The operational use of remote sensing images to map coffee fields in large areas, in order to obtain reliable and timely agricultural statistics, is still under development. Several factors raise difficulties to correctly identify and map the coffee crop. However, technological advances observed in the last years, in terms of image acquisition with improved quality and greater quantity, as well as the development of new tools for image analysis, is favoring the development of an operational method that may contribute to establish the official coffee crop statistics in Brazil. Therefore, the present work has the objective to describe the method and to present the result of the coffee fields mapped in the states of Minas Gerais and São Paulo using remote sensing images and geoprocessing techniques. The methodological approach consisted of four steps: a) image restoration and georeferencing; b) non-supervised classification; c) visual interpretation on the computer screen to suppress omission and commission errors; and d) coffee crop area estimation. The results indicated that the applied method was appropriate to map the coffee crop in the states of Minas Gerais and São Paulo.
remote sensing; satellite images; geoprocessing; agricultural statistics
