Design and development of a portable Vis-NIR spatially resolved spectroscopic device for nondestructive and rapid evaluation of growth stress in standing trees

• A multifiber-based Vis-NIR spatially resolved spectroscopic device was developed. • Hinoki cypress trees, growing on a slope within the field, were sampled. • New device was tested to gather light scattering and absorption data from trees. • Measured spectra were correlated to growth stress and wood anatomical features. • Machine learning method achieved promising performance on growth strain prediction. Growth stress (growth strain) is caused by internal stresses that develop within the wood during expansive growth and maturation. Although growth stress evaluations are critical for predicting and mitigating problems such as warping, cracking, and deformation during processing and use, conventional methods have several disadvantages, including notably high operational costs, time-consuming, and relatively destructive. This study aimed to develop a nondestructive, portable, and rapid technique that can evaluate local growth stress in standing trees. A multifiber-based visible-near-infrared (Vis-NIR) spatially resolved spectroscopic (SRS) device was, therefore, designed and used to collect light scattering and absorption information from standing trees. Principal component analysis revealed that the collected Vis-NIR SRS spectra were highly correlated with anatomical features related to growth stress, such as the difference in light scattering (particularly at approximately 846 nm) and moisture content (approximately 970 nm) inside the wood, as well as the wood surface color (approximately 660 nm). The determination coefficient for the cross-validation set was found to be 0.84 with a root mean square error of 333 με using partial least squares regression analysis. In addition to the high prediction accuracy achieved by machine learning algorithms, the results of Monte Carlo modeling of photon migration in 3D wood structure models also demonstrated that the Vis-NIR SRS approach has great potential for on-site applications, offering a new and cost-effective method for evaluating growth stress in green trees.