The latest update to the database \u2013 explored in new research published in journal Scientific Data<\/a> \u2013 includes data up to and including the year 2024.<\/p>\n It reveals that, once the data from smaller fires is included, fire emissions sit at roughly 3.4bn tonnes of carbon (GtC) annually \u2013 significantly higher than previous estimates.<\/p>\n It also shows that carbon emissions from fires have remained stable over the past two to three decades, as rising emissions from forest fires have been offset by a decline in grassland fire emissions.\u00a0<\/p>\n The database update also illustrates how the amount of area burned around the world each year is falling as expanding agriculture has created a fragmented landscape and new restrictions on crop residue burning have come into force.\u00a0<\/p>\n Landscape fires<\/p>\n Fire events vary widely in cause, size and intensity. They take place across the globe in many types of landscapes \u2013 deserts and ice sheets are the only biomes that are immune to fire.<\/p>\n When vegetation burns, it releases greenhouse gas emissions, which contribute to global warming. It also releases pollutants that cause local air pollution and, on a global scale, have a cooling effect on the climate.<\/p>\n Forest fires often generate considerable media attention<\/a>, especially when they threaten places where people live.\u00a0<\/p>\n However, the forest fires that make the news represent just a small fraction of all fires globally.\u00a0<\/p>\n More than 95% of the world\u2019s burned area occurs in landscapes with few trees, such as savannahs and grasslands.\u00a0<\/p>\n Fires have helped maintain tropical savannah ecosystems for millions of years. Savannahs have the perfect conditions for fire: a wet season which allows grasses and other \u201cfuels\u201d to grow, followed by an extended dry season where these fuels become flammable.\u00a0<\/p>\n Historically, these fires were ignited by lightning. Today, they are mostly caused<\/a> \u2013 intentionally or accidentally \u2013 by humans.<\/p>\n And yet, despite their prevalence, these fires receive relatively little media attention. This is not surprising, as they have been part of the landscape for so long and rarely threaten humans, except for their impact on air quality.<\/p>\n Fires also occur in croplands. For example, farmers may use fire to clear agricultural residues after harvest, or during deforestation to clear land for cultivation.\u00a0<\/p>\n The term \u201clandscape fires\u201d is increasingly used to describe all fires that burn on land \u2013 both planned and unplanned.\u00a0<\/p>\n (The term \u201cwildfire\u201d, on the other hand, covers a subset of landscape fires which are unplanned and typically burn in underdeveloped and underinhabited land.)\u00a0<\/p>\n Calculating the carbon emissions of landscape fires is important to better understand their impact on local air quality and the global climate.\u00a0\u00a0<\/p>\n New data<\/p>\n In principle, calculating carbon emissions from fires is straightforward. The amount of vegetation consumed by fire \u2013 or \u201cfuel consumption\u201d \u2013 in one representative \u201cunit\u201d of burned area has to be multiplied by the total area burned.\u00a0<\/p>\n Fuel consumption can be determined through field measurements and satellite analysis.<\/p>\n For example, the burned area of a relatively small fire can be measured by walking around the perimeter with a GPS device. Fuel consumption, meanwhile, can be derived by measuring the difference in amount of vegetation before and after a fire, something that is usually only feasible with planned fires.\u00a0<\/p>\n In practice, however, fires are unpredictable and highly variable, making accurate measurement difficult.<\/p>\n To track where and when fires occur, researchers rely on satellite observations.\u00a0<\/p>\n For two decades, NASA<\/a>\u2019s MODIS satellite sensors<\/a> have provided a continuous, global record of fire activity. To avoid too many false alarms, the algorithms these satellites use are built in a way so fires are flagged only when they burn an entire 500-metre grid cell.\u00a0<\/p>\n However, this approach misses many smaller fires \u2013 resulting in conservative estimates of total burned area.\u00a0<\/p>\n The latest update to the GFED includes, for the first time, finer-resolution satellite data, including from the European Space Agency<\/a>\u2019s \u201csentinel missions<\/a>\u201d.<\/p>\n This data shows that fires too small to be picked up by a satellite with a 500-metre spatial resolution are extremely common. So common, in fact, that they nearly double previous estimates<\/a> of global burned area.\u00a0<\/p>\n The data shows that, on average, 800 hectares of land \u2013 an area roughly the size of Australia \u2013 has burned annually over the past two decades.<\/p>\n The map below shows the frequency of fires around the world. Regions shaded in dark red burn, on average, 50-100% each year. In other words, fires occur annually or biannually.\u00a0 Regions in dark blue, on the other hand, are those where fires occur, but are very infrequent. Most regions fall in between these extremes.<\/p>\n The map shows that the areas most prone to fire are largely found in the world\u2019s savannah and agricultural regions.\u00a0\u00a0<\/p>\n Falling burned area<\/p>\n Over recent decades, the total burned area globally each year has been declining<\/a>.\u00a0<\/p>\n This is largely due to land-use change in regions which used to have frequent fires.<\/p>\n For example, savannah is being converted to croplands in Africa. This transforms a frequently burning land-use type to one that does not burn \u2013 and creates a more fragmented landscape with new firebreaks which limit the spread of fire.<\/p>\n The decline in burned area is also due to the introduction of more stringent air quality regulations limiting crop residue burning in much of the world, including the European Union.<\/p>\n The amount of \u201cfuel\u201d \u2013 or biomass \u2013 in a unit area of land varies greatly. Arid grasslands are biomass-poor and, therefore, produce less carbon emissions when burned, whereas fuel consumption in tropical forests with peat soils is extremely high.<\/p>\n Maps of carbon emissions from fires closely resemble maps of burned area. However, they typically highlight biomass-rich areas, such as dense forests.<\/p>\n This is illustrated in the map below, which shows how fires in regions coloured dark red on the map produce, on average, 1,000-5,000 grams of carbon per square metre. In these places, much more carbon is lost during fires than gained through photosynthesis.\u00a0<\/p>\n Meanwhile, much of the world\u2019s savannah regions are coloured in yellow and orange on the map, indicating that fires here produce between 100-500 grams of carbon per square metre.<\/p>\n Rising forest fire carbon emissions<\/p>\n The boost in fire emissions captured by the latest version of the GFED is most pronounced in open landscapes, including savannahs, grasslands and shrublands.<\/p>\n Forest fire emissions, on the other hand, have barely changed in the updated version of the database. This is because most forest fires are relatively large and were already well captured by the coarse resolution satellite data used previously.\u00a0<\/p>\n However, the trend in forest fire emissions is sloping upwards over the study period.<\/p>\n Overall, current estimates \u2013 which take into account the new data from smaller fires \u2013 suggest that, over 2002-22, global fire emissions averaged 3.4GtC per year.\u00a0<\/p>\n This is roughly 65% higher than estimates set out in the previous update<\/a> to the GFED, which was published in 2017.\u00a0<\/p>\n For comparison, today\u2019s fossil fuel emissions are around 10GtC per year.\u00a0<\/p>\n Comparisons between fire and fossil fuel carbon emissions are somewhat flawed, as much of the carbon released by fires is eventually reabsorbed when vegetation regrows.\u00a0<\/p>\n However, this is not the case for fires linked to deforestation or the burning of tropical peatlands, where regrowth is either much slower \u2013 or non-existent, if forests are converted to agriculture. These fires account for roughly 0.4GtC each year \u2013 just less than 12% of total fire emissions \u2013 and contribute directly to the long-term rise in atmospheric carbon dioxide (CO2).<\/p>\n The traditional view of forest fires as \u201ccarbon-neutral\u201d is increasingly uncertain as the climate changes due to human activity. Longer fire seasons<\/a>, drier vegetation<\/a> and more lightning-induced ignitions<\/a> are increasing<\/a> fire frequency in many forested regions.\u00a0<\/p>\n This is most apparent in the rapidly-warming boreal forests of the far-northern latitudes. The year 2023 saw the highest emissions<\/a> ever recorded by satellites in boreal forests, breaking a record set just two years before<\/a>.\u00a0<\/p>\n Moreover, the fires in boreal forests are becoming more intense<\/a> \u2013 meaning they burn hotter and consume a larger fraction of vegetation. This, in turn, jeopardises the recovery<\/a> of forests.\u00a0<\/p>\n In cold areas, fires also cause permafrost to break down faster<\/a>. This happens because fires remove an organic soil layer that has an insulating effect which prevents permafrost thaw.\u00a0<\/p>\n The map below shows the dominant fire type in different regions of the world, including boreal forest fires (dark green), cropland fires (red), open savannah (darker yellow) and woody savannah (brown).\u00a0<\/p>\n Changing \u2018pyrogeography\u2019<\/p>\n Thanks to more precise satellite data we now know that fire emissions are higher than we thought previously, with the new version of GFED having 65% higher overall fire emissions than its predecessor.\u00a0<\/p>\n However, all evidence suggests that emissions from fires have been stable over the past two to three decades. This is because an increase in forest fire emissions is being offset by a decline in grassland fire emissions.\u00a0<\/p>\n The world\u2019s changing \u201cpyrogeography\u201d is illustrated in the bar chart below, which breaks down annual fire emissions across different types of biome.<\/p>\n It shows how low-intensity grassland fires with modest fuel consumption \u2013 represented in yellow and brown \u2013 have declined over time, while high-intensity forest fires \u2013 illustrated in green colours \u2013 are becoming more prominent, albeit with substantial variability in emissions year-on-year.<\/p>\n![\"This](\"https:\/\/www.newsbeep.com\/nz\/wp-content\/uploads\/2025\/12\/GFED5_Fig1_map_BA.png\")
Global distribution of the average burned area over 2002-22, expressed as a percentage of the land area in each 0.25 by 0.25 degree grid cell. Based on the GFED<\/a> dataset. Credit: Chen et al. (2023<\/a>)<\/p>\n![\"This](\"https:\/\/www.newsbeep.com\/nz\/wp-content\/uploads\/2025\/12\/GFED5_Fig2_map_EM.png\")
Fire carbon emissions, in grams of carbon per square metre. Based on the GFED<\/a> emissions dataset. Credit: Van der Werf et al. (2025<\/a>)<\/p>\n![\"This](\"https:\/\/www.newsbeep.com\/nz\/wp-content\/uploads\/2025\/12\/image2.png\")
Dominant fire type around the world, based on total carbon emissions. Cropland fires are in red, woody savannah in brown, open savannah in dark yellow, grassland and shrubland in light yellow, peatland in black, tropical forest in aquamarine, temperature forest in mid-green and boreal forest in dark green. Credit: GFED5<\/a><\/p>\n![\"This](\"https:\/\/www.newsbeep.com\/nz\/wp-content\/uploads\/2025\/12\/image4.png\")
Annual emissions across various fire categories, where yellow-brown represents savannah and grassland, orange cropland, black peatland and various shades of green the different forest-fire types. Credit: GFED5 <\/p>\n