Wet-pocket classification in Abies lasiocarpa using spectroscopy in the visible and near infrared range

Abstract

The potential of visible and near infrared (Vis-NIR) spectroscopy to distinguish wet-pockets from normal subalpine fir (Abies lasiocarpa Hook) wood was evaluated. Two specimen classes were used, namely, wood with more than half of the surfaces covered by wet-pockets (WW), and wood completely free of wet-pockets (NW). A partial least square (PLS) regression model was derived and calibrated to predict moisture content ranging from 0 to 210%, and its usefulness for moisture-based sorting of green lumber was assessed. Samples were sorted into the two classes after Vis-NIR scanning via two models: (1) soft independent modeling of class analogy (SIMCA) and (2) PLS discriminant analysis. The SIMCA model using second derivatives and wavelengths spanning 650 to 1150 nm successfully classified 98% of WW and NW in the green state, while it resulted in misclassification of 96% of the specimens after air-drying. The discriminant PLS model using wavelengths spanning 650–1150 nm, correctly classified WW and NW 96% in the green state and 100% after air-drying, respectively. These results clearly demonstrate the applicability of Vis-NIR spectroscopy to discriminate wet-pockets from normal wood.