Retrieval of Aerosol Characteristics from Bidirectional, Spectral and Polarization Measurements of POLDER/ADEOS

Kurzfassung

An algorithm for the retrieval of aerosol and cloud characteristics from bidirectional spectral polarization measurements of POLDER over ocean using neural network is presented. Radiative transfer simulations for an atmosphere -ocean surface - system were performed. The simulations were used to train a neural network. This trained neural network was applied to POLDER data to derive aerosol and cloud optical thickness and the aerosol type characterized by the refractive index and the mode radius of the cloud droplet size distribution.