Premium
A comparative study to distinguish the vineyard of origin by NIRS using entire grapes, skins and seeds
Author(s) -
FerrerGallego Raúl,
HernándezHierro José Miguel,
RivasGonzalo Julián C,
EscribanoBailón M Teresa
Publication year - 2012
Publication title -
journal of the science of food and agriculture
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.782
H-Index - 142
eISSN - 1097-0010
pISSN - 0022-5142
DOI - 10.1002/jsfa.5842
Subject(s) - vineyard , linear discriminant analysis , sample (material) , wine , partial least squares regression , mathematics , tracing , computer science , pattern recognition (psychology) , statistics , artificial intelligence , horticulture , chemistry , biology , food science , chromatography , operating system
Abstract BACKGROUND: Interest in high‐quality products with a clear geographical origin is increasing. For the wine industry and market sector, identity preservation is of fundamental importance owing to the large number of geographical classifications. Nowadays, there is a growing demand for analytical methods for tracing grapes and wines. In the oenological sector, infrared spectroscopy is becoming an attractive tool allowing simultaneous measurement of several analytical parameters and enabling real‐time decision making. RESULTS: Discriminant partial least squares, a supervised pattern recognition technique, was employed to discriminate between vineyards of origin using the near‐infrared spectra of intact grapes, skins or seeds. In order to compare the three sample presentations, a receiver operating characteristic curve was used. The best results were obtained using intact grape seeds, with prediction rates of samples correctly classified of about 95%, although the good results obtained with entire grapes (about 93% of samples correctly classified) and the simplicity of use of the fibre optic probe could advise using entire grape presentation for comprehensive studies. CONCLUSION: The procedure reported here seems to have excellent potential for a fast and reasonably inexpensive analysis of the origin of samples. It is noted that such classification can be made at any time of ripening. This paper provides information of interest to develop new and extensive models in the future. © 2012 Society of Chemical Industry