Three sensors were tested for non-destructive and real-time monitoring of wood log moisture content (MC), air-dry density and basic specific gravity (BSG): a near-infrared (NIR) spectrometer, a time-resolved NIR instrument and a ground penetrating radar (GPR). Statistical and physical models were established to relate the NIR and GPR data to the wood properties investigated. The physical models involved the computation of intermediate physical properties: the optical coefficients and the relative permittivity for the NIR and GPR models, respectively. Additionally, the ability of these sensors to characterize properties from thawed and frozen wood, softwood and hardwood, sapwood and heartwood as well as from the cross section (CS) and through the bark (TB) of the logs was also investigated. The best estimation accuracies were achieved with the statistical models using partial least squares (PLS) regression. MC was best predicted using the GPR (with a root mean square error [RMSE] of 7% and a coefficient of determination [R2 ] of 0.95), while BSG was best predicted using the NIR spectrometer (RMSE = 0.019, R2 = 0.78). Air-dry density was estimated with a RMSE of 0.047 g·cm·3 (R2 = 0.56) using time-resolved NIR analysis and one single wavelength (846 nm). The statistical models were capable of handling the influence of the log state, which was not taken into account in the physical models. However, the physical models presented are more easily transferable among sensors. Using the GPR, larger wood volumes are also sampled than when using the NIR systems. This work demonstrates that portable sensors based on NIR and GPR technologies could be used to determine MC and BSG of logs in the field. Such sensors could be used to reduce energy consumption, reduce waste, increase product quality and decrease production costs in the forest industry.
Zircon - This is a contributing Drupal Theme Design by WeebPal.
