Thomas, C; Briottet, X; Santer, R (2013). Remote sensing of aerosols in urban areas from very high spatial resolution images: application of the OSIS code to multispectral PELICAN airborne data. INTERNATIONAL JOURNAL OF REMOTE SENSING, 34(3), 919-937.
New developments in satellite and airborne remote-sensing instruments with metric spatial resolutions allow more precise study of cities. For studies involving the radiative characterization of urban features, knowledge of the atmosphere and particularly of aerosols is required to perform atmospheric compensation of the images. A new characterization tool called OSIS (Observation of Shadows for Aerosol Inversion over 3D Scenes) has been developed, which is adapted to urban remote-sensing images of metric spatial resolution acquired in the visible and near-infrared spectral domains. It is based on the observation of shadow/sun transitions and allows characterization of the aerosol optical properties. A theoretical sensitivity study of this code has shown a precision of the same order of magnitude as most of the satellite products for aerosol optical thickness retrieval. The goal of this article is to test this tool with real remote-sensing images. The data used are PELICAN (Plateforme et Logiciels Informatiques de Cameras Aeroportees Numeriques) airborne multispectral images of 20 cm spatial resolution acquired during the 2009 MUSARDE (Multi-angular Urban Remote Sensing Acquisitions for Reflectance Retrieval in a 3D Environment) campaign over Toulouse. First, the OSIS tool and the measurement campaign are presented in the article. Then, the data processing and the results are discussed based on the budget error of OSIS corresponding to the MUSARDE data. The study shows that the mean retrieved aerosol optical thickness is in agreement with this budget error with an average RMS absolute error of 0.05-0.06 depending on the wavelength. Thus, this study shows the potential of OSIS with a high accuracy of aerosol optical thickness retrieval, even in such complex areas as cities. It also presents recommendations for the use of this tool, such as the use of satellite images that are better calibrated, and/or advice for selecting the pixels of interest.