A generic platform for hyperspectral mapping of wood

Abstract

Sustainable use is essential to guarantee future availability of wood. Structural and chemical mapping can contribute to fit-for-purpose usage, as accurate and detailed knowledge of the material enables guided utilization and optimal performance of wood. The platform presented in this paper, illustrated on poplar disks for early selection, collects data in the NIR, visual and X-ray spectrum in maximum 30 min per wood disk. Flatbed scanning proves to be a cheap and fast technique to evaluate basic physical properties such as cross-sectional area and wood density. X-ray CT scanning is used for density mapping adding information on density variation between and within disks. Chemometric modeling of infrared spectra is used to map density, tension wood and lignin content. X-ray- and NIR-based density mapping showed good correlation, although NIR-based maps do not show the same level of detail as X-ray images do due to lower resolution. Results of NIR-based hyperspectral mapping illustrate that tension wood zones were denser and contained less lignin which corresponds with the existence of the G-layer described in the literature. In all, the combination of high-end tools together with simple tools such as flatbed scanning allows for high-throughput and high-resolution quantitative mapping of some of the main properties of poplar wood. Furthermore, infrared scanning can be used to map density, tension wood and chemistry without the need for more complex, expensive and/or time-consuming methods, yet with less accuracy and lower resolution, thus being applicable as single tool for breeding selection.