THE DERIVATIVE RATIO ALGORITHM - AVOIDING ATMOSPHERIC EFFECTS IN REMOTE-SENSING

The usual goal of aircraft and satellite remote sensing is to extract information which is directly related to ground targets, in spite of atmospheric degradation, which often complicates target identification and classification. Most simple algorithms for spectral target recognition are empirical and parameterize the shape (slope, curvature, etc.) of the remotely observed spectra. These algorithms succeed to the extent that the shape of the observed spectra is characteristic of the target; they fail when spectral variations in the solar irradiance or atmospheric parameters overwhelm those of the target reflectance. It is possible to define an analytical algorithm based on derivatives of the radiative transfer equation. The advantage of this approach is that it allows definition of conditions under which a derivative algorithm will be insensitive to atmospheric effects, and it allows estimation of expected errors. This paper describes the development of the "derivative ratio algorithm," based on derivatives of a simple radiative transfer equation. The limiting conditions of the algorithm are derived and demonstrated using examples of reflectance spectra of turbid water and an ash leaf. For these example targets, the algorithm indicates that some spectral features do survive the trip through the atmosphere and are recognizable using ratios of the spectral derivatives. The most detectable spectral features tended to be those that spanned bandwidths substantially larger than the minimum bandwidth tested (10 nm).