Frost event detection and impact modelling within the Destination Earth On-Demand Extremes framework

Spring-time frost events pose a significant risk for the agricultural sector. Vineyards, apricots and other crops might be damaged by freezing temperatures after vegetation season onset. Frost events occur commonly during cold outbreaks in April or May after prolonged periods of relatively high air temperatures in late-winter or early-spring. The damage can be significantly reduced by suitable measures, provided a reliable forecast is available. As frost intensity is often highly spatially variable, its forecast can benefit from hectometric-scale numerical atmospheric modelling. We present results related to frost events detection and impact modelling within the Destination Earth On-Demand Extremes (DEODE) project. The detection scheme uses the ECMWF ensemble forecasts of 2-m air temperature and cloud cover to identify regions of potential high-impact frost events. The Global Digital Twin 2-m air temperature sums since the start of the year are used to delimitate areas where vegetation is not active yet. These areas are masked from the risk assessment. Finally, the frost damage risk is expressed in 0–5 scale and may serve as guidance for hectometric-scale simulation domains. After a hectometric simulation is completed, the output is ingested into a downstream frost impact model. The impact model estimates current development phase of several crops and corresponding temperature thresholds for frost damage. These thresholds are compared with hectometric-scale forecasted temperatures to obtain the magnitude of excess of critical temperature for each crop. The impact model has been evaluated during several pilot frost events in the Czech Republic and Spain.