Real-time monitoring of chromatic and phenolic dynamics of vinification employing UV-Vis spectroscopy, Python and chemometrics

Increasing consumer demand for high-quality wines required the introduction of more comprehensive monitoring indicators for quality management, necessitating winemakers to strive for rapid, real-time, and affordable monitoring tools during winemaking. This study investigated the feasibility of cost-efficient and easy-to-implement ultraviolet-visible (UV-Vis) spectroscopy combined with chemometric methods to predictively monitor the evolution of wine color and phenolics throughout commercial fermentation. Fermentations on a winery scale were carried out with a daily analysis of the UV-Vis spectra, color, and phenolics. Furthermore, characteristic wavelengths were established by a self-compiled Python program to reduce redundancy in spectra. The optimized partial least squares (PLS) model predicted well with a residual prediction deviation (RPD) greater than 2.54, and the robustness was further verified by a robust external Bland-Altman method. UV-Vis spectroscopy can achieve objective monitoring of the practical fermentation profile, which can better help winemakers make decisions to improve wine quality and consistency.