MULTI-SENSOR DATA TO RECOGNIZE AND MAP SEDIMENTARY ENVIRONMENTS IN THE AMAZONAS RIVER LOWLAND
Multi-sensor data to recognize and map sedimentary environments in the Amazonas River lowland
DOI:
https://doi.org/10.5016/geociencias.v39i2.13500Abstract
This work proposes the use of Synthetic Aperture Radar (SAR) images that operate in the microwave range, as well as optical images that operate in Visible-Near Infrared (VNIR) to recognize and map the main river environments of the Lower Amazon. The test area (Ilha Grande do Tapará, corresponding to the central sector of the Lower Amazon River) was chosen with the purpose of increasing the accuracy and quality of control points used in the interpretation of the lowland environments. The mapping of the sedimentary features was based on unsupervised classifications and visual interpretation of the optical images, SAR and the SPC-SAR image resulting from the synergism between the initial images. The results allowed the identification of mud, sand, tree vegetation and aquatic pastures, resulting in the association of environments of lakes, dikes, fluvial channels, river deltas and fluvial beach. The general accuracy indices obtained by the classifiers were 93.33% and 0.90 for the optical sensor image; 93.33% and 0.89 for the SAR image; and 78.25% and 0.63 for the SPC-SAR product. The results indicate that the use of the associated data provides us with important and accurate information about the plain and also that the integrated product contributed to the mapping of the studied area.