Effects of change in spectral resolution and channel position on radiative transfer model-based atmospheric correction techniques applied to imaging spectrometers

Removing atmospheric effects is the first step before extracting physical parameters from remote sensing data. Imaging spectrometers (IS), with contiguous narrow bands, can give access to some of the parameters required to constrain radiative transfer codes (RTC) used to model radiance at the sensor. Since existing and future IS have different characteristics, the author presents some preliminary results on the impact of spectral resolution and band position on some of the steps involved in RT-based atmospheric correction: estimation of apparent reflectance, retrieval of atmospheric water vapor, target elevation, estimation of atmospheric optical depth. Radiance spectra of various land surfaces (vegetated areas, bare soils, etc.) under various atmospheric conditions are simulated using the MODTRAN2 RTC and “degraded” to mimic existing and future airborne and spaceborne IS. Preliminary results at this stage confirm that coarser spectral resolution lessens the accuracy of the retrieved parameters. They also demonstrate that for estimation of atmospheric water vapor and target elevation, bands have to be centered at the maximum absorption for a better sensitivity and that spaceborne instruments give better results due to the increased path length. However, increased path length is a limitation for the estimation of apparent reflectance. An accurate estimate of the visibility seems to be impossible without a previous knowledge of surface reflectance or the presence of a dark target such as a body of water in the scene