Applicazione di sensori prossimali e algoritmi di computer vision per il monitoraggio dello stress idrico in vite: accuratezza, limiti e prospettive di spazializzazione

The increasing climatic variability affecting Italian viticulture requires innovative tools capable of monitoring vine water status with high spatial and temporal resolution. Within this context, the ViViS project aims to evaluate whether a fixed, proximal sensing system, specifically the WCam system developed by CET Electronics, can reliably detect spatial differences in vine water status and support precision irrigation strategies. Building on the technological framework developed in the previous IRRIVISION project, the present study investigates the spatial representativeness of fixed vision sensors by comparing WCam-derived indices with distributed physiological measurements collected across two contrasting vineyard plots (A and B) cultivated with Vitis vinifera cv. Glera. The VCam monitoring system integrates stereo-vision for leaf inclination estimation, infrared thermography for canopy temperature assessment, electronic leaves for thermal references, soil sensors, and a local weather station. Throughout the 2025 season, stem water potential (Ψstem), stomatal conductance, leaf temperature, and PSII photosynthetic efficiency (PhiPSII) were measured to validate WCam outputs. The first part of the season revealed clear physiological differences between the two plots: Plot B consistently exhibited more negative Ψstem values, higher leaf temperatures, and subtle differences in leaf inclination, all of which were accurately captured by the WCam system. These early divergences confirmed that the fixed sensor could detect intra-field spatial variability linked to soil properties. However, after mid-July, high rainfall events (167.8 mm were registered between July and early September; 375 mm total seasonal precipitation) homogenized soil moisture across the vineyard, causing a natural convergence of physiological responses between plots. The WCam system reproduced this convergence accurately, demonstrating that its lack of discrimination in this period was not a technological limitation but rather a reflection of the environmental conditions. Across the entire dataset, leaf temperature, both measured and WCam-estimated, emerged as the variable most strongly correlated with Ψstem, while leaf inclination showed significant relationships mainly at night or when expressed as day–night variation. These results confirm that the WCam system can retrieve physiologically reliable indicators of vine water status.