Abstract: This paper used spatial descriptive statistics,spatial interpolation,spatial autocorrelation,standard deviation ellipse,and gravity center analysis methods to explore and compare pollution types,development trends,agglomeration,and migration characteristics of air quality in China from an annual,quarterly,monthly,and daily perspective.Many studies have focused on the spatiotemporal evolution pattern of air quality in China.However,there are only a few articles that show the highest historical statistical average values in northwestern region,especially high PM_2.5 anomalies in Xinjiang.However,the time and causes of such high anomalies,pollution factors,and people’s contribution to PM_2.5 have not been discussed.Air quality research mainly uses remote sensing inversion data and nearearth monitoring data.In the desert,arid,and semi-arid regions of western China,dust is often removed  during remote sensing inversion data processing,and the concentration of respirable suspended particulates is obviously underestimated.The study results of near-earth monitoring data are greatly affected by the number of monitoring stations and their regional distribution.Air quality monitoring stations in China are mainly concentrated in the highly urbanized eastern-central regions,while the western region is rarely included in air quality monitoring.There are gaps in research in northwestern region,especially in the Tarim Basin,due to some flaws in data itself and limited access.Since 2016,the National Environmental Monitoring Center of China has expanded its coverage to more western cities and can to obtain several times the previous air quality monitoring data.This paper used the near-earth air quality monitoring data from 1 484 stations in 366 cities in China from 2016 to 2017 to maximize coverage of the western region and to identify pollutants in northwest hotspots.The study found that the western region could be divided into three sub-regions based on interpolation.The coastal and mountainous areas show a significant positive response to excellent air quality areas .Shanxi is the largest province with pollution aggravation mainly affected by SO₂ concentration,while Beijing and Henan are the provinces that obviously improved pollution mainly by treating the adverse impact of PM_2.5.The overall distribution pattern is dominated by western Xinjiang and Hebei-Shandong-Henan,forming a significant dual nuclear high-value clustering model.The Hu Huanyong line is the boundary between east-west China’s air pollution and the Yangtze River is the North-South boundary line.The air quality clustering feature in the north is obviously worse than that in the south,and in the east,it is significantly denser than that in the west.Agglomeration is greatly affected by temperature zoning.The overall air quality distribution is oriented in the NE-SW direction,and the gravity transfer center is distributed in Henan Province,mainly moving to the northeast.This distribution change redefines its seasonal differentiation.The main pollution sources in the whole country are PM₁₀ and PM_2.5. PM₁₀ is mainly distributed in Xinjiang,and is strongly affected by natural factors due to dust pollution.PM_2.5 is mainly distributed in Central China,Northern China,and Northern Jiangsu,which are mainly affected by man-made economic activities.In addition,as for urban composite  air pollution,the mechanism of interaction between air pollution in the city and its surrounding is a further issue to be explored in the future.

Key words: air quality index, primary pollutants, space-time pattern, gravity shift