Global aerosol vertical structure analysis by clustering gridded CALIOP aerosol profiles with fuzzy k-means

Aerosol vertical structures are critical to understanding distribution and source-sink patterns of aerosol on a large scale. In this study, we carried out spatial clustering analysis for 10-years long CALIOP aerosol profiles with a fuzzy k-means (FKM) method. Raw and normalized data sets were both classified into three representative clusters. Raw aerosol profiles of original data described both aerosol density and structure patterns, which were classified into polluted cluster, medium cluster and clean cluster with visual inspection. The mean aerosol extinction coefficient values in near surface from large to small respectively belonged to polluted cluster, medium cluster and clean cluster. As altitude increased, mean aerosol extinction coefficients of polluted cluster were in rapid decline trend from surface to upper atmosphere. In comparison, there was a slower decrease speed for the aerosol extinction coefficient values of mean aerosol profiles of the other two clusters. Aerosol profiles clusters using normalized data could be used to describe aerosol vertical structure patterns. Normalized aerosol profiles were classified into boundary-layer concentrated cluster (boundary cluster), vertically even distributed cluster (v-even cluster) and surface-layer concentrated cluster (surface cluster). The boundary cluster was stable in the low atmosphere with a decline trend upwards, which was spatially corresponds to strong anthropogenic emission and dust regions. The mean normalized extinction coefficient values of v-even cluster were relatively stable in a large vertical range (about 4 km) at regions with relatively weak wind fields. The coefficients of surface cluster were mainly distributed in the near surface with mostly in coastlines and low aerosol optical depth (AOD) regions. The cluster analysis of CALIOP aerosol profiles provided general patterns for global distributions of aerosol profile density and structure.