Scalable early detection of grapevine virus infection with airborne imaging spectroscopy.

Abstract

The US wine and grape industry suffers $3B in damages and losses annually due to viral diseases such as Grapevine Leafroll-associated Virus Complex 3 (GLRaV-3). Current detection methods are labor intensive and expensive. GLRaV-3 undergoes a latent period in which the vines are infected but do not yet display visible symptoms, making it an ideal model to evaluate the scalability of imaging spectroscopy-based disease detection. We deployed the NASA Airborne Visible and Infrared Imaging Spectrometer Next Generation (AVIRIS-NG) to detect GLRaV-3 in Cabernet Sauvignon grapevines in Lodi, CA in September 2020. Foliage was removed from the vines as part of mechanical harvest soon after imagery acquisition. In both Sept. 2020 and 2021, industry collaborators scouted 317ac on a vine-by-vine basis for visible viral symptoms and collected a subset for molecular confirmation testing. Grapevines identified as visibly diseased in 2021, but not 2020, were assumed to have been latently infected at time of acquisition. We trained spectral models with random forest and synthetic minority oversampling technique to distinguish non-infected and GLRaV-3-infected grapevines. Non-infected and GLRaV-3 infected vines could be differentiated both pre- and post-symptomatically at 1m through 5m resolution. The best-performing models had 87% accuracy distinguishing between non-infected and asymptomatic vines, and 85% accuracy distinguishing between non-infected and asymptomatic + symptomatic vines. The importance of non-visible wavelengths suggests this capacity is driven by disease-induced changes to overall plant physiology. Our work sets a foundation for using the forthcoming hyperspectral satellite Surface Biology and Geology for regional disease monitoring.