Proposta de um sistema hierárquico de classificação de cobertura da terra baseado na análise de imagens de satélite.

ABSTRACT -Land cover maps are essential to conduct programs of environmental monitoring, biodiversity conservation and land occupation. In Brazil, there are no specific rules to guide the development of such maps. Consequently, there is a tendency of each institution to create its own system to attend its specific needs. This study presents a new land cover classification system based on a single date, multispectral Landsat image and suitable for scales varying from 1.1,000,000 to 1:100,000. Visual image interpretation and field inspection allowed the identification of four spectral land cover classes at the 1:1,000,000 scale: water bodies, natural vegetation covers, planted vegetation covers, constructed surfaces and others. At the 1:500,000, 1:250,000 and 1:100,000 scales, these classes were splited into 10, 19 and 22 subclasses, respectively. Results showed that the Landsat imagery is useful for land cover mapping proposals in the exploratory, surveying and semidetailed level