Prediction of non-recoverable collapse in Eucalyptus globulus from near infrared scanning of radial wood samples

Near infrared (NIR) spectroscopy calibrations was used to predict radial profiles of cellulose content, wood density, cellulose microfibril angle (MFA) and modulus of elasticity (MOE) in 20-year-old plantation Eucalyptus globulus to identify non-recoverable collapse zones associated with tension wood. Radial (cambium-to-pith) wood cores were extracted at a height of 1.0 m from trees selected to represent a range of silvicultural treatments. NIR spectra were measured at 1 mm intervals along the radial-longitudinal face of each core after drying to 12% equilibrium moisture content (EMC) at 40 °C. Tangential shrinkage was measured at eight points along each core, following steam reconditioning and re-drying to 12% EMC. Additional cores from 20 of the sample trees were collected. Radial profiles of density, MFA and MOE were obtained for wood strips prepared from these cores, using the SilviScan 3 wood assessment system. Trait profiles were matched to radial NIR scans of these cores, enabling the development of NIR calibrations using partial least squares (PLS) regression. These, and an existing NIR calibration for cellulose content, were used to predict the radial profiles of the four wood properties for the first set of cores. Predicted wood properties were then related to actual tangential shrinkage measurements and the occurrence of visible bands of non-recoverable collapse. A regression model was developed to reliably predict regions of non-recoverable collapse from NIR-predicted cellulose content and MOE. Micrography of stained wood sections indicated that the collapse was caused by the presence of tension wood.