Abstract: It is an urgent affair to think up a quick and precise method in the identification of tea varieties. A rapid tea identification method by near infrared reflectance spectroscopy coupled with pattern recognition based on SIMCA was proposed in this paper. In the spectra region between 6500cm-1 and 5300cm-1, four predictive models of Longjing tea, Biluochun tea, Qihong tea and Tieguanyin tea were built separately by the standard normal variate (SNV) preprocessing method with SIMCA pattern recognition method. The results showed that four models are the best when 4, 5, 2 and 3 principal components were used separately in building models. Under theα=5% significance level, the identification rates of four models for the unknown samples are 90%, 80%, 100% and 100% in turn by means of NIR wave lengths, while, the rejection rates of four models are all 100%. A new idea by the quick and precise identification of tea was offered in this paper.

Key words: tea, near-infrared spectroscopy, Soft Independent Modelling of Class Analogy(SIMCA), identification