Impact of wood surface quality on tropical species identification using benchtop and portable NIR spectrometers

Abstract

The identification of wood in tropical forests is a challenge due to the existence of many species with similar anatomical characteristics, which makes visual differentiation difficult. Traditional techniques, such as anatomical characterization and analysis of organoleptic characteristics, are effective, but slow and dependent on the knowledge of anatomists. Near-infrared spectroscopy (NIR), combined with multivariate statistical techniques, has shown promising results in the accurate classification of wood. The aim of this study was to evaluate the impact of wood surface quality on the performance of NIR instruments in identifying tropical wood species. Sixteen tropical timber species of commercial interest were selected from fields and log yards in the Lavras micro-region in Minas Gerais and their identification was confirmed by comparison with standard samples from the wood anatomy laboratory's xylotheque. The samples were prepared using different tools. NIR spectra were recorded with portable and bench NIR instruments on the cross-sectional surfaces of the wood samples in five situations: (1) field conditions (untreated), (2) chainsaw, (3) circular saw, (4) band saw and (5) sandpaper. Principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were used to evaluate the NIR signatures. The models fitted by PLS-DA with cross-validation showed high success rates, with classifications ranging from 95.3 % to 99.2 % for untreated, circular sawed, band sawed and sanded surfaces. Samples whose surfaces were prepared with a chainsaw resulted in less accurate classifications: 88.7% for benchtop NIR sensors and 92.8% for portable NIR sensors. These results highlight the potential of NIR spectroscopy for classifying tropical woods, even when there are variations in the surface.