ABSTRACT
The aim of this work was to evaluate the potential of Landsat-TM5 and Rapideye images for the mapping of forests with monodominance of aroeira in the municipality of Tumiritinga, Minas Gerais state, Brazil. We assessed different combinations involving bands of multispectral image, principal components, and normalized difference vegetation index (NDVI) for the supervised classification of Rapideye and Landsat/TM5 images. The thematic maps were evaluated by Kappa and conditional Kappa indexes for “aroeira” soil use class and by the analysis of Confusion Matrix. The best performance for the Rapideye classification showed Kappa 80 and Conditional Kappa 90 indexes. The best Landsat-TM5 classification produced Kappa 80 and conditional Kappa 76 indexes. The thematic map produced shows that 22% of the municipality is under the occupation of aroeira in monodominance. The study showed that both images can be used for mapping land use and land cover in Tumiritinga, and that Rapideye images produced best performance to mapping aroeira monodominance. The thematic map produced shows that 22% of the municipality is under the occupation of M. urundeuva in monodominance, and that this species is not endangered in Tumiritinga, Minas Gerais state.
Keywords:
forest monodominance; remote sensing; image classification