ABSTRACT
Using a Machine Learning Model (MLM) to decouple meteorological parameters, this paper quantified true impacts of emission reduction by pollution sources resulting from COVID-19 on air quality in Xianyang. Compared with the non-epidemic scenario, the results showed that concentrations of PM2.5, PM10, SO2, NO2, and CO in Xianyang had significantly decreased by 19.3%, 26.0%, 13.4%, 60.1% and 9.1%, respectively, with NO2 decreasing the most, SO2 and CO decreasing slightly, and O3 increased by 50.9% conversely. Under the condition that both primary emission and precursors of secondary particulate matter decreased, the concentration of PM2.5 dropped lower than expected, and O3 increased though, showing the complexity of PM2.5 and O3 control, in the meanwhile implying that the impact of operating pollution sources during the epidemic on air quality was greater than malfunctioned sources, and official regulations to restrict and suspend production in factories (similar to the impact of the pandemic) had limited improvement on air quality. In the future, emphases should be put on the treatment of operating pollution sources during the pandemic such as scattered coal and biomass combustion, heat production and supply, and crude oil processing and petroleum product manufacturing. © 2021, Editorial Board of China Environmental Science. All right reserved.