Evaluation of adjacency effect for MIVIS airborne images

This work is aimed to atmospherically correct remote sensing data in the solar spectral domain (Visible and Near Infrared) allowing the better assessment of the surface spectral material characteristics. This was obtained by the inversion of the radiative transfer equation for at-sensor signal. In order to detect targets with peculiar spectral characteristics, the atmospheric correction has to take into account the diffuse radiation that constitutes a significant component to the at sensor radiance. The effect of this component (namely adjacency effect), which tends to mask the pixel seen by the sensor, derives principally from the atmospheric scattering due to the aerosol loading in the scene. At this purpose an algorithm based on 6S calculation was defined to derive the direct and diffuse component of the radiation required to determine the contribution to the pixel reflectance related to the surrounding pixels. The developed algorithm allowed the assessment of this environmental contribution besides the pixel reflectance. Such application, on airborne hyperspectral sensor MIVIS (Multispectral Infrared and Visible Imaging Spectrometer) scenes, leads to obtain accurate pixel reflectance if compared with ground measurements acquired within testing areas. This work shows how adjacency effect has a significant role in the correction of remote sensing data, especially if acquired by an airborne hyperspectral sensor. The preliminary analysis of the results have highlighted that the adjacency effect is not negligible, mainly when pixels in the scene are spectrally heterogeneous.