This study used Attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy and Fourier transform near-infrared (FT-NIR) spectroscopy with principal component regression (PCR) and partial least squares regression (PLS) to build hardwood prediction models. Wet chemistry analysis coupled with high performance liquid chromatography (HPLC) was employed to obtain the chemical composition of these samples. Spectra loadings were studied to identify key wavenumber in the prediction of chemical composition. NIR-PLS and FTIR-PLS performed the best for extractives, lignin and xylose, whose residual predictive deviation (RPD) values were all over 3 and indicates the potential for either instrument to provide superior prediction models with NIR performing slightly better. During testing, it was found that more accurate determination of holocellulose content was possible when HPLC was used. Independent chemometric models, for FT-NIR and ATR-FTIR, identified similar functional groups responsible for the prediction of chemical composition and suggested that coupling the two techniques could strengthen interpretation and prediction.
This research addressed a rapid method to monitor hardwood chemical composition by applying Fourier transform infrared (FT-IR) spectroscopy, with particular interest in model performance for interpretation and prediction. Partial least squares (PLS) and principal components regression (PCR) were chosen as the primary models for comparison. Standard laboratory chemistry methods were employed on a mixed genus/species hardwood sample set to collect the original data. PLS was found to provide better predictive capability while PCR exhibited a more precise estimate of loading peaks and suggests that PCR is better for model interpretation of key underlying functional groups. Specifically, when PCR was utilized, an error in peak loading of ±15 cm−1 from the true mean was quantified. Application of the first derivative appeared to assist in improving both PCR and PLS loading precision. Research results identified the wavenumbers important in the prediction of extractives, lignin, cellulose, and hemicellulose and further demonstrated the utility in FT-IR for rapid monitoring of wood chemistry.
This paper addresses the precision in factor loadings during partial least squares (PLS) and principal components regression (PCR) of wood chemistry content from near infrared reflectance (NIR) spectra. The precision of the loadings is considered important because these estimates are often utilized to interpret chemometric models or selection of meaningful wavenumbers. Standard laboratory chemistry methods were employed on a mixed genus/species hardwood sample set. PLS and PCR, before and after 1st derivative pretreatment, was utilized for model building and loadings investigation. As demonstrated by others, PLS was found to provide better predictive diagnostics. However, PCR exhibited a more precise estimate of loading peaks which makes PCR better for interpretation. Application of the 1st derivative appeared to assist in improving both PCR and PLS loading precision, but due to the small sample size, the two chemometric methods could not be compared statistically. This work is important because to date most research works have committed to PLS because it yields better predictive performance. But this research suggests there is a tradeoff between better prediction and model interpretation. Future work is needed to compare PLS and PCR for a suite of spectral pretreatment techniques.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.