A rapid and nondestructive method for the prediction of lignin content in tropical Amazon woods using FT-NIR spectroscopy

Abstract

Nondestructive methods have been used to predict different technological characteristics of wood. The study evaluated multivariate PLS (partial least squares) models for Klason lignin prediction in forest species native to the Amazon using Fourier transform near-infrared spectroscopy (FT-NIR). Samples of 40 species of commercial wood (Amazonas/Brazil) were obtained in the form of discs at breast height, and wedges in the sapwood-pith direction were extracted. Lignin quantification (reference) was performed, and NIR spectra were obtained on the surface of the radial plane of the samples. A matrix was built with reference data × spectra, and PLS models were built and evaluated. The combination of chemometric data produced 150 predictive models in the bands 7,400–5,823, 7,332–5,187, and 7,000–4,100 cm⁻¹. Using figures of analytical merit (precision), 28 models were selected when they presented R²c > 0.85 (calibration determination coefficients) and R²v > 0.50 (validation). The PLS 4 model (2nd derivative, 8 latent variables) was considered the most robust in the study (R²v = 0.93; SE = 0.01%, standard error; RMSEP = 3.52%, root mean squared error of prediction). The results indicate the use of FT-NIR spectroscopy to predict Klason lignin in Amazonian woods, confirming the efficiency of the tool in producing fast, accurate, nondestructive results without the generation of chemical residues, where this estimate can be useful for studies of forestry, ecology, forest management, and wood technology.