How the index is built
The fire weather index originates with the Canadian forest service and is today the system in common use across Europe; EFFIS and GWIS compute it for the whole continent. It is not a single measurement but a chain of six quantities.
Three moisture codes describe how dry three different fuel layers are. FFMC stands for the fine litter at the surface, which responds to weather within hours. DMC stands for the loose organic layer below it, with a memory of days. DC stands for the deep, compacted organic soil, which needs weeks to become moist again, it is the reason a heatwave keeps having effects long after the first rain.
FFMC and wind produce the Initial Spread Index (ISI): how fast a newly ignited fire spreads. DMC and DC produce the Buildup Index (BUI): how much fuel is available at all. ISI and BUI together give the FWI, the intensity to be expected from a fire under the conditions of the day.
From number to danger class
The FWI is a dimensionless number. The six classes samoEKO draws follow the thresholds of the EFFIS legend service; the names come from samoEKO, because a situation picture has to name a class that a fire station understands.
Because that constitutes a modification in the sense of the licence, the attribution carries the addition “modified”.
| Class | Threshold |
|---|---|
| low | < 11.2 |
| moderate | 11.2-21.3 |
| high | 21.3-38.0 |
| very high | 38.0-50.0 |
| extreme | 50.0-70.0 |
| very extreme | > 70.0 |
The thresholds are taken unchanged from the EFFIS legend service, the names are from samoEKO. There is no ready-made class layer, the classification is done inside samoEKO.
Two computations, two answers
EFFIS and GWIS deliver the same index but from different weather models: EFFIS from the Meteo-France model at 0.1 degree grid width and three forecast days, GWIS from the ECMWF model at 0.0714 degrees and nine days.
The results diverge markedly. On 2 August 2025, at 17.5 to 18.0 degrees east and 43.5 to 44.0 degrees north, an EFFIS value of 34.1 stood against a GWIS value of 12.6, one computation reports “moderate”, the other “low”. That is not an error but the difference between two weather forecasts.
For the situation picture one hard rule follows: the two sources are never mixed in one view, and which one is shown is written on the display.
What the index cannot do
The FWI describes weather and fuel moisture, not ignition. Whether somebody sets a fire, whether lightning strikes, whether roadside mowing throws a spark, of that it knows nothing. An extreme index without an ignition source remains without consequence; a moderate one with an ignition source can become a disaster.
Two technical limits come on top. From the fourth forecast day onwards EFFIS returns a zero image instead of an error: anyone checking only the status code writes “index 0, no danger” into the situation picture every day, samoEKO therefore tests for entirely empty rasters. And in the delivered raster the value 0 is ambiguous, because sea and a genuine index of 0 cannot be distinguished without an accompanying land mask; that mask is still missing.
Finally, the computed index lags the weather: after rain it stays high for days although the real danger has dropped immediately. That is why the precipitation radar sits beside it.