Search for a command to run...
Abstract Timely plant disease forecasts are essential for effective fungicide use, yet standard weather-driven models treat crop canopies as spatially uniform and overlook within-field variability that can be captured by remote sensing. Using the sugar beet–Cercospora leaf spot (CLS) pathosystem, we test whether linking remotely sensed leaf area index (LAI) can be used to initialise a CLS negative‑prognosis model in a spatially explicit manner and thereby improve disease monitoring and warning. LAI was mapped from multispectral satellite (10 m), super-resolved satellite (1 m) and airborne (1 m) imagery on a regular grid across a field, converted to canopy-closure dates for each grid cell, and used to initialise a classic CLS negative-prognosis model. Unlike the classic CLS model, which produces a single, spatially averaged epidemic onset date for the area covered by the driving weather station, our approach produces a distribution of earliest estimated epidemic onset dates (EEEOs) across grid cells, reflecting spatial heterogeneity in canopy development and epidemic risk. EEEO was evaluated against ground-derived epidemic onset (EO) dates and represents the earliest date on which a CLS epidemic could occur once negative prognosis no longer applies. In non-inoculated areas, EO occurred 156.9 days after sowing (DAS), whereas EEEO occurred substantially earlier: 117.4 DAS for satellite, 114.0 DAS for super-resolved satellite, and 103.8 DAS for airborne imagery, with similar patterns in inoculated plots. Spatially explicit EEEO were earlier than those from the conventional field-average CLS model, indicating that accounting for within-field canopy heterogeneity can accelerate and localise disease warnings. This proof-of-concept framework captures sub-field variability missed by regional-scale prognosis and can guide more targeted scouting and fungicide use.