Comparing different techniques of satellite imagery classification to Mineral Mapping Pegmatite of Muiane and Naipa (Mozambique)

• Tipo: Artigo em Livro de Atas de Conferência Internacional
• Data: 2019

Autores

Gemusse, U (Autor) Outra; Alexandre Lima (Autor) FCUP; Ana Teodoro (Autor) FCUP

Ata de Conferência Internacional

Título: EARTH RESOURCES AND ENVIRONMENTAL REMOTE SENSING/GIS APPLICATIONS X

SPIE Conference on Earth Resources and Environmental Remote Sensing/GIS Applications X held at SPIE Remote Sensing Conference

Strasbourg, FRANCE, SEP 10-12, 2019

Indexação

ISI Web of Knowledge - 6 Citações

Scopus - 10 Citações

Outras Informações

ID Authenticus: P-00R-FE1

DOI: 10.1117/12.2532570

Abstract (EN): Several scientific studies with different concept on the mapping of pegmatites have been done in Muiane and Naipa (Mozambique) region. However, none of the studies compare different satellite data and different remote sensing classification algorithms. This study aims to compare the land cover/use classification maps and their accuracies considered sentinel-2, aster, and Landsat OLI imagery. The algorithms employed to evaluate the pegmatites location at Naipa and muiane in alto ligonha pegmatite district were minimum distance (MinD), spectral angle mapper (SAM), and maximum likelihood (ML). The identified features of landscape characteristics selected includes 8 class (kaolinite; montmorillonite; water; built up; bare soil; grasslands; shrubs; isolated bush). The results showed that SAM and MinD algorithms are appropriate for mineralogical mapping validated with ground truth data and geological maps. A kappa index of 0.85 and an overall accuracy (OA) of 80% was obtained for SAM algorithm, and a kappa of 0,80 and OA of 90% for the MinD algorithm. The classification of the images using SAM and mind showed better results for the clays (kaolinite, montmorillonite) visible in both classifications, has also been tested unsupervised classifications or criteria determined by the geologist using an input training dataset in the case of supervised classifications.

Idioma: Inglês

Tipo (Avaliação Docente): Científica

Nº de páginas: 10