Cluster analysis of ultrafine particle size distribution based on long-term measurement at SORPES in Yangtze River Delta of China

This work presents an overview of more than 3-year measurements of ultrafine particle size distributions (PSD, 6-800 nm) at a sub-urban site, the Station for Observing Regional Processes of the Earth System (SORPES) located in Nanjing, Yangtze River Delta of China. For the purpose of understanding the temporal variations as well as the source of ultrafine particles, k-means cluster analysis was applied and seven clusters were finally separated from the PSD data set. PSD spectra, hourly and seasonal frequencies of occurrence, the data of pollutant gases, PM_2.5 and chemical composition, as well as metrological parameters were used to define and interpret each cluster. In general, four clusters (i.e. C1, C2, C3, and C4) were related with new particle formation (NPF) and growth, but they were at different stages or with different intensity, accounting for 20.4% of total PSD data; One (i.e. C5) was attributed to clean regional background, with high level of relative humidity, accounting for 37.9% of total data; Two clusters (i.e. C6 and C7) were polluted clusters, accounting for 41.6% of total data. Two extremely polluted episodes in C6 and C7, respectively, one was caused by fossil fuel combustion and another was caused by biomass burning, were selected as case study. The PSD of two polluted episodes was totally different, even though the mass concentration was similar. The associations between each clusters were analyzed, showing that the contribution of NPF to pollution and the conversion between clean and pollution because of accumulation and wet deposition.