Abstract:

Mesoscale climate models provide indispensable tools to understand land-atmosphere interactions over urban regions. However, uncertainties in urban canopy parameters (UCPs) and parameterization schemes lead to degraded representation of the drag effect in complex built terrains. In particular, for the widely applied single-layer urban canopy model (SLUCM) coupled with the Weather Research and Forecasting (WRF) model, near-surface horizontal wind speed is known to be overestimated systematically. In this study, idealized large eddy simulations (LES) and WRF-SLUCM simulations are conducted to study the separate effect of UCPs and aerodynamic parameterization on atmospheric boundary layer processes and rainfall variabilities in Phoenix, Arizona. For LES that explicitly resolves surface geometry, significant differences between three-dimensional (3D) versus two-dimensional (2D) representation of urban morphology are found in the surface layer and above. When surface drag is parameterized following SLUCM, surface morphologies have little impacts on the mean momentum transfer. WRF-SLUCM simulation results, incorporated with 3D urban morphology data, indicate that simply refining the frontal area index will reduce the surface drag, which further amplifies the systematic positive bias of SLUCM in predicting horizontal wind speed. Replacing the drag parameterization in SLUCM by LES-based aerodynamic parameters has evident impacts on near-surface wind speed. The impact of urban roughness representation becomes the most evident during rainfall periods, due to the important role of surface drag in dictating moisture convergence. Our study underlines that apart from intensive efforts in obtaining detailed UCPs, it is also critical to enhance the urban momentum exchange parameterization schemes.