Predicting extractives, lignin, and cellulose contents using near infrared spectroscopy on solid wood in Eucalyptus globulus

Near infrared reflectance (NIR) spectroscopy can be used to reliably predict both the physical and chemical wood properties of Eucalyptus. However, studies have been based on ground wood, which is costly and time-consuming to obtain. Predicting wood traits from NIR spectral data taken from solid wood would greatly increase the speed and cost-effectiveness of this procedure. Existing ground wood calibrations were evaluated for the prediction of wood chemistry from NIR spectral data taken from solid wood. Extractives, acid-soluble lignin, and Klason lignin contents were poorly predicted. Total lignin and cellulose contents showed moderate relationships between laboratory values and the NIR predicted values. NIR calibrations were further developed specifically for predicting wood chemistry from solid wood. All calibrations had high R2 values from 0.72 to 0.88, and standard errors of calibration were less than 1.37%. Calibration validation produced high correlation coefficients between predicted and laboratory values for extractives, Klason lignin, total lignin, and cellulose contents with R2 values ranging from 0.67 to 0.87. Acid-soluble lignin content was poorly predicted. This study showed that NIR analysis on solid wood of E. globulus could be reliably used to predict extractives, lignin, and cellulose contents. It also determined that existing ground wood calibrations, although they could give crude estimates of the wood chemistry values, would need to be re-developed for accurate predictions from solid wood.