To explore the application of high-temporal-resolution data in PM2.5 source apportionment during air pollution events, ambient air PM2.5 components were continuously monitored in urban Nanjing from January to December, 2017. Commercially available instruments for continuous measurements were deployed to obtain hourly concentrations of elements, water-soluble ions, and carbonaceous components of PM2.5. Data for 15 elements and 5 bulk components during three pollution events(firework combustion during the Spring Festival, a spring sandstorm, and a winter haze event) and across the whole year comprised four datasets for source apportionment using positive matrix factorization(PMF), and the distribution of factor/source contributions and estimations of average concentrations of characteristic components were compared based on different input datasets(PMFfirework-sand-haze and PMFfull-year). The results showed that the identified factors/sources, factor profiles, and contributions differed largely between PMFfirework-sand-haze and PMFfull-year solutions. For example, the relative average contribution of the firework combustion factor derived from the PMFfull-year solution(was 1.50%) was far less than that of the PMFfirework solution. The dust factor had an average contribution of 8.51% in the PMFsand solution, which was approximately double that of the PMFfull-year solution. This might be explained by the fact that PMF assumes unvaried source compositions during the measurement campaign, meaning that the source apportionment results based on long-term observations will include bias due to changes in emission sources. Furthermore, during the firework combustion event, the estimated average concentration of K from the PMFfirework solution[(1.32±1.17) μg·m-3, P=0.64]was closer to measured value[(1.36±1.19) μg·m-3]than that of the PMFfull-year solution[(1.16±1.19) μg·m-3, P=0.0090]. For the sand storm event, the concentrations of Fe, Si, and Ti were significantly underestimated by the PMFfull-year solution[(0.061±0.042)-(1.06±0.65) μg·m-3, P<0.05], while their peak concentrations agreed well between the PMFsand estimations and the observations. During the winter haze event, all PM2.5 bulk components were well estimated by both the PMFfull-year and PMFhaze solutions. Based on these results, PMF source apportionment results based on continuous measurement data during pollution events can reasonably reflect short-term variations in characteristic PM2.5 components and their sources, which can improve the timeliness of air pollution source apportionment.
Keywords: PM2.5; PMF source apportionment; continuous monitoring; contribution distribution; pollution event.