Assessment of vineyard performance to predict winegrape quality

Abstract

In many situations, winegrapes are grown by grapegrowers and then sold to winemakers to make wine. Developing quantifiable measures of winegrape quality is seen as beneficial to winemakers and grapegrowers to increase clarity and accountability between both parties. Currently vineyard performance in any given vintage can be assessed by price paid for the fruit at market, past performance, objective and subjective measures or a combination of these. Past vineyard studies have shown that defining winegrape quality is not a simple exercise and determining what vineyard measurements to take, when to take them and how to interpret their influence has been the subject of wide-ranging research. One of the aims of this study was to develop models, on a commercial scale, for predicting winegrape quality from known vineyard performance measures. Two winegrape cultivars commonly grown in Australia; Shiraz and Cabernet Sauvignon were investigated. This research developed a methodology to assess winegrape quality in a commercial situation in the McLaren Vale, Langhorne Creek and Adelaide Hills wine regions using vineyard measures, assessments of canopy architecture and berry composition. Known vine performance measures were taken at key phenological growth stages and then assessed for their ability to predict winegrape quality. Two models for predicting winegrape quality were developed - a growing season (GS) and a harvest (HRV) model. The GS prediction model used image analysis of canopy architecture and vineyard observations taken up to 50% veraison (EL 35, Coombe 2004). This was done to assess if early measures could predict wine quality and therefore allow time for grapegrowers to adjust their practices before harvest. The HRV prediction model combined image analysis of canopy architecture with berry composition measurements (total tannin, anthocyanins and phenolics) up until the harvest period. Results of the trial on Shiraz showed winegrape quality prediction by the HRV models were better than the corresponding GS models as measures of grape composition at harvest were good predictors of winegrape quality. When Cabernet Sauvignon was assessed by the same methodology the HRV models and the corresponding GS models had comparable predictive abilities. This research showed that models of winegrape quality can be developed in commercial vineyards by combining canopy architecture measurements with grape berry composition. Based on the results with these grape varieties, using this methodology, winegrape quality modelling could be adapted for use in other regions and on other varieties.