[Characteristics of Two Pollution Episodes Before and After City Heating in Beijing from February to March of 2019].

The characteristics of meteorological conditions and pollutant concentrations were analyzed based on two pollution episodes before and after city heating in Beijing during February to March of 2019. The backward trajectory and WRF-CAMx models were used to analyze the evolution of pollutants before and after city heating, and the influences of meteorological conditions, regional transport, and secondary transformation on the episodes were discussed. There was little difference in the average ρ(PM2.5) between February 21-24 (episode 1) and March 18-20 (episode 2), with concentrations of 100.1 µg·m-3 and 97.2 µg·m-3, respectively. However, compared with that of episode 2, in episode 1 the average peak value was higher with two peak stages, the diurnal variation was clearer, and the process developed much more rapidly. Moreover, episode 1 was regional pollution, while episode 2 was more related to local pollution in Beijing. The SO2 concentrations in both episodes were not higher than 16 µg·m-3, thereby indicating the effectiveness of coal-burning treatment and other measures. In addition, two peaks occurred in the diurnal fluctuation of SO2 in episode 1, whereas only one peak occurred for episode 2. In episode 1, the CO concentration was high and the ratio of ρ(CO)/ρ(SO2) increased around February 22-23 (phase 1); moreover, the pollutant concentrations in the central and southern areas of the Beijing-Tianjin-Hebei region and those in the background sites located in the southern part of the Beijing plain were higher than those in the urban area, thereby indicating that the diffusion conditions of episode 1 were unfavorable and the first PM2.5 peak was mainly affected by regional transport. A high ratio of ρ(PM2.5)/ρ(CO) in episode 2 suggested a slightly larger proportion of secondary generation for PM2.5, whereas higher ratios of ρ(NO2)/ρ(CO), ρ(SO2)/ρ(CO), and ρ(SO42-)/ρ(PM2.5) in episode 2 and the similar SOR value to that of episode 1 demonstrated that episode 1 was more advantageous for gas phase transformation and episode 2 was more affected by the coal industry. Phased analysis of episode 1 showed that the indicators of second generation for PM2.5 in phase 2 (around February 23-24) of episode 1 and episode 2 were similar, and both were higher than that in phase 1 of episode 1, which implied that the second PM2.5 peaks of episode 1 and episode 2 were mainly related to local emissions and chemical conversion. Both WRF-CAMx with and without assimilation experiments could better reproduce the temporal variation in pollutants, and the correlation between the simulation and observations increased but with lower values after assimilation. The model performance for the PM2.5 trend simulation significantly increased with data assimilation, and the simulated lower NO2 in February and higher NO2 in March as well as the overestimated SO2 were also improved. In addition, the pollutant concentration simulation in Beijing was more sensitive to that of Hebei in episode 1, which suggested that episode 1 was more affected by regional transport. The simulation ability for the rapid growth of pollutants needs to be promoted, and the response of pollutant types to emission reduction and the feedback related to the atmospheric oxidant and aerosol properties may be important for the simulation effect, which all require further study.