Source Apportionment, Concentration Prediction, And Meteorological Effect Of Particulate Matter (PM₁₀) In Urban Air

Atmospheric particulate matter (PM) has adverse impacts on the global climate, urban visibility, and particularly human health. An increasing number of short-and long-term epidemiological studies have shown a strong association between PM mass concentration and morbidity and mortality due to respiratory and cardiovascular diseases. With the rapid development of industrialization and urbanization, the urban air pollution in China is becoming more serious, and PM is still the primary pollutant in urban air. It is very important to study the sources of particulate air pollution and its influence factors for developing effective control strategy.This study takes Changsha for example, atmospheric PM10 samples were collected during July and October in 2008, and chemical compositions were analyzed by Wave Dispersive X-ray fluorescence (WD-XRF). Source apportionment study was carried out on PM10 and its elemental concentrations of Na, Mg, Al, Si, P, S, C1, K, Ca, Ti, Mn, Ni, Cu, Zn, Pb and Fe. The principle component analysis (PCA) was applied to source apportionment. Meteorological effect on PM10 was studied on real-time monitored PM10 and meteorological data in 2008. And artificial neural network (ANN) models for prediction of PM10 hourly concentrations were developed basing on relations between PM10 concentration and meteorological factors.From this study we can get the following main results. First, local soil dust and remote coal combustion & secondary aerosols were the most important sources for atmospheric PM10 in the suburban area of Changsha, and meteorological effects were different from each other. The local soil dust was affected by temperature, while long range transport coal combustion& secondary aerosols was dominated by northwesterly wind. Second, meteorological conditions as well as pollution sources do effect on PM concentrations. The mixing layer height and wind speed are very important for particle dispersion, and they are negatively correlated with PM10 concentration. Low mixing layer height and low wind speed are causes for PM10 episode. Third, the ANN models for prediction of PM10 hourly concentrations can be developed basing on relations between PM10 concentration and meteorological factors. ANN models show better results than traditional multiple linear regression models. The ANN models for prediction of PM10 hourly concentrations work well for PM10 episode as well as for general pollution.