Characterization and source apportionment of PM_2.5 based on error estimation from EPA PMF 5.0 model at a medium city...

Heze city, a medium-size city in Shandong province, Eastern China. Ambient PM_2.5 samples were collected in urban area of Heze from August 2015 to April 2016, and chemical species and sources of PM_2.5 were investigated in this paper. The results indicated that the average concentration of PM_2.5 was 100.9 μg/m³ during the sampling period, and the water-soluble ions, carbonaceous species included elemental carbon (EC) and organic carbon (OC), as well as elements contributed 32.7-51.7%, 16.3% and 12.5%, respectively, to PM_2.5. Pearson's correlation analysis showed that the existing form of NH₄⁺ was more complex and diverse in spring/summer, and ammonium nitrate, ammonium sulfate and ammonium hydrogen sulfate might be major form of NH₄⁺ in autumn/winter. Correlation analysis between PM_2.5 and SO₄^2-/NO₃^-, PM_2.5 and OC/EC during different seasons suggested that mobile sources might make an important impact on the increase of PM_2.5 concentrations in spring/summer, and stationary sources might play a critical role on the increase of PM_2.5 concentrations in autumn/winter. Seven factors were selected in Positive Matrix Factorization (PMF) models analysis based on the Error Estimation (EE) diagnostics during different seasons. Secondary source had the highest contribution to PM_2.5 in Heze for the whole year, and followed by coal combustion, vehicle exhaust, soil dust, construction dust, biomass burning and metal manufacturing, and their annual contributions to PM_2.5 were 26.5%, 17.2%, 16.5%, 11.5%, 7.7%, 7.0% and 3.8%, respectively. The air masses that were originated from Mongolia reflected the features of large-scale and long-distance air transport; while the air masses that began in Jiangsu, Shandong and Henan showed the features of small-scale and short-distance. Shandong, Henan and Jiangsu were identified as the major potential sources-areas of PM_2.5 by using potential source contribution function (PSCF) and concentration-weighted trajectory (CWT) models.