On-the-go hyperspectral imaging for monitoring grape maturity in Tempranillo: supporting adaptive harvest decisions under climate variability

Abstract

Climate change is disrupting grapevine phenology and threatening terroir-driven wine typicity, challenging the definition of optimal harvest timing. This study develops and validates an on-the-go dual-range hyperspectral imaging (HSI) workflow for non-destructive, real-time monitoring of grape maturity in Vitis vinifera L. cv. Tempranillo. The system integrates two push broom cameras—VIS-NIR (400–1000 nm) and SWIR (900–1700 nm)—mounted on a mobile platform operating under field conditions. Over seven sampling dates from veraison to harvest (August–September 2025), hyperspectral images were collected in a commercial vineyard in Logroño (La Rioja, Spain). Corresponding reference analyses included total soluble solids (TSS), pH, titratable acidity (TA), tartaric acid, malic acid, yeast-assimilable nitrogen (YAN), anthocyanins, and total phenolic index (TPI).

Partial least squares (PLS) regression models were developed to predict grape composition parameters from processed hyperspectral data. Model performance was assessed through cross-validation, exhibiting very good to excellent predictive performance for key technological maturity traits: TA (R²cv = 0.96), malic acid (R²cv = 0.94), pH (R²cv = 0.93), and TSS (R²cv = 0.88). YAN (R²cv = 0.71) achieved good quantitative accuracy, while tartaric acid (R²cv = 0.58), anthocyanins (R²cv = 0.60), and TPI (R²cv = 0.64) provided satisfactory qualitative discrimination among maturity classes. The high performance for malic acid highlights the relevance of including the SWIR spectral range, where absorption features of organic acids are more pronounced. This work represents, to the best of our knowledge, the first field-based demonstration of YAN prediction from hyperspectral data, expanding the analytical scope of proximal sensing for viticulture.

The proposed on-the-go HSI framework offers a practical, scalable tool for adaptive harvest decision-making under increasing climate variability. It supports the maintenance of varietal typicity and wine style consistency by facilitating timely, data-driven harvest scheduling. Future work will extend model validation across seasons and sites and integrate automated segmentation and machine learning algorithms for operational deployment.

References

Bertoglio, R., Piliego, M., Guadagna, P., Gatti, M., Poni, S., & Matteucci, M. (2024). On-the-go table grape ripeness estimation via proximal snapshot hyperspectral imaging. Computers and Electronics in Agriculture, 226, 109354. https://doi.org/10.1016/j.compag.2024.109354

Cervera, M. T., Cabezas, J. A., Rodríguez-Torres, I., Chavez, J., Cabello, F., & Martínez-Zapater, J. M. (2002). Varietal diversity within grapevine accessions of cv. Tempranillo. Vitis, 41(1), 33-6. http://hdl.handle.net/10261/52050

Dale, L. M., Thewis, A., Boudry, C., Rotar, I., Dardenne, P., Baeten, V., & Pierna, J. A. F. (2013). Hyperspectral imaging applications in agriculture and agro-food product quality and safety control: A review. Applied Spectroscopy Reviews, 48(2), 142-159. https://doi.org/10.1080/05704928.2012.705800

Fernández-Novales, J., Barrio, I., & Diago, M. P. (2021). Non-invasive monitoring of berry ripening using on-the-go hyperspectral imaging in the vineyard. Agronomy, 11(12), 2534. https://doi.org/10.3390/agronomy11122534

Goetz, A. F., Vane, G., Solomon, J. E., & Rock, B. N. (1985). Imaging spectrometry for earth remote sensing. Science, 228(4704), 1147-1153. DOI: 10.1126/science.228.4704.1147

Gowen, A. A., O’Donnell, C. P., Cullen, P. J., Downey, G., & Frias, J. M. (2007). Hyperspectral imaging–an emerging process analytical tool for food quality and safety control. Trends in Food Science & Technology, 18(12), 590-598. https://doi.org/10.1016/j.tifs.2007.06.001

Gutiérrez, S., Tardaguila, J., Fernández‐Novales, J., & Diago, M. P. (2019). On‐the‐go hyperspectral imaging for the in‐field estimation of grape berry soluble solids and anthocyanin concentration. Australian Journal of Grape and Wine Research, 25(1), 127-133. https://doi.org/10.1111/ajgw.12376

Ibáñez, J., Muñoz-Organero, G., Zinelabidine, L. H., de Andrés, M. T., Cabello, F., & Martínez-Zapater, J. M. (2012). Genetic origin of the grapevine cultivar Tempranillo. American Journal of Enology and Viticulture, 63(4), 549-553. DOI: 10.5344/ajev.2012.12012

OIV. (2024). Compendium of International Methods of Wine and Must Analysis (International Organisation of Vine and Wine: Dijon, France).

OIV. (2017). Vine varieties distribution in the world (International Organisation of Vine and Wine: Paris, France).

Ramos, M. C., & de Toda, F. M. (2020). Variability in the potential effects of climate change on phenology and on grape composition of Tempranillo in three zones of the Rioja DOCa (Spain). European Journal of Agronomy, 115, 126014. https://doi.org/10.1016/j.eja.2020.126014

Shenk, J. S., & Westerhaus, M. O. (1995). Routine operation, calibration, development and network system management manual. NIRSystems Inc., Silver Spring, MD, USA, 3.

Tardaguila, J., Stoll, M., Gutiérrez, S., Proffitt, T., & Diago, M. P. (2021). Smart applications and digital technologies in viticulture: A review. Smart Agricultural Technology, 1, 100005. https://doi.org/10.1016/j.atech.2021.100005

Van Leeuwen, C., Destrac-Irvine, A., Dubernet, M., Duchêne, E., Gowdy, M., Marguerit, E., Pieri, P., Parker, A., de Rességuier, L., & Ollat, N. (2019). An update on the impact of climate change in viticulture and potential adaptations. Agronomy, 9(9), 514. https://doi.org/10.3390/agronomy9090514

Ye, W., Xu, W., Yan, T., Yan, J., Gao, P., & Zhang, C. (2023). Application of near-infrared spectroscopy and hyperspectral imaging combined with machine learning algorithms for quality inspection of grape: a review. Foods, 12(1), 132. https://doi.org/10.3390/foods12010132

Acknowledgments

This study is framed within the HyperGrape project (PID2023-150555OB-I00, funded by MCIU/AEI/10.13039/501100011033 and by FEDER, EU).