Available AVIRIS hyperspectral data are in radiance and to comparison with laboratory data, the atmospheric correction is necessary. The aim of this paper was to analyze the influence of atmospheric corrections methods and conversion to TOA reflectance in the determination of the spectral index RCGb by means SWIR data of AVIRIS sensor, in Sao João D''Aliança, Goiás State, Brazil. Was applied five different methods – Green, FLAASH, QUAC, IARR and Flat Field – and was applied the RCGb index to reflectance images and radiance. The RCGb index values of the six images we compared by Pearson correlation coefficient. Results showed high correlation between methods, with variations of 0,045. However, the best performance methods were QUAC and Green, with spatial distribution of RCGb index values closer to reality. The Flat Field method showed the worst performance, with lowest proportionality, underestimating levels of gibbsite in relation to kaolinite, being a method not recommended to determination of the RCGb index in AVIRIS images.