Research on the influencing factors of urban atmospheric dust reduction based on geographic detector and random forest model: a case study of “2+26” city

In order to study the regional distribution and influencing factors of atmospheric dustfall in the “2+26” cities, this paper uses geographic detectors and random forest methods to analyze the temporal and spatial distribution of dustfall and its main driving factors. The results show that the dustfall in the “2+26” cities in spring exceeds the standard extremely seriously, reaching 11.3 t·km−2·30 d−1 in May, and the exceedance rate reaches 25.56%. Dustfall decreases in summer and autumn, and the lowest dustfall in December reaches 4.7 t·km−2·30 d−1. There are large differences in the average annual dustfall consumption of the “2+26” cities, and cities such as Yangquan and Taiyuan exceed the standard seriously. Among them, Yangquan and Taiyuan are 11.2 t·km−2·30 d−1 and 10.3 t·km−2·30 d−1, respectively, with exceedance rates of 24.44% and 14.44%. According to the analysis of atmospheric dustfall in 2019 and 2020, industrial smoke emissions and population density are the main socioeconomic influencing factors, with q values of 0.837 and 0.736 in 2019, and 0.606 and 0.684 in 2020. Precipitation and green area are key influences among natural factors, with q values of 0.701 and 0.663 in 2019, and 0.643 and 0.571 in 2020. The random forest model reveals the relative importance of different factors on atmospheric dustfall. Industrial smoke and dust emissions, population density, precipitation, relative humidity and green area are the main factors affecting urban atmospheric dustfall. Industrial smoke and dust emissions and population density factors are the main factors that increase dustfall; precipitation and relative humidity contribute to the sedimentation of particulate matter, and green area plays an effective role in reducing suspended particulate matter. Other factors have relatively little impact on atmospheric dustfall. Future research should strengthen the in-depth analysis of these factors to improve the predictive ability of the model and provide a more accurate reference for dust control.