Analysis Of The Temporal And Spatial Characteristics Of Urban Air Quality In My Country Based On Multivariate Functional Data Clustering

In today's society,China's air pollution situation is very serious,seriously jeopardizing the physical quality of people.With the advancement of industrialization,in China prevention and control pressure in has also increased.Therefore,understanding the status of environmental quality and conducting research has become an urgent task,and it is also of great significance to urban development.In the past,study of air quality has certain limitations based on traditional data types,and it's prone to over-reliance on classical assumptions and loss of information.Nowadays,with the advent of the big data era,functional data has emerged,and domestic functional data clustering method for air quality is still in its infancy.Therefore,based on clustering method of multivariate function data,this paper studies the air quality spatio-temporal characteristics analysis and main change modes of 365 cities nationwide from the new perspective of functional data,and the overall type of urban air pollution in China.The classification of the typical pollution area,and finally the K-means cluster analysis of functional data for visual comparison analysis,the relevant content is as follows.Based on the monthly air quality data of 365 cities in China from 2015 to 2017,this paper selects the six main pollution factors of air in China: fine particulate matter,inhalable particulate matter,sulfur dioxide,nitrogen dioxide,carbon monoxide and ozone.Firstly,B-spline basis function is selected to fit the original discrete data smoothly,and then,based on functional principal component analysis,the monthly mean concentration curves of pollutants in 365 cities in China are reduced in functional form,and the functional principal component score matrix is obtained.Then,in the clustering research method,the multivariate functional Gaussian mixture model and the multivariate functional K-means method are used respectively.In this paper,the cluster analysis of urban air quality in China is carried out,and characteristics and changes of different pollution status of air quality in different cities in China during the past three years are described.Using R,the clustering results of 365 urban air pollution types in China are visually realized on the map,and industrial sulfur dioxide and industrial dust emissions are added to assist the analysis of the clustering results.Finally,ARI,the evaluation index of the accuracy of the two clustering methods,is compared by simulation,and representative cities are selected for empirical analysis.Finally,according to the difference between the results of the two methods,select representative cities for comparative analysis,summarize the advantages and disadvantages of the two methods,and finally give some measures and suggestions to improve the pollution status,and formulate different policies according to different clusters.Measures to provide relevant decision-making basis and theoretical support for future environmental governance.The results show that air pollution indicators of 365 sample cities in China from 2015 to 2017 have significant spatial and temporal distribution characteristics.In terms of time characteristics,it is closely related to seasonal variation factors.The three typical pollution areas have obvious similarities in time distribution.Their characteristics can be summarized as follows: the particulate matter pollution is serious in winter heating season;the particulate matter pollution is reduced in summer,but ozone pollution is serious and aggravates year by year.Geographically,through the functional cluster analysis of six pollutant concentration curves,three types of typical pollution areas were divided.From the point of view of geographic location,the trend of gradual variation from outside to inside showed obvious spatial differences.In addition,compared with K-means clustering based on functional data,the clustering based on functional data model takes into account the regularity of six pollution indicators changing with time and the correlation between pollution indicators,so in theory the clustering results are more accurate and practical.Finally,it's verified by simulation and empirical analysis.