Source apportionment of PM_2.5 during different haze episodes by PMF and random forest method based on hourly measured...

Hourly measured PM_2.5-bound species, gases, and meteorological data were analyzed by the PMF receptor model to quantify source contributions, and by the random forest to estimate decisive factors of variations of PM_2.5, sulfur oxidation ratio (SOR), and nitrogen oxidation ratio (NOR) during different haze episodes. PM_2.5 variation was influenced by CO (17%), SO₂ (19%), NH₃ (12%), O₃ (10%), air pressure (P, 9.9%), and temperature (T, 10%) during the whole period. SOR was determined by SO₂ (15%), temperature (T, 9.8%), relative humidity (RHU, 15%), and pondus hydrogenii (pH, 35%), and NOR was influenced by NO_x (19%), O₃ (14%), NH₃ (13%), and RHU (15%). Three types of pollution episodes were captured. Process I was characterized by high CO (contributing 40% of PM_2.5 concentration variation estimated by the random forest) due to coal combustion for heating during winter in northern China. According to the PMF, coal combustion (32%) and secondary sources (38%) were both the most important contributors in the first stage, and then, when the RHU increased to above 80%, the highest contribution was from secondary sources (40%). Process II was during the Spring Festival and was characterized by 8.8 μg m^-3 firework contribution. High SO₂ during this process, especially on the CNY's Eve, was observed due to the firework displays, and SO₂ gave a high contribution (24%) to PM_2.5 variation. Process III showed high ions and high RHU in summer with sulfate and nitrate contributing 44% and 22%, respectively. Furthermore, meteorological parameters and NH₃ play a key role on SOR and NOR.