Millions of premature mortalities are caused by the air pollution of fine particulate matter with aerodynamic diameters less than 2.5 μm (PM2.5) globally per year. To effectively control the dominant emission sources and abate air pollution, source apportionment of PM2.5 is normally conducted to quantify the contributions of various sources, but the results of different methods might be inconsistent. In this study, we dynamically harmonized the results from the two dominant source apportionment methods, the source-oriented and receptor models, by updating the emission inventories of primary PM2.5 from the major sectors based on the Bayesian Inference. An adjoint model was developed to efficiently construct the source-receptor sensitivity matrix, which was the critical information for the updates, and depicted the response of measurements to the changes in the emissions of various sources in different regions. The harmonized method was applied to a measurement campaign in Beijing from January to February 2021. The results suggested a significant reduction of primary PM2.5 emissions in Beijing. Compared with the baseline emission inventory of 2017, the primary PM2.5 emissions from the local residential combustion and industry in Beijing had significantly declined by about 90 % during the investigated period of the year, and the traffic emission decreased by about 50 %. The proposed methods successfully identified the temporally dynamic changes in the emissions induced by the Spring Festival. The methods could be a promising pathway for the harmonization of source-oriented and receptor source apportionment models.
Keywords: Bayesian inference; Emission inventory; Receptor model; Source apportionment; Source-oriented model.
Copyright © 2022 The Authors. Published by Elsevier B.V. All rights reserved.