Evaluation of wood variation based on the eigenvalue distribution of near infrared spectral matrix
A method to evaluate the wood variation based on the eigenvalue analysis for the near infrared spectral matrix is presented. The set of eigenvalues calculated from the variance-covariance matrix is treated as the Hamiltonian, which represents the energy eigenstate of the wood, and the wood variation is discussed from the viewpoints of thermodynamics and statistical mechanics. To determine the validity of this idea, two sample groups, one having a high and the other having a low modulus of elasticity ( Efr), are prepared, because they obviously have different molecular configurations in the cell wall. The eigenvalues of the high Efr group are widely distributed compared with those of the low Efr group. The probability corresponding to each energy eigenstate of the low Efr group is flatly distributed compared with that of the high Efr group. These results indicate that the low Efr wood has a more disordered structure than the high Efr wood. The Helmholtz free energy is higher in the high Efr group; in contrast, the entropy is higher in the low Efr group. The results obtained in this study are consistent with the previous knowledge with regard to the relationship between the mechanical properties and the microscopic structure of wood. Hence, the eigenvalues obtained from the NIR spectral matrix provide useful information to assess the variation and stability of wood.