Open Journal Systems

COMBINAÇÃO DE IMAGENS RADARSAT E LANDSAT-5 TM PARA FINS DE MAPEAMENTO GEO-AMBIENTAL

Alcides Ferreira da Silva Junior

Abstract



O uso combinado de imagens de satélites de diferentes sensores tem
proporcionado diversas pesquisas e soluções para o mapeamento de muitas áreas.
Esse trabalho mostra a aplicabilidade da combinação de imagens óticas e de radar
para fins de mapeamento geo-ambiental. Para tal, utilizou-se imagens dos satélites
Landsat-5 TM (bandas 3, 4 e 5) e RADARSAT (banda C). Como estudo de caso,
tomou-se uma área pertencente ao município de Dourados-MS, onde pode-se
distinguir tipos bem distintos de classes, tais como: solo exposto, áreas de cultivo
agrícola, pastagem, área urbana, vegetação remanescente, várzea, corpos d'água etc.
Verificou-se que o uso combinado dessas imagens apresentou melhorias na
interpretação dos dados, no processamento digital e na acurácia temática da imagem
classificada. As classes mais beneficiadas foram: pastagem, área de plantio direto,
reflorestamento e várzea; as áreas de solo exposto e corpos d'água mostraram-se
bem classificadas sem o auxílio da banda radar; verificou-se também que a imagem
RADARSAT isoladamente mostrou-se eficiente na diferenciação entre áreas
urbanas e corpos d'água.

Combination of optic and radar imagery for geo-environmental mapping

Abstract


The combination of different satellite sensors imagery provides solution
for mapping in many areas. This work presents an application of the combination of
optic and radar imagery for geo-environmental mapping. For such purpose Landsat
TM5 and RADARSAT imagery were used. This project was developed in the city
of Dourados, in the Pantanal area (Central part of Brazil), where different types of
classes could be distinguished such as bare soil, agricultural area, pasture, urban
area, natural vegetation, floodplain, water etc. The combined use of Landsat and RADARSAT imagery has improved the accuracy of the processed data mainly in
the land use thematic map. An improvement was noticed in the following classes:
pasture, soil preparation, reforestation and wetlands. In addition, bare soil and water
bodies have been properly classified using only Landsat imagery. The use of
RADARSAT imagery by itself has proved to be efficient in the selection of urban
areas and water bodies.