Prediction of wood density independently of moisture conditions using near infrared spectroscopy

Near infrared (NIR) spectra obtained from 100 Japanese larch (Larix kaempferi) wood samples containing various amounts of moisture were used to examine the effect of moisture conditions on the accuracy of predicting wood density. Partial least squares regression (PLS-R) analysis was performed to predict wood density under air dry (DEN_ar), water impregnated (DEN_wi) and oven dry (DEN_ov) conditions. The NIR spectra varied with the moisture conditions of the wood, where the characteristic absorbance bands in the vicinity of 7320 cm(-1) (1366nm), 7160 cm(-1) (1400 nm) and 7000cm(-1) (1428 nm) were related to cellulose and water. The spectral differences between high- and low-density samples varied depending on the moisture conditions; high-density samples showed low absorbance values at 7160 cm(-1) when wet and showed high absorbance values at 7320cm(-1) and 7000cm(-1) when dry. DEN_ar, DEN_wi and DEN_ov could be predicted using spectra collected from the corresponding moisture conditions [coefficient of determination (R-2)=0.79-0.89; standard error of prediction (SEM= 24-26 kg m(-3)]. Prediction of DEN_ar and DEN_ov could also be achieved using spectra collected from various moisture conditions (R-2=0.86-0.87, SEP=22 kg m(-3)). The loadings from PLS-R analysis indicated that the absorption bands in the vicinity of 7320cm(-1), 7160 cm(-1) and 7000cm(-1) played an important role in predicting wood density. NIR spectroscopy has the potential to predict wood density independently of the moisture content of the sample.