Reconhecimento e mapeamento de gêneros de mangue a partir de dados espectrorradiométricos e imagens Ikonos na Ilha de Marajó – Pa

Abstract

Ikonos II satellite high resolution images combined with radiometric data have been successfully used for mapping and spectral characterization of coastal environments, especially mangroves. This research aimed to map the coastal environments and recognize the different types of mangroves in Soure (Marajó Island/PA), using botanical inventory data and reflectance of vegetation. The study methods included automated classification of the Ikonos multispectral images and in situ reflectance measurements of the vegetation leaves with a spectrum radiometer. Seven distinct environments were identified in the coastal plain: grasslands, old sandy barrier, beach, water mass, flooded area, terrace swamp and mangrove. The Kappa index and global accuracy in the general classification of the coastal resources are 94% and 95% respectively. The use of object-oriented classification of Ikonos images together with spectrum radiometer field data allowed to distinguish between three types of mangrove, including Laguncularia, Avicennia and Rizhophora. The Kappa index and global accuracy for mangrove classification in this study were 70 % and 78.29% respectively. Evidencing that the classification based on field data was efficient and reliable.