An evaluation of atmospheric correction techniques using the spectral similarity scale

The spectral similarity scale (SSS) is used to evaluate the ATREM and ACORN hyperspectral atmospheric correction software techniques. The SSS evaluates spectra by evaluating the shape and the brightness between pairs of spectra in a hyperspectral data set. An AVIRIS observation of corn crops in Shelton, Nebraska with ground truth is used for this evaluation. Initial results show that it is possible for atmospheric correction techniques to add many "false" spectral features that were not present in the original observation. A correct atmospheric correction of a data set increases the spectral contrast of some data and reveals other subtle spectral features. The ACORN software provides a superior correction to ATREM in terms of removing gaseous spectral features such as that of water