摘要
针对京津冀地区多年来重工业较多、结构性污染突出等问题,该文充分利用多期扬尘地表和工业企业污染源、交通网络、地理国情地表覆盖数据、气象和地形数据,结合MODIS AOD产品和环境监测数据,采用主成分分析和最佳子集回归方法优选预测变量,构建估算PM2.5和PM10浓度的地理加权回归模型,实现京津冀地区2013、2015和2017年PM2.5/PM10年均浓度空间分布模拟制图,分析PM2.5/PM10年均浓度时空分布。实验结果表明,PM2.5和PM10浓度估算模型的决定系数R2分别为0.76和0.86,平均相对预测误差分别为10.87%和13.54%。
Aiming at the problem that Beijing-Tianjin-Hebei region had become one of the most serious area of air pollution in China,this paper made full use of multi-temporal dust surfaces data,industrial pollution sources data,transportation network,land cover,meteorological and topographic data,combined with MODIS AOD product and air quality monitoring data,principal component analysis and optimal subset regression were utilized to optimize the prediction variables to construct a geographically weighted regression model to estimate PM2.5 and PM10 concentration,the spatial distribution of PM2.5/PM10 was simulated,and the spatial-temporal change was analyzed.The experimental results showed that the determination coefficients R2 of PM2.5 and PM10 concentration estimation models are 0.76 and 0.86 respectively,and the average relative prediction errors are10.87% and 13.54%,respectively.
作者
桑会勇
李爽
魏英策
翟亮
SANG Huiyong;LI Shuang;WEI Yingce;ZHAI Liang(Chinese Academy of Surveying& Mapping,Beijing 100036,China)
出处
《测绘科学》
CSCD
北大核心
2019年第6期317-323,共7页
Science of Surveying and Mapping
基金
科技部重点研发计划项目(2017YFB0503502)
中国测绘科学研究院基本科研业务费项目(7771820)
地球观测与时空信息科学国家测绘地理信息局重点实验室开放基金项目(777182102)
关键词
大气污染
浓度估算
地理加权回归
主成分分析
air pollution
concentration estimation
geographic weighted regression
principal component analysis
