Javaux, MathieuClaessens, OliverOliverClaessens2025-05-142025-05-142025-05-142024https://hdl.handle.net/2078.2/36550Recent decades have seen a dramatic transformation in Belgian viticulture, with vineyard areas expanding more than fivefold and achieving record-breaking wine production, especially in 2022 - a year of exceptional drought. This study addresses the crucial challenge of accurately predicting vine water status. Indeed, while a certain level of water deficit can be beneficial for quality wine production, severe water shortages can adversely affect both yield and quality. Therefore, the adaptation of viticultural practices and precise water status assessment is paramount. Focusing on two non-irrigated vineyards in Wallonia with grass-covered inter-rows, this study employed high-resolution data acquired through UAV remote sensing to produce vine water status maps. The methodology utilized multispectral and thermal imagery, alongside in-situ vine water status measurements, collected during the notably dry period of July and August 2022. The study explored the effectiveness of the unsupervised k-means in isolating vine-specific information. It also assessed the capability of various multispectral and thermal vegetation indices in predicting vine water status. Indices related to leaf chlorophyll content, particularly those combining near-infrared and red edge bands, showed correlation with vine water status, suggesting their potential as reliable indicators. A predictive model built using stepwise linear regression was developed, accounting for 67% of the observed variance in Ψstem measurements. However, its applicability was limited to a single season, pointing to the need for recalibration or incorporation of multi-seasonal data for enhanced robustness. Finally, the analysis of the maps generated in this study underscored the extensive spatial and temporal variability of water status within individual vineyard parcels. At Vignoble du Château de Bousval, a general correlation was observed between spatial variability in water status and the thickness of the loam layer. In contrast, at Domaine W, the analysis indicated distinct water stress levels between Chardonnay and other grape varieties. This leads to uncertainties regarding the adaptability of the model to different grape varieties and raises questions about whether the observed differences in water status are due to inherent variations in the hydraulic behavior of these varieties.UAVHigh resolutionVineyardWater statusSpatial variabilityMultispectralThermaltext::thesis::master thesisthesis:44028