Résumé
Dimensional stability, along with the natural durability and colour of the wood, is one of the most important characteristics of teak used as timber. However, it is very time-consuming to take measurements of this kind. For the purposes of selection for the production of improved varieties, the number of samples to be measured rapidly exceeds the capacity of a traditional laboratory.
Near-infrared spectroscopy, based on a set of reference data, is a tool enabling many of the chemical properties of wood to be predicted and the number of laboratory measurements to be reduced exponentially. The issue here is a question of checking the effectiveness of NIRS tool to build models and predict the shrinkage and fibre saturation point of teak wood from Togo.
The results show the possible use of NIRS to measure the dimensional stability of teak wood and that it is appropriate to choose the type of wood and type of surface to be measured by NIRS. The best prediction models for radial and tangential shrinkage and fibre saturation point give R-2 values of 0.72, 0.83 and 0.87 respectively with ratios of performance deviation of 1.8, 2.4 and 2.8.
Consequently, after verification on other sets of teak samples, which may or may not be included in the prediction model, NIRS can be used to predict shrinkage and fibre saturation point values accurately for a large number of samples, making it possible to include these characteristics in the selection criteria for classifying wood and high throughput phenotyping.