New PLS analysis approach to wine volatile compounds characterization by near infrared spectroscopy (NIR)

This work aims to explore the potential of near infrared (NIR) spectroscopy to quantify volatile compounds in Vinho Verde wines, commonly determined by gas chromatography. For this purpose, 105 Vinho Verde wine samples were analyzed using Fourier transform near infrared (FT-NIR) transmission spectroscopy in the range of 5435 cm^-1 to 6357 cm^-1. Boxplot and principal components analysis (PCA) were performed for clusters identification and outliers removal. A partial least square (PLS) regression was then applied to develop the calibration models, by a new iterative approach. The predictive ability of the models was confirmed by an external validation procedure with an independent sample set. The obtained results could be considered as quite good with coefficients of determination (R²) varying from 0.94 to 0.97. The current methodology, using NIR spectroscopy and chemometrics, can be seen as a promising rapid tool to determine volatile compounds in Vinho Verde wines.