Objective  Contents of minerals and rare-earth elements in the alpine black teas from Shouning, Fujian were analyzed for grade classification.

  Methods  On 60 Jinmudan alpine black tea of various grades, sensory evaluation and chemical analysis were performed. A chemometric method was applied on the data to construct a grade discrimination model for the tea.

  Results  Based on the analytical data, 15 minerals detected in the specimens were ranked by their contents as K＞Ca＞Mg＞Mn＞Al＞Fe＞Na＞Zn＞Cu＞Pb＞Sn＞Se＞As＞Cd＞Hg. Among them, K, Ca, Mg, and Mn were more abundant and accounted for 96.78% of the total. The rank on the 15 rare-earth elements was Ce＞La＞Y＞Nd＞Gd＞Pr＞Dy＞Er＞Yb＞Sm＞Eu＞Ho＞Lu＞Tb＞Tm, with Ce, La, Y, and Nd being the dominant at contribution rate of 82.44%. For constructing a grading model, 14 characteristic elements, including Cu, Al, Sn, Ce, Tm, Lu, Se, Mn, Dy, Ho, Gd, Tb, Er, and Yb, were selected by the analysis of variance (ANOVA) combined with receiver operating characteristic curve (ROC curve) (P＜0.05, AUC＞0.7). A discriminant model of multi-layer perceptual neural network with a discriminant rate of 90.48% on the training group and 94.44% on the test group was established.

  Conclusion  K, Ca, Mg, Mn, and other minerals were richly found in the Shouning Jinmudan alpine black teas. The constructed multivariate statistical model based on the chemical composition could be used to classify the grades of the teas.