Examining the Role of ATR-FT-IR Spectroscopy and Machine Learning in Wood Forensics, Part 2

Illegal logging is a growing issue globally. Along with the illegal timber trade, both these devastating practices are posing global threats to forest conservation and management efforts (1). To combat these threats, wood forensics has emerged as a burgeoning field to help identify wood and help criminologists determine whether the wood taken was illegal. Wood forensics uses several tools, including artificial intelligence (AI), to characterize wood species and identify illegal timber (1,2).

Historically, various analytical techniques have been employed in wood forensics, each with its own set of advantages and limitations. For example, gas chromatography–mass spectrometry (GC–MS) has been used for identifying wood species and their geographic origin by detecting specific volatile compounds (3). However, GC–MS faces challenges related to the insufficient volatility of certain wood components, as well as the complexity and destructiveness of the analytical process (3).

Summer sunny forest | Image Credit: © aksol - stock.adobe.com

Another technique that is used for this purpose is X-ray diffraction (XRD, primarily utilized to examine crystalline materials in wood, but often is hindered by the labor-intensive sample preparation and the potential lack of distinct crystalline structures in certain wood components (3). Infrared (IR) and Raman spectroscopy have also been employed to identify functional groups within wood, offering insights into its chemical composition (3). These techniques are valued for their non-destructive nature and rapid data acquisition capabilities, making them suitable for high-throughput analysis. However, Raman spectroscopy is limited by issues such as fluorescence interference and sensitivity to moisture (3).

In a recent study from Microchemical Journal, Vishal Sharma of Panjab University (Chandigarh, India) and his team demonstrated the potential of attenuated total reflectance Fourier transform infrared (ATR-FT-IR) spectroscopy to overcome the limitations seen in traditional methods (3).

ATR-FT-IR allows for the detailed characterization of wood components by analyzing spectral data, revealing unique chemical signatures associated with different wood species (3). The use of a diamond crystal in the ATR accessory enables straightforward scanning of samples with high moisture content or aqueous solutions, requiring minimal sample preparation (3). This method facilitates the utilization of the entire mid-infrared (mid-IR) spectrum, providing a comprehensive analysis without the limitations imposed by conventional IR spectroscopy (3).

Sharma’s study also shows the value of integrating machine learning with ATR-FT-IR spectroscopy. The research team used tree-based supervised machine learning algorithms, including Decision Tree, Extra Tree, Random Forest, and CatBoost classifiers, to enhance the accuracy of wood species discrimination (3). These algorithms were particularly effective in deciphering complex patterns within ATR-FT-IR spectra, allowing for the precise identification of wood species (3).

A critical aspect of the study is the application of Isolation Forest (iForest) for identifying and removing outliers, which are subsequently excluded from supervised modeling (3). This step is crucial in ensuring the reliability of the data used in machine learning. Another useful tool is principal component analysis (PCA). It is valuable in this application area because it can reduce dimensionality and streamline the analysis process (3).

The study demonstrated that machine learning techniques with ATR-FT-IR spectroscopy can differentiate between Eucalyptus, Dalbergia, and Populus wood species. The ability to accurately and efficiently identify wood species on-site, using portable ATR-FT-IR instruments, could change the way law enforcement agencies and timber industries operate. By facilitating rapid and reliable wood identification, this technology could play a pivotal role in curbing illegal logging and promoting sustainable forestry practices (3).

Moreover, the study's focus on wood species prevalent in the Indian timber industry highlights the broader applicability of this approach. The methods developed in this research could be adapted for use in other regions and industries, contributing to global efforts to combat deforestation and illegal timber trade (3).

By integrating ATR-FTIR spectroscopy with machine learning, this study provides a better tool for combating illegal logging and promoting sustainable forestry practices. The practical implications of this research are far-reaching, offering new possibilities for law enforcement, timber industries, and environmental conservation efforts worldwide (3). As technological progress continues, the future of wood forensics looks promising, with the potential to make a substantial impact on global efforts to protect our forests and natural resources (3).

References

(1) Grant, J.; Chen, H. K. Using Wood Forensic Science to Deter Corruption and Illegality in the Timber Trade. World Wildlife Fund. Available at: https://www.worldwildlife.org/pages/tnrc-topic-brief-using-wood-forensic-science-to-deter-corruption-and-illegality-in-the-timber-trade#:~:text=One%20of%20the%20most%20promising,product%20(see%20Figure%202). (accessed 2024-09-02).

(2) Henderson, M. Forestry Forensics: Using AI to Identify Wood. Mississippi State University. Available at: https://www.cfr.msstate.edu/news/news_article.asp?guid=817 (accessed 2024-09-03).

(3) Sharma, A.; Garg, S.; Sharma, V. ATR-FTIR Spectroscopy and Machine Learning for Sustainable Wood Sourcing and Species Identification: Applications to Wood Forensics. Microchem. J. 2024, 200, 110467. DOI: 10.1016/j.microc.2024.110467