Submicron particles in polluted regions have received much attention because of their influences on human health and climate. A k-means clustering technique was performed on a dataset of particle number size distributions (PNSD) that was obtained over more than 3 years in the Yangtze River Delta (YRD) region of East China. With simultaneous measurements of meteorological conditions, trace gases and aerosol compositions, 7 clusters were categorized and interpreted. Cluster 1 and cluster 2, which accounted for 9.9% of the total PNSD data, were attributed to new particle formation and vehicle exhaust emissions with different intensities; Cluster 3 and Cluster 4, which accounted for 10.5% of the total PNSD data, were related to the growth of nucleation mode particles; Cluster 5, which accounted for 37.9% of the total data, was attributed to the humid YRD background; and Cluster 6 and Cluster 7, which accounted for 41.6% of the total dataset, were both pollution-related clusters with similar mass concentrations but completely different PNSD. Although the PM_2.5 mass concentrations were somewhat similar, the particle number concentrations of the accumulation mode particles could vary by more than one order of magnitude from the urban background cluster to the pollution-related clusters. The cluster proximity diagram and conversion flow chart of clusters clearly show the influence of new particle formation and growth on haze, as well as the conversion between background and polluted conditions. This study highlights the importance of PNSD for understanding urban air quality and recommends the clustering technique for analyzing complex PNSD datasets.

新粒子形成,气溶胶粒径分布,外场观测,聚类分析