Volume XLII-3/W9
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-3/W9, 239–243, 2019
https://doi.org/10.5194/isprs-archives-XLII-3-W9-239-2019
© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XLII-3/W9, 239–243, 2019
https://doi.org/10.5194/isprs-archives-XLII-3-W9-239-2019
© Author(s) 2019. This work is distributed under
the Creative Commons Attribution 4.0 License.

  25 Oct 2019

25 Oct 2019

AN AEROSOL TYPE CLASSIFICATION METHOD BASED ON REMOTE SENSING DATA IN GUANGDONG, CHINA

Y. C. Zheng1,2, L. L. Li3, and Y. P. Wang1 Y. C. Zheng et al.
  • 1Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, Guangzhou, China
  • 2University of Chinese Academy of Sciences, Beijing, China
  • 3College of Environmental Science and Engineering, Zhongkai University of Agricultural and Engineering, Guangzhou, China

Keywords: Aerosol Type Classification, K-means Clustering Algorithm, Mahalanobis Distance, Aerosol Optical Depth, Angstrom Exponent, Ultraviolet Aerosol Index

Abstract. This paper provides an aerosol classification method based on remote sensing data in Guangdong, China in year 2010 and 2011. Aerosol Optical Depth, Angstrom Exponent and Ultraviolet Aerosol Index, as important properties of aerosols, are introduced into classification. Data of these three aerosol properties are integrated to establish a 3-dimension dataset, and k-means clustering algorithm with Mahalanobis distance is used to find out four clusters of the dataset, which respectively represents four aerosol types of urban-industrial, dust, biomass burning and mixed type. Prior knowledge about the understanding of each aerosol type is involved to associate each cluster with aerosol type. Temporal variation of the aerosol properties shows similarities between these two years. The proportion of aerosol types in different cities of Guangdong Province is also calculated, and result shows that in most cities urban-industrial aerosols takes the largest proportion while the mixed type aerosols takes the second place. Classification results prove that k-means cluster algorithm with Mahalanobis distance is a brief and efficient method for aerosol classification.