NASA has organized a worldwide Aerosol Robotic Network (AERONET) of Sun photometers. These well characterized instruments (CIMEL Sun photometer) make aerosol optical depth measurements based on the transmission of the direct sunlight through the atmosphere in 5 and 8 spectral bands. AERONET aerosol retrievals are limited to clear sky conditions (no clouds or no cirrus clouds) and retrieve aerosol optical depth. To retrieve other parameters such as aerosol size distribution, single scattering albedo, refractive index, etc., retrievals based on algorithms from Nakajima [2] and Dubovik [1] are used. Since the algorithms are limited to clear sky conditions and not available to the public, we decided to write our own retrieval algorithm which uses the solar principal plane and almucantar scans from the CIMEL instrument. We studied a variety of radiative transfer methods that are able to simulate ground measurements by computing the hemispherical sky radiance. Nor various reasons we decided to use the Santa Barbara DISORT Atmospheric Radiative Transfer (SBDART) code [4,5] and combined it with IDL's Amoeba (down-hill simplex method) [3] to retrieve five parameters: (1) water cloud optical depth, (2) ice cloud optical depth, (3) cloud water droplet radius, (4) horizontal visibility (related to aerosol optical depth), and (5) RH factor which couples relative humidity and aerosol size distribution, from principal plane and almucantar scans. The SBDART computed sky radiances are very close to the measured sky radiances 40 degrees away from the solar aureole since the Discrete Ordinates Radiative Transfer (DISORT) algorithm 161 is not accurate in the forward scattering direction. In this paper we will show results of the inversion of AERONET data using SBDART.
