Europe faces up to tenfold increase in extreme fires in a warming climate

Wildfires are a crucial ecological component, with ~3% of global land area burned yearly. In Europe, ~96% of ignitions are human-caused, and suppression is often used to protect people and assets. Fire spread and intensity are governed primarily by weather, vegetation, and topography. Recent decades have seen more destructive wildfires, with three of the worst European fire years (by burned area) occurring since 1980 within the last six years. Wildfires are emerging in new places due to increased vegetation flammability, occurring outside traditional seasons and at higher intensities, with a small number of megafires responsible for ~97% of burned areas and significant secondary impacts (erosion, deforestation, carbon loss) and risks to society and infrastructure. Despite many studies on climate-induced fire danger trends, a quantitative pan-European assessment of how temperature and precipitation—alone and combined—alter the probability of extreme wildfire events remains limited. This study fills that gap by conducting a sensitivity analysis using FWI derived from ERA5 to build two-dimensional impact response surfaces (IRS) linking projected changes in temperature and precipitation to the probability and duration of extreme fire weather. The study assumes relative stability in ignition sources and suppression/land-use practices, noting that increases in fire risk and fuel availability (e.g., from rural abandonment) will challenge control measures. The analysis focuses on a latitudinal gradient across three IPCC regions (NEU, CEU, SEU) and five empirically identified fire-regime clusters (boreal, temperate, semi-arid, alpine, Mediterranean). Under SSP2-4.5, Europe’s annual temperature is projected to rise by ~2 °C (2036–2065) and ~3 °C (2070–2099), up to ~4 °C under SSP5-8.5, with consequential impacts on landscape flammability.

Prior work documents increasing global wildfire danger linked to anthropogenic warming and declines in fuel moisture, with extreme events concentrated among a few large fires and expanding outside traditional seasons. Studies have examined regional fire danger trends in Europe (e.g., Mediterranean), the importance of vapor pressure deficit (VPD), and links between FWI and burned area, especially for large events. There is evidence that suppression efforts, ignition patterns (including lightning), and land-use changes modulate realized fire activity. However, a comprehensive Europe-wide, quantitative assessment of how combined temperature and precipitation changes affect the probability of extreme fire weather had been lacking; this work addresses that through an IRS-based sensitivity framework integrating ERA5-based FWI and CMIP6 projections.

Data and baseline: The study uses ERA5 reanalysis to force the ECMWF Global ECMWF Fire Forecast (GEFF) model to compute daily Fire Weather Index (FWI) for 1981–2010 (April–October focus). CMIP6 multi-model ensemble projections under SSP2-4.5 provide future climate change signals. Clustering of fire regimes: K-means clustering (optimal K via silhouette criterion with manual refinement) was applied to FWI time series and an intensity metric (max FWI × range) to delineate five fire regions (boreal, temperate, semi-arid, alpine, Mediterranean) with similar season length and intensity. Definition of extremes and season length: For each region, an extreme fire weather threshold was defined as the 20-year return period FWI (RP20), estimated via Weibull plotting position using annual maxima over 1981–2010. Fire season length was defined as the number of days per year with FWI above half the RP20 threshold. Incremental climate perturbations and IRS: To link climate change to fire weather responses, ERA5 daily series were perturbed using CMIP6-derived monthly change signals. Temperature perturbations were additive from +0 to +6 °C in +1 °C steps. Precipitation perturbations were multiplicative from 0.6 to 1.6 in 0.2 steps. Monthly CMIP6 signals were mapped to daily via upsampling using ERA5. For each (ΔT, ΔP) combination, FWI was recomputed and (i) the exceedance probability of RP20 and (ii) the excess season length (days above half-RP20) were estimated. The two-dimensional impact response surfaces (IRS) plot these responses, with continuous curves interpolated between grid points. Projection sampling: Probability density functions of projected ΔT and ΔP from the CMIP6 SSP2-4.5 ensemble were overlaid on IRS for short-term (2036–2065) and long-term (2070–2099) periods, with mean outcomes highlighted. Spatial aggregation summarized the share of land area in probability/season-length categories for each region. Assumptions: Ignition patterns and suppression/land use remain broadly similar; FWI is used as a proxy for potential fire activity, acknowledging stronger linkage during extreme conditions and potential fuel limitations. Uncertainties in precipitation projections and hydrological processes were noted.

- Five fire regions with distinct RP20 FWI thresholds and current season lengths were identified (examples from Fig. 1): Boreal RP20 ≈ 19 (≈35 days above half-threshold), Temperate ≈ 28 (≈38 days), Semi-arid ≈ 45 (≈55 days), Alpine ≈ 15 (≈25 days), Mediterranean (MED) ≈ 63 (≈115 days). The MED region currently has the longest and most intense fire season. - Sensitivity to temperature and precipitation varies by region (IRS results): • MED: Highly temperature-sensitive; a +1 °C increase nearly doubles the probability of RP20 exceedance (from 5% to ~10%), while precipitation has weaker influence; ~+20% precipitation is needed to offset +1 °C warming. Similar but less pronounced behavior occurs in the semi-arid region. • Boreal, Temperate, Alpine: Probability increases accelerate beyond ~+2 °C. A −20% precipitation change doubles the probability (5%→10%) even with modest warming; temperature effects are smaller below +2 °C. - Season lengthening: Reaching +2 °C warming is associated with a nearly uniform ~+10 days extension of the fire season across Europe. The operational burden differs: northern European countries could see ~30% more days requiring fire assistance, while southern countries see ~8% more. - Short-term (2036–2065) probabilities under SSP2-4.5 (most likely outcomes): • Across most regions, ~66% of the landscape shows a 5–10% annual probability of RP20 events. • MED is higher risk: ~40% of area at 10–15%, ~11% at 20–25%, and ~5% at 25–50% probability. - Long-term (2070–2099): • Medium-risk areas expand in all regions; in MED, medium-risk area triples; in other regions, medium-risk area roughly doubles (from ~6% to ~12%). • MED fire season increases by at least one week; by century’s end, ~20% of MED could see +20 days. Northern Europe also lengthens by ~1 week. Accounting for moisture/VPD could imply even longer seasons. - Spatial extent: >50% of Europe’s land area is projected to at least double the probability of extreme events (5%→10%). Localized hotspots, especially across the MED (eastern Turkey, Greece, southern Italy, Iberian Peninsula, southern France, northern Africa), could experience up to a tenfold increase (5%→50%). An Atlantic corridor in central/northern Europe also shows elevated risk; northern Europe generally increases to ~20% probability (one-in-five-year frequency), but remains precipitation-sensitive. - Season extension patterns: Short-term +12–16 days in Iberia, eastern Spain, southern France, Greece, western Turkey; central Europe +4–8 days. Long-term: central Europe adds ~+4 more days; MED +12–22 days. By century’s end, up to ~18% of southern Europe may face catastrophic fire conditions as frequently as every other year. - Uncertainties: Precipitation projections remain more uncertain than temperature due to hydrological and cloud-process complexities; NEU shows high temperature variability with some models indicating increased precipitation.

The analysis links extreme fire weather (RP20 FWI) to increased probabilities and longer seasons under warming, with the strongest absolute effects in the Mediterranean where FWI thresholds conducive to megafires (FWI >50) are common. In boreal/temperate/alpine regions, while absolute probabilities remain lower, increases become substantial beyond +2 °C and under drying, potentially overwhelming local suppression capacity during peak years. The 2018 Scandinavian fires illustrate how reaching RP20 conditions can strain resources even where fires are usually manageable (FWI ~20). The MED region faces the largest amplification due to warming and drying, confirming the need for enhanced preparedness and mitigation. Boreal and temperate regions exhibit high sensitivity to precipitation deficits, and projected increases in lightning could raise ignitions, compounding risk. Europe’s shared response mechanisms (e.g., RescEU) can reduce casualties and improve efficiency, but the expected doubling or more of extreme-event probabilities, tenfold hotspots in the south, and continent-wide season lengthening may stress suppression systems and budgets. The results argue for a strategic shift toward proactive risk mitigation (fuel management, land-use planning, community preparedness) alongside international resource sharing. These findings provide actionable risk information by quantifying both probability increases and their spatial extent, offering national and regional planners a framework to anticipate future assistance needs, allocate resources, and plan adaptation under a warming climate.

This study provides a pan-European, quantitative sensitivity assessment of extreme fire weather to temperature and precipitation changes by coupling ERA5-based FWI with CMIP6 projections via impact response surfaces. It shows that: - Southern Europe is poised for the largest increases in extreme fire probabilities, with localized tenfold rises and substantial lengthening of the fire season; - A +2 °C warming threshold triggers marked increases in central and northern Europe, especially under drying; - More than half of Europe will at least double the probability of extreme fire weather, with widespread season extensions that will strain suppression capacity. The IRS framework enables direct translation of climate projections into probabilistic fire-weather impacts, supporting planning and adaptation decisions. Future research should incorporate dynamic vegetation and fuel feedbacks, changes in ignition/suppression practices, explicit moisture/VPD coupling, alternative emissions scenarios, and higher-resolution regional modeling to refine local risk estimates and management strategies.

- FWI is a proxy for potential fire activity and not a direct measure; its correlation with observed fires is strongest during extreme conditions and weaker in fuel-limited systems. - Assumes relatively stable ignition sources (human behavior) and no major changes in suppression capacity or land use; real-world changes could alter outcomes. - Vegetation dynamics and post-fire recovery are not modeled; significant fuel depletion or changes in fuel structure could reduce realized fire activity regardless of FWI. - CMIP6 precipitation projections are more uncertain than temperature due to spatial/temporal variability and hydrological process modeling challenges. - Coarse-resolution CMIP6 outputs are upsampled using ERA5; downscaling and perturbation choices may influence results. - Moisture/VPD effects are discussed but not fully integrated into the IRS beyond temperature/precipitation perturbations; considering them could lengthen projected seasons further. - Results focus on SSP2-4.5; other scenarios may yield different magnitudes and spatial patterns of risk.