Abstract
The objective of this research was to calibrate and compare the ability of near infrared (NIR) and Fourier transform infrared reflectance (FT-IR) to predict strand density. Four hundred and fifty flakes were randomly selected, measured for density, and then 99 flakes (33 = low, 33 = medium and 32 = high density) were selected for calibration and cross-validation through the predicted sum of squares (PRESS) method. This research found that calibration equations from NIR spectra (R2 = 0.82; RMSEP=0.059 g cm−3) was superior to equations developed from FT-IR (R2 = 0.68; RMSEP=0.076 g cm−3) when the same number of factors were used. Using a first-derivative pre-treatment, five factors were necessary to predict strand density from NIR spectra despite a random tracheid (wood cell) orientation. Lignin-associated wavelengths (range, 1660–1670 nm) provided the most leverage in predicting strand density although crystalline and amorphous components of cellulose also played a role. The prediction of density above 0.65 g cm−3 was inaccurate for both NIR- and FT-IR-based models. The bias was attributable to the additional absorbance that occurred at higher densities.
© 2010 IM Publications LLP
PDF Article
More Like This
Cited By
You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
Contact your librarian or system administrator
or
Login to access Optica Member Subscription
Add to BibSonomy