Credit: Getty Images.
The second annual State of Wildfires report published today (16 October) in Earth System Science Data by an international research project, highlights how climate change is fuelling an increase in the frequency and severity of extreme wildfires, escalating risks to people and nature.
Co-led by ECMWF, the UK Centre for Ecology & Hydrology (UKCEH), the UK Met Office, and the University of East Anglia (UEA), the project brings together expertise from 60 institutions across 20 countries globally to track, understand and predict global wildfire activity.
Covering the period March 2024 to February 2025, the report documents when and where extreme wildfire events took place, assesses their drivers and predictability, and analyses the potential influence of climate change and land use.
Key findings from this year’s report
The global analysis highlights that the fire season was marked by a paradox: less land burned, but more carbon was released than in most years on record.
Between March 2024 and February 2025, fires scorched about 3.7 million square kilometres worldwide – around nine percent below the long-term average – yet emitted over 8 billion tonnes of carbon dioxide into the atmosphere, around ten percent above the average since 2003. The reason lies in where the fires burned: dense, carbon-rich forests across the Americas and other high-biomass regions. These intense blazes pushed global fire emissions higher than the annual fossil fuel output of more than 200 countries. While wildfires flared across continents, it was the Americas that dominated the season, with a string of exceptional and destructive events driving much of the global anomaly.
This year’s report also documents that around 100 million people were exposed to wildfires globally, with hotspots in India, the Democratic Republic of the Congo, and China. Despite this, formal displacement was low (~20,000 people), highlighting a gap between exposure and documented disruption.
Wildfires also caused 215 billion USD losses in built assets, with India, the U.S., China, and South Africa most affected. Recorded direct losses totalled 57 billion USD, dominated by fires in California, where fires occurred in very high-value areas.
Looking into specific events, the international team of researchers evaluated the causes and predictability of four major wildfire episodes and the likelihood that they were driven by climate change. These events occurred in key regions of global importance during the 2024–25 season:
- Northeast Amazonia (Jan–Mar 2024)
- Pantanal–Chiquitano (Jul–Sep 2024)
- Southern California (Jan 2025)
- Congo Basin (Jul–Aug 2024)
Prolonged drought was the dominant fire enabler in the three tropical regions, while in California, a combination of extreme heat, strong winds, and dense, dry vegetation were dominant enablers.
ECMWF’s contribution
Climate change is driving increasingly frequent extreme fires, with major consequences for ecosystems and society. The State of Wildfires Project goes beyond documenting this shift. It uncovers mechanisms, highlights future risks, and develops tools that will protect us.
“We must move from observing what is happening to anticipating what comes next and how it will impact our society,” said Francesca Di Giuseppe, State of Wildfires Project co-lead and Principal Scientist at ECMWF.
ECMWF led the analysis on the drivers and predictability of extreme wildfires and generated air-quality datasets used by collaborators to quantify the health impacts from smoke and particulate pollution.
Since 2018, ECMWF has delivered fire danger forecasts through the Copernicus Emergency Management Service (CEMS). Building on this, the Centre has in recent years pioneered novel approaches to fire prediction, using artificial intelligence. One such innovation utilised in the report is the Probability of Fire (PoF) model, an output of the ECMWF SPARKY fire model. This machine learning system merges weather forecasts with information on vegetation dryness and flammability, lightning strikes, population density, and road networks to predict how many fire hotspots may ignite in a given area.
As Joe McNorton, scientist at ECMWF and contributor to the report, explains: “A step-change in our ability to predict wildfires is underway, driven by advances in tracking how living and dead vegetation fuels evolve on the landscape. We can now anticipate major fire events with far greater accuracy than ever before – in some cases, even a month in advance.”
Using this model, the team identified the respective roles of weather, vegetation density, and ignition sources in determining the most extreme events.
In all four regions, exceptionally fire-prone weather was the primary driver of extreme outcomes, with vegetation fuel conditions set by prior seasons playing a crucial role. The dryness of vegetation also played a critical role during the extreme wildfires in Amazonia and the Congo Basin, where abnormally dry forests and wetlands allowed fires to spread faster and further.
Air quality impact assessment
To understand how smoke from wildfires affected air quality and the impact on human health, ECMWF scientist Enza Di Tomaso worked with collaborators Matt Jones (University of East Anglia (UEA), UK), Guilherme Mataveli (National Institute for Space Research (INPE), Brazil) and Douglas Hamilton (North Carolina State University, USA) on an air-quality impact assessment.
Data was used from the Global Fire Assimilation System (GFAS) within the EU Copernicus Atmosphere Monitoring Service (CAMS; implemented by ECMWF). GFAS combines satellite observations of fire intensity with information on vegetation type to estimate how much smoke and pollution each fire releases.
These emissions were then run through the ECMWF Integrated Forecasting System extended with modules of atmospheric composition (IFS-COMPO). This system simulates chemical reactions, source and sink processes for the main reactive trace gases and aerosol species, and models how weather systems transport those pollutants through the atmosphere, providing a detailed picture of air quality across the globe.
To identify the specific impact of the wildfires, two sets of model simulations were run: one including emissions from the “local fires”, and one without them. The difference between the two revealed how much pollution could be directly attributed to the fires themselves. The analysis focused on fine particulate matter (PM2.5) – microscopic particles small enough to penetrate the lungs and potentially enter the bloodstream, posing a major health risk.
The outputs were then combined with population data to estimate exposure and assess the number of days on which people were exposed to air pollution. In the Pantanal–Chiquitano, concentrations of PM2.5 were found to exceed the World Health Organization’s (WHO) daily standard safety limit of 15 µg/m³ on most days between July and October 2024, with local fires responsible for an additional 16 days of poor air quality.
Enza Di Tomaso, ECMWF scientist and contributor to the report, explained: “By comparing model runs with and without local fire emissions, we can see how fires degrade air quality. A collaborative approach allows us to quantify the contribution of wildfires to pollution exposure. At the peak of the fire season in the Pantanal-Chiquitano in September, wildfires accounted for almost 60 per cent of all airborne pollution experienced by the average person in the region.”
Future fire seasons
Events occurring since March 2025 will be the focus of next year’s State of Wildfires report, but several extreme wildfire events occurring during the northern hemisphere spring and summer have already caused major and widely reported disruption, and the scientific community has already begun to disentangle the causes.
Record-breaking fires in South Korea, Türkiye, Spain, and Portugal collectively burned more than 700,000 hectares and displaced tens of thousands of residents. Analyses by the World Weather Attribution group found that the hot, dry, and windy conditions driving these events were strongly influenced by climate change, with fire weather in South Korea twice as likely, in Türkiye up to ten times more likely, and in Spain and Portugal up to forty times more likely than in a pre-industrial climate.
In the future, extreme fires will become more frequent as the climate continues to warm. “ECMWF’s expertise in weather prediction, air-quality monitoring, and fire forecasting will be vital in helping societies anticipate these events and reduce their devastating impacts.” Francesca Di Giuseppe concludes.
State of Wildfires Project
The State of Wildfires Project was established to systematically track and analyse global wildfire activity.
It brings together satellite observations and expert knowledge to track extreme wildfires, understand their causes and assess predictability under future climate scenarios. Visit the State of Wildfires project website to find out more.