MULTITEMPORAL ANALYSIS OF SAR IMAGES FOR DETECTION OF FLOODED AREAS IN PANTANAL/ANALISE MULTITEMPORAL DE IMAGENS SAR PARA DETECCAO DE AREA INUNDADAS NO PANTANAL

• Published in: Ra'e ga 2019-09, Vol.46 (3), p.88
• Main Authors: de Almeida Pereira, Gabriel Henrique, Junior, Clovis Cechim, Fronza, Giovani, Deppe, Flavio Andre Cecchini
• Format: Article
• Language: Spanish
• Subjects: Biomes; Equipment and supplies; Geospatial data; Image processing; Precipitation (Meteorology); Radar systems; Remote sensing; Sensors
• ISSN: 1516-4136
• DOI: 10.5380/raega

The Pantanal is one of the most important and preserved biomes in Brazil. This region is annually flooded due to episodes of precipitation along the Paraguay River and its tributaries. Understanding the dynamics of flooding is extreme important since it influences the entire Pantanal ecosystem. Remote Sensing data is an alternative to the identification of flooded areas and their changes in different periods. Among the possible sensors capable of mapping these flooded areas Radar sensor is one of the most attractive--mainly due to the low influence of cloud cover and atmospheric conditions, allowing imaging in dry or rainy seasons. For this work, Radar images from Sentinel 1 satellites for the years 2016, 2017, and 2018 were used. All available data from these years for the study area were used to generate images that represent the seasonality in the region for each year. In total, 1141 Sentinel 1 radar images were processed. The processing of such amount of data was possible through Google Earth Engine platform, which is capable of robust processing of a large amount of data, especially Remote Sensing data. At the end, it was possible to generate images that represent the seasonality of each year. It was also possible to compare the years, highlighting the differences between flooded areas indicating the periods of major precipitation. Keywords: Remote Sensing; image processing; time series images; seasonality; wetlands. O Pantanal e um dos biomas mais importantes e conservados do Brasil. Esta regiao e alagada anualmente em periodos chuvosos. O entendimento da dinamica de inundacao e de extrema importancia, pois influencia todo o ecossistema do Pantanal. Para identificar as areas inundadas e suas alteracoes em diferentes periodos anualmente, devido a grande extensao territorial do bioma, surge como alternativa o uso de dados de Sensoriamento Remoto. Dentre os possiveis sensores capazes de mapeamento destas areas inundaveis se destaca o radar, principalmente pela baixa influencia de cobertura de nuvens, permitindo imageamento em epocas de chuva ou seca, independente das condicoes climaticas. Desta forma, para este trabalho, utilizou-se imagens de radar do satelite Sentinel 1, para os anos de 2016, 2017 e 2018. Todos os dados disponiveis destes anos para a area de estudo foram utilizadas para geracao de imagens que representam a sazonalidade na regiao. Ao todo, 1141 imagens de radar do Sentinel 1 foram processadas. O processamento de tal quantidade de dados