Satellites have been watching Europe for over twenty years. From that archive we rebuilt every individual fire: where it was, how long it lasted and how fiercely it burned. Some of what comes out is surprising, and on one point it runs straight against what most people assume.
The fire season is not everywhere in summer. Per region, fires larger than 100 hectares only, summed over all measured years.
The red bar marks the peak month. In Spain, Portugal and France there are more large fires in March than in August: that is spring burning of vegetation, a documented practice. The August fires are bigger and fiercer, though. In north-west Europe the peak falls in April, when last year\u2019s dead grass is bone dry and the trees have no leaves yet. In June, exactly when most people think of fire risk, the count is low.
| Size | Count | Share | Area |
|---|
More than half of all "fires" consist of a single satellite observation. We draw a heat zone of roughly 50 hectares around it; that is our own lower bound, not a measured perimeter. Usually it is agricultural burning or a brief roadside flare-up. At the same time a fraction of a percent accounts for nearly half the total. Counting fires measures something different from counting hectares.
| Nearest town | Year | ha | days | MW |
|---|
The hectares are the measured heat zone, not the burned area, so treat them as an upper bound. Fires that smouldered weakly for months are left out: those are almost always field fires that clump together in our measurement into one seemingly enormous blaze.
| Year | Count | > 100 ha | Trend | Area |
|---|
Since 2021 satellites detect about a third fewer fires in Europe than in the decade before. That is not a measurement error: two independent satellite systems with different resolutions show exactly the same pattern. But fewer fires is not the same as less risk. Counts say nothing about scale: according to EFFIS, 2022 was the second-worst fire season in the EU since records began in 2000, surpassed only by 2017, with fewer but far larger fires.
| Year | VIIRS | MODIS | Ratio |
|---|
The decline since 2021 is not a measurement error. VIIRS and MODIS are independent instruments with very different resolutions: with 375-metre pixels VIIRS sees more than twice as many fires as MODIS with its one-kilometre pixels. Yet their ratio has stayed stable for fifteen years, including the year of the break. If something were wrong with the measurements, that ratio is exactly what would have shifted.
| Country | Count | Trend | Peak month |
|---|
Counts say little about wildfire danger here. Countries differ greatly in how much agricultural burning takes place, and in these figures that is indistinguishable from wildfire. Russia and Ukraine top the list with a March and April peak: that is crop residue burning before sowing, and it accounts for nearly half of everything we measure in Europe.
The source is the public NASA FIRMS archive, in which satellites have recorded heat sources on the earth\u2019s surface since 2000. We regrouped those individual detections into separate fires, using the same method as the live map, so that history and present mean the same thing.
Everything NASA flags as industry, volcano or offshore is filtered out, as are known false heat sources from our own masks. Counting starts in 2012: before that only the older MODIS sensor was watching, with one-kilometre pixels, and it missed smaller fires. A continuous line from 2000 would suggest a rise in risk that comes purely from the camera.
Main limitation: a "fire" here is a group of detections close together in space and time. In areas where burning happens on a large and scattered scale, separate fires can merge into one large measured event. That is why the largest-fires list is filtered strictly, and why the country figures carry the caveats they do.
All figures on this page are calculated by us from raw satellite data; they are not official statistics. Where we refer to outside sources, they are named.