基于协同Kriging插值和首尾分割法的PM2.5自然城市提取
Classification of PM2.5 for natural cities based on co-Kriging and head/tail break algorithms

PM2.5空气污染问题目前是社会关注热点以及学术研究重点。该文对PM2.5污染的自然城市提取进行了研究，结合PM2.5的站点监测数据和气溶胶遥感数据并采用协同Kriging插值实现了PM2.5数据空间化，然后采用首尾分割分类方法实现了PM2.5污染分布的分类和污染自然城市的提取。对中国大陆PM2.5自然城市的提取结果进行了分析和讨论。结果表明：采用适当的分割阈值，首尾分割分类方法可以有效进行PM2.5污染自然城市提取工作，有助于决策者合理划分PM2.5联合治理的区域范围。
Abstract：PM2.5 air pollution is now a hot topic in both social and academic circles. This study investigated the classification of natural cities based on PM2.5 concentrations in Mainland China. Firstly, the PM2.5 data obtained at monitoring stations and aerosol optical depths (AOD) obtained by remote sensing were fused to yield more accurate PM2.5 spatial distributions using a co-Kriging algorithm. Then, the PM2.5 concentrations were classified using the head/tail break clustering algorithm to identify natural cities with high PM2.5 pollution levels. Distribution of natural cities was also analyzed. The results show that the head/tail break algorithm with an appropriate segmentation threshold can efficiently identify natural cities with high PM2.5 concentrations. These classification results can guide policy makers to divide the country into several areas for pollution control.