In the heart of Portugal’s Douro Wine Region, a revolution is brewing, one that could redefine how we understand and manage grape maturation. Renan Tosin, a researcher from the Department of Geosciences, Environment and Spatial Planning at the University of Porto and INESC TEC, has pioneered a novel approach to precision viticulture that promises to enhance wine quality and production efficiency. His work, published in the journal ‘Smart Agricultural Technology’ (Intelligent Agricultural Technology), introduces a tomography-like (TL) method that could transform the way winemakers approach their craft.
Imagine being able to peer into the very heart of a grape, understanding its internal composition without ever touching it. This is the power of the TL method, which uses a Vis-NIR point-of-measurement sensor to penetrate grape tissues non-destructively. By employing visible and near-infrared light, the sensor provides spectral data that predict key physicochemical properties, such as soluble solids content (SSC), weight-to-volume ratio, chlorophyll, and anthocyanin levels across the grape’s skin, pulp, and seeds. “This method allows us to capture the dynamic process of grape maturation in a way that was previously impossible,” Tosin explains.
Over two growing seasons (2021–2022), Tosin and his team monitored grape maturation, collecting data at six post-veraison stages. The results were astonishing: detailed metabolic maps that revealed significant variations in SSC, chlorophyll, and anthocyanin levels across different vineyard zones. These maps, integrated with topographical factors like altitude and NDVI-based vigour assessments, provide a comprehensive view of how environmental variability impacts grape maturation.
The implications for the wine industry are profound. By generating high-throughput data, the TL method enables winemakers to implement site-specific management (SSM) practices. This means tailored interventions based on metabolic profiles, rather than relying solely on cultivar characteristics. “We can now guide winemakers to make data-driven decisions, aligning vineyard practices with specific wine quality goals,” Tosin notes.
The study also developed predictive models using random forest (RF) and self-learning artificial intelligence (SL-AI) algorithms. The RF model showed remarkable robustness, achieving stable predictions with an R² of at least 0.86 and a mean absolute percentage error (MAPE) of no more than 33.83%. These models, combined with the metabolic maps, offer a powerful tool for precision viticulture, enhancing both wine quality and production efficiency.
But how might this research shape future developments? The potential is vast. As Tosin puts it, “This is just the beginning. The TL method can be adapted for other crops, and the principles behind it can be applied to various agricultural sectors.” Imagine a future where farmers can non-destructively monitor the internal composition of their crops, tailoring management practices to specific needs. This could lead to more sustainable farming, reduced waste, and higher yields.
The TL method, as detailed in ‘Intelligent Agricultural Technology’, is more than just a scientific breakthrough; it’s a glimpse into the future of agriculture. By providing actionable insights and supporting site-specific management, it empowers winemakers and farmers to make informed decisions, ultimately enhancing the quality and efficiency of their products. As we stand on the brink of this agricultural revolution, one thing is clear: the future of farming is data-driven, and the TL method is leading the way.
