The study, led by the University of East Anglia, was published in Science. The researchers used machine learning to identify parts of the world - called pyromes - where climate factors have the strongest control on forest fires. In one of largest extratropical pyromes, which spans boreal forests in Eurasia and North America, emissions from forest fires nearly tripled between 2001 and 2023 amounting to an additional half a billion tonnes of CO2 per year.
This increase was linked to a rise in fire-favourable weather - periods of hot, dry and windy conditions -, drought, and increased vegetation productivity (more vegetation means more fuel for wildfires). These changes are linked to high rates of warming and more frequent heatwaves and droughts in northern boreal forests.
Driven by climate change
While the emissions from northern boreal forest fire increased, the emissions from tropical forest pyromes declined. This is related to reduced deforestation fires in moist tropical forests and increased fragmentation of dry tropical forests due to agriculture and other land uses. Overall, CO2 emissions for forest fires globally increased by 60% during the study period, with the most significant contributions coming from extratropical regions.
Veraverbeke: 'Using more than 2 decades of satellite data, we show that forest fires have increased in mid-latitudes and northern regions. This may seem intuitive to many, as fire records seem to be broken every summer, such as in Greece, California, Canada or Siberia in recent years. However, this is the first time that this increase has been demonstrated at a global scale. We also show that this increase is driven by climate change.'
Machine learning
Machine learning was key to unlocking new patterns in the shifting global geography of forest fires. The team used it to group global forest ecoregions into 12 distinct pyromes, where forest fire extent depends on similar sets of controls, such as climatic, human, and vegetation controls.
By grouping forest ecoregions in this way, the researchers aimed to better understand the regional variations in fire drivers and their sensitivity to climate change, which allowed them to isolate the effects of climate change from other influencing factors such as land use and vegetation productivity.
Increase in severity
Forests are of worldwide importance for carbon storage, with their growth helping to remove CO2 from the atmosphere and reduce rates of global warming. However, the study, which involved an international team from the UK, the Netherlands, the US, Brazil, and Spain, also reveals a worrying increase in the severity and spread of forest fires as the climate warms.
The carbon combustion rate, a measure of fire severity based on how much carbon is emitted per unit of area burned, increased by almost 50% across forests globally. The researchers warn that further expansion of forest fires can only be averted if the primary causes of climate change are tackled.
'Forests are sensitive to this increase in fires,' says Veraverbeke. 'It often takes decades for a forest to recover after a fire, and sometimes that recovery does not occur. If there is no recovery and forest fires occur more often, the greenhouse gases released during those fires can further exacerbate climate change.'
The contribution of Sander Veraverbeke was funded by the European Research Council through a Consolidator grant under the European Union’s Horizon 2020 research and innovation program.
