Hyperspectral imaging for precision viticulture

Abstract

Precision viticulture aims to optimize vineyard management by monitoring and responding to variability within vine plots. This work presents a comprehensive study on the application of hyperspectral imaging (hsi) technology for monitoring purposes in precision viticulture.  Authors explore the deployment of hsi sensors on various platforms including laboratory settings, terrestrial vehicles, and unmanned aerial vehicles, facilitating the collection of high-resolution data across extensive vineyard areas. The technology’s capabilities are highlighted through its application in detecting early signs of diseases such as grapevine leafroll and red blotch viruses, evaluating water stress for informed irrigation, and mapping grape composition for precise canopy management and harvest timing. In the disease detection studies reported here, hsi coupled with machine learning demonstrated the ability to detect vines infected with viruses with 76% accuracy at the pre-symptomatic stage and 87% accuracy at the symptomatic stage compared to molecular tests and surpassing visual assessment by experts. In irrigation scheduling tasks, hsi data from nir to swir wavelengths helped manage irrigation by predicting vine water potential (ψstem) with significant accuracy (r2 = 0.54, rmse = 0.11 mpa). In on-the-go grape ripening monitoring, hsi on board of a utv was able to non-invasively assess grape composition, achieving high predictive accuracies for key grape parameters brix ( r2 = 0.91, nrmse of 7% ), ph (r2 = 0.90, nrmse = 6 %), ta (r2 = 0.85, nrmse of 7.5 (g/l)) , and total anthocyanins (r2 = 0.91, nrmse of 7 (mg/g berry fm)), thereby enhancing vineyard management by understanding spatial variability in grape attributes. The paper concludes with discussions on future directions of hsi in viticulture, emphasizing the need for automated data processing techniques and integration with other precision agriculture tools, underscoring the transformative potential of hyperspectral imaging in modernizing vineyard practices.

Sensoristica iperspettrale per la viticoltura di precisione

La viticoltura di precisione mira a ottimizzare la gestione dei vigneti monitorando e rispondendo alle variazioni all’interno delle parcelle. Questo lavoro presenta uno studio completo sull’applicazione della tecnologia di imaging iperspettrale (hsi) a scopo di monitoraggio nella viticoltura di precisione. Gli autori esplorano il dispiegamento di sensori hsi su varie piattaforme, inclusi ambienti di laboratorio, veicoli terrestri e veicoli aerei senza pilota, facilitando la raccolta di dati ad alta risoluzione su vaste aree di vigneto. Le capacità della tecnologia sono evidenziate attraverso la sua applicazione nel rilevamento precoce di malattie come i virus dell’accartocciamento fogliare e del red blotch, nella valutazione dello stress idrico per un’irrigazione informata, e nella mappatura della composizione dell’uva per una gestione precisa della chioma e della tempistica del raccolto. Negli studi di rilevamento delle malattie riportati qui, l’hsi abbinato all’apprendimento automatico ha dimostrato la capacità di rilevare viti infette da virus con una precisione del 76% allo stadio pre-sintomatico e dell’87% allo stadio sintomatico rispetto ai test molecolari e superando la valutazione visiva degli esperti. Nei compiti di programmazione dell’irrigazione, i dati hsi dalle lunghezze d’onda nir a swir hanno aiutato a gestire l’irrigazione prevedendo il potenziale idrico delle viti (ψstem) con un’accuratezza significativa (r2 = 0.54, rmse = 0.11 mpa). Nel monitoraggio della maturazione dell’uva on-the-go, l’hsi a bordo di un utv è stata in grado di valutare in modo non invasivo la composizione dell’uva, raggiungendo alte precisioni predittive per parametri chiave dell’uva brix (r2 = 0.91, nrmse del 7%), ph (r2 = 0.90, nrmse = 6%), ta (r2 = 0.85, nrmse di 7,5 (g/l)), e antociani totali (r2 = 0.91, nrmse del 7 (mg/g berry fm)), migliorando così la gestione del vigneto comprendendo la variabilità spaziale nelle caratteristiche dell’uva. Il documento si conclude con discussioni sulle future direzioni dell’hsi nella viticoltura, sottolineando la necessità di tecniche automatizzate di elaborazione dei dati e l’integrazione con altri strumenti di agricoltura di precisione, sottolineando il potenziale trasformativo dell’imaging iperspettrale nella modernizzazione delle pratiche viticole.

Imagerie hyperspectrale pour la viticulture de précision

la viticulture de précision vise à optimiser la gestion des vignobles en surveillant et en répondant aux variations au sein des parcelles. ce travail présente une étude complète sur l’application de la technologie d’imagerie hyperspectrale (hsi) à des fins de surveillance dans la viticulture de précision. les auteurs explorent le déploiement de capteurs hsi sur diverses plateformes, y compris les environnements de laboratoire, les véhicules terrestres et les véhicules aériens sans pilote, facilitant la collecte de données à haute résolution sur de vastes zones de vignobles. les capacités de la technologie sont mises en évidence par son application dans la détection précoce de maladies telles que les virus du red blotch et de l’enroulement, l’évaluation du stress hydrique pour une irrigation informée, et la cartographie de la composition du raisin pour une gestion précise de la canopée et du timing de la récolte. dans les études de détection de maladies rapportées ici, l’hsi couplée à l’apprentissage automatique a démontré la capacité de détecter les vignes infectées par des virus avec une précision de 76 % au stade pré-symptomatique et de 87 % au stade symptomatique par rapport aux tests moléculaires et en surpassant l’évaluation visuelle par des experts. dans les tâches de programmation de l’irrigation, les données hsi des longueurs d’onde nir à swir ont aidé à gérer l’irrigation en prédisant le potentiel en eau de la vigne (ψstem) avec une précision significative (r2 = 0.54, rmse = 0.11 mpa). dans le suivi de la maturation des raisins en continu, l’hsi embarquée sur un utv a pu évaluer de manière non invasive la composition des raisins, atteignant de hautes précisions prédictives pour des paramètres clés du raisin tels que le brix (r2 = 0.91, nrmse de 7%), le ph (r2 = 0.90, nrmse de 6%), l’acidité totale (r2 = 0.85, nrmse de 7.5 (g/l)), et les anthocyanes totaux (r2 = 0.91, nrmse de 7 (mg/g berry fm)), améliorant ainsi la gestion des vignobles par la compréhension de la variabilité spatiale des attributs du raisin. le document se conclut par des discussions sur les orientations futures de l’hsi en viticulture, en soulignant le besoin de techniques de traitement automatisé des données et d’intégration avec d’autres outils d’agriculture de précision, mettant en avant le potentiel transformateur de l’imagerie hyperspectrale dans la modernisation des pratiques viticoles.

Publication date: November 18, 2024

Issue: OIV 2024

Type: Article

Authors

Luca Brillante¹, Eve Laroche-Pinel¹

¹ Department of Viticulture and Enology, California State University Fresno – 2360 E Barstow Ave, Fresno, United States of America

Contact the author*

Tags

IVES Conference Series | OIV | OIV 2024

Citation

Related articles…

Pesticide – Free viticulture: towards agroecological wine-producing socio-ecosystems

Can we cultivate grapevine without pesticides? This is a huge challenge for this emblematic crop, which is one of the largest users of plant protection products. Pesticides are mainly used to protect the vine against leaf diseases (powdery mildew, mildew, black-rot), even in organic farming, which uses copper in particular. What are the research avenues that can help eliminate pesticides today?

Use of glutathione and a selected strain of metschnikowia pulcherrima as alternatives to sulphur dioxide to inhibit natural tyrosinase of grape must and prevent browning

The enzymatic browning of grape must is still a major problem in oenology today [1] being particularly serious when the grapes have been infected by grey rot [2]. Browning is an oxidation process that causes certain foods to turn brown, which often leads to them being rejected by consumers [3]. This is a particular problem in the case of wine, because grape must is very vulnerable to enzymatic browning [4].

Oenological potential of wines and agronomical characterisation of grapes from five white resistant Italian varieties at Serra Gaúcha, Southern Brazil

Rio grande do sul is the main grape producing state in Brazil, with the largest wine-growing area, responsible by 90% of the national production of wines and grape juices. Serra Gaúcha is the main vitivinicultural region, where around 15% of the area is destined to produce wines from vitis vinifera L. grapes. This region presents high rainfall during the grape maturation cycle, a factor that leads to great risk of attacks by fungal pathogens. the use of resistant varieties can reduce the cost and quantity of spraying, improving wine quality, focusing on a sustainable vitiviniculture.

Multi-mineral wine profiling and Artificial Intelligence: Implementing the signatures of each wine to train algorithms to meet the new challenges facing the wine industry

Multi-mineral wine profiling and artificial intelligence: implementing the signatures of each wine to train algorithms to meet the new challenges facing the wine industry. Although their quantity is minimal, minerals are essential elements in the composition of every wine. Their presence is the result of complex interactions between factors such as soil, vines, climate, topography, and viticultural practices, all influenced by the terroir. Each stage of the winemaking process also contributes to shaping the unique mineral and taste profile of each wine, giving each cuvée its distinctive characteristics.

Produce wines with no quantifiable phytosanitary residues – Impact of washing grapes?

Consumer expectations are increasingly shifting towards “residue-free wines.” However, from an analytical standpoint, “zero” does not exist. Laboratories often use the quantification limits of analysis methods to signify ‘zero.’ Improved techniques now allow for the quantification of levels that were previously undetectable. This is why we prefer to use the term “unquantifiable residue” rather than “absence of residues.”