A Chemometric Approach to Timber Species Classification Using Near-Infrared Spectroscopy (NIRS) and SIMCA

Abstract

This study investigates the potential of using Soft Independent Modeling of Class Analogy (SIMCA) in conjunction with Fourier transform Near-Infrared Spectroscopy (NIRS) as a non-destructive and effective alternative method for classifying four commercially significant Indian timber species: Cedrus deodara (CD), Dalbergia sissoo (DS), Shorea robusta (SR), and Tectona grandis (TG). NIR absorbance spectra were collected using an FT-NIR spectrophotometer and SIMCA models developed through the Unscrambler 10.1 software. Class-specific Principal Component Analysis (PCA) models were employed to construct the classification algorithm. The developed training models achieved perfect classification performance, recording 100% accuracy for TG, CD, DS, and 97.5% for SR. Validation using independent wood samples yielded 100% prediction accuracy for all four species. Additionally, Eucalyptus tereticornis (ET), which was not part of the model calibration, was accurately recognised as distinct, confirming the model’s specificity. The NIRS-SIMCA combination thus provides a dependable and efficient solution for timber species determination, enhancing quality assurance and regulatory adherence in the wood industry.