Influence and prediction of PM2.5 through multiple environmental variables in China

Abstract

Fine particulate matter (PM2.5) is an important indicator to measure the degree of air pollution. With the pursuit of sustainable development of China's economy and society, air pollution has been paid more and more attention. The spatial distribution of PM2.5 is affected by multiple factors. In this study, we selected Normalized Difference Vegetation Index (NDVI), precipitation, temperature, wind speed and elevation data to analyze the impact of each variable on PM2.5 in different regions of China. The results show that the high-value areas of PM2.5 were mainly concentrated in the North China Plain, the middle and lower reaches of the Yangtze River Plain, the Sichuan Basin, and the Tarim Basin. PM2.5 showed an upward trend in North China, Northeast China and Northwest China, while in most of South China, especially the Sichuan Basin, PM2.5 showed a downward trend. Therefore, the northern region of China needs to take measures to curb the growth of PM2.5. In Northwest China, wind speed and temperature had a greater impact on PM2.5. In North China, wind speed had a greater impact on PM2.5. In southern China, temperature and NDVI had a greater impact on PM2.5. The deep learning model can better simulate the spatial distribution of PM2.5 based on the selected variables. The clustering effect of single variable is better than multivariate spatial information clustering based on principal component analysis (PCA). It is difficult to explain which variable has the greatest impact on PCA clustering. This study can provide an important reference for PM2.5 prevention and control in different regions of China.

Keywords: Cluster; Correlation; Deep learning; PM2.5; Regression.