Hyperspectral imaging and Raman spectroscopy, nondestructive methods to assess wine grape composition

Abstract

Grape composition is of high interest for producing quality wines. For that, grape analyses are necessary, and they still require sample preparation, whether with classical analyses or with NIR analyses. The aim of the study was to test the ability of two nondestructive analyses, directly on grapes, hyperspectral imaging (HSI) and Raman spectroscopy to assess their composition.
For that, 7 grape varieties were analyzed for 2 vintages. Each grape was characterized by its technological ripening (levels of sugars, organic acids and pH) and its phenolic ripeness (total phenolic, total flavonoids, total anthocyanins contents, as well as extractable phenolic, extractable flavonoids, extractable anthocyanins, values obtained from a model wine maceration from skins, and color intensity). Spectra were recorded on 100 and 40 fresh berries per date and variety respectively with hyperspectral imaging and Raman. Raw data underwent different pretreatments (SNV, 1st and 2nd derivative) and PLS-R were then realized in order to provide models to assess grape composition.
The results showed that the 1st derivative data pretreatment generated better models and was then kept for all following analyses. Both methods, Raman spectroscopy and hyperspectral imaging, showed good ability to assess technological ripening parameters (sugar and acid contents) as well as phenolic content (TPI, Total Phenolics, Total Anthocyanins, Total Flavonoids and their extractable equivalents) (with globally R² > 0.81). However, it was not possible to reach the color intensity of grapes.
Even if both methods have the potential to assess wine grape quality on 11 important parameters, the quality of the models generated in our study was dependent on the quality parameter, the type of grapes (color) and the method, except for fructose, TSS and Extractable Anthocyanin contents, which were equivalent. Thus, the glucose concentration and the Total Phenolic Index (TPI) were better assessed by Raman spectroscopy, whereas Extractable Phenolics content was better estimated by HSI for both white and red grapes as well as Total  Anthocyanin content. Tartaric acid, Total Flavonoids, Color Intensity and extractable Flavonoids were better assessed by HSI for red grapes but by Raman for white grapes.
The quality of the generated models was yet dependent on the color of grapes and the parameter considered. More data would be necessary to strengthen the models but the proof of concept was successful with this study