versão impressa ISSN 0120-9965
A method for identifying rice crops through Landsat 7 ETM+ and ASTER satellite images was developed in some areas of the departments of Cundinamarca and Tolima (Colombia). The method integrates image preprocessing, creation of NDVI and average texture data based masks by means of fuzzy logic; digital processing through principal component analysis; endmember extraction through n-dimensional display; SAM classification of coverage areas, and evaluation of results. The method allowed the identification of rice crops with a global accuracy of more than 70% and appropriate kappa values (i.e., ranging from 0.45 to 0.74). Based on the results, the method can be said to constitute a good approach to the generation of cropping area information intended to support decision making in agriculture
Palavras-chave : NDVI; endmember; SAM; average texture.