Aerosol over land with MERIS, present and future
Université du Littoral,
32 avenue Foch,
62 930 Wimereux,
Aerosol remote sensing over land is based on the use of pixels covered by vegetation. Because of the absorption of the photosynthesis pigments, these pixels are quite dark in the blue and in the red. POLDER was used to define a set of representative 20 biomes for which LUTs (Look Up Table) of surface reflectances implemented in the MERIS software. A clear limitation to the initial MERIS process was the limited spatial coverage of DDV (Dense Dark Vegetation) pixels selected by a spectral index: the ARVI (Atmospheric Resistant Vegetation Index.) That the reason why, the DDV concept has been extended to include less dark pixels, LARS (Land Aerosol Remote Sensing) for the aerosol remote sensing. For these LARS pixels, a simple linear regression can be applied to the surface reflectance versus the ARVI in the MERIS band used for aerosol remote sensing. The salient result is that the LARS coverage is good enough to generate a level 3 aerosol product over land.
On this level 3, we can clearly see the transition between the initial biomes. Global MODIS surface albedo maps were use to regenerate a new set of LUTs. We report here the methodology used to get these new LUTs and the resulting modification brought in the algorithm. We also present the results of the evaluation exercise (visual inspection of the images) and of the validation (use of ground based aerosol optical thickness measurements).
It is time at this stage of the algorithm development to propose the inclusion of the aerosols in the atmospheric correction scheme as well as to suggest the best way to generate a level 3 product.