Classifying waste wood from Amazonian species by near-infrared spectroscopy (NIRS) to improve charcoal production

Abstract

Solutions to differentiate wood wastes in the storage yard of a charcoal production unit in Amazonia, where identification is often mistaken due to broad morphological similarity of species, can improve the efficiency of the carbonization process. This study aimed to develop multivariate models to quickly identify the wood wastes of 12 tropical species based on the spectral signature in the near-infrared (NIR) region, to improve the control of raw material used in charcoal production. The spectral data were subjected to principal component analysis (PCA) and partial least squares–discriminant analysis (PLS-DA). Spectra acquired from the transverse surface of the wood yielded clearer clusters in the PCA score plot. However, the PLS-DA model fitted with the first derivative of the spectra measured on the radial surface of the wood showed the highest rate of correct classification (97.9%) of the 12 species. Thus, the results proved that the technique is reliable and fast for differentiating wood from branches of several species native to Amazonia, especially to group similar wood species for charcoal production.