Application Of Complex Network Theory In The Analysis Of Air Quality Index Of Major Cities In China

The rough development model of China in the past few decades has brought about a tremendous impact on the environment while promoting rapid economic and industrial development.Air pollution is deteriorating and seriously threatening people’s daily life and health.Currently,air pollution has caused widespread environmental and public health problems and aroused significant attention around the world.This paper conducts a descriptive statistical analysis on the data of AQI（air quality index）and various air pollutants of 35 urbans in China from March 5,2015 to December 31,2017.We find that the primary air pollutants in all cities are particulate pollutants like PM2.5 and PM10.The evolutionary trend shows that the air quality index and other atmospheric pollutants,except for O₃,have a high concentration level in winter,followed by spring and autumn,and the lowest in summer,presenting a clear U-shaped feature,while the change of O₃ is just the opposite.From the overall level,the air pollution level of northern cities in China is obviously higher than that of southern cities,that is,the pollution level of northern cities is higher.Various pollutants in the atmosphere will transfer under complex meteorological conditions,thus affecting the air quality in other regions.That is to say,the air quality in different regions affects each other.However,this correlation does not appear immediately,there exists a time lag effect.In this paper,the time lag effect is introduced to measure the correlation of the air quality index of different cities.A directed and weighted air quality index correlation network model is built by using complex network theory.Topological analysis shows that the air quality of southern cities has less influence on northern cities than that of northern cities.Furthermore,the cluster behavior during the evolution of the network is analyzed based on the percolation theory.The results show that the abrupt phase transition usually occurs between three to six weeks ahead of the peak or valley point of the evolutions of AQIs mean for highly polluted region,which suggests that this event can make an alarm.Then,the cross entropy is applied to investigate the robustness of the percolation analysis results,and it is found that the cross entropy value between the sequence of the largest size change of the giant cluster and the sequence of air quality index in high-pollution regions is exactly the minimum when the warning signal is issued.This paper describes the correlation between air quality in different regions and identifies the peak and valley points in the evolution of the average of AQIs for the high-pollution region.This research not only enriches the theories and methods of air quality prediction,but also provides a new perspective for researchers to conduct research on air quality from a broader dimension.Meanwhile,this work is helpful for the environmental protection department to further understand the evolution trend of air quality from different perspectives,and formulate reasonable and effective measures for the prevention and control of air pollution.