An AERONET-based aerosol classification using the Mahalanobis distance. (September 2016)

• Record Type: Journal Article
• Authors: Hamill, Patrick; Giordano, Marco; Ward, Carolyne; Giles, David; Holben, Brent
• Is Part Of: Atmospheric environment. Volume 140(2016)
• Page Start: 213
• Page End: 233
• Publication Date: 2016-09
• Subjects: Atmospheric aerosols -- Aerosol typing -- AERONET -- Mahalanobis distance -- Seasonal aerosol variation -- High AOD events; Air -- Pollution -- Periodicals; Air -- Pollution -- Meteorological aspects -- Periodicals; 551.51
• DOI: 10.1016/j.atmosenv.2016.06.002
• Languages: English
• ISSNs: 1352-2310

Abstract

Abstract: We present an aerosol classification based on AERONET aerosol data from 1993 to 2012. We used the AERONET Level 2.0 almucantar aerosol retrieval products to define several reference aerosol clusters which are characteristic of the following general aerosol types: Urban-Industrial, Biomass Burning, Mixed Aerosol, Dust, and Maritime. The classification of a particular aerosol observation as one of these aerosol types is determined by its five-dimensional Mahalanobis distance to each reference cluster. We have calculated the fractional aerosol type distribution at 190 AERONET sites, as well as the monthly variation in aerosol type at those locations. The results are presented on a global map and individually in the supplementary material. Our aerosol typing is based on recognizing that different geographic regions exhibit characteristic aerosol types. To generate reference clusters we only keep data points that lie within a Mahalanobis distance of 2 from the centroid. Our aerosol characterization is based on the AERONET retrieved quantities, therefore it does not include low optical depth values. The analysis is based on "point sources" (the AERONET sites) rather than globally distributed values. The classifications obtained will be useful in interpreting aerosol retrievals from satellite borne instruments. Highlights: Aerosols characterized as Urban-Industrial, Biomass, Dust, Mixed and Maritime. Aerosol typing using AERONET retrieved parameters. Seasonal variation in aerosol type at AERONET sites. Use of Mahalanobis distance to identify type of individual AERONET measurements.