Scalable early detection of grapevine virus infection with airborne imaging spectroscopy

Abstract

Viral diseases, including Grapevine Leafroll-associated Virus Complex 3 (GLRaV-3), cause $3 billion in damages and losses to the United States wine and grape industry annually. GLRaV-3 has a well-studied, year-long latent period in which vines are infectious but do not yet display visible symptoms, making it an ideal model pathosystem to evaluate the scalability of symptomatic and asymptomatic imaging spectroscopy-based disease detection. Plant disease causes physiological and chemical changes to occur locally and systemically throughout a plant, which imaging spectroscopy can detect both directly and indirectly. Reliable and scalable disease detection during the latent period would greatly reduce management costs, as current detection methods are entirely ground-based, labor-intensive, and expensive. Here, we use data collected in September 2020 by the NASA Airborne Visible/Infrared Imaging Spectrometer Next Generation (AVIRIS-NG) to detect GLRaV-3 in Cabernet Sauvignon grapevines in Lodi, CA. During September 2020 and 2021, industry collaborators scouted 317 acres of Vitis vinifera winegrapes for visible disease symptoms, and collected a subset for confirmation molecular testing at a commercial facility. Grapevines identified as visibly diseased in 2021 were assumed to have been latently infected (asymptomatic) during the September 2020 AVIRIS-NG data collection. We combined random forest with synthetic minority oversampling technique (SMOTE) to train multiple spectral models able to distinguish between non-infected (NI) and GLRaV-3-infected grapevines. We observed clear spectral differences that allowed for differentiation between NI and GLRaV-3 infected vines both pre- and post-symptomatically at 1m through 5m resolution. Our two best performing models had 87% accuracy (0.73 Kappa) distinguishing between NI and asymptomatic (aSy), and 85% accuracy (0.71 Kappa) distinguishing between NI and (aSy + symptomatic [Sy]) respectively. We hypothesize these spectral differences are linked to changes in overall plant physiology induced by disease, as visible foliar symptoms were restricted to the lower canopy.

Towards Cloud-Native, Machine Learning Based Detection of Crop Disease with Imaging Spectroscopy

“…D. Predicted disease incidence raster for non-infected vs asymptomatic vines (top) and probability for each class (bottom). Figure taken from (Romero Galvan et al, 2022)…”

“…The AVIRIS-NG acquistions captured vineyards roughly a week after the disease incidence coordinates were recorded. For more detailed description of disease incidence collection, validation, and aggregation, see Romero Galvan (Romero Galvan et al, 2022). In brief, disease incidence data was collected at seven vineyards and covered 7 acres (ac) of red grape variety Aglianico, 204ac of Cabernet Sauvignon, and 57ac of Petite Sirah in the city of Lodi, CA by visual inspection for foliar symptoms (”scouting”) by expert teams with a subset sent for external validation via molecular testing in September of 2020 and 2021.…”

“…Following common convention of training ML models (Gholamy et al, 2018), our Random Forest models (RFm) were trained through a 70/30 training/validation split. The models were trained on spatially-resampled three-meter SI as Romero Galvan found to be optimal for GLRaV-3 detection in grapevine(Romero Galvan et al, 2022). In total, 268 acres were classified capturing 80% of disease incidence in the field.…”

“…Plant disease changes how solar radiation interacts with leaves, canopy, and plant energy balance, all of which is captured in SI. This underlying capacity is what enables airborne imaging spectroscopy to non-destructively detect biotic stress in both natural and agroecosystems asymptomatically (Zarco-Tejada et al, 2018, 2021; Sapes et al, 2022; Romero Galvan et al, 2022). Disease detection with SI has benefited greatly from data processing using artificial intelligence, specifically machine learning (ML) (Jiménez-Brenes et al, 2019; Hruŝka et al, 2018).…”

“…First, we aim to provide a general, cloud-based platform for training and testing SI-informed ML models for non-intrusive and scalable crop disease management. To that end, we showcase new techniques for processing (2.4) and refining spectroscopic imagery (2.3) in a novel agricultural application: grapevine leafroll associated virus complex 3 (GLRaV-3) detection in wine grape (2.1) (Romero Galvan et al, 2022). The second goal is to provide a general platform for researchers to explore the implications of future satellite missions, including SBG and beyond, whose data scales and resolutions are yet to be finalized, for actionable disease detection and decision making.…”