Description
Idea
Food crises start in the field, but authorities usually detect them only after harvest, when the product is already in the food chain. In 2012, a severe drought in Romania (Romania risk) and the Balkans favoured aflatoxin-producing fungi in maize. The contamination was found months later in feed and milk, triggering 17 EU RASFF alerts, product recalls and farmer protests in 2013. The warning signs were visible from space during the growing season (aflatoxin incident in Maize in Europe).
Today the information needed to act early exists, but it is fragmented: satellite imagery, weather data, RASFF notifications, veterinary bulletins and news, in three languages, on both sides of the EU’s eastern border. Inspectors and crisis managers have neither the time nor the tools to combine them.
Solution
FloRisk is a space-enabled early warning system for food supply resilience in the Lower Danube – Black Sea region (Romania and southern Ukraine).
- In-season crop phenology per field. Sentinel time series detect, as the season unfolds, when each wheat, maize or sunflower field enters its critical window (flowering and grain filling).
- Stress in the critical window. Each crop has its own alarm condition: heat and drought for maize and sunflower, warm and wet weather for wheat.
- Two risk levels. Production risk (delayed, stressed or failed crops) and quality risk (aflatoxins in maize, Fusarium/DON in wheat), with explainable scores and uncertainty.
- From score to action. A prioritised sampling list for food safety inspectors, field checks for paying agencies and insurers, and situation reports generated automatically in Romanian, Ukrainian and English.
What brings this idea?
Existing tools are regional (50 km grids), retrospective (annual phenology products) or focused on other regions. FloRisk combines in-season, crop-specific phenology at 10 m with weather stress and turns it into field-level decisions. One engine, different rules per crop: the same year can trigger a wheat alert in May and a maize alert in August.
From data to decisions - Making space data speak
Machine learning fills the gaps that make satellite monitoring fail in a crisis: it uses radar to see crops under summer clouds, maps crops where no official parcel data exists (as in southern Ukraine) and predicts when each field will flower. Language models connect these signals to RASFF alerts and agricultural news, answer questions in plain language and write situation reports in Romanian, Ukrainian and English. Every figure comes from the models, not the language model, and every decision stays with a human.
Who we serve?
Public authorities responsible for civil protection, agriculture and food safety, who need to act before a crisis rather than after it. Insurers and paying agencies, who need faster, evidence-based assessment of crop losses. Grain traders, feed producers and port operators, who need to know the risk behind every shipment. At EU level, FloRisk supports the shared effort to keep Europe’s food supply secure and its eastern border resilient.
Pilot area
The Galați – Reni – Izmail triangle on the Danube (Romania, Ukraine and Moldova): farmland on both banks, the same weather shocks and the grain ports through which Ukrainian exports reach the EU.
Proposed EU space technologies
| Source | What we use | Value for FloRisk |
|---|---|---|
| Copernicus Sentinel-2 | NDVI/EVI/NDRE time series, 10 m, every 5 days | Core signal for crop phenology and stress at field level |
| Copernicus Sentinel-1 | SAR backscatter (VV/VH) | Works through clouds, fills gaps in July–August, the critical window for maize and sunflower |
| Copernicus Sentinel-3 | Land surface temperature | Detects heat stress during flowering |
| Copernicus Land Monitoring Service – HR-VPP | Vegetation phenology 2017 onwards, 10 m | Multi-year baseline of the “normal season” for each field |
| Copernicus Climate Change Service – ERA5 | Air temperature, precipitation, history | Weather stress in the critical window, climate baseline |
| Copernicus Data Space Ecosystem + openEO | Cloud processing | Time series per field computed server-side, scalable to the whole region |
| ESA APEx algorithms / ESA WorldCereal | Ready-made crop and land-cover maps | Crop masks, also where no public parcel data exists |
| Galileo | Precise positioning | Field app for inspectors and sampling teams to verify alerts on the ground |
Copernicus is free, open and covers both sides of the border with the same quality. Ground data does not stop at the Danube, and neither does the risk. Combining optical, radar and thermal data makes the system reliable in the cloudy summer weeks when decisions matter most.
Challenge to solve: Space-enabled crisis response and civil resilience
FloRisk addresses Challenge 2: Space-enabled crisis response and civil resilience, by treating food supply as critical infrastructure.
Early warning instead of late response. Authorities get weeks or months to act: targeted sampling, faster harvest and drying, market and stock planning, before contaminated or missing food becomes a public crisis.
Cross-border resilience at the EU’s eastern border. Romania has helped export almost 29 million tonnes of Ukrainian grain through Constanța since 2022 [Romania help ...]. FloRisk maps risk on both sides of the Danube, so inspectors and traders know the risk level of the origin region of grain in transit.
Shared situational awareness. Multilingual reports (RO/UA/EN) give Romanian, Ukrainian and EU actors the same picture at the same time.
Water dimension
Drought is the main driver of the risks FloRisk monitors. The system tracks water stress in crops (weather plus satellite signals) and can extend to Danube water levels and irrigation canals with Sentinel-1/2. The same drought that damages harvests also lowers the river used to ship grain and reduces water for irrigation. This links food security directly to the protection and management of water resources in the region.