CCD/CBERS-2 images in the spectral bands of CCD2, CCD3 and CCD4 of the years 2004 and 2005, from Mirante do Paranapanema - SP (Brazil), were transformed into surface reflectance images using the 5S atmospheric correction model and radiometrically normalized. The main objective was to spectrally characterize pastures of Brachiaria brizantha in the flowering phase, exempt and infected with the disease "mela-das-sementes da braquiária" making it possible its detection through the comparison among the SBRF - Surface Bidirectional Reflectance Factor values. At the same time, it was aimed to evaluate the effectiveness of the CCD/CBERS-2 images to obtain spectral responses of pastures. The healthy and diseased vegetation canopies were identified through the analysis of the surface reflectance values and the observed data of the Accumulative Crop Water Stress Index (ACWSI) obtained from the study area. The results indicated that the main differences were the decrease of the reflectance in the CCD3 band and the increase of the reflectance in the CCD4 band in the diseased areas. Using the CCD/CBERS-2 data associated with the ACWSI, the methodology was effective to discriminate the vegetation canopies infected with the "mela-das-sementes da braquiária".
remote sensing; Brachiaria brizantha; ACWSI
