Análise do coeficiente de retroespalhamento e classificação do uso do solo de áreas inundadas no Pantanal Norte – MT, por meio de imagens alos-palsar

• Ano de defesa: 2013
• Autor(a) principal: Oliveira, Gabriel Vitoreli de
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Mato Grosso; Brasil; Instituto de Ciências Humanas e Sociais (ICHS); UFMT CUC - Cuiabá; Programa de Pós-Graduação em Geografia
• Palavras-chave em Português: Áreas úmidas; Imagens de radar; Sensoriamento remoto; CNPQ::CIENCIAS HUMANAS::GEOGRAFIA; Wetlands; Radar images; Remote sensing
• Link de acesso: http://ri.ufmt.br/handle/1/1254

Wetlands perform a vital ecological role in the maintenance of local and global ecosystems, also appear as one of the most fragile ecosystems on Earth, they are highly susceptible to human actions. Despite this susceptibility, there is a strong deficiency in monitoring these areas, which requires further study in this theme. Currently, remote sensing has become an indispensable tool in the natural resources monitoring. Depending on the imaging characteristics, radar is presented as an efficient alternative for mapping wetlands, since it uses wavelengths which penetrate vegatation and interact with soil and water depth. In this context, this work aims to assess the potential of ALOS/PALSAR (L-band, HH polarization) in the delineation of flood due to the soil different classes, and assess the potential of contextual classifier ICM-MAXVER in classification of land use of Pirizal region in Poconé- MT, Panatal biome. Were extracted and analyzed backscattering coefficient values of ALOS/PALSAR in flooded and not flooded areas. We sought to develop a classification model from the backscatter coefficient values for identifying these areas, with data validation observed in the study area. Was performed a radar image classification with the contextual classifier MAXVER-ICM to attest the ALOS/PALSAR image’s potential in discrimination of different land uses in the grid Pirizal. The results showed that the radar images in the L band are effective in identifying flooded and not flooded forests. The graphs and tables generated from the logistic regression indicated that this sensor presents limitations to separate flooded fields to fields during the dry season. The contextual classifier showed excellent results in delineation of the different land use classes in the study area, with a Kappa index of 0.977 and global accurancy of 98.30%.