Open AccessHyperspectral remote sensing as an alternative to estimate soil attributes
Author(s) -
José Alexandre Melo Demattê,
Marcelo Rodrigo Alves,
Bruna Cristina Gallo,
Caio Troula Fongaro,
Arnaldo Barros e Souza,
D.J. Romero,
Marcus Vinicius Sato
Publication year - 2015
Publication title -
ciência agronômica/revista ciência agronômica
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.424
H-Index - 26
eISSN - 1806-6690
pISSN - 0045-6888
DOI - 10.5935/1806-6690.20150001
Subject(s) - hyperspectral imaging , soil science , calibration , environmental science , soil test , remote sensing , statistical model , soil water , mathematics , statistics , geology
Minimizing environmental impacts and increasing crop productivity depend mainly on the knowledge of chemical, physical and mineralogical characteristics of the soil attributes. However, traditional methods are timeconsuming and costly. The objective of this study was to determine and validate a method to quantify soil attributes using UV-Vis-NIR Spectroscopy as an alternative to conventional methods of soil analyses. The work comprised two main phases: (1) creation and calibration of statistical models to determine the soil attributes derived from spectral data extracted from soil samples collected in area 1, (2) validation of statistical models in area 2 and correlations between the estimated and observed values (conventional method) for each soil attribute. The equations of the attributes Fe2O3, Al2O3, and clay reached R2 > 0.80 and may be applied to a different database than the one that was used to generate the equations, provided that they belong to the same study site