UAV-BASED MULTISENSOR DATA FUSION FOR REAL-TIME MONITORING OF DURUM WHEAT WATER STRESS IN SEMI-ARID MEDITERRANEAN ENVIRONMENTS

This study presents an integrated UAV-based precision agriculture system combining multispectral and thermal remote sensing 
with ground-based physiological measurements to monitor durum wheat responses to water stress in semi-arid Mediterranean environments. Three durum wheat varieties were cultivated under four irrigation regimes (rainfed, 100%, 75%, and 50% evapotranspiration) during 20232024 in Bordj Bou Arreridj, Algeria. A DJI Matrice 300 RTK equipped with multispectral, thermal, and RGB cameras acquired weekly data. Key findings: (1) NDVI strongly correlated with above-ground biomass (r² = 0.84, p < 0.001); (2) Crop Water Stress Index (CWSI) from thermal imagery exhibited robust correlation with soil moisture (r² = 0.82, p < 0.001); (3) UAV monitoring provided 7–10 days advance warning compared to visual assessment; (4) Novel red-edge spectral-enzymatic correlations enabled non-invasive stress phenotyping; (5) Machine learning models predicted grain yield with 85% accuracy at four weeks pre-harvest; (6) Spatial management zones identified 15–25% water savings potential while maintaining 92–95% yield.