Avaliação de Índices de Vegetação Para a Discriminação de Alvos Agrícolas em Imagens de Satélites

Vegetation indices have been used for vegetation mapping and for vegetation parameter estimation. The objective of this paper is to present an evaluation of vegetation indices to verify which are best fit for mapping different agricultural and natural vegetation surfaces. ETM+ Landsat 7 images were transformed into reflectances, corrected for atmospheric effects and NDVI, PVI, SAVI and WDVI vegetation indices were calculated. Statistics, histogram and visual verification of the indices images showed that PVI has a better performance than the others. Comparisons between the indices showed that PVI had a slightly better performance than NDVI and a large difference in relation to SAVI and WDVI. Despite not having the best performance, NDVI continues to be an attractive vegetation index as it does not need soil reflectance data. Further research is necessary with classification trials between these indices. Palavras-chave: Vegetation indices, vegetation mapping, remote sensing, indices de vegetacao, mapeamento da vegetacao, sensoriamento remoto.