Wine characterisation according to geographical origin using analysis of mineral elements and rainfall correlation of oxygen isotope values

The wine industry has developed rapidly; however, wine fraud is a potential risk for consumers. In China, methods for detecting wine authenticity are far from perfect. To reduce the risk of counterfeit wines, Inductively Coupled Plasma-Mass Spectrometry and Isotope Ratio Mass Spectrometry were used to geographically classify 104 wines from four major production areas. In this paper, the naturally distributed characteristics of thirty-eight mineral elements contents and the effect of rainfall on the oxygen isotope values in wine were investigated. The result of delta O-18 ranged from -13 parts per thousand to 7 parts per thousand. The oxygen isotope of wine water in Northwest China is obviously more positive than that in South China. Linear discriminant analysis (LDA) showed 88.5% classification accuracy in training set and 81.7% in the cross-validation result. An artificial neural network (ANN) model determined origin of the wine with higher accuracy than LDA model. Furthermore, delta O-18 and Sr/Rb are important recognition elements in ANN, and the accuracy of region recognition can reach 90.9%. Thus, mineral elements and isotope ratios are effective in contributing to wine authenticity control in wine origin.