A Fast Algorithm for Retrieving Maps of Atmospheric Pollution by Fine Particulate Matter from Multispectral Satellite Images

Abstract

A new algorithm for retrieving the total content of atmospheric fine particulate matter (FPM) (particles with a size of less than 1.0 and 2.5 μm) from multispectral satellite images in visible and IR regions of the electromagnetic spectrum is described. The algorithm is based on the regression relations between the top of atmosphere (TOA) reflectance, microphysical parameters of aerosol, and geometrical parameters of the satellite scene. The regression equations are built based on the calculated TOA brightness in the spectral channels of the satellite instrument for the ensemble of randomly generated parameters of the atmospheric radiative transfer model and geometrical parameters of the satellite scenes. Subsequently, this allows real-time mapping of the atmospheric FPM pollution directly from the satellite images without solving ill-posed inverse problems based on the solar radiation transfer in the atmosphere and aerosol light scattering. The algorithm suggested is implemented and tested for the MERIS (Medium Resolution Imaging Spectrometer) satellite instrument. Comparison of the MERIS-retrieved total content of the atmospheric FPM with AERONET (Aerosol Robotic Network) data shows a standard deviation of ~0.5 μg/cm². The applicability of the algorithm developed to real-time monitoring of the regional and transboundary transport of the atmospheric FPM pollutants during wildfires is demonstrated.