Prediction of chemical properties in different tissues of white spruce (Picea glauca) and lodgepole pine (Pinus contorta) biomass using near-infrared spectroscopy

Conifers such as white spruce (Picea glauca) and lodgepole pine (Pinus contorta) represent a major biomass resource for Canadian wood industries, yet residues, including branches and trunk components, remain underutilized. This study aimed to quantify extractives and structural cell wall components across multiple woody tissues of white spruce and lodgepole pine (seven tissue types—bark, branches, heartwood, sapwood, wood, roots, and foliage), and to develop a rapid, non-destructive method for predicting lignocellulosic composition using near-infrared spectroscopy coupled with chemometric modeling to support efficient biomass conversion. Chemical composition and surface chemistry were characterized using both destructive, according to the Technical Association of the Pulp and Paper Industry (TAPPI) and non-destructive (near-infrared spectroscopy, NIRS) methods. Principal component analysis (PCA) effectively distinguished biomass types, explaining 99% of NIR spectral variance. Partial least squares (PLS) regression models demonstrated strong predictive performance for wood components, with subset models for wood (mixture of heartwood and sapwood), heartwood, and sapwood achieving R² values of 0.89–0.99, ratio of performance to deviation (RPD) of 3.68–4.08, and root mean square error (RMSE) of 0.19–1.74%. Models for bark, foliage, and branches showed moderate accuracy. These results highlight that biomass subdivision significantly enhances predictive accuracy for chemical property characterization, enabling more efficient utilization of conifer residues in chemical, bioenergy, and bioproduct applications.