Comparison of vegetation indices and image classification methods for mangrove mapping at semi-detailed scale in southwest of Rio de Janeiro, Brazil
Nenhuma Miniatura disponível
Data
2023-04-01
Orientador
Coorientador
Pós-graduação
Curso de graduação
Título da Revista
ISSN da Revista
Título de Volume
Editor
Tipo
Artigo
Direito de acesso
Resumo
This paper presents the mangrove mapping carried out in the Rio de Janeiro City, Brazil, using two remote sensing data processing approaches in order to evaluate their potentialities as a complementary tool for oil spill sensitivity mapping. Ten vegetation indices were computed using the Landsat 8 imagery available in Google Earth Engine, and subsequently their spectral patterns were classified through three supervised and five unsupervised methods. Additionally, one pre-processed Landsat 8 OLI bands composition were classified by these eight classification algorithms. To role as a ground-truth for the comparison of 88 automatically produced maps, a mangrove map was prepared based on the methodological guidelines of Oceanic Atmospheric Administration of United States of America for Environmental Sensitivity Index. The best results were presented by Cobweb unsupervised classification of Mangrove Vegetation Index, properly identifying a great mangrove habitats diversity, such as inland brackish, riverine fringe and seaward forests.
Descrição
Palavras-chave
Idioma
Inglês
Como citar
Remote Sensing Applications: Society and Environment, v. 30.