Joint prediction of potato yield and nitrogen status at row closure using multi-task deep learning and UAV-derived LiDAR–multispectral data

• Fused UAV multispectral and LiDAR data quantified potato canopy structure and petiole nitrate-N at row closure. • Joint modelling of yield and petiole nitrate-N improved predictive performance compared with single-task approaches. • LiDAR height and structural metrics added complementary information to multispectral predictors. • Single-date UAV data at the row-closure stage were sufficient for nitrogen prediction but less informative for yield. • Feature attribution analyses indicated physiologically consistent relationships between canopy structure and nitrogen status.