Borde, R, Ramon, D, Schmechtig, C, Santer, R (2003). Extension of the DDV concept to retrieve aerosol properties over land from the Modular Optoelectronic Scanner (MOS) sensor. INTERNATIONAL JOURNAL OF REMOTE SENSING, 24(7), 1439-1467.
Air pollution and aerosol characteristics are of considerable interest to studies in climate change and air quality. Although the reference methodology will undoubtedly remain in situ measurements, the synoptic view and detailed spatial coverage potentially offered by space sensors would constitute some very desirable complementary information, even if of a qualitative nature. Effective retrieval of information about the atmospheric and surface characteristics necessitates an atmospheric correction step, which takes into account the effects of multiple scattering. These are greatly dependant on aerosol properties. Here we address the possibility of remote sensing the aerosols over land, mostly with space sensors currently in orbit that are imagers in the visible and the near-infared. One popular method is the concept of dark dense vegetation (DDV). The dark targets in the blue and red allow retrieval of the spectral aerosol path radiance and inference of the aerosol loading and size distribution, once the refractive index is known. For the Medium Resolution Imaging Spectrometer (MERIS) land product algorithm, they are selected by a thresholding on Atmospherically Resistant Vegetation Index (ARVI) (> similar to0.78). However, as shown on Modular Optoelectronic Scanner (MOS) images acquired over western Europe, pure DDV targets are sparse (< 1% of land pixels) and of little use for aerosol characterization, at least for Europe. This problem is tackled by extending the DDV concept to brighter targets that have a lower ARVI (down to 0.6). It is shown that these new targets have a reflectance in the red that is very well correlated with the ARVI, and a quite constant reflectance in the blue. The idea of using a DDV reflectance in the red dependant on ARVI in the aerosol retrieval step is tested on several MOS images, on which an atmospheric correction algorithm similar to the one developed for MERIS is applied. The result is that aerosol optical properties are now retrieved over 10% of the land area with little loss of accuracy compared to pure DDV. The proposed ARVI-red reflectance empirical relationship also includes correction for the adjacency effects that arise when a dark area is surrounded with some brighter surfaces.