Abstract
On axial and radial surfaces of larch heartwood Fourier-transform near infrared (FT-NIR) spectra were acquired with the aim of finding spectral differences and classifying between the three larch species (Larix decidua, L kaempferi and L. eurolepis). Detailed investigation of the second derivative of the spectra revealed small differences according to the species in the small region from 6700 cm−1 to 6300 cm−1, allowing discrimination of the spectra according to species with principal components analysis (PCA), cluster analysis and soft independent modelling of class analogy (SIMCA), whereas unmodified spectra over the whole range (10000 cm−1 to 5100 cm−1) did not. Clear clustering according to species was found in the scores plots of the first and second principal components in PCA and in the dendrograms of cluster analyses. SIMCA allowed classifying 100% of the unknown spectra to the right species using spectra from axial surfaces and 87% using spectra from radial surfaces. The study shows that detailed investigation of FT-NIR spectra combined with multivariate data analysis is able to reveal differences and classification according to species and the potential of finding information not visible at first sight.