Utilization of remote sensing technology to detect riesling vineyard variability

Abstract

Context and purpose of the study – Vineyard blocks can vary spatially with respect to several viticulturally significant qualities such as soil variables, vine vigor, vine physiology, yield components, and berry composition. The ability to detect this variation enables the application of precision viticulture, whereby intra‐vineyard variability can be readily identified and corresponding responses can be made. Although it has been well established that this variation can exist, its detection is often difficult, with vineyard blocks spanning large areas and variation occurring over several variables. The aim of this project was to determine if remote and proximal sensing technologies could be used to detect this vineyard variation in six Ontario Riesling vineyards over a 3‐year period.

Material and methods – Six commercial Riesling vineyards across the Niagara Peninsula in Ontario, Canada were selected and 80‐100 grapevines, in a ≈8 m x 8 m grid pattern, were identified and geolocated. From these vines, the following variables were measured in 2015‐2017: soil moisture, vine water status (leaf water potential, leaf ψ; leaf stomatal conductance, gs), vine size, yield components, berry composition, winter hardiness, and grapevine leaf roll‐associated virus (GLRaV) infection. Furthermore, two sensing technologies—a ground‐based red/green/blue (RGB) proximal sensing system (GreenSeeker), and an unmanned aerial vehicle (UAV) with two sensors (RGB and thermal), collected electromagnetic reflectance from each vineyard block. These data were transformed into normalized difference vegetation index (NDVI). Lastly, replicate wines were made from grapes harvested from areas of low vs high NDVI. Wines were subjected to sensory sorting and the sorting data were subjected subsequently to correspondence analysis, creating a Chi‐square metric map that displayed the wines and their descriptors on a descriptor‐based space. The overall hypothesis was that maps produced from NDVI data could be used to detect variation in other variables such as leaf ψ, gs, berry composition, and GLRaV status, as well as implicate wine quality.

Results – NDVI maps demonstrated similar spatial configurations to maps of yield, vine size, berry weight, water status, and berry composition. Spatial zones corresponding to high NDVI were associated with zones of high vine water status, vine size, yield, titratable acidity (TA) and low soluble solids and terpene concentration. NDVI data as well as vine size, leaf ψ, gs, GLRaV infection, winter hardiness, and berry composition consisted of significant spatial clustering within the vineyard. Both the proximal and UAV technologies produced maps of similar spatial distributions; however, the GreenSeeker NDVI data provided more significant relationships with agricultural data compared to the UAV NDVI. Direct positive correlations were observed between NDVI vs. vine size, leaf gs, leaf ψ, GLRaV infection, yield, berry weight, and TA and inverse correlations with soluble solids and terpene concentration. Wines created from areas of high vs low NDVI differed slightly in basic wine composition (pH, TA, ethanol). Sensorially, panelists were often able to distinguish between wines made from high vs. low NDVI zones and associate those wines with specific descriptors. Ultimately, remote sensing demonstrates the ability to consistently detect areas within a vineyard differing in several important variables, which have implications for vine physiology, berry composition, and wine sensory attributes.