Validation of atmospheric correction algorithm ATCOR

Atmospheric correction of satellite images is necessary for many applications of remote sensing, i.e. computation of vegetation indices and biomass estimation. The largest uncertainty in atmospheric correction arises out of spatial and temporal variation of aerosol amount and type. Therefore validation of aerosol estimation is one important step in validation of atmospheric correction algorithms. Our ground-based measurements of aerosol-optical thickness spectra (AOT) were performed synchronously to overpasses of satellites Rapid-Eye and Landsat. Validation of aerosol retrieval by the widely used atmospheric correction tool ATCOR1,2 was then realized by comparison of AOT derived from satellite data with the ground-truths. Mean uncertainty is ΔAOT550 ≈ 0.04, corresponding approximately to uncertainty in surface albedo of Δρ ≈ 0.004. Generally, ATCOR-derived AOT values are mostly overestimated when compared to the ground-truth measurements. Very little differences are found between Rapid-Eye and Landsat sensors. Differences between using rural and maritime aerosols are negligible within the visible spectral range.