Due to climate change, heatwaves and prolonged periods of drought are more frequent and cause serious consequences to yield and berry composition of grapevine (Vitis vinifera L.). In response to this challenge, machine learning model was built to predict sunburn damages on the berries. The trial was conducted over two years (2022–2023) in a not irrigated vineyard of cv. Sangiovese, trained to vertical shoot positioning (VSP) spur pruned cordon. The vineyard was monitored from veraison to harvest with a weather station and thermocouples connected to a wireless sensor network (WSN). The evolution of the sunburn damages was visually evaluated twice a week. The damages appeared soon after veraison and the severity of the symptoms increased when heatwaves occurred. Weather station data including air temperature, solar radiation and relative humidity were analysed and used to build prediction models for sunburn damage. Ten parameters were derived from raw data to supply the prediction models of neural network (NN) and Support Vector Machine (SVM) optimised with gamma tuning. The NN achieved 90.32% accuracy in cross-validation, followed by SVM with 86.22% using radial kernel. The machine learning model was created using TensorFlow framework and it is available in the mobile phone application SHEET which will alert grape growers about the risk of sunburn damages on their orchards.

Allegro, G., Filippetti, I., Pastore, C., Sangiorgio, D., Valentini, G., Bortolotti, G., et al. (2025). Prediction of berry sunburn damage with machine learning: Results on grapevine (Vitis vinifera L.). BIOSYSTEMS ENGINEERING, 250(February 2025), 62-67 [10.1016/j.biosystemseng.2024.12.006].

Prediction of berry sunburn damage with machine learning: Results on grapevine (Vitis vinifera L.)

Allegro Gianluca;Ilaria Filippetti;Chiara Pastore;Daniela Sangiorgio;Gabriele Valentini;Gianmarco Bortolotti;
2025

Abstract

Due to climate change, heatwaves and prolonged periods of drought are more frequent and cause serious consequences to yield and berry composition of grapevine (Vitis vinifera L.). In response to this challenge, machine learning model was built to predict sunburn damages on the berries. The trial was conducted over two years (2022–2023) in a not irrigated vineyard of cv. Sangiovese, trained to vertical shoot positioning (VSP) spur pruned cordon. The vineyard was monitored from veraison to harvest with a weather station and thermocouples connected to a wireless sensor network (WSN). The evolution of the sunburn damages was visually evaluated twice a week. The damages appeared soon after veraison and the severity of the symptoms increased when heatwaves occurred. Weather station data including air temperature, solar radiation and relative humidity were analysed and used to build prediction models for sunburn damage. Ten parameters were derived from raw data to supply the prediction models of neural network (NN) and Support Vector Machine (SVM) optimised with gamma tuning. The NN achieved 90.32% accuracy in cross-validation, followed by SVM with 86.22% using radial kernel. The machine learning model was created using TensorFlow framework and it is available in the mobile phone application SHEET which will alert grape growers about the risk of sunburn damages on their orchards.
2025
Allegro, G., Filippetti, I., Pastore, C., Sangiorgio, D., Valentini, G., Bortolotti, G., et al. (2025). Prediction of berry sunburn damage with machine learning: Results on grapevine (Vitis vinifera L.). BIOSYSTEMS ENGINEERING, 250(February 2025), 62-67 [10.1016/j.biosystemseng.2024.12.006].
Allegro, Gianluca; Filippetti, Ilaria; Pastore, Chiara; Sangiorgio, Daniela; Valentini, Gabriele; Bortolotti, Gianmarco; Kertész, István; Lien Le Phuo...espandi
File in questo prodotto:
File Dimensione Formato  
Allegro et al. (2025).pdf

accesso aperto

Tipo: Versione (PDF) editoriale
Licenza: Licenza per Accesso Aperto. Creative Commons Attribuzione (CCBY)
Dimensione 2.71 MB
Formato Adobe PDF
2.71 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11585/999087
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 0
  • ???jsp.display-item.citation.isi??? ND
social impact