Statistical Modelization of the Descriptor “Minerality” Based on the Sensory Properties and Chemical Composition of Wine

• Main Authors: Elvira Zaldívar Santamaría, David Molina Dagá, Antonio T. Palacios García
• Format: Article
• Language: English
• Published: MDPI AG 2019-11-01
• Series: Beverages
• Subjects: minerality; partial least squares regression; predictive model; white wine; red wine
• Online Access: https://www.mdpi.com/2306-5710/5/4/66

When speaking of “minerality” in wines, it is common to find descriptive terms in the vocabulary of wine tasters such as flint, match smoke, kerosene, rubber eraser, slate, granite, limestone, earthy, tar, charcoal, graphite, rock dust, wet stones, salty, metallic, steel, ferrous, etc. These are just a few of the descriptors that are commonly found in the tasting notes of wines that show this sensory profile. However, not all wines show this mineral trace at the aromatic and gustatory level. This study has used the statistical tool partial least squares regression (PLS) to mathematically model the attribute of “minerality” of wine, thereby obtaining formulas where the chemical composition and sensory attributes act jointly as the predictor variables, both for white wines and red wines, so as to help understand the term and to devise a winemaking approach able to endow wines with this attribute if desired.