What happened
In August 2023, a wildfire near Alexandroupolis in the Evros region of north-eastern Greece burned for more than a week. On 29 August, the EU's civil protection service said it was the largest wildfire in the EU since 2000, when the European Forest Fire Information System (EFFIS) began recording data, and put the burned area at over 810 square kilometres, as reported by Al Jazeera.
The European Commission's Joint Research Centre summarises each fire season in its annual report. Its report on the 2023 season combines official figures from contributing countries with the fire-danger analysis and burned-area mapping from EFFIS. It is the reference to use for official numbers.
Why fires like this are hard to plan for
A large wildfire is not one problem. It is several connected ones:
- Weather and fuel: heat, dryness and wind interacting with vegetation and terrain.
- People: villages, farms, tourists and seasonal workers, all in different places at different times.
- Infrastructure: the roads used for evacuation, power lines, water supply and hospitals.
- Time: a fire's front can move faster than plans written for "average" conditions.
Each of these is usually managed in a separate system by a separate organisation. A digital twin's job is to bring them together in one model.
What a twin could add
We want to be careful here: we are not claiming that any tool would have prevented a specific fire. What a regional fire twin could do is help planners ask better questions before the season:
- Exposure maps: which settlements and roads sit in the most likely fire paths under typical summer winds?
- Evacuation stress tests: if a given road is cut, how long would it take to move people out by the remaining routes?
- Fuel management priorities: where would clearing vegetation reduce risk most for the most people?
- After-action learning: replay past fires in the model, compare with what happened, and improve both the model and the plans.
The data already exists
Much of what such a twin needs is already public: EFFIS fire-danger forecasts and burned-area maps, Copernicus satellite imagery, land-cover data, road networks and census statistics. The missing piece is joining them around the specific questions of a region or municipality. That is the kind of pilot Kassandra Labs is looking for.