基于地理探测器的广州市大气PM2.5浓度驱动因素分析

PM_2.5变化的驱动因素是大气PM_2.5研究的重要内容.为了揭示PM_2.5污染的特点及其驱动影响因子，以广州市为例，采用地理探测器方法探测自然因素（包括平均降水量、平均温度、平均气压、平均相对湿度、平均风速、植被指数）与社会经济因素（包括人口密度、国内生产总值、工业总产值、人均公园绿地面积、公交车辆数、电力消费量）对2015年广州市ρ（PM_2.5）变化的影响机制与差异.结果表明:①基于因子探测分析发现，对ρ（PM_2.5）变化影响最大的前三位驱动因素分别为植被指数、公交车辆数与电力消费量，对应的因子影响程度指标值分别为0.51、0.46、0.40.②基于生态探测分析发现，植被指数与其他自然因素（如平均温度、平均降水量、平均气压等）对ρ（PM_2.5）空间分布的影响均存在显著差异，与所有社会经济因素对ρ（PM_2.5）空间分布的影响均不存在显著差异；除植被指数外，公交车辆数与其他自然因素及社会经济因素对ρ（PM_2.5）空间分布的影响均存在显著差异.③基于交互探测分析发现，所有影响因素（包括自然因素与社会经济因素）对ρ（PM_2.5）变化的交互作用均大于单一影响因素的独自作用，其中平均降水量与平均气压交互作用后对ρ（PM_2.5）变化的影响最大.研究显示，自然因素（尤其是植被指数、平均降水量）及自然因素与人为活动（如交通出行、电力消费等）交互效应对广州市ρ（PM_2.5）的变化起决定性作用. 

Analysis of Driving Factors of Atmospheric PM2.5 Concentration in Guangzhou City Based on Geo-Detector

The driving factors of atmospheric PM_2.5 concentration is of great significance to PM_2.5 study. In order to identify the characteristics and driving factors of PM_2.5 pollution, this paper takes the Guangzhou City as a case to detect the impact mechanisms that natural factors (including average precipitation, average temperature, average pressure, average relative humidity, average wind speed,NDVI) and socioeconomic factors (including population density, GDP, industrial output value, per capita park greenspace area, the number of public transport vehicles, the total consumption of electric power) impose on PM_2.5 concentration in 2015 based on Geo-Detector method. The key results were as follows:(1)Factor detector results reveal that NDVI is the dominant contributor to PM_2.5 concentration variation, followed by the number of public transport vehicles and the total consumption of electric power,with respective q value of 0.51, 0.46, 0.40. (2)Based on the ecological detector, it is found that there are significant difference in the impact of NDVI with other natural factors (average temperature, average precipitation, average air pressure, etc.), but there is no significant difference in the impact of NDVI with all social factors,except for NDVI and the average wind speed. Besides, there is a significant difference in the impact of the number of public transport vehicles on the distribution of PM_2.5 concentration with other natural and socioeconomic factors. (3)The interaction detector results indicate that the impact of a combination of any natural factor with socioeconomic factor on the distribution of PM_2.5 concentration is far more powerful than that of any individual factor. Among all the factors, the impact of a combination of the average precipitation with average pressure on the distribution of PM_2.5 concentration is the most significant. The study shows that the natural factors (especially NDVI, average precipitation) and the effect of interaction between natural factors and human activities,such as traffic travel and electricity consumption, play a decisive role in the impact on the distribution of PM_2.5 concentration in Guangzhou City.