Prediction of sensory attributes in winemaking grapes by on-line near-infrared spectroscopy based on selected volatile aroma compounds

Abstract

Aroma represents an important quality aspect for wine. The aroma of different grapes and wines is formed by the varying composition and concentrations of numerous aroma compounds, which result in different sensory impressions. The analysis of aroma compounds is usually complex and time-consuming, which requires the development of rapid alternative methods. In this study, grape mash samples were examined for aroma compounds, which were released under tasting conditions. A selection of the determined aroma compounds was grouped according to their sensory characteristics and calibration models were developed for the determination of sensory attributes by near-infrared (NIR) spectroscopy. The calibration models for the selected sensory attributes “fruity,” “green,” “floral” and “microbiological” showed very high prediction accuracies (0.979 < R2C < 0.996). Moreover, four different grape model solutions, whose compositions were based on the results from GC–MS-based analysis of the grape mash samples, were examined in a sensory evaluation. Despite large variation of the single values, the averaged values of the given scores for intensity of odour and taste showed differences between the model solutions for most of the evaluated sensory attributes. Sensory analysis remains essential for the evaluation of the overall aroma; however, NIR spectroscopy can be used as an additional and more objective method for the estimation of possible desired or undesired flavour nuances of grape mash and the quality of the resulting wine.Graphical