Aerosol properties from automated shadow analysis

A new technique for retrieval of aerosol properties from analysis of satellite-measured radiance is presented. The physical basis of the technique was developed by Vincent (2006) where he described the analysis of shadows to derive aerosol optical depth values that are valid at the wavelength of the spectral band of the image. For multispectral imagers, an optical depth is retrieved for each image wavelength. This work extends the application with a fully automated approach that identifies and analyzes shadows throughout the image. Analysis of imagery from cities and other structures are described. AERONET sun photometers are used for ground truth. Initial results indicate consistency between retrieval values of optical depth and ground truth is less than 0.05. The speed of the analysis suggests approximately 10,000 shadows can be examined per minute of run time on 8 parallelized processors.