Relationship between sensory analysis and near infrared spectroscopy in Australian Riesling and Chardonnay wines

The aim of this work was to investigate the relationship between sensory analysis and visible and near infrared spectroscopy in two Australian white wine varieties (namely unwooded Chardonnay and Riesling). A total of 120 samples (2 varieties x 3 replicates x 20 commercial labels) were scanned in transmission mode in the visible and near infrared range (400-2500 nm). For the purposes of this study four aroma descriptors generated from a sensory descriptive analysis study: estery, honey, passionfruit, lemon-citrus and two palate properties: overall flavour and sweetness were selected for investigation. Calibration models between sensory properties and spectra were developed using partial least squares regression (PLS1 and PLS2) with cross validation. The correlation coefficients (R-cal) were greater than 0.70 for estery, lemon and honey, and less than 0.50 for passionfruit, overall flavour and sweetness in both calibration and cross validation. Therefore, this work demonstrates that some relationships between sensory data and both visible and near infrared spectra exists to assess sensory properties in the white wine varieties. Further work will be carried out with a larger set of data including additional sensory properties in different white wine varieties in order to validate the method. (c) 2005 Elsevier B.V. All rights reserved.