FeuxFrance
Is it getting worse?
The question deserves an answer with figures, and with its limits. Here is the longest series available, year by year, and the calendar of its seasons.
Sensor
Terra and Aqua. The series starts in 2003: Aqua was only launched in May 2002, and the earlier years have one daily overpass instead of two.
The two sensors cannot be compared with each other: MODIS sees at 1 km, VIIRS at 375 m, so VIIRS detects more small fires. Each series is consistent within itself, and that is all we ask of it.
2003β2026, year by year
2026 is still running: its bar is not comparable to the others.
| Year | Points chauds |
|---|---|
| 2003 | 3,519 |
| 2004 | 482 |
| 2005 | 557 |
| 2006 | 1,079 |
| 2007 | 998 |
| 2008 | 1,234 |
| 2009 | 536 |
| 2010 | 889 |
| 2011 | 799 |
| 2012 | 1,503 |
| 2013 | 776 |
| 2014 | 1,166 |
| 2015 | 1,218 |
| 2016 | 1,202 |
| 2017 | 1,839 |
| 2018 | 682 |
| 2019 | 2,056 |
| 2020 | 1,457 |
| 2021 | 1,404 |
| 2022 | 2,819 |
| 2023 | 1,418 |
| 2024 | 972 |
| 2025 | 1,565 |
| 2026 (en cours) | 2,199 |
2,199 hotspots as of July 25, 2026, against a full year for the other bars. The comparison to date is on the home page.
Is the season getting longer?
A year can count as many fires as another while spreading them very differently. This curve shows the pace of accumulation: if the recent lines rise earlier, the season starts earlier.
Cumulative hotspots day by day, one line per year. MODIS sensor, the only one comparable across the whole period.
The 2026 curve stops on July 25, 2026: that is the last day loaded for this sensor, not the end of the season.
| Year | Total |
|---|---|
| 2003 | 3,519 |
| 2004 | 482 |
| 2005 | 557 |
| 2006 | 1,079 |
| 2007 | 998 |
| 2008 | 1,234 |
| 2009 | 536 |
| 2010 | 889 |
| 2011 | 799 |
| 2012 | 1,503 |
| 2013 | 776 |
| 2014 | 1,166 |
| 2015 | 1,218 |
| 2016 | 1,202 |
| 2017 | 1,839 |
| 2018 | 682 |
| 2019 | 2,056 |
| 2020 | 1,457 |
| 2021 | 1,404 |
| 2022 | 2,819 |
| 2023 | 1,418 |
| 2024 | 972 |
| 2025 | 1,565 |
| 2026 | 2,199 |
Over the period, mid-season arrived 8 days earlier on average between the first and second half of the years observed. This is an average over few years: a single unusual season shifts it.
What that means, year by year
Hectares measured by EFFIS, COβ modelled by Copernicus from the satellite detections. 2026 is still running. See the formulas
| Year | Hectares burned | COβ (kt) |
|---|---|---|
| 2026 as of July 25 | 96,926 | 8,398 |
| 2025 | 36,951 | 1,733 |
| 2024 | 12,268 | 570 |
| 2023 | 22,350 | 934 |
| 2022 | 66,392 | 4,830 |
| 2021 | 30,652 | 1,516 |
| 2020 | 14,547 | 946 |
| 2019 | 43,602 | 2,018 |
Where these figures come from
- NASA FIRMS Β· the hotspots, VIIRS and MODIS, in near real time and in archive.
- EFFIS, Copernicus Β· the measured burned areas and the emissions.
- BDIFF Β· the official French database, published a year behind.