Séries temporais de imagens de Sensoriamento Remoto aplicados à identificação de culturas agrícolas: contribuição metodológica para análise do uso e cobertura da terra
Carregando...
Data
2021-12-01
Autores
Guedes Junior, Edvaldo
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
O estudo do uso e cobertura da terra é importante para o entendimento dos sistemas terrestres e da dinâmica geográfica resultante da interação entre o homem e a natureza. Como resultado desta interação, observamos padrões de uso e cobertura da terra como áreas agrícolas, florestas e centros urbanos que configuram as paisagens terrestres, de forma que nestes recortes geográficos ocorrem as complexas relações de troca entre matéria e energia que sustentam a vida na Terra. Como contribuição para o entendimento dos padrões de uso e cobertura da terra, esta pesquisa utilizou técnicas de classificação supervisionada e extração de índice de vegetação em séries temporais de imagens dos sistemas orbitais LANDSAT e SENTINEL-2 para avaliar a dinâmica do uso e cobertura da terra na região centro leste do Estado de São Paulo. As imagens dos sistemas orbitais foram processadas a partir do sistema de geoprocessamento na nuvem “Google Earth Engine” (GEE), sendo geradas as classes de uso e cobertura da terra compostas por áreas urbanas, vegetação natural, pastagens, lavouras de ciclo anual, lavouras de ciclo perene, solo exposto e corpos d’ água. O processo de classificação das imagens orbitais teve frequência mensal abrangendo o período de janeiro a dezembro de 2020 e foi estatisticamente validado através do índice kappa, matriz de confusão e pelos índices de acurácia do produtor e consumidor. Os resultados do processo de classificação do uso da terra permitiram observar uma dinâmica espacial dominada por culturas agrícolas de ciclo anual, onde a cana-de-açúcar é a cultura dominante, exercendo grande influência no aspecto ambiental da área de estudo.
The study of land use and land cover is important for the understanding of land systems and the geographic dynamics resulting from the interaction between man and nature. As a result of this interaction, we observe patterns of land use and land cover such as agricultural areas, forests and urban centers that configure terrestrial landscapes, where take a place the complex relations exchange between matter and energy that sustain life on Earth. As a contribution to understand land use and land cover patterns, this research used supervised classification techniques and vegetation index extract by time series images from the LANDSAT and SENTINEL-2 orbital systems to assess the dynamics of land use and land cover in the central eastern region of the São Paulo State. The images of the orbital systems were processed using the “Google Earth Engine” (GEE) cloud geoprocessing system, generating land use and land cover classes composed of urban areas, natural vegetation, pastures, annual crops, perennial crops, bare soil and water. The classification process of orbital images had a monthly frequency covering the period from January to December 2020 and was statistically validated through the kappa index, confusion matrix and the producer and consumer accuracy indices. The results of the land use classification allowed us to observe a spatial dynamic dominated by annual crops, where sugarcane is the dominant crop with great influence on environmental aspect of the study area.
The study of land use and land cover is important for the understanding of land systems and the geographic dynamics resulting from the interaction between man and nature. As a result of this interaction, we observe patterns of land use and land cover such as agricultural areas, forests and urban centers that configure terrestrial landscapes, where take a place the complex relations exchange between matter and energy that sustain life on Earth. As a contribution to understand land use and land cover patterns, this research used supervised classification techniques and vegetation index extract by time series images from the LANDSAT and SENTINEL-2 orbital systems to assess the dynamics of land use and land cover in the central eastern region of the São Paulo State. The images of the orbital systems were processed using the “Google Earth Engine” (GEE) cloud geoprocessing system, generating land use and land cover classes composed of urban areas, natural vegetation, pastures, annual crops, perennial crops, bare soil and water. The classification process of orbital images had a monthly frequency covering the period from January to December 2020 and was statistically validated through the kappa index, confusion matrix and the producer and consumer accuracy indices. The results of the land use classification allowed us to observe a spatial dynamic dominated by annual crops, where sugarcane is the dominant crop with great influence on environmental aspect of the study area.
Descrição
Palavras-chave
Uso e cobertura da terra, Sensoriamento remoto, Geoprocessamento, Land use, Land cover, Remote sensing, Geographic information systems