Correction of moisture effects on near infrared calibration for the analysis of phenol content in eucalyptus wood extracts

Methods based on near infrared spectroscopy used to assess wood properties are susceptible to variations in physical parameters (temperature, grain size, etc.). As wood is a hygroscopically sensitive material, we studied the effects of moisture on near infrared absorbance and calibration to accurately determine the application potential of this technique under routine. A collection of Eucalyptus urophylla x E. grandis hybrid wood pieces were analysed to obtain reference calibration of polyphenol contents in wood extracts via NIR spectra acquired under constant moisture conditions. Other specimens from the same source were assessed to obtain spectra for eight moisture contents spanning a broad variation range. The effects of moisture on absorption and on estimates based on a reference model were analysed. An increase in moisture content prompted a rise in near infrared absorption over the entire spectrum and for water O-H absorption bands. The polyphenol content estimates obtained by assessing specimens against the reference calibration at variable moisture contents revealed prediction bias. Five correction methods were then tested to enhance the robustness relative to moisture. In-depth calibration and external parameter orthogonalization (EPO) were found to be the most efficient methods for offsetting this factor.