Landscape Change Detection Using Fuzzy Decision Making and Grfm Jers-1 Data Classification on Wetlands in the Central Amazonia, Brazil

1. ABSTRACT This abstract presents an approach for oil spill environmental sensitivity evaluation in landscape modifications within half hydrological cycle over the wetland in the Western Amazonia, Brazil, using JERS-1 SAR seasonal classified images and a fuzzy logic method. A pair of multitseasonal set of JERS-1 images (1995-low water season and 1996-high water season) obtained in The Global Rain Forest Mapping (GRFM) Project in Western Amazonia, Brazil, presented the identification of large extension of wetlands in a region characterized by the high sensitivity to oil spill [1][2]. The oil spill hazard information derived from JERS-1 SAR data is straightforward to interpret and constitutes a representation of the original Environmental Sensitivity Index (ESI) product conceived by PETROBRAS [3]. The GRFM data set is organized in semi-continental synthetic aperture radar image mosaics with a 100 m nominal resolution. The global, continuous and weather-independent coverage provided by the SAR system onboard the JERS-1 satellite enable end users to monitor rapidly changing conditions in cloud-covered, rain forest regions. Potential applications for environmental protection in Amazonia include the understanding and the management of flood hazards. The basis for the definition of a sensitivity index to oil spills in Western Amazonia is the relationship between physical and biological characteristics of the fluvial environment. These habitats were used to rank the overall sensitivity of fluvial areas in Western Amazonia to spilled oil as part of the Environmental Sensitivity Index (ESI) [3]. The most sensitive fluvial environment for oil spills was given an index of 10b, the least sensitive, an index value of 1. Visual inspection of the high flood GRFM JERS-1 SAR mosaic indicates a widespread occurrence of double bounce reflections of flooded forests that is characterized by a strong radar return. Such a backscatter mechanism occurs when the JERS-1 SAR beam reflects off both the tree trunks and the smooth reflective surface of the water (Figures 1A and 1B). JERS-1 SAR images therefore demonstrate that the study area was extensively covered by flooded forests in May 1996, which is the most oil-sensitive habitat (ESI Ranking = 10b). Because of its ability to deal with imprecise, uncertain or ambiguous data or relationships among data sets, fuzzy logic methods has been applied in a wide range of ecological and environmental issues, including environmental impact evaluation. The objective of this paper is to present an approach to extract membership values of environment sensitivity in regional scale directly from dual …