Abstract: Grasping the spatial and temporal evolution of PM_2.5 pollution is the basis and prerequisite for its targeted management. The annual average PM_2.5 concentration data of each prefecture-level city in Hubei Province from 2001-2020 were extracted from the raster data of PM_2.5 earth surface concentration, and spatial autocorrelation and kernel density estimation methods and Geodetector were used to study the spatial and temporal distribution characteristics of PM_2.5 and the characteristics of meteorological driving factors. The main results of the study are as follows: The rate of compliance with the secondary standard for PM_2.5 concentration in major cities across Hubei Province appears to be increasing. The average annual PM_2.5 concentration in each city gradually increased from 42.47 to 62.94 from 2001 to 2013, and decreased from 62.94 to 32.85 from 2013 to 2020. The kernel density estimation shows that the annual average PM_2.5 concentrations in the cities were gradually dispersed before 2013, while the annual average PM_2.5 concentrations in the cities after 2013 were gradually concentrated in the lower concentration intervals. Taking Wuhan and other central cities as the dividing line, there is a gradient of high to low PM_2.5 concentrations in Hubei Province in both east and west directions, and the concentration in the west is lower than that in the east. The dispersion effect in areas of higher concentrations diminishes gradually after 2013. There was a significant positive aggregation effect in the spatial distribution of PM_2.5, Qianjiang, Xiantao and Tianmen basically showed high-high aggregation characteristics, while Enshi Tujia and Miao Autonomous Prefecture and Shennongjia Forest Area all showed low-low aggregation characteristics. Very few cities showed high-low or low-high aggregation characteristics. Geodetector shows that the meteorological factors had a significant effect on the PM_2.5 concentration. The average explanatory degree of various meteorological factors on PM_2.5 concentration is ranked as follows: wind speed (0.798) > temperature (0.752) > humidity (0.727) > sunshine (0.694) > precipitation (0.639). The dominant driving factors were different in different years, and temperature was the dominant driving factor before 2010, while wind speed was the dominant driving factor after 2010.

Key words: PM_2.5, temporal and spatial patterns, driving factors, spatial autocorrelation, kernel density estimation, Geodetector