4) Dr. Tarin Paz-Kagan

Effective monitoring of plant water status (PWS) is critical for optimizing irrigation and sustaining productivity in commercial orchards. Stem water potential (SWP) and trunk growth rate (TGR) are reliable indicators of water stress; however, labor-intensive measurement methods limit their application. This study develops and validates a scalable framework to model seasonal dynamics of SWP and TGR and predict yield in almond orchards by integrating UAV-based multi-sensor imagery with meteorological and irrigation data. Over two growing seasons (2022–2023), multispectral, thermal, and LiDAR data were collected for 60 trees, alongside physiological and environmental measurements. Random Forest models achieved high predictive accuracy, including robust performance in the independent test set (Plot A: R² = 0.79 for SWP, 0.78 for TGR, and 0.87 for yield), demonstrating strong generalizability across orchard plots. Canopy temperature, red-edge vegetation indices, and the crop water stress index were key predictors of SWP, while irrigation volume and chlorophyll-related indices influenced TGR. Yield prediction was strongly driven by early-season SWP, TGR, and canopy nitrogen content. Fuzzy clustering of SWP and TGR seasonal curves revealed distinct stress-response groups, with membership in the most stressed cluster associated with a yield reduction of up to 76%. These findings highlight the potential of combining UAV imagery, physiological indicators, and machine learning to deliver spatially explicit, tree-level predictions of water stress and yield. The framework provides actionable insights for precision irrigation and resource management in perennial orchard systems.