Scientists test wildfire prediction model during High Fens blaze

Updated 18 August 2026, 18:00 CEST — Scientists tested a wildfire-prediction model on Tuesday, 18 August, using the continuing fire in the High Fens in eastern Belgium as a real-world case, and the first comparison between its projections and the fire’s observed spread produced promising results, VRT NWS reported. The researchers have not yet published a full validation or established that the system is ready for operational firefighting.

The test seeks to determine whether computer modelling can give emergency commanders an earlier indication of the direction and pace of a vegetation fire. Such tools combine information about weather, terrain and combustible material to simulate how flames spread. Their value rests on speed and local accuracy: a forecast that arrives early enough can help services position crews, protect access routes and identify exposed areas.

The High Fens present an unusually demanding test. The protected plateau contains heath and peatland, where fire behaviour differs from that of conventional woodland. KU Leuven research led by Jonas Mortelmans found that hydrology plays a particularly important role in peat fires and that systems designed for ordinary forests do not fully represent peatland conditions. The research produced a peat-specific adaptation of the Canadian Fire Weather Index, although its improvement over the standard index was described as modest.

That distinction matters because a surface perimeter does not always reveal what is happening below ground. Dry peat stores carbon-rich organic matter and can continue smouldering after visible flames subside. Researchers therefore need to compare the model not only with the mapped burn area, but also with field observations, moisture conditions and any underground combustion.

The present fire follows repeated official warnings about the plateau’s vulnerability. Wallonia’s environment administration restricted access during elevated fire danger in April 2026, explaining that dry vegetation and adverse weather allow fires to ignite and spread rapidly. A Walloon government warning issued in 2025 also recalled that the April 2011 High Fens fire burned about 1,200 hectares, then the largest recorded in Wallonia.

The immediate result is encouraging but preliminary. Researchers must now disclose the model’s inputs, forecast horizon and error margins, and test whether it performs consistently as wind and moisture change. Emergency services will determine separately whether its output is reliable and fast enough to support decisions during future fires.