Spatial and Temporal Variabilities of PM2.5 Concentrations in China Using Functional Data Analysis

As air pollution characterized by fine particulate matter has become one of the most serious environmental issues in China, a critical understanding of the behavior of major pollutant is increasingly becoming very important for air pollution prevention and control. The main concern of this study is, within the framework of functional data analysis, to compare the fluctuation patterns of PM2.5 concentration between provinces from 1998 to 2016 in China, both spatially and temporally. By converting these discrete PM2.5 concentration values into a smoothing curve with a roughness penalty, the continuous process of PM2.5 concentration for each province was presented. The variance decomposition via functional principal component analysis indicates that the highest mean and largest variability of PM2.5 concentration occurred during the period from 2003 to 2012, during which national environmental protection policies were intensively issued. However, the beginning and end stages indicate equal variability, which was far less than that of the middle stage. Since the PM2.5 concentration curves showed different fluctuation patterns in each province, the adaptive clustering analysis combined with functional analysis of variance were adopted to explore the categories of PM2.5 concentration curves. The classification result shows that: (1) there existed eight patterns of PM2.5 concentration among 34 provinces, and the difference among different patterns was significant whether from a static perspective or multiple dynamic perspectives; (2) air pollution in China presents a characteristic of high-emission "club" agglomeration. Comparative analysis of PM2.5 profiles showed that the heavy pollution areas could rapidly adjust their emission levels according to the environmental protection policies, whereas low pollution areas characterized by the tourism industry would rationally support the opportunity of developing the economy at the expense of environment and resources. This study not only introduces an advanced technique to extract additional information implied in the functions of PM2.5 concentration, but also provides empirical suggestions for government policies directed to reduce or eliminate the haze pollution fundamentally.