In the Bordeaux region of southwestern France, numerous vineyards are turning finicky grapes into bold blends of red wine. Some bottles sell for thousands of dollars each. Prestigious chateaus pride themselves on the mysterious combination of soils, microclimates, traditional methods and known as terroir that make their wines exceptional.
“This is one of those terms that the wine industry likes to keep a bit mysterious, it’s part of the magic of wine,” said Alex Pouget, a computational neuroscientist at the University of Geneva.
Dr. Pouget seeks to apply chemical precision to this phenomenon.in study In a paper published Tuesday in the journal Communication Chemistry, he and his colleagues described a computer model that can pinpoint which Bordeaux wineries are producing wine based solely on its chemical composition. The model also predicted the wine’s year of production, known as its vintage, with about 50% accuracy.
Although wine enthusiasts often claim to be able to tell the difference between top estate wines, he says they rarely conduct blind tasting tests. “People have been making these claims for decades, but we’ve never really had objective measurements to show that this is true,” he says.
Dr. Pouget grew up in Paris in a family that drank only Bordeaux (“You’re pretending Burgundy doesn’t exist,” he said). As a young neuroscientist in the late 1980s, he studied the brain using machine learning, a type of artificial intelligence that identifies patterns in large datasets. Although he believed these methods could be useful to the wine industry, he was unable to test the idea for 30 years.
He teamed up with Stéphanie Marchand of the Institute of Grape and Wine Sciences in Bordeaux. He has created a database of 80 wines of various vintages from seven chateaus. The database stored each wine’s chemical signature gleaned from gas chromatography, an old and inexpensive method for breaking down substances into their molecular components.
The researchers trained an algorithm to look for common patterns in the wine’s chemical fingerprints. They were shocked by the results. The model grouped the wines into distinct clusters that matched the geographic location of the Bordeaux region. This showed that, as winemakers have argued for centuries, the particularities of each winery greatly influence the chemistry of the wines produced there.
The winery gave the researchers permission to study their wines on the condition that they not be named. Dr. Pouget said that all wines are part of famous wines. 1855 Bordeaux classificationA ranking established by Napoleon III to promote the best Bordeaux wines.
Dr. Pouget was surprised that the wine producers did not want to be named because the results of the study reinforced the idea that their wines were special. “There’s scientific evidence that it makes sense to charge people money because you’re creating something unique,” he said with a laugh.
The independent researchers said the study is part of recent research using machine learning to decipher terroir. “This is where the field is headed and we need to go to make sense of the wealth of data,” said David Jeffrey, an expert in wine chemistry at Australia’s University of Adelaide.
For example, he used machine learning to Classifying Shiraz wine From Australia’s Barossa Valley.
Dr. Jeffrey said this approach is “not just about what makes a good wine chemically.” The model could help growers adjust their growing and winemaking methods to preserve product characteristics even if they experience years of unpredictable weather. “This is especially important in the face of a changing climate,” he said.
Another application of these models could lead to eradication, experts say. scam, fairly common in expensive wines. The producer adjusts the bottle, label Add a cork to make copying difficult.
“If you have any doubts about the origin of a wine, you can probably tell whether the wine is fake by analyzing the wines sourced from that winery as a benchmark,” says Head of Viticulture and Enology. , said Cornelis van Leeuwen. At Bordeaux Science Agro.
Dr van Leeuwen, who was not involved in the study, said the approach would likely work in any wine region, as long as the model was trained on a wide range of wines from different producers and vintages. However, an open question is whether the model will maintain its accuracy after several years, he said.
Pouget, who has an extensive wine collection, hopes to repeat the study with some of his favorite varieties from the Châteauneuf-du-Pape region in south-east France.
But among the best wines, age is probably more important than region, he says.
“I only drink old wine,” he said. “I think it’s a bit criminal for people under the age of 15 to drink alcohol.”