) model due to difficulties in modeling the turbulent transport of surface fluxes. The Weather Research and Forecasting model (WRF) has provided many PBL schemes that may feature a non‐local transport component driven by large eddies or a one‐and‐half order turbulence closure model, but few of them possess the two features at once. In the present study, a turbulent kinetic energy (TKE)‐based eddy diffusivity/viscosity method is integrated into the non‐local Asymmetric Convective Model version 2 (ACM2) PBL scheme and implemented in WRF. The original first‐order eddy‐diffusivity term in ACM2 is discarded and an extra prognostic equation for TKE, which considers the tendency of TKE by buoyancy, wind shear, vertical transport, and dissipative processes, is supplied to calculate the diffusivity/viscosity. Non‐local transport is modeled the same as ACM2 using the transilient matrix method. Idealized tests using prescribed surface heat flux and roughness length are performed. TKE‐ACM2 displays advantages over the PBL scheme developed by Bougeault and Lacarrère (hereinafter referred to as Boulac) and ACM2 in the wind speeds (WS) profile because it better matches large‐eddy simulations results in the surface momentum flux. Real case simulations show that TKE‐ACM2 generally outperforms in the diurnal vertical profiles of WS under stable conditions. TKE‐ACM2 also produces a better alignment under moderately unstable conditions in the early nighttime at the urban LiDAR station. However, the model exhibits discrepancies more apparently under a more unstable condition during the winter daytime.
Read full abstract