Multiple technical observations of the atmospheric boundary layer structure of a red-alert haze episode in Beijing

Abstract

The study and control of air pollution involves measuring the structure of the atmospheric boundary layer (ABL) to understand the mechanisms of the interactions occurring between the atmospheric boundary layer and air pollution. Beijing, the capital of China, experienced heavy haze pollution in December 2016, and the city issued its first red-alert air pollution warning of the year (the highest PM2.5 concentrations were later found to exceed 450 µg m−3). In this paper, the vertical profiles of wind, temperature, humidity and the extinction coefficient (reflecting aerosol concentrations), as well as ABL heights and turbulence quantities under heavy haze pollution conditions, are analyzed, with data collected from lidar, wind profile radar (WPR), radiosondes, a 325 m meteorological tower (equipped with a 7-layer ultrasonic anemometer and 15-layer low-frequency wind, temperature, and humidity sensors) and ground observations. The ABL heights obtained by three different methods based on lidar extinction coefficient data (Hc) are compared with the heights calculated from radiosonde temperature data (Hθ), and their correlation coefficient can reach 72 %. Our results show that Hθ measured on heavy haze pollution days was generally lower than that measured on clean days without pollution, but Hc increased from clean to heavy pollution days. The time changes in friction velocity (u*) and turbulent kinetic energy (TKE) were clearly inversely correlated with PM2.5 concentration. Momentum and heat fluxes varied very little with altitude. The nocturnal sensible heat fluxes close to the Earth surface always stay positive. In the daytime of the haze pollution period, sensible heat fluxes were greatly reduced within 300 m of the ground. These findings will deepen our understanding of the boundary layer structure under heavy pollution conditions and improve the boundary layer parameterization in numerical models.