DERIVATION OF SCALED SURFACE REFLECTANCES FROM AVIRIS DATA

Analysis of high resolution imaging spectrometer data requires a thoro ugh compensation for atmospheric absorption and scattering. A method f or retrieving ''scaled surface reflectances,'' assuming horizontal sur faces having Lambertian reflectances, from spectral data collected by the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) (Vane, 198 7) is presented here. In this method, the integrated water vapor amoun t on a pixel by pixel basis is derived from the 0.94-mum and 1.1 4-mum water vapor absorption features. The transmission spectrum Of water v apor (H2O), carbon dioxide (CO2), ozone (O3), nitrous oxide (N2O), car bon monoxide (CO), methane (CH4), and oxygen (O2) in the 0.4-2.5 mum r egion is simulated based on the derived water vapor value, the solar a nd the observational geometry, and through use of narrow band spectral models. The scattering effect due to atmospheric molecules and aeroso ls is modeled with the 5S computer code (Tanre et al., 1986). The AVIR IS radiances are divided by solar irradiances above the atmosphere to obtain the apparent reflectances. The scaled surface reflectances are derived from the apparent reflectances using the simulated atmospheric gaseous transmittances and the simulated molecular and aerosol scatte ring data. The scaled surface reflectances differ from the real surfac e reflectances by a multiplicative factor. In order to convert the sca led surface reflectances into real surface reflectances, the slopes an d aspects of the surfaces must be known. For simplicity, the scaled su rface reflectance is simply referred to as the ''surface reflectance'' in this article. The method described here is most applicablefor deri ving surface reflectances from AVIRIS data acquired on clear days with visibilities 20 km or greater. More rigorous atmospheric radiative tr anfer modeling is required in order to derive surface reflectances fro m AVIRIS data measured on hazy days.