桉树材性性状近红外预测模型的建立

试验以采集的100份桉树不同组合杂交子代的木芯及木粉样品作为研究对象，以常规方法测定所取木材样品的木材密度、纤维长度和纤维宽度并用 Polychromix 手持式近红外仪采集了自然风干状态木粉的近红外光谱信息。光谱数据的处理及建模用 Unscrambler 9.7软件完成。建模结果显示：木材密度、纤维长度和纤维宽度的预测精度均可达90%以上。建模过程中，木材密度较纤维长度和纤维宽度所需的校正样本集数量要多，说明要达到一定的预测精度，纤维长度和纤维宽度其所建模型的预测范围会相应变小。

NIRS Calibration for Predicting Wood Properties of Eucalyptus

100 increment cores and wood powder samples from a range of Eucalyptus taxa were collected in order to develop a Near Infra Red Spectral (NIRS) calibration for predicting wood properties of such species. Measurements of wood density, fiber length and fiber width of samples were first obtained from the increment cores using conventional methods. A Polychromix Phazir was then used to collect the near infrared spectra of the air-dried wood powder samples. The software Unscrambler 9.7 was used for data analyses and developing calibration models. The accuracy the calibrations obtained for predicting wood density, fiber length and width were all greater than 90%. The number of samples used for the calibration for NIR prediction of wood density was higher than for other traits, fiber length and width.