The analysis of hyperspectral image data requires either the application of a radiative transfer model to correct radiance data from the sensor to reflectance or the simultaneous acquisition of surface spectral reflectance data to obtain correction factors for the solar irradiance and atmospheric transmission and scattering. The transmission, and in some cases the scattering, is highly location dependent because the major absorber is water vapor, a poorly mixed gas. Therefore, radiative transfer modeling is required to extend point measurements to the rest of the image. Several models have been developed for atmospheric correction and four of them, ATREM (Gao et al., 1993), HATCH (Qu et al., 2003), ACORN (Miller, 2002) and FLAASH (Matthew et al., 2000) have gained prominence. The latter two are commercially available. In this paper we describe the results of using model and measured reflectances propagated to the top of the atmosphere using MODTRAN4 (Adler-Golden et al., 1999) and retrieved using the above models for different precipitable water vapor values. In addition, we have applied the models to AVIRIS and Hyperion data in order to compare the average reflectances obtained with each model, and compared the results for the derivation of spectral reflectance under a variety of precipitable water vapor conditions.
展开▼