The mangroves are important ecosystems characteristic from tropical and subtropical region located into transition zones between terrestrial and marine environment, with tide influence and presenting typical vegetation. Facing their ecological function, we may emphasize the fact of this environment of presenting appropriate conditions of reproduction and habitat to lots of animals, also it protect the coastline, softening the erosion impact. This study aims to perform the mapping and the changes detection of mangroves areas of Brazilian’s South and Southeast coast, between the years 2008 and 2016, from multi-sensor data (SAR Images, Optical and Digital Elevation Models) using the object-oriented classification approach (GEOBIA) determine the reduction of an image into homogeneous regions (Images objects) by grouping sets of pixels that have similar characteristics (spectral aspects, color, shape, compactness and softness). Several studies about mangrove mapping using digital images processing techniques were performed to the studied area, but the object-oriented approach represents an alternative and advancement at classification of images obtained by different sensors. As results, it was observed that in 2008 Brazilian’s South Southeast coast had 849,55 km² of mangrove areas and 806,64 km² in 2016. This demonstrate 42,91 km² of mangrove area net loss. Among the 5 mapped states, only the state of Paraná presented an increase in total mangrove areas between 2008 and 2016. From the analyses of changes detection it was evidenced a total increase of 138,53 km², a total erosion of 181,44 km² and a mangrove area of 668,11 km² remained unchanged. The validation of the classification occurred through statistical analysis of two confusion matrices (2008 and 2016) containing the rights and wrongs of the classification. The confusion matrix for the year 2008 presented global accuracy indexes = 0.92, Kappa index = 0.84, and Tau index = 0.84. In 2016 they presented global accuracy indexes = 0.93, Kappa index = 0.86, and Tau index = 0.86. These results demonstrate the effectiveness of the use of object oriented classification for the mapping and analysis of the dynamics of the southern and southeastern mangroves of the country, reaching levels between very good and excellent. In relation to the year 2008, when we associated the data obtained in this work with data from Nascimento Júnior (2016) and Pereira (2015) we reached 13,790 km2 of mangrove forests mapped using homogeneus methodology (GEOBIA).
|
