366
T. Ma et al.
a cost-effective densitometer by a continuous NIR single wavelength laser source
and an avalanche photodiode module. It could obtain a good calibration result with
a conventional X-ray densitometer (RMSECV = 0.046 g cm
−3 ).
16.2.4 Wooden Anatomical Features
Wood has a porous three-dimensional structure. It is composed mostly of elongated
cells that are parallel along the tree. A basic understanding of the wooden anatomical
features is essential. Hein has developed NIR models to predict the microfibril angle
(MFA) of Eucalyptus wood [22]. The RMSE between NIR predicted and X-ray
diffraction derived values was 1.3°. Inagaki et al. demonstrated high-quality results
when utilizing NIR to predict the fiber length of Eucalyptus solid [23]. Isik et al.
examined NIRS to predict wood cell wall thickness, coarseness, air-dry density, MFA,
and modulus of elasticity (MOE) of loblolly pine [24]. Furthermore, it suggested that
NIRS can be utilized for screening loblolly pine progeny tests for surrogate wood
traits.
16.2.5 Wood Mechanical Properties
The mechanical properties of wood are its fitness to resist outside forces. Knowledge
of these properties is very important in the wood industry. However, conventional
measurement methods are mostly destructive and time-consuming. Many studies
have shown that wood mechanical properties could be evaluated by NIRS assisted by
chemometrics. Wood density and the cellulosic feature are important in constructing
prediction models from the viewpoint of the chemical absorption band. Horvath
et al. utilized NIRS to predict the green modulus of elasticity (MOE) and green
ultimate compression strength (UCS) of 1- and 2-year-old transgenic and wild-type
aspen. Calibration results showed a well-predicted UCS (R
2
= 0.91, RMSEP =
1.04 MPa) and green MOE (R
2
= 0.78, RMSEP = 538 MPa) [25]. Scimleck et al.
examined to predict MOE, density, and modulus of rupture (MOR) simultaneously by
diffuse NIR reflectance collected from the transverse surface of Pernambuco blocks
[26]. Calibration results showed that the density prediction had the highest accuracy,
followed by MOE, which results in MOR were pore. They suggested the presence of
extractives may weaken the NIR-based calibration models. Watanabe et al. developed
PLS-R-based calibration models for rapidly evaluating longitudinal growth strain
(LGS) [27]. The LGS is one of the most important wood quality indices, and high
levels easily cause end splitting. NIR spectra and LGS were measured from the
peripheral locations of three Sugi green logs. The spectra with higher LGS tended to
be lower absorbance may be caused by the chemical and physical properties related to
the LGS. The calibration model achieved good accuracy (R
2 was 0.61 with a RMSEP
of 0.015%). Kobori et al. and Sofianto et al. tested to measure NIR spectra from the
T. Ma et al.
a cost-effective densitometer by a continuous NIR single wavelength laser source
and an avalanche photodiode module. It could obtain a good calibration result with
a conventional X-ray densitometer (RMSECV = 0.046 g cm
−3 ).
16.2.4 Wooden Anatomical Features
Wood has a porous three-dimensional structure. It is composed mostly of elongated
cells that are parallel along the tree. A basic understanding of the wooden anatomical
features is essential. Hein has developed NIR models to predict the microfibril angle
(MFA) of Eucalyptus wood [22]. The RMSE between NIR predicted and X-ray
diffraction derived values was 1.3°. Inagaki et al. demonstrated high-quality results
when utilizing NIR to predict the fiber length of Eucalyptus solid [23]. Isik et al.
examined NIRS to predict wood cell wall thickness, coarseness, air-dry density, MFA,
and modulus of elasticity (MOE) of loblolly pine [24]. Furthermore, it suggested that
NIRS can be utilized for screening loblolly pine progeny tests for surrogate wood
traits.
16.2.5 Wood Mechanical Properties
The mechanical properties of wood are its fitness to resist outside forces. Knowledge
of these properties is very important in the wood industry. However, conventional
measurement methods are mostly destructive and time-consuming. Many studies
have shown that wood mechanical properties could be evaluated by NIRS assisted by
chemometrics. Wood density and the cellulosic feature are important in constructing
prediction models from the viewpoint of the chemical absorption band. Horvath
et al. utilized NIRS to predict the green modulus of elasticity (MOE) and green
ultimate compression strength (UCS) of 1- and 2-year-old transgenic and wild-type
aspen. Calibration results showed a well-predicted UCS (R
2
= 0.91, RMSEP =
1.04 MPa) and green MOE (R
2
= 0.78, RMSEP = 538 MPa) [25]. Scimleck et al.
examined to predict MOE, density, and modulus of rupture (MOR) simultaneously by
diffuse NIR reflectance collected from the transverse surface of Pernambuco blocks
[26]. Calibration results showed that the density prediction had the highest accuracy,
followed by MOE, which results in MOR were pore. They suggested the presence of
extractives may weaken the NIR-based calibration models. Watanabe et al. developed
PLS-R-based calibration models for rapidly evaluating longitudinal growth strain
(LGS) [27]. The LGS is one of the most important wood quality indices, and high
levels easily cause end splitting. NIR spectra and LGS were measured from the
peripheral locations of three Sugi green logs. The spectra with higher LGS tended to
be lower absorbance may be caused by the chemical and physical properties related to
the LGS. The calibration model achieved good accuracy (R
2 was 0.61 with a RMSEP
of 0.015%). Kobori et al. and Sofianto et al. tested to measure NIR spectra from the