Combining micro-densitometry and near-infrared spectroscopy in a unique tool for predicting of wood and pulp physical properties

摘要

In a recent maritime pine (Pinus pinaster Ait) breeding program, different techniques were tested as rapid predictors of wood and pulp quality, including near-infrared spectroscopy (NIRS) and micro-densitometry (MDM). Among different wood and pulp traits, we were interested in assessing the strength properties of high Kappa number unbleached kraft pulp. The possibility of coupling the two techniques to improve the prediction of physical property quality was investigated. Among the different tests performed, the best way of getting the information from MDM profiles as the first discriminator of wood and pulp quality was to take the ratio between the average wood density of the portion of the profile above the threshold 500 g/dm~3 and the total wood density. The method can even be simplified for set of samples of the same age, by taking the number of annual rings with a density higher than 500 kg/m~3 (called here the V500 parameter). Two classes of V500 were arbitrarily adopted and separate NIRS calibrations were established for each class. Their performances were compared to the calibrations obtained with the whole set of samples. A clear improvement in the quality of prediction was observed . R~2 of V500 calibrations were quite higher than the overall model and both the errors of prediction and the residual standard deviations were lower. Although higher precision can be obtained with extracted samples, this time-consuming step can be avoided with moderate loss of prediction quality. Indeed, better results were systematically obtained for untreated samples and separate NIRS calibrations than the single NIRS calibrations obtained from extractive-free samples.