Unless effective corrections can be applied, satellite remote sensing data will remain modified by the absorption and scattering effects of the atmosphere through which the electromagnetic radiation must pass, between the Sun, the ground and the sensor. The true reflectance of the land will not be recoverable, and multi-temporal datasets will not be comparable as a result of the variability of the atmosphere. This article presents a method of removing atmospheric effects from satellite remote sensing images for low-reflectance areas, such as water, where the atmosphere accounts for the majority of the at-satellite measured radiance in the visible bands. The method uses visibility observations to select a reference image for the area of interest. The reflectance of the dark target is calculated after atmospheric correction from the reference image, and is used in conjunction with Turner and Spencer's atmosphere-surface simulation (Turner and Spencer, 1972) and Forster's method (Forster, 1984), to correct the remainder of the images. The method is applied to three large water treatment reservoirs to the west of London.