AVALIAÇÃO DE SOLOS DA REGIÃO CENTRO-OESTE DO PARANÁ/BRASIL: UM ESTUDO ESPECTROSCÓPICO E MULTIVARIADO

Abstract

The purpose of this work was use spectroscopic techniques with multivariate analysis to determine the grain size fractions, the type of management and content of C, H and N of soils from Midwest region of Paraná/Brazil, specifically in the Guarapuava and Irati cities. The soils were analyzed by three techniques spectroscopic: Energy Dispersive X-Ray Fluorescence, Infrared and Photoacoustic. Multivariate analysis such as Principal Components (PCA) for exploratory analysis of soils, and regression by Partial Least Squares (PLS) for estimating the texture and the contents of C, H and N in the soils from the spectroscopic data were used. For all spectroscopic techniques in the analysis PCA was able to distinguish between soils of the two regions of collection, and in some cases also allowed to obtain a distinction between the type of management and soil texture. Prediction models for calculations of the contents of C, H and N in soils, using spectroscope data were not adequate. PLS results can be used for get soil texture. The advantages of the spectroscopic analysis associated with PCA and PLS are reduced analysis time, low cost, not require the use of reagents in addition to being non-destructive compared to conventional techniques.