Sensitivity of urban boundary layer simulation to urban canopy models and PBL schemes in Beijing

Mesoscale models with urban canopy models (UCM) have been increasingly used to study urban boundary layer processes. Using the data from a high-resolution Doppler lidar, automatic weather stations (AWS), and a flux tower located in the urban site, we assessed the performance of the urbanized Weather Research and Forecasting (WRF) model through three urban canopy models (the single-layer UCM, and the multi-layer BEP and BEM models) and four planetary boundary layer (PBL) schemes (the non-local first-order YSU, SH and ACM2 schemes, as well as the local TKE-based BouLac scheme) for one cloudy and one clear sky days. Results show that the WRF-Urban generally overestimates the sensible heat flux and underestimates the latent heat flux. The simulated 2-m temperature and 10-m wind speed are more sensitive to UCMs than to PBL schemes. Using the BouLac PBL scheme and the multi-layer BEP generates the best agreement with AWS observations. Simulations with the multi-layer BEM produce the highest mixing-layer heights. The convective boundary layer (CBL) from the single-layer UCM experiment develops at the slowest pace when compared with other two multi-layer UCMs. When the single-layer UCM is used, simulations with the non-local mixing YSU, SH and ACM2 schemes perform better than the TKE-based scheme (BouLac) for representing the CBL structure. Additionally, the scale-aware SH scheme considering the effect of grid resolution on the vertical dimension, simulates the potential temperature profiles that are closest to observations.