Non ― supervised classification of aerosol mixtures for ocean color remote sensing

摘要

Satellite ocean-color algorithms generally use aerosol-mixture models to estimate and remove the atmospheric contribution to the measured signal. These models, based on aerosol samples, may or may not be realistic. In atmospheric correction, we are more interested in the optical behavior of the aerosols through the entire atmosphere. Comparisons of SeaWiFS-derived and measured aerosol optical thickness have revealed a systematic underestimation of the Angstrom coefficient, suggesting that the reference models may not be representative of actual conditions. To investigate the adequacy of the models and ultimately to improve atmospheric correction, we analyze atmospheric optics data collected by the AERONET project under a wide range of aerosol conditions at coastal and island sites. Using non-supervised classification techniques (self-organized mapping, hierarchical clustering), we determine the natural distribution of retrieved aerosol properties of the total atmospheric column, i.e., the volume size distribution function and the refractive index, and more importantly identify clusters in this distribution. These clusters may be used as new aerosols mixtures in radiative transfer algorithms. We compare the clusters with the SeaWiFS reference models and, through application examples, conclude about their potential to improve atmospheric correction of satellite ocean color.