A Geospatial Model for Identifying High-Risk Locations for Downy Mildew (Plasmopara viticola) Infestation in Vineyards of Greece

Downy mildew (Plasmopara viticola) poses a major and recurring threat to Greek viticulture, yet existing point-based forecasting models require in-vineyard stations, limiting scalability in fragmented landscapes. This study introduces a spatially explicit model (MeteoGrape) using one fully equipped reference meteorological station plus eight distributed sensors across an 85 km2 area in Kavala, Greece. The model is structured in three phases. In Phase A, a single reference station was paired with eight low-cost distributed sensors to reconstruct hourly temperature and relative humidity data through regression correction and radial basis function interpolation, generating a 342-cell grid at 0.005° resolution. During Phases B and C, deterministic epidemiological rules were applied to simulate oospore development, with accumulated degree-hours and humidity exposure converted into spatial risk classifications. Cross-validation (leave-one-sensor-out) confirms meteorological reliability. The model captured an elevated risk period beginning on 16 May, preceding the regional advisory bulletin (23 May), and mapped the spatial distribution of accumulated risk through late May. Validation supports temporal consistency at the regional scale, while fine-scale spatial accuracy is identified as a subject for future field-based evaluation. The framework demonstrates the feasibility of extending established point-based disease models into spatially explicit risk maps under limited meteorological infrastructure.