Taking grapevine yield forecasting into the digital age

Abstract

The technology of Near Infrared Spectroscopy (NIRS) can be a powerful analytical technique for the rapid analysis of various compositional parameters in wine grapes, must and grapevine tissues. This study investigated the feasibility of using NIRS technology to non-destructively predict grapevine bud fruitfulness. Discriminant Analysis models to categorise buds were prepared using Principle Component Analysis reduced spectra and were compared against microscopic bud dissection data and actual fruitfulness measured in the field at flowering.

Summary

Accurate predictions were obtained for the bud fruitfulness of both Pinot Noir and Chardonnay varieties over 4 consecutive seasons, utilising three different instruments. As indicated by good prediction accuracy values, portable NIRS technology has the potential to predict grapevine bud fruitfulness. In the final year of prediction the Bruker MPA instrument employed in the laboratory demonstrated an accuracy of 97.32% when predicting the fruitfulness of Pinot Noir and Chardonnay buds. In the field scanning buds in situ, the ARCoptics Rocket portable instrument demonstrated an accuracy of 100% for predicting bud fruitfulness of Pinot Noir buds in 2018 and an accuracy of 79.65% in 2019 when predicting fruitfulness of Pinot Noir and Chardonnay buds. In 2018 when there were sufficient unfruitful buds to create a category of their own and when calibrations were prepared using ‘yes’ or ‘no’ for presence of inflorescences the prediction rate was 80% correct. This high accuracy for predicting unfruitful buds would be particularly valuable for industry.

Overall the challenges with the technique include orientation of the probe, confusion due to compound buds, and confusion created by the shoot apex contributing to the spectral signal.Software developed to use alongside the technology has demonstrated proof of concept that the tools can be employed in the vineyard to scan a representative number of buds and predict the level of fruitfulness to inform pruning decisions on a block level. The software can be found at https://digiviti.indicium.cloud/. A video clip to assist with demonstrating the ease of application of the research for industry was also produced.

Digital technology will increasingly support decision making for management and this project has demonstrated the potential use of NIRS in supporting pruning decisions in vineyards for the wine industry.