REMOVAL OF ATMOSPHERIC INFLUENCES ON SATELLITE-BORNE IMAGERY - A RADIATIVE-TRANSFER APPROACH

Three atmospheric correction methods for satellite-borne imagery are described and compared. The main problem considered is to determine the reflectance of a small Lambertian target which is located on a uniform Lambertian background. The principal differences between the three methods for solving the problem are that i) in the first method the background reflectance is used, but not the apparent background reflectance (the classical method), ii) just the opposite is done in the second method, and iii) both reflectances are used in the third method, thereby eliminating the often inaccurate path radiances. These correction methods may not only be used for atmospheric correction itself, but also to evaluate the use of approximative radiative transfer calculations in atmospheric correction schemes. As an example, we evaluate the Sobolev Modified Approximate Radiative Transfer (SMART) method employing numerical values obtained with the exact doubling method. It is found that the accuracies of the derived target reflectances depend significantly on the correction method used, the actual target and background reflectances, the optical properties of the atmospheric constituents, and the viewing direction considered. It is shown that the first method yields, generally, poorer results for the target reflectance than the second and third method. It is tentatively concluded that replacement of SMART by a more accurate algorithm significantly improves the derived target reflectance only if the aerosol phase function is known with sufficient accuracy.