The effect of natural and socioeconomic factors on haze pollution from global and local perspectives in China

Analyzing the factors that cause haze and the regional differences in the influence of factors on haze is the premise and critical to precise prevention and control of haze pollution. This paper explores the global effects of haze pollution drivers and the spatial heterogeneity of factors on haze pollution using global and local regression models. The results show that, from a global perspective, a 1 μg/m 3 increase in the average PM 2.5 concentration of a city’s neighbors will increase the city’s PM 2.5 concentration by 0.965 μg/m 3 . Temperature, atmospheric pressure, population density, and green coverage of built-up areas are positively associated with haze, while GDP per capita is the opposite. From a local perspective, each factor has different influencing scales on haze pollution. Specifically, technical support is on a global scale, and for every 1 unit increase in technical support level, the PM 2.5 concentration will decrease by 0.106–0.102 μg/m 3 . The influencing scales of other drivers are local. In southern China, the concentration of PM 2.5 decreases by 0.001–0.075 μg/m 3 for every 1 °C increase in temperature, while in northern China, the concentration of PM 2.5 increases by 0.001–0.889 μg/m 3 . In the region around the Bohai Sea in eastern China, the concentration of PM 2.5 will decrease by 0.001–0.889 μg/m 3 for every 1 m/s increase in wind speed. Population density positively impacts haze pollution, and the impact intensity gradually increases from 0.097 to 1.140 from south to north. For every 1% increase in the proportion of the secondary industry in southwest China, the PM 2.5 concentration will increase by 0.001–0.284 μg/m 3 . For cities in northeast China, for every 1% increase in the urbanization rate, the PM 2.5 concentration will decrease by 0.001–0.203 μg/m 3 . These findings help policymakers develop targeted joint prevention and control policies for haze pollution, considering regional differences.