Back to Search Start Over

Spectral classification of soils: A case study of Brazilian flooded soils

Authors :
Marcos Rafael Nanni
Mônica Sacioto Chicati
Everson Cezar
Marcelo Luiz Chicati
Roney Berti de Oliveira
Source :
Remote Sensing Applications: Society and Environment. 6:39-45
Publication Year :
2017
Publisher :
Elsevier BV, 2017.

Abstract

Studies on the spectral behavior of Brazilian soils and their possible applicability to increase productive efficiencies are still scarce, mainly for paddy soils, present in a small part of the territory and yet very important for agricultural production in the country. The objective of work was to obtain parameters for classification of paddy soils using spectral data obtained by diffuse spectroradiometry. The surface and subsurface soil samples were collected by using a 90 m 2 sampling grid with equidistant points. Spectral analysis was performed in a laboratory by using a spectroradiometer with 1 nm resolution at 350–1100 nm and 2 nm resolution at 1100–2500 nm. The evaluation of spectral curves was performed on the basis of four different procedures. The statistical analysis was performed by using the procedures STEPDISC, STEPWISE, and DISCRIM of Statistical Analysis System (SAS) software. The results show that soil discriminations by the spectral data obtained in the laboratory were successful. However, the model that tests the procedure for soil discrimination has low accuracy of 47.79% when compared with other methods.

Details

ISSN :
23529385
Volume :
6
Database :
OpenAIRE
Journal :
Remote Sensing Applications: Society and Environment
Accession number :
edsair.doi...........4aed65a3b7ded4ab7816d2bf2bae1ca5