Introduction

Wildfires are among the most destructive natural hazards globally, causing severe environmental, economic, and social impacts (e.g.,1). Although fire activity is shaped by multiple factors, including fuel availability, land management, and ignition sources, certain weather conditions can allow fires to spread rapidly and become uncontrollable. “Fire weather” conditions, typically characterized by high temperatures, low humidity, dryness, and strong winds, substantially increase wildfire risk and determine how fast a fire can intensify and spread2. Extremely severe fire-weather conditions (hereafter referred to as extreme fire weather) can facilitate the transition of a wildfire into a high-intensity, erratic, and self-sustaining state, commonly referred to as an extreme fire3.

There is growing evidence that climate change is making fire weather conditions more extreme (e.g.,3,4,5). In recent decades, regions around the world including South America6,7, North America8,9, and Australia10 have experienced a notable increase in the frequency of extreme fire weather conditions. Climate-driven fire weather likely contributed to the catastrophic and deadly wildfire events recently recorded in different parts of the world (e.g.,11,12,13,14).

Southern Europe, characterized by warm, dry summers and highly heterogeneous landscapes, is particularly vulnerable to wildfires15. Recurrent heatwaves16 and persistent droughts17,18 have amplified fire risk across the region19, even as expanded fire-suppression efforts have reduced ignitions and, in many areas, lowered total burned area (e.g.,20,21,22). Nevertheless, fire-suppression has proven insufficient to prevent recent catastrophic wildfires (e.g.,23,24,25), including the Larouco fire, which burned more than 40,000 ha in August 2025 and became the largest wildfire ever recorded in Galicia, northwestern Spain (Fig. S1). The 2025 fire season in Spain was the most destructive in more than three decades, with approximately 350,000 ha burned26.

In addition to increasing warming pressure, fire risk in Southern Europe is influenced by large-scale climate modes. The most influential climate mode on Earth is the El Niño–Southern Oscillation (ENSO), which modulates global atmospheric circulation through planetary wave trains and affects precipitation, temperature, and storm tracks across all continents27. ENSO has been shown to influence fire risk in large parts of South America6,7, North America9, Australia10,28, Africa29, and Asia30. In Europe, ENSO’s influence is often mediated through regional circulation patterns such as the North Atlantic Oscillation (NAO) (e.g.,31), which has been linked to fire activity in parts of Southern Europe (e.g.,32,33). Despite growing recognition of these teleconnections, their influence on fire-prone conditions in Southern Europe remains poorly studied.

Here, we combine wildfire statistics from Portugal, Spain, France, Italy, and Greece with ERA5 reanalysis34, and large-scale climate indices (ENSO, NAO) to examine recent changes in fire risk across the region. As a metric of fire risk, we used the Fire Weather Index (FWI), which combines temperature, precipitation, wind, and fuel moisture35, all of which are influenced by anthropogenic warming and natural climate modes. We show that the FWI is becoming progressively more intense in Southern Europe, with the Iberian Peninsula exhibiting some of the steepest trends. The detected FWI intensification is consistent with broader changes in temperature and precipitation patterns. We also show that FWI variability across parts of Southern Europe is correlated with fluctuations in ENSO and the NAO, highlighting the influence of large-scale climate variability on fire-weather conditions across the region.

Results

2025 Season

Although there is no universally accepted definition of the wildfire season across Southern Europe, statistics from the European Forest Fire Information System (EFFIS)36 show that the period of highest wildfire activity (hereafter referred to as fire season) generally extends from March 1 to October 31. During the peak of the 2025 fire season, record-breaking temperatures were observed across much of the region (Fig. 1a). August featured the third major heatwave of the summer in southwestern Europe. This event began in late July in Portugal and then spread to Spain and southern France in early August37. Driven by a persistent high-pressure ridge (i.e., a heat dome), the August heatwave became the most intense on record in Spain, lasting 16 days (3–18 August)38. It led to a national mean temperature anomaly of + 4.6 °C and, at its peak, resulted in the hottest ten-day period recorded in Spain since at least 195038.

Fig. 1
Fig. 1
Full size image

A lengthy spell of blistering heat fueled major wildfires in August 2025. a) Air temperature in August 2025 relative to the 1981–2010 mean. b) Precipitation in August 2025 relative to the 1981–2010 mean. c) Fire Weather Index (FWI) in August 2025 relative to the 1981–2010 mean. d) Black Carbon Optical Depth (BCOP) on 17 August 2025. The extreme BCOP values over the northwestern Iberian Peninsula are attributable to widespread wildfires. Near-surface air temperature (t2m) and precipitation (tp) data come from the ERA5 reanalysis34. The BCOD map is based on data from NASA’s Goddard Earth Observing System (GEOS) model, GEOS-CF v1.065. Plots were generated using Python’s Matplotlib library (version 3.4.3)66, available at https://matplotlib.org/3.4.3/contents.html.

Wildfires raged across Southern Europe amid the lengthy spell of intense heat and severe drought. Soaring temperatures and dry soils (resulting from summer precipitation deficits close to 100%; Fig. 1b) led to extreme fire-weather conditions across much of the region. In the northwestern Iberian Peninsula, the FWI reached values up to 100% above the long-term average (Fig. 1c). These conditions followed an unusually wet spring across Iberia, which had stimulated vegetation growth and increased fuel loads, priming the landscape for an active fire season. The abrupt shift from a wet spring to an extremely hot and dry summer fueled wildfires that burned more intensely and spread more rapidly, endangering lives, devastating wildlife and property, and severely degrading air quality (Fig. 1d). According to statistics from EFFIS36, wildfires burned hundreds of thousands of hectares across Southern Europe in 2025, making it one of the most severe fire seasons on record.

The Iberian Peninsula was the hardest-hit region. In Spain, 2025 was the most destructive year in more than three decades, with about 350,000 ha burned26. This included the Larouco fire, which burned more than 40,000 ha and became the largest ever recorded in Galicia, northwestern Spain (Fig. S1). Neighboring Portugal also experienced extreme fire conditions, with more than 250,000 ha burned39, its worst since the record-breaking 2017 season, when more than half a million hectares were destroyed25. Together, the fires in Spain and Portugal (i.e., the Iberian Peninsula) accounted for about 80% of the total burned area in Southern Europe in 2025 (Fig. S2a). The Iberian Peninsula contributes disproportionately to European fire activity, often accounting for the majority of the burned area (Fig. S2a) and the number of fires (Fig. S2b) detected by satellites in Southern Europe.

Latest year seen in context

In recent years, summers across the Iberian Peninsula have experienced increasing temperatures and persistent drought conditions. August 2025 was among the warmest on record, with intense heatwaves pushing daily mean temperatures far above normal (Fig. S3a) and making summer (June-July-August) 2025 the warmest summer on record in Iberia (Fig. S3b). These extremes reflect a broader climate-driven trend: six of the ten largest positive temperature anomalies in the Iberian Peninsula have occurred in the past decade (Fig. S3b), and the frequency of heatwaves (e.g.,40) and “warm” days (according to the 90th percentile) has increased sharply in recent decades (Fig. S3c). Although no substantial long-term trends have been observed in spring or annual precipitation (Fig. S3d), summer rainfall has declined across Spain and Portugal from 1971 to 2025 (Fig. S3e, f). Notably, Iberia has experienced summer precipitation deficits in 20 of the past 25 years, and summer 2025 ranks among the ten driest summers of the past fifty years (Fig. S3e, f). The combination of abundant fuel (resulting of the 2025 wet spring; Fig. S3d), exceptional heat, and severe summer drought fueled the intense fires that raged across northwestern Spain last year.

The intense fire activity observed in 2025 is also consistent with a rebound in burned area across Southern Europe that began seven years ago. Since the early 1980s, the annual burned area in countries such as Spain and Italy declined steadily, reaching record lows in 2018 (Figs. S4a-b). However, since 2018, Southern Europe has experienced years of enhanced fire activity, including two years (2022 and 2025) in which more than half a million hectares burned (Fig. S4c). Other Southern European countries, namely Portugal, France, and Greece, have also experienced elevated fire activity in recent years (Fig. S4d-f).

Despite the recent rebound, the burned area in Southern Europe remains lower than in the 1980s. With the exception of Portugal, the annual burned area in each Southern European country has been lower over the last decade than during the 1981–2010 period (Fig. S5a). Particularly large declines have been observed in Spain, France, and Italy, where burned area has decreased by 25%, 41%, and 31%, respectively (Fig. S5a). In Spain, the annual average burned area was approximately 170,000 ha during 1981–2010 but declined to about 127,000 ha over the last decade, largely due to improvements in fire prevention and suppression. These reductions underscore the overall success of fire-management and containment efforts. Since the 1980s, Southern Europe has strengthened fire suppression capacity through better-trained brigades, expanded aerial resources, rapid initial-attack strategies, and, in some countries, fuel-break networks and public awareness initiatives (e.g.,20,21,22).

Strengthened fire suppression capacity helps explain why the number of fires has also declined consistently across Southern Europe in recent decades. In each Southern European country, the annual number of fires during the last decade was at least 40% lower than the 1981–2010 average (Fig. S5b). This decline is consistent with a reduction in human-caused ignitions (that account for the majority of fires41) driven by stricter regulations, increased public awareness, rural depopulation, and improved surveillance systems (e.g.,20). The reductions have become markedly pronounced over the past 25 years. Spain and Portugal, for example, have reduced fire counts by more than half: each now typically registers fewer than 10,000 fires per year, compared with more than 20,000 and 30,000 fires, respectively, in the early 2000s (Fig. S6a, b). Even during the highly active 2025 year, the combined number of fires in Spain and Portugal was only slightly above 15,00026,39. While the reduction has been less pronounced, Italy, France, and Greece have also experienced declines in fire numbers over the past two decades (Fig. S6c-e).

Southern Europe is becoming more flammable

The FWI has increased considerably in recent decades across Southern Europe (Figs. 2a and S7). The rise has been particularly pronounced in parts of France and Spain, where over the last decade the summer FWI was up to 50% higher than during 1981–2010. These increases are consistent with other climate-related trends. As in the rest of the world, summer temperatures have risen steadily in Southern Europe. In southern Spain, southern France and in most of Italy, temperatures were, on average, up to 2 °C higher over 2016–2025 compared to the long-term mean (Fig. 2b). At the same time, parts of the region have experienced substantial drying, with summer precipitation in southwestern Iberia nearly 50% lower than the 1981–2010 average. As shown in Fig. 2c, the changes in summer precipitation are not uniform across Southern Europe. In fact, parts of southern Italy and Greece appear to have gained precipitation in recent decades, which may have contributed to the reductions in both burned area and fire counts observed there (Fig. S5-S6).

Southern Europe has experienced a marked increase in the frequency of fire-prone summer days. The rise in extreme fire-weather conditions (according to the 90th percentile) has been steep across most of the region (Figs. 2d and S8). While extreme fire weather conditions generally occurred on fewer than 10 days per summer during 1981–2010 (Fig. S9a, upper panel), they have reached up to about 25 summer days in parts of each of the countries in our study area over the past decade (Fig. S9a, lower panel). Summer “warm” conditions (according to the 90th percentile) have also increased by up to 150% in Mediterranean Europe over the same period (Fig. 2e). While “warm” conditions generally occurred on fewer than 10 days per summer during 1981–2010 (Fig. S9b, upper panel), they have reached up to about 25 summer days in parts of the Mediterranean Europe over the past decade (Fig. S9b, lower panel). Increases in the number of flammable summer days have also been observed in Greece and southern Italy (Fig. S2d), one of the few regions in southern Europe where summer precipitation has risen (Fig. 2c) and the number of summer days without precipitation has slightly decreased (Fig. 2f) from 1981 to 2010 to 2016–2025.

Although the FWI depends on several meteorological variables, it is strongly influenced by temperature and precipitation (e.g.,42). Accordingly, the spatial patterns of change in FWI between 1981 and 2010 and 2016–2025 (Fig. 2a) roughly mirror those of temperature change over the same period (Fig. 2b) across Mediterranean Europe. Likewise, changes in the number of flammable days (Fig. 2d) tend to generally follow the patterns observed in the number of warm days (Fig. 2e). Changes in precipitation (Fig. 2c) and in the number of dry days (Fig. 2f) may have helped moderate the increases in FWI and flammable days in Greece, while exacerbating the rise in these parameters in areas of France that lie north of its Mediterranean coast (Fig. 2a). The strong sensitivity of the FWI to temperature and precipitation helps explain why the changes in fire risk observed across Mediterranean Europe in recent decades are closely associated with changes in temperature (Fig. S10a) and inversely associated with changes in precipitation (Fig. S10b). Although the FWI is also dependent on surface wind speed (Fig. S10c), wind speeds in Southern Europe have shown relatively small changes (less than ± 10%) in recent decades (Fig. S11).

Fig. 2
Fig. 2
Full size image

Most of Southern Europe has become more flammable. Change from 1981–2010 to 2016–2025 in the a) summer Fire Weather Index (FWI), b) summer air temperature, c) summer precipitation, d) summer flammable days, e) summer warm days, and f) summer dry days. A day is classified as warm or flammable if its mean temperature or Fire Weather Index (FWI), respectively, exceeds the 90th percentile of the daily baseline climatology. A day is considered as dry if no precipitation occurs. The climatology is based on daily temperature, and daily FWI over the 1981–2010 reference period (see Methods). Daily near-surface air temperature (t2m), precipitation (tp), and FWI (fwinx) data come from the ERA5 reanalysis67,34. Plots were generated using Python’s Matplotlib library (version 3.4.3)66, available at https://matplotlib.org/3.4.3/contents.html.

The rise in fire risk is likely related to broader atmospheric circulation changes

The NAO index43 characterizes the strength and position of the westerly winds and storm tracks, which in turn influence heat and moisture transport, precipitation, and temperature across the North Atlantic–European region. It measures the sea-level pressure difference between the Azores High and the Subpolar Low44. In its negative phase, pressures are lower than normal in the central North Atlantic and higher at high latitudes (Fig. 3a), while the positive phase shows the opposite pattern. In Southern Europe, a negative NAO index is often associated with hotter and drier summers45,46.

The frequent negative NAO values observed during recent fire seasons coincided with increased fire risk in Southern Europe. The NAO index, averaged over summer and autumn (June–July–August-September–October–November, JJASON), has been negative for 25 of the past 30 years (Fig. 3b, upper panel). As a result of these frequent negative values, we found a significant negative trend (p < 0.05) in the summer and autumn NAO index over the period 1960–2025 (Fig. 3b, upper panel), in contrast to the positive trend observed in the NAO index averaged over winter and spring (Fig. 3b, lower panel). While the positive winter and spring NAO index has been linked to milder winters in Northern Europe (e.g.,47), the negative summer and autumn NAO index is associated with atmospheric blocking that suppress rainfall, intensify drought, and increase the likelihood of heatwaves, soil-moisture depletion, and fire-prone conditions in Southern Europe (e.g.,45,46).

Fig. 3
Fig. 3
Full size image

The escalation of fire risk observed in Southern Europe in recent decades coincided with frequent negative values of the summer NAO index. a) Sea Level Pressure (SLP) anomaly averaged from June to August 2023, relative to the 1981–2010 mean. The lower-than-normal pressures in the central North Atlantic and higher-than-normal at high latitudes are consistent with the negative values of the NAO index observed in summer 2023. b) Time series of the North Atlantic Oscillation (NAO) index averaged over summer and autumn (June-July-August-September-October-November, JJASON) (upper plot) and averaged over winter and spring (December-January-February-March-April-May, DJFMAM) (lower plot). The NAO index values come from NOAA’s Climate Prediction Center (CPC): https://www.cpc.ncep.noaa.gov/products/precip/CWlink/pna/nao.shtml. SLP (msl) data come from come from the ERA5 reanalysis34. Plots were generated using Python’s Matplotlib library (version 3.4.3)66, available at https://matplotlib.org/3.4.3/contents.html.

Fire weather variability in Southern Europe is influenced by NAO and ENSO

The NAO varies across multiple timescales (including daily, weekly, seasonal, and even interannual) and so does its influence on fire weather in Southern Europe. A negative NAO phase often brings hotter and drier conditions across the region, whereas a positive phase typically produces the opposite pattern (e.g.,45,46). Accordingly, we found a significant anticorrelation (p < 0.05) between FWI and the NAO index in parts of Greece, Italy, France (Corsica), and Spain (Fig. 4a). The influence of the NAO across most of Southern Europe helps explain the fire weather synchronicity in region (e.g.,48). Interestingly, as shown in Fig. 4a, the correlation between the NAO index and fire weather is not significant in the northwestern Iberian Peninsula (the region most affected by the record-breaking 2025 fires).

Fig. 4
Fig. 4
Full size image

The variability of fire risk in parts of Southern Europe is associated with NAO and ENSO. Pearson correlation between the summer (JJA) Fire Weather Index (FWI) and the a) JJA NAO index; b) JJA sea surface temperature (SST) anomalies in the Niño 1 + 2 region (in the eastern tropical Pacific). c) Time series of annual burned area in the Iberian Peninsula (black dotted line), JJA FWI values averaged across the Iberian Peninsula (red line), and JJA SST anomalies in the Niño 1 + 2 region (black line). The burned area is correlated with the FWI (R = + 0.4), and with SST anomalies in the Niño 1 + 2 region (R = -0.3). All time series were linearly detrended prior to the correlation analysis. For panels (a) and (b), temporal autocorrelation was accounted for using effective sample sizes derived from the lag-1 autocorrelation of the paired time series. Stippling indicates correlations that are statistically significant at the 0.05 significance level. The Supplementary Material (Fig. S14) presents equivalent versions of panels (a) and (b) in which statistical significance is additionally adjusted for multiple testing using the Benjamini–Hochberg false discovery rate (FDR) correction at the 0.05 significance level. Data over the period 1982–2025 were used. FWI (fwinx) data come from come from the ERA5 reanalysis67. The NAO index values come from NOAA’s Climate Prediction Center (CPC) available at https://www.cpc.ncep.noaa.gov/products/precip/CWlink/pna/nao.shtml while SST anomalies in the Niño Regions come NOAA’s CPC available at https://www.cpc.ncep.noaa.gov/data/indices/wksst8110.for. Burned area in c) was computed aggregating estimates reported by the national agencies of Portugal and Spain. The sources of all wildfire statistics used in this study are listed in the Data Availability section. Plots were generated using Python’s Matplotlib library (version 3.4.3)66, available at https://matplotlib.org/3.4.3/contents.html.

ENSO influences climate on seasonal to interannual timescales, with teleconnections that evolve over months to years. Although SST anomalies in the tropical Pacific Niño regions do not directly control European climate, they can act as important modulator (e.g.,49). NOAA’s Climate Prediction Center (CPC) currently monitors several Niño regions, including Niño 1 + 2 (in the eastern tropical Pacific) and Niño 3.4 (in the central tropical Pacific) (Fig. S12a). El Niño and La Niña events are classified based on SST anomalies exceeding ± 0.5 °C in the Niño 3.4 region50. Several record-breaking fire seasons in Spain, for example, such as in 1985, 1989, 2017, 2022, and 2025 (Fig. S4a), were preceded by La Niña conditions (Fig. S12b). Motivated by this apparent association, we investigated the relationship between ENSO variability and fire-weather conditions across Southern Europe. Although significant correlations were identified between SST anomalies in the Niño 3.4 region and FWI across parts of Southern Europe, including Sicily and southern Greece, no significant relationship emerged between Niño 3.4 SST anomalies and FWI over the Iberian Peninsula (Fig. S13).

The interannual variability of fire weather in the Iberian Peninsula appears to be linked to SSTs in the Niño 1 + 2 region (Fig. 4b). We found significant correlations (p < 0.05) between Niño 1 + 2 SSTs and the FWI across parts of northwestern Iberia, the area most affected by the record-breaking 2025 fires (Fig. 4b). This relationship is consistent with the observed correlation between the burned area in the Iberian Peninsula (Spain + Portugal) and the SST anomalies in the Niño 1 + 2 region (Fig. 4c). These results provide further evidence of a statistical association between ENSO and the fire weather in relevant parts of Southern Europe.

It should be noted that the statistical significance of the correlations shown in Figs. 4a-b and S13 is sensitive to methodological choices. To illustrate this sensitivity, Fig. S14 presents the results of an additional analysis in which the Benjamini–Hochberg false discovery rate (FDR) correction was applied at the 0.05 significance level to account for multiple testing. Although the overall correlation patterns remain largely unchanged, the FDR correction considerably reduced the spatial extent of the statistically significant correlations shown in Figs. 4a-b and S13.

Discussion

The 2025 Iberian fire crisis is consistent with a trend toward more intense and frequent fire-weather extremes across Southern Europe. Our results show that rising temperatures are intensifying fire weather in the region, making extreme FWI values increasingly frequent. The increase in extreme fire-weather conditions has been steep across most of the Iberian Peninsula, a sub-region that contributes disproportionately to Southern Europe’s total burned area and that has experienced not only hotter summers but also recurrent summer droughts in recent decades. Whereas extreme fire weather typically occurred on fewer than ten summer days during 1981–2010, parts of Iberian Peninsula have experienced up to 25 such extreme summer days over the past decade. In contrast, FWI increases have been more moderate in southern Italy and Greece, consistent with localized upward trends in summer precipitation that have partly offset the effects of rising temperatures.

The intensification of fire weather over recent decades is consistent with broader changes in atmospheric circulation. Reflecting these circulation shifts, we found a negative trend in the NAO index averaged over summer and autumn. In Southern Europe, a negative NAO phase is often associated with hotter and drier summers (e.g.,45,46). A trend toward weaker summer–autumn westerlies (i.e., increasingly negative NAO conditions) has been linked to Arctic amplification (e.g.,51). Rapid Arctic warming reduces the meridional temperature gradient, particularly in late summer and autumn, which may promote more frequent atmospheric blocking and heatwaves (e.g.,52). However, these linkages remain uncertain, as does the future evolution of the summer NAO. Current climate model projections nonetheless suggest increased NAO variability and more frequent extreme phases (both positive and negative) under continued warming53.

Interannual variability in Southern European fire weather is also associated with tropical Pacific variability. Several record-breaking fire seasons in the region (including 1985, 1989, 2013, 2017, 2022, and 2025) were preceded by La Niña conditions. Accordingly, we found that SST anomalies in the Niño 1 + 2 region (in the eastern tropical Pacific) are anticorrelated with the FWI across parts of Southern Europe, especially in northwestern Iberia, the area most severely affected by the 2025 fires. In other words, active fire seasons in parts of Southern Europe are often associated with Eastern Pacific La Niña events. In the context of increasing fire risk, these correlations may provide a basis for future predictive applications. ENSO-related conditions may also help explain the recent escalation in Southern European fire activity. Although there is no long-term SST trend, negative anomalies in the tropical Pacific have been unusually common in recent years, with La Niña conditions present in 7 of the last 10 years. Yet future ENSO response to continued greenhouse-gas forcing remains uncertain (e.g.,54).

Despite increasing climatic pressure toward higher fire risk, Southern Europe has made substantial progress in fire management and suppression. That explains why the rise in extreme fire-weather conditions over recent decades has not been accompanied by a proportional rise in fire activity. Burned area remains lower than in the 1980s in all Southern European countries except Portugal, and the number of fires has declined consistently across the region. In each Southern European country, the annual number of fires during the past decade was approximately 40% lower than the 1981–2010 average. This apparent contradiction reflects management improvements that have counteracted climate-driven increases in fire weather. Since the 1980s, Southern Europe has strengthened fire suppression capacity through better-trained brigades, expanded aerial resources, rapid initial-attack strategies, and, in several countries, fuel-break networks and public awareness campaigns (e.g.,20–22). Yet, intensive suppression has also promoted fuel accumulation in many landscapes. This “suppression paradox” creates conditions in which fewer but more extreme and fast-spreading megafires can occur (e.g.,55).

Europe has shifted toward fewer, but more intense and larger events. In Western Mediterranean Europe, a small number of wildfire events are responsible for most of the total burnt area, and this concentration is increasing over time56. Although improvements in prevention and suppression have reduced ignitions in Southern Europe, the underlying climatic conditions that favor fire have continued to worsen, particularly in the Iberian Peninsula. Since the 2018 record low, Spain has experienced several active fire seasons, including two years (2022 and 2025) in which more than a quarter of a million hectares burned. Recent active fire seasons suggest that climate-driven extreme fire-weather conditions are starting to surpass the capacity of existing wildfire suppression systems, far earlier than earlier projections suggested (e.g.,57). Under climate-driven extreme fire-weather conditions, fires that escape initial attack can quickly become catastrophic. Unfortunately, climatic pressure on fire regimes is expected to intensify further58 and Europe is expected to face up to a tenfold increase in extreme fires this century59.

Effective adaptation may require integrated strategies that combine improved fire management with climate-resilient land-use planning. A number of approaches have been proposed to reduce fire severity, including prescribed burning60, shifting from maximum to moderate suppression61, progressive suppression strategies that allow low-intensity fires to spread under safe conditions55, and adaptive risk-informed management frameworks62. Future fire management strategies in Southern Europe must, however, be tailored to reflect not only evolving climatic trends but also local ecological and socio-environmental contexts.

Methods

Study region and period

Our study region encompasses Southern Europe, specifically Portugal, Spain, France, Italy, and Greece. Wildfire statistics used in this study were obtained from the national reporting agencies of these countries. These datasets provide multi-decadal records of two key metrics (annual burned area and annual number of fires) which were used to characterize long-term changes in fire activity. These national wildfire datasets are compiled using country-specific reporting methodologies and definitions, making direct harmonization across countries challenging. To avoid introducing uncertainties associated with differing reporting practices, our analyses were generally performed separately for each country and did not rely on aggregated national wildfire records.

However, note that when discussing fire activity at the scale of Southern Europe (Fig. S4c), we used burned area estimates from the European Forest Fire Information System (EFFIS), which provides a harmonized dataset based on a consistent methodology across national boundaries. Although the EFFIS record is shorter than the national datasets (covering only the period 2006–2025), its standardized methodology ensures comparability among countries and therefore provides a robust basis for regional-scale analyses.

Moreover, when discussing fire activity at the scale of the Iberian Peninsula over the period 1982–2025 (Fig. 4c), we aggregated the burned-area estimates reported by the national agencies of Portugal and Spain. While minor methodological differences may exist between the two national reporting systems, we do not expect these differences to considerably affect the characterization of interannual burned-area variability at the Iberian Peninsula scale. The annual burned area in both countries frequently exceeds 100,000 ha, such that the aggregated series is assumed to be largely insensitive to small methodological inconsistencies.

The sources of all wildfire statistics used in this study are listed in the Data Availability section.

Gridded climate maps

For the color-coded maps shown in Figs. 1 and 3, and 4, the spatial domain was selected to focus on the Mediterranean and fire-prone regions of Southern Europe: 35°N-48°N, 12°W-25°E. This domain encompasses the entirety of Portugal, Spain, Italy, and Greece, while including the southern and southwestern regions of France, where wildfire activity has historically been concentrated.

Note that the color-coded maps shown in Figs. 1 and 3, and 4 and in the Supplementary Material are not based on national wildfire statistics, but rather on spatially consistent and harmonized datasets across national boundaries, such as the ERA5 reanalysis34 (see details below).

Fire Weather Index (FWI)

The FWI is a metric of fire danger derived solely from weather data and is computed from daily noon observations of wind speed, 24-hour accumulated precipitation, temperature, and relative humidity. As described by Van Wagner35, these meteorological variables are used sequentially to calculate the Fine Fuel Moisture Code (FFMC), which represents moisture conditions in shaded litter fuels; the Duff Moisture Code (DMC), which reflects the moisture content of decomposed organic material beneath the litter; and the Drought Code (DC), which characterizes deep soil drying. These components are then combined to obtain the Initial Spread Index (ISI), integrating fine fuel moisture and surface wind speed to estimate fire spread potential; the Buildup Index (BUI), which uses the DMC and DC to represent the available fuel for combustion; and finally, the FWI, which integrates ISI and BUI to provide an overall indicator of fire risk.

Here we used daily values of the FWI from the ERA5 reanalysis34 (see details below) aggregated according to the temporal scale of interest (e.g., monthly, seasonal, multi-month, or annual averages).

Reanalysis Data

For the selected spatial domain (35°N-48°N, 12°W-25°E), we used daily meteorological variables from the ERA5 reanalysis dataset, including near-surface air temperature (t2m), precipitation (tp), and FWI (fwinx). ERA5 is produced by the European Centre for Medium-Range Weather Forecasts (ECMWF) and is provided on a regular 0.25° × 0.25° latitude–longitude grid. In addition, and beyond the selected spatial domain, gridded ERA5 sea-level pressure (SLP) (msl) fields were used to generate Fig. 3a.

Although ERA5 reanalysis data are available from the 1970s onward, all correlation analyses involving reanalysis variables were restricted to the period beginning in the early 1980s, when the reanalysis is generally considered to be more reliable due to the increasing availability of satellite observations. Daily values aggregated according to the temporal scale of interest (e.g., monthly, seasonal, multi-month, or annual averages). Spatial averaging was applied when assessing fire-weather conditions at a regional scale, such as over the Iberian Peninsula. The sources of ERA5 data are listed in the Data Availability section.

North Atlantic Oscillation (NAO)

The NAO operates primarily on synoptic to seasonal timescales, with strong variability from days to weeks but also meaningful seasonal and interannual components. To assess the impact of the NAO on fire weather, we compared changes in the FWI with changes in the NAO index. The latter measures the sea-level pressure difference between the Azores High and the Subpolar Low44. In its positive phase, pressures are higher than normal in the central North Atlantic and lower at high latitudes, while the negative phase shows the opposite pattern. These phases drive basin-wide shifts in the strength and position of the North Atlantic jet stream and storm track, influencing large-scale heat and moisture transport44.

The NAO index is computed by projecting the NAO loading pattern to the daily anomaly 500 millibar height field over 0–90°N. The NAO loading pattern has been chosen as the first mode of a Rotated Empirical Orthogonal Function (EOF) analysis43 using monthly mean 500 millibar height anomaly data from 1950 to 2000 over 0–90°N latitude.

Although reliable NAO index values are available for earlier periods (see, for example, the time series presented in Fig. 3b), in order to ensure consistency, all analyses involving NAO data (including correlations with ERA5-derived FWI values across the selected spatial domain) were restricted to the period beginning in the early 1980s. Daily NAO index values were obtained from NOAA’s CPC (link provided in the Data Availability section) and aggregated according to the temporal scale of interest (e.g., monthly, seasonal, multi-month, or annual averages).

El Niño-Southern Oscillation (ENSO)

ENSO influences climate on seasonal to interannual timescales. To assess the impact of ENSO on the fire weather, we compared changes in the FWI with changes in the sea surface temperature (SST) in two El Niño regions: the Niño 3.4 region (5°N-5°S, 170°W–120°W) and the Niño 1 + 2 region (0–10°S, 90°W–80°W) (Fig. S12a). The phases of ENSO (i.e., El Niño and La Niña) are driven by the strength of trade winds27. During El Niño events, trade winds weaken, and warm water accumulates off the South American west coast. During La Niña events, trade winds strengthen, increasing upwelling and bringing cold, nutrient-rich water to the surface.

To evaluate statistical associations between ENSO variability and fire-weather conditions across Southern Europe, SST anomalies in the Niño regions were compared with ERA5-derived FWI values across the study domain. Weekly SST anomalies for the Niño regions, available since 1982, were obtained from NOAA’s CPC (link provided in the Data Availability section) and aggregated to the temporal scale of interest (e.g., monthly, seasonal, multi-month, or annual averages).

Analysis of extremes

Building on earlier work6,7, we used a 30-year reference period (1981–2010) and applied a 15-day rolling window to ERA5-derived values of the daily mean temperature (T), and FWI, generating datasets of 450 values for each calendar day. The mean of each dataset was used as the daily climatological baseline, and daily anomalies (for T, and FWI) were computed as deviations from this baseline. Based on the resulting anomaly distributions, we identified:

  • “Flammable” days: days exceeding the 90th percentile of the FWI anomaly distribution for the reference period; and.

  • “Warm” days: days exceeding the 90th percentile of the T anomaly distribution.

Regarding precipitations, here we considered a day to be “dry” if no precipitation occurred.

Statistical Tests

Pearson correlation analyses were used to evaluate relationships between fire-weather conditions (i.e., FWI), climate indices (i.e., NAO and ENSO SST), and wildfire activity (e.g., burned area). Prior to the correlation analyses, all time series were linearly detrended to ensure that the reported relationships primarily reflect interannual variability rather than potential common long-term trends. Unless otherwise stated, statistical significance was defined as p < 0.05.

Correlation analyses, including between FWI and both NAO and ENSO SST (Fig. 4a and b, and S13), also accounted for temporal autocorrelation through effective sample-size adjustments based on the lag-1 autocorrelation of the paired time series. This approach allowed the persistence characteristics of the climate signals to be retained while providing a more conservative estimate of statistical significance. The adjustment is particularly relevant for ENSO-related analyses, as ENSO exhibits substantial temporal persistence on seasonal to interannual timescales. In contrast, the NAO generally exhibits relatively weak year-to-year persistence, with lag-1 autocorrelation that is often small and, in many cases, statistically indistinguishable from white noise (e.g.,63,64).

Furthermore, to assess the sensitivity of the correlations between FWI and both NAO and ENSO to different methodological choices, we conducted an additional analysis in which, beyond detrending and accounting for temporal autocorrelation, we also apply the Benjamini–Hochberg false discovery rate (FDR) correction at the 0.05 significance level to account for multiple testing. The FDR correction reduces the likelihood of identifying spurious significant correlations arising from the large number of simultaneous grid-point tests. In practice, and as shown in Fig. S14, applying the FDR correction reduces the spatial extent of statistically significant correlations, leaving the overall correlation patterns largely unchanged.