Abstract
Accelerating climate change poses unprecedented challenges to viticulture in Mediterranean basins, where increasing frequencies of prolonged droughts and extreme heatwaves threaten both grape yield and wine quality. Conventional physiological measurements of vine water status, such as stem water potential using pressure chambers, are labor-intensive, destructive, and lack the spatial resolution required for precision irrigation management. This study presents the development, calibration, and field validation of an unmanned aerial vehicle (UAV)-based hyperspectral imaging platform specifically engineered for the early detection of water deficit stress in commercial vineyards (Vitis vinifera L. cv. Tempranillo and Syrah). Flying at an altitude of 50 m with an ultra-compact pushbroom sensor (400–1000 nm across 150 contiguous spectral channels), the system captured sub-decimeter spatial resolution data synchronized with ground-truth physiological measurements throughout critical phenological stages. By coupling narrow-band physiological reflectance indices, notably the Photochemical Reflectance Index (PRI) and the Water Index (WI), with machine learning regression architectures, the model predicted leaf stem water potential (Ψstem) with high accuracy (R² = 0.84, RMSE = 0.14 MPa). The spatial-temporal mapping successfully identified incipient canopy water stress up to nine days prior to the manifestation of visible symptoms or traditional broad-band vegetation index degradation. These findings demonstrate that lightweight airborne hyperspectral remote sensing provides a non-invasive, scalable decision-support framework that optimizes vineyard water-use efficiency and advances climate-resilient viticultural sustainability.
