A technique for aerosol retrieval over land from meris data

In this paper we present an aerosol retrieval over land algorithm specifically designed for the ENVISAT MERIS sensor, with the aim of the atmospheric correction of the data. The fundamental basis lies on a multiple parameter inversion for spectrally heterogeneous sets of five pixels in a window inside the image, considering invariant the atmosphere in this window. In the modellization of the Top of Atmosphere (TOA) radiances, the reflectance in the surface level for each pixel is modelled as a linear combination of two typical vegetation and soil spectra. In the atmospheric contribution the information relative to gaseous abundances and its vertical profiles is fixed, whilst four parameters are left free for aerosol characterization. These are the aerosol optical depth in 550nm and the percentage of the four aerosol species implemented in the 6S code (dust-like, water soluble, oceanic and soot types). As a result of this TOA MERIS data inversion, proportions of the two surface endmembers are retrieved for each one of the five reference pixels, and the sought aerosols parameters for the atmospheric window. For an initial validation of the algorithm it has been check with MODTRAN4 MERIS simulated TOA spectra, retrieving both atmospheric and surface parameters in good agreement with forward conditions. The application to the atmospheric correction of real MERIS images is also satisfactory, although a final version of the algorithm is still under development