Spatio-temporal clustering model based on angular depth for regional division of PM2.5 in China

PM2.5, which refers to particles with an aerodynamic diameter of 2.5 micrometers or less. The spatial distribution of PM2.5 within China exhibits inherent heterogeneity on a national scale, yet within certain regional confines, a semblance of homogeneity can be observed. The mitigation of the detrimental impacts of PM2.5 necessitates the formulation of geographically tailored regulatory measures by policymakers, predicated upon a comprehensive clustering analysis of PM2.5 concentration data across the nation. The execution of such an analysis, however, is not without its challenges, given the complex and often noisy nature of the data. In this scholarly work, a robust clustering framework is proposed, grounded in a novel depth-based concept capable of effectively managing outliers in both magnitude and form, while also integrating spatial dimensions. The application of this framework to a dataset of PM2.5 concentrations has delineated ten distinct regions within China, each characterized by unique PM2.5 distribution profiles. The homogeneity within these clusters has been substantiated, and the demarcation of cluster boundaries is discernible. Such delineation facilitates the development of localized emission reduction policies by policymakers and fosters the establishment of regional collaborative frameworks aimed at the management of air pollution in China.