Comparison of atmospheric modelling versus empirical line fitting for mosaicking HYDICE imagery

This letter compares two approaches for determining surface reflectance from hyperspectral airborne imagery (HYDICE) radiance values, and discusses the implications for mosaicking multiple flight lines of HYDICE imagery. Empirical line correction, using calibration panels followed by histogram matching to an adjacent image, produced results comparable with that obtained from applying the ATREM model (Gao et al. 1993, Remote Sensing of Environment, 44, 165-178) to the two flight lines individually. Percentage errors were greatest for low albedo materials and for the long wavelength portions of the spectra. Overall, differences in reflectance values between adjacent frames of 2-5% in the visible/near-infrared to more than 25% in the mid-infrared can be expected.