Influence of moisture on the identification of tropical wood species by NIR spectroscopy

Abstract Solutions for species discrimination are important for monitoring native timber harvesting. Near-infrared (NIR) spectroscopy has shown promise for identifying wood species in real time. The influence of moisture content on the model’s performance for classifying wood is not well known. The objective was to evaluate the effect of wood moisture on the predictive capacity of the models for species discrimination based on NIR spectra using a benchtop and a portable spectrometer. First, NIR signatures were collected on the radial face of wood specimens at equilibrium moisture content (EMC) of 11 native species from Amazonia using both equipments. After saturation, new spectra were collected at the maximum moisture condition and subsequently at every 10 % of the water mass loss during drying. Partial least squares discriminant analysis (PLS-DA) was developed to discriminate the timber species according to their spectral signatures. Principal component analysis of the spectral data obtained in EMC was able to discriminate the species depending on the density gradient of the specimens. Moisture had no significant impact on the spectral signal. The PLS-DA models successfully classified unknown wood samples by species with over 91 % accuracy, regardless of moisture content. Both NIR devices show strong potential for use in forest inspections.