GTWR模型联合地理探测的长三角地区PM_(2.5)驱动因素研究

Research on the Driving Factors of PM_(2.5) in the Yangtze River Delta Based on GTWR Model Combined with Geographical Detector

采用2001~2020年全球高精度PM_(2.5)卫星反演数据集,结合空间统计分析、地理探测器和时空地理加权模型(GTWR)等方法,探讨了长三角地区人口社会经济发展因素及气象因素对PM_(2.5)的影响程度及时空异质性。结果表明:(1) GTWR模型考虑了时间尺度,在拟合效果方面优于传统的全局回归(OLS)和地理加权回归(GWR)模型;(2)地理探测器的探测q值能较为准确的定量说明各驱动因素对PM_(2.5)的影响程度。在不同时期各因素q值呈现波动式变化,其中年末人口、公路货运量、年末用电量、气温的影响保持稳定增长趋势;(3)GTWR拟合结果显示,各因素对PM_(2.5)的影响均存在正负效应,其中年末人口和气温正向效应显著,而公路货运量、年末用电量和相对湿度负向效应显著,其余因素均存在不同程度的波动变化。此外回归系数的区域性变化特征表明各驱动因素具有明显的时空异质性。

Based on the global high-precision PM_(2.5) satellite inversion dataset from 2001 to 2020,in conjunction with spatial statistical analysis,geographic detector,and the geographically and temporally weighted regression(GTWR),this study examined the extent of influence and spatiotemporal heterogeneity of human socioeconomic and meteorological factors on PM_(2.5) in the Yangtze River Delta region.The findings were as follows:(1)The GTWR considered the temporal scale and demonstrated superior fitting performance compared to traditional global regression(OLS)and geographically weighted regression(GWR)models.(2)The geographical detector's q-value provided an accurate and quantitative explanation of the influence of each driving factor on PM_(2.5).The q-values of different factors exhibited fluctuating patterns across various time periods.Notably,the impact of population,road freight volume,electricity consumption,and temperature at the year-end maintained a consistent upward trend.(3)The GTWR fitting results indicated that various factors had both positive and negative effects on PM_(2.5).Specifically,year-end population and temperature had significant positive effects,while road freight volume,electricity consumption,and relative humidity at the year-end had significant negative effects.Other factors demonstrated varying degrees of fluctuation.Moreover,the regional variation in regression coefficients suggested distinct spatial and temporal heterogeneity among the driving factors.