Seasonal statistical analysis of the impact of meteorological factors on fine particle pollution in China in 2013–2017

Abstract

Based on long-term PM2.5 data observed at high temporal and spatial resolution, the relationships between PM2.5, primary emission, and weather factors in China during four seasons were examined using statistical analysis. The results reveal that primary emission plays a decisive role in the spatial distribution and seasonal variability of PM2.5, except in western China, where PM2.5 is controlled by dust weather. In addition to the accumulation of primary emissions, unfavorable meteorological conditions for the diffusion of air pollution lead to the occurrence of PM2.5 pollution. The significant dynamic factors affecting PM2.5 concentration are surface wind speed, planet boundary layer height, and ventilation coefficient, especially in winter. The ventilation coefficient is inversely correlated with PM2.5. Better ventilation is more favorable for the dilution and outflow of local PM2.5. However, in spring and autumn, ventilation coefficient and PM2.5 are positively correlated over the southern regions with low emission, indicating that ventilation also affects the inflow of PM2.5 from outside the region. Wind shear, 850 hPa divergence, and vertical velocity have insignificant effects on the long-term variations in PM2.5. The significant thermal factor is 850 hPa temperature in winter, except in the Pearl River Delta and Xinjiang regions. In spring, the influence of each thermal factor is weak. In summer, the influences of temperature and humidity are more significant than in spring. In autumn, the influence of humidity is relatively obvious, compared with other thermal factors. The correlation coefficients between multi-factors regressed and observed PM2.5 concentrations pass the 95% confidence test, and are higher than that of single-factor regression over most regions. The observed data from December 2016 to February 2017 were chosen to test the regression equation. The test result reveals that the regression equation is effective for predicting PM2.5 concentrations over regions with high primary emission.