Multispectral drone imaging for field-scale yield variability mapping in precision agriculture

Field-scale yield variability is a defining feature of modern crop production, yet many fertilization and irrigation decisions are still made uniformly, ignoring within-field heterogeneity. This study evaluates multispectral drone imaging as a practical tool for mapping yield variability at field scale and for supporting site-specific management decisions. Highresolution multispectral imagery was acquired at key growth stages over commercial cereal fields, processed into calibrated reflectance and vegetation index maps, and linked to georeferenced yield data from combine harvesters. A set of empirical and machine-learning models was developed to relate multispectral metrics to final yield and to delineate stable management zones. Model performance, spatial resolution, positional accuracy and operational requirements were compared with those of satellite-based approaches and traditional scouting. Results show that multispectral drone imaging captures fine-scale yield patterns with sufficient accuracy to inform variable-rate strategies, while remaining compatible with typical farm logistics and digital agriculture platforms.