Open AccessUse of proximal sensor for soil classes separation applying Principal Component Analysis (PCA)
Author(s) -
Valéria Ramos Lourenço,
David Bruno de Sousa Teixeira,
Carlos Alexandre Gomes Costa,
Calors Alberto Kenji Taniguchi
Publication year - 2020
Publication title -
journal of hyperspectral remote sensing
Language(s) - English
Resource type - Journals
ISSN - 2237-2202
DOI - 10.29150/jhrs.v10.3.p130-137
Subject(s) - principal component analysis , spectroradiometer , separation (statistics) , environmental science , soil test , soil science , spectral analysis , remote sensing , soil water , mathematics , geology , reflectivity , statistics , physics , spectroscopy , quantum mechanics , optics
The spectrally active components of the soil allow the realization of integrative analyzes of soil aspects such as their classification. The purpose of this study was to evaluate the separation of soil classes from spectral reflectance data using principal components analysis (PCA). The study was carried out in the Aiuaba Experimental Basin located in the municipality of Aiuaba, Ceará, Brazil. Soil samples were collected in Ustalfs, Ustults and Ustorthents profiles. The samples were submitted to spectral analysis by a spectroradiometer and, subsequently, to PCA. Principal components were used to identify which of them contribute more significantly to the separation of the soil classes analyzed, based on their relationship with the soil attributes using a two-dimensional graphical analysis. From the examination of spectral behavior data of the different soil classes, the use of PCA allowed the separation of the classes Ustorthents, Ustalfs and Ustults from each other.