Identifying Wood Species Using a New Predictive Model

Researchers reporting in Microchemical Journal have demonstrated how one-class classification model based on data-driven soft independent modeling of class analogy (DD-SIMCA) in MATLAB achieved an authentication precision between 99.1% and 100% when identifying wood species in the Amazon.¹ The findings from this study reveal that the DD-SIMCA classification model could improve sustainable forest management.

What spectroscopic techniques were used in the study?

In the abovementioned study, which was led by Cristiano S. do Nascimento at the National Institute for Research in the Amazonia, examined how to better identify tree species using spectroscopic methods.¹ Species identification has been an ongoing challenge in the Amazon because of the fact that the Amazon has expansive biodiversity.² To address this problem, researchers integrated Fourier transform near-infrared (FT-NIR) spectroscopy with chemometric modeling to identify wood species and predict key material properties without damaging samples.

The work addresses a longstanding challenge in the Amazon rainforest. This challenge is that large volumes of fallen timber remain underutilized because accurately identifying species is difficult in remote environments.¹ By applying FT-NIR spectroscopy to ten taxa across five botanical families, the researchers identified distinct spectral fingerprints, particularly in regions associated with lignin and hemicellulose composition.¹

What benefit did the data-driven soft independent modeling of class analogy (DD-SIMCA) have on classifying tree species?

The DD-SIMCA model not only achieved a high authentication precision, but it accomplished this feat even when analyzing chemically similar species. Once species identity was established, the same spectral data set was processed through partial least squares regression (PLSR) models developed in TQ Analyst software to estimate physical, chemical, and mechanical properties of the wood.¹

How did the predictive models used in the study perform?

The predictive models showed strong performance, with coefficients of determination generally exceeding 0.88 and low prediction errors.¹ These metrics suggest the approach could serve as a rapid screening tool for determining potential end uses of fallen timber, supporting more efficient allocation and reuse.

The novelty of the work lies in combining species authentication and quality prediction into a single analytical workflow, expanding on previous efforts that treated these steps separately. The integrated approach is designed to support decision-making in sustainable forestry and circular bioeconomy initiatives by enabling data-driven valuation of a largely overlooked resource.¹

What were the limitations of this study?

There were several key limitations of this study. For example, logistical constraints hampered the team’s ability to expand their sample set for this study. As a result, the models were developed using a limited data set and validated internally.¹ As a result, further studies incorporating larger and more diverse samples will be required to confirm robustness and ensure applicability across broader geographic and environmental conditions.

“While this study successfully demonstrates the feasibility of the method, rigorous external validation with larger, independent datasets is required before full implementation,” the authors wrote in their study.¹

If validated at scale, the method could provide forestry professionals and policymakers with a practical, non-destructive tool to unlock the economic and environmental value of naturally fallen trees, reducing waste while supporting conservation-focused resource management strategies.¹

References

1. Do Nascimento, C. S.; de Andrade, J. C.; da Silva, C. E.; et al. A Non-destructive Approach Using Spectral Fingerprinting and Chemometrics in the Authentication and Quality Prediction of Fallen Tree Wood. Microchem. J. 2026, 225, 117971. DOI: 10.1016/j.microc.2026.117971
2. Hadlich, H. L.; Schongart, J.; Wittmann, F.; et al. Exploring the Potential of Field Spectroscopy for Tree Species Identification in Different Amazonian Forest Ecosystems. Glob. Ecol. Conserv. 2025, 64, e03970. DOI: 10.1016/j.gecco.2025.e03970