This study investigates the role of atmospheric rivers (ARs) in modulating precipitation and wildfire activity across South America using the European Centre for Medium-Range Weather Forecasts Reanalysis version 5 (ERA5) together with Moderate Resolution Imaging Spectroradiometer (MODIS) burned area, Normalized Difference Vegetation Index (NDVI) and Normalized Difference Water Index (NDWI) observations. The results reveal widespread declines in AR frequency and AR-associated precipitation across much of the continent, particularly during March–May (MAM) and June–August (JJA) over the Brazilian savanna (Cerrado) and Chaco, regions characterized by extensive burned area and recurrent wildfire activity. Spatial and temporal analyses reveal a dual role of ARs in regional fire dynamics. On short timescales (0–3 months), enhanced AR-associated precipitation suppresses fire activity by increasing fuel moisture and reducing incoming surface solar radiation through greater cloud cover. At longer lags (3–6 months), enhanced moisture availability promotes vegetation growth and fuel accumulation, producing dry biomass that can subsequently increase wildfire susceptibility during the following dry season. In central Brazil, burned area peaks and negative precipitation–fire correlations occur at lags of approximately 3–4 months, highlighting the delayed response of ecosystem productivity and fire occurrence to AR variability. Variations in AR activity are accompanied by large-scale circulation anomalies, including upper-level trough–ridge patterns and changes in Integrated Vapor Transport (IVT), which modulate regional moisture transport, precipitation, and the subsequent vegetation–fire response. These findings demonstrate that future changes in AR frequency and intensity may substantially influence the hydroclimate, vegetation productivity, and wildfire regimes of South America, with important implications for fire early-warning systems, ecosystem management, and climate adaptation strategies.
Recent extreme heatwaves in eastern China have caused escalating socio-economic and environmental vulnerabilities. This study identifies distinct sub-seasonal variability in summer heatwaves, with the North China Plain (NCP) prominently affected in June and the Yangtze River Valley (YRV) during July–August (JA). The Tibetan Plateau atmospheric heat source (TP AHS) exhibits corresponding spatial differences, displaying significant interannual co-variability with regional heatwave features. During June, an intensified TP AHS is accompanied by a southward displacement of the upper-level westerly jet stream, driving differential thermal advection and deep subsidence over the NCP. Spatially overlapping with this aloft downwelling, the westward expansion of the western Pacific subtropical high (WPSH) aligns with a positive horizontal temperature advection center over the NCP, closely tied to an anomalous mid-to-lower tropospheric cyclonic circulation along the northern flank of the WPSH. During JA, the enhanced TP AHS corresponds to a northward jet stream shift and a northwestward relocation of the South Asian High (SAH), establishing anomalous upper-level convergence over the YRV. These multi-scale circulation configurations, together with a V-shaped isentropic downglide structure radiating from the TP, form a coherent dynamic framework driving persistent subsidence over the YRV. Additionally, these dynamically aligned subsidence zones feature prominent cloud reduction that enhances surface downward solar radiation and sensible heat flux. Through a land–atmosphere coupling loop, these localized diabatic feedbacks structurally match the enhanced intensity and persistence of regional extreme heatwaves. These findings demonstrate that the TP AHS exhibits a month-dependent configuration structurally coupled with sub-seasonal heatwaves across eastern China, co-varying with 35–51% of the extreme heatwave days. This provides process-level insights into the physical linkages between upstream TP thermal anomalies and downstream thermodynamic extremes.
This study assesses the relationships of ARs (ARs) and extratropical cyclones with relative thermal degree day (TDDrel) variability and precipitation across the Northern Hemisphere high latitudes (40–90°N). Using two AR detection frameworks together with independent cyclone and hydroclimatic datasets, we show that AR frequency explains a substantially larger fraction of TDDrel variability than cyclone frequency across most seasons. Enhanced AR occurrence is consistently associated with positive TDDrel anomalies and increased precipitation, highlighting the importance of poleward moisture transport for high-latitude thermodynamic and hydrological variability. The diagnosed relationships are sensitive to AR detection methodology. The Takahashi and Sakazaki framework produces stronger and more spatially coherent relationships between AR activity, TDDrel, and precipitation than the Guan and Waliser framework, demonstrating that AR detection methodology can substantially influence estimates of AR-related hydroclimatic impacts. The strongest relationships occur over North America and Eurasia, with weaker but still widespread responses across Arctic land regions. Although ARs and extratropical cyclones are dynamically connected, cyclone frequency exhibits a comparatively weaker and more spatially heterogeneous relationship with TDDrel, indicating that cyclone occurrence alone does not fully characterize the thermodynamic variability associated with organized poleward moisture transport. Overall, the results identify AR frequency as an important statistical predictor of high-latitude thermodynamic variability and emphasize the combined thermodynamic and hydrological influence of ARs. These historical relationships provide a basis for investigating how future changes in AR frequency, intensity, and persistence may affect high-latitude hydroclimate under continued Arctic warming.
The canonical relationship between the El Niño-Southern Oscillation (ENSO) and tropical cyclone (TC) activity in the western North Pacific (WNP) is well established, yet its stability across different ENSO temporal evolutions remains unclear. Using ERA5 reanalysis datasets, ERSSTv5, IBTrACS TC records (1965–2023), and the Dynamic Genesis Potential Index (DGPI), we demonstrate that this relationship substantially weakens during successive ENSO events. While Reversal ENSO events trigger coherent basin-wide cyclonic anomalies and significantly elevate JAS ACE, Isolated events produce a weaker but still positive TC response, successive events induce a unique cyclone–anticyclone dipole over the WNP. This dipole structure exerts opposing dynamical influences that effectively cancel each other out within the primary TC genesis region, resulting in a non-significant ACE response. Factor decomposition of the DGPI reveals that this cancellation is primarily driven by the competition between suppressed low-level vorticity and enhanced mid-tropospheric ascending motion. High-resolution climate projections from CMIP6/HighResMIP suggest a ∼30% increase in the frequency of successive ENSO events under future warming scenarios. Our findings imply that the increasing prevalence of Successive ENSO will reduce the seasonal predictability of WNP TCs, necessitating the integration of ENSO temporal typology into operational forecasting frameworks.
Heatwaves (HW) heavily affect several sectors of the economy and society in the European and Mediterranean regions, and particularly in the Po Valley area. Their frequency and intensity are projected to increase, and a thorough understanding of how this occurs in relation to rising global temperatures and to the projected changes in other weather and climate variables is fundamental for developing effective adaptation plans. In this study, developed within the Climate Intelligence (CLINT) project, we propose an application of the storylines technique, empowered by the definition of shared global thermodynamical conditions across the simulations (Common Temperature States, CTS) and by the implementation of an ML-based driver-oriented validation for the used CMIP6 models. This allows us to produce a set of physically sound storylines, where the increase of heatwave-related indices is inspected under specific conditions in the evolution of other drivers, researching proximate drivers and teleconnections. Our results highlight relevant contributions coming from precipitation depletion over three different Euro-Mediterranean domains (Northern Africa, Western and Eastern Europe), which is associated with enhanced sensible heat fluxes and co-occurs with the largest increases in the HW Magnitude Index and the number of HW days. Furthermore, the increase of these indices is found to be connected with different confidence levels to tropical modes such as El Niño Southern Oscillation and the convective activity in the Western Pacific, whose upper-level divergence triggers planetary wave trains. By contrast, the pressure gradients in the North Atlantic appear to be positively correlated with HW in Northern Italy, but are of secondary importance with respect to precipitation decrease. These findings demonstrate how the use of Machine Learning tools (here, the Spatio-Temporal Cluster-Optimised Feature Selection (STCO-FS) framework) can reinforce the storylines approach and advance scientific understanding of extreme events in a changing climate.
Extreme rainfall events (EREs, ≥ 80 mm day-1) in Taiwan were examined for their spatiotemporal characteristics from 1960 to 2022 using an event-based tracking method. Climatologically, Taiwan experiences approximately 11 EREs annually, with the highest frequency and magnitude occurring during the tropical cyclone (TC) season (July–October, 56%), followed by the early summer season (May–June, 33%). Spatially, EREs exhibit a west-to-east migration from spring to winter, corresponding to the monsoonal transition, with higher-magnitude events predominantly concentrated in the central and southern mountainous regions.EREs demonstrate significant variability on both interannual and secular timescales. Their interannual frequency is strongly modulated by ENSO-related sea surface temperature anomalies. During the summer of El Niño developing years, an anomalous cyclonic circulation in the western North Pacific coincides with an enhanced monsoon flow and higher frequency of landfalling TCs, thereby favoring more frequent EREs. In contrast, the subsequent anticyclonic circulation anomaly in autumn is accompanied by fewer ERE occurrences, with this reduction persisting into the summer of the following decaying years. The long-term evolution of EREs exhibits a significant upward trend in both frequency and intensity. This enhancement is closely associated with a deceleration in TC translation speeds at landfall and a higher proportion of westward-tracking storms making landfall in central Taiwan. Additionally, intensified TC-related ascent is linked to this trend.
The definition of spatio-temporal extreme coastal water level events along the coast is critical for flood hazard assessments. Traditional site-by-site analyses of extreme events at large scales (thousands of kilometers) often oversimplify spatial dependencies and neglect the temporal evolution of storms across space, leading to unreliable hazard estimates. In this study, we implement a spatio-temporal statistical model that integrates wave and storm surge contributions to estimate water level footprints and their associated flooding over large spatial scales. To translate these water levels into flood areas, we employ a hybrid calculation approach that combines physics-based hydrodynamic modelling with machine learning. Demonstrated in the Bay of Biscay (1230 km of shoreline), we reproduced historical events in both time and space and generated large synthetic databases of extreme events. The synthetic database enables the analysis of extreme event progression along the coast, and allows for a more spatially-coherent determination of flooded areas return levels by incorporating spatial dependencies. The results revealed notable differences between the event-based approach (coherent spatio-temporal phenomena) and the standard site-by-site analysis, particularly at larger spatial scales. In the Bay of Biscay, our comparison shows that spatially uniform 100-year return levels are substantially less frequent when spatial consistency is enforced. For atmospheric water levels, comprising storm surge and wave setup but excluding astronomical tide, the spatially uniform 100-year return-level footprint is expected to occur approximately once every 494 years when spatial dependence is considered. Separately, for flooded areas computed under a high-tide scenario, the spatially uniform 100-year flooded-area estimate corresponds to an occurrence of approximately 175 years. These findings highlight the importance of considering the spatio-temporal evolution of extremes to improve large-scale coastal flood hazard assessments and provide more spatially coherent hazard information to support coastal adaptation and resilience.
In this study, we present the first characterization of compound wind and precipitation extremes (CWPEs) over southeastern South America (SESA). The spatial and seasonal variability of individual and compound extreme occurrences was assessed using ERA5, ERA5-Land, MSWX, WFDE5 and NASA POWER datasets during 1981–2019. Wind extremes (WEs) over SESA occur mostly in spring and winter. Over eastern SESA, precipitation extremes (PEs) are frequent throughout the four seasons, while over northern and western SESA, their frequency is higher in summer. CWPEs are frequent over eastern SESA, where more than 40% of PEs are also CWPEs. Around 40% of CWPEs occur in spring. These results are congruent across the five datasets considered but some differences are observed over southern Brazil and the pre-Andean region. The mean synoptic conditions associated with widespread CWPE events over SESA are consistent with a low-level jet advecting moisture from lower latitudes toward SESA, favoring instability conditions that are also enhanced by the upper-level jet dynamics. We used Self-Organizing Maps to reduce and organize the synoptic environments associated with widespread CWPEs, and then Ward clustering to obtain four representative weather types (WTs). The resulting WTs significantly enhance CWPE, PE and WE occurrences in sectorized areas of SESA.
The occurrence of extreme cold events is expected to decrease with climate change but they will remain an issue to face for the society due to their impact on health, agriculture and infrastructure. Even if these events are less and less frequent, the surge of such event questions the citizens and disturb their perception of climate change. Using a very simple asymmetric indicator called TTA (Temperature Tail Asymmetry) adapted for cold and warm extremes analysis, we analyze such “outliers” or “historical” events compared to their “usual” extremes taken as a reference. We find that particularly in the mid-latitudes of the north hemisphere, historical spells of extreme cold are inherently less frequent than exceptional heat waves. This feature is disconnected from climate change, it is an intrinsic characteristic of the temperature fluctuations due to the property of the atmospheric general circulation, polar vortexes characteristics and geographical patterns.
Marine heatwaves (MHWs) in the Northwest Pacific (NWP) have intensified in recent decades, yet the mechanisms responsible for persistent upper-ocean warming remain poorly understood. Previous studies primarily focused on the record-breaking July 2022 surface MHW, whereas the warming in 2022 persisted throughout the year and extended well below the mixed layer. Here, I investigate the 2022 annual upper-ocean (0–100 m) MHW east of the Kamchatka Peninsula using the Estimating the Circulation and Climate of the Ocean technology involving the MITgcm adjoint framework. Adjoint sensitivity and perturbation analysis are used to identify the dominant forcing pathways. The results show that thermodynamic forcing, including near-surface turbulent heat fluxes and radiative fluxes, contributes primarily through local air–sea heat exchange on seasonal timescales. In contrast, zonal wind forcing exhibits sensitivity several years prior to the event, indicating the importance of ocean dynamical adjustment, subsurface heat storage, and ocean memory. Perturbation experiments further demonstrate that wind-induced responses extend throughout the upper ocean rather than remaining confined to the surface. Compared with Northeast Pacific MHWs, NWP MHWs exhibit stronger wind sensitivity and greater state dependence. These results suggest that persistent upper-ocean MHWs in the NWP are governed by a combination of thermodynamic forcing and delayed ocean dynamical processes. The findings highlight the importance of subsurface ocean memory for long-duration MHWs and demonstrate the value of adjoint methods for identifying their precursor pathways. The results should, however, be interpreted in light of the coarse model resolution and the linear assumptions inherent in the adjoint framework.
As climate extremes intensify globally, understanding the role of urban greening in shaping compound hot-dry events (CHDEs) is critical for designing climate-resilient cities. Here, we have integrated the maximum daily temperature with the monthly standardized precipitation evapotranspiration index (SPEI) to examine how urban vegetation modulates CHDE frequency and intensity across China from 2000 to 2024. Results show that the mean annual urban leaf area index (LAI) is 1.77, while CHDEs occurred on average 12 times per year. In contrast, CHDE intensity remained relatively stable during the study period, with an average annual value of 0.018. CHDE frequency and intensity were generally higher in northern cities. However, the strongest negative association between LAI and CHDEs was concentrated in central China, particularly for CHDE intensity. Precipitation predominantly controlled CHDEs frequency, whereas potential evapotranspiration were the primary drivers of intensity. Within the major urban agglomerations, the dominant drivers of CHDE frequency varied substantially, whereas CHDE intensity was more consistently dominated by temperature. These findings highlight the crucial role of urban greening in mitigating compound hot-dry risks and provide useful insights for developing climate-adaptive urban strategies under future warming.
Under ongoing global warming, record-breaking heavy snowfall occurred in Japan in February 2025. During that winter, the sea surface temperatures (SSTs) around Japan were markedly higher than the climatological mean. This study investigates the impacts of global warming and anomalously high SSTs on heavy snowfall in Japan in February 2025 using the pseudo global warming method and SST sensitivity experiments. Precipitation increases by 7.8%, while snowfall decreases by 2.0%, owing to past global warming over land areas below 100 m along the Sea of Japan coast. Further global warming, assuming 2 K and 4 K warmer conditions relative to the pre-industrial period, intensifies precipitation, whereas snowfall decreases by 9.3% and by a substantial 42.6%, respectively. In contrast, the high SST anomalies enhance precipitation and snowfall by 21.8% and 17.5%, respectively, over land areas below 100 m. The anomalously high SSTs not only increase atmospheric moisture but also enhance atmospheric instability, resulting in a marked increase in precipitation and snowfall. By comparison, global warming induces a slight atmospheric stabilization, which suppresses the precipitation increase expected under enhanced moisture and warming. The southeastern part of Japan's northern island, where record-breaking heavy snowfall occurred on February 3, 2025, also shows increased snowfall attributable to global warming and a high SST anomaly. In this region, the high SST anomaly strengthens the horizontal thermal contrast between land and the ocean, thereby enhancing the ascent of moist air along the boundary between inland areas and the ocean and resulting in further enhanced precipitation.
Air temperatures have risen worldwide as a consequence of ongoing climate warming, with particularly strong increases observed in Europe.This study presents a comprehensive assessment of long-term changes in air temperature extremes over Romania during 1901–2023 using the homogenized RoClimHom daily dataset from 156 meteorological stations. A suite of 18 ETCCDI indices was analysed to evaluate trends in percentile-based, threshold-based, absolute, and duration-related temperature extremes, and their relationship with large-scale atmospheric circulation. Results reveal a robust and spatially coherent warming signal across all categories of extremes, with a marked acceleration after the 1980s. Cold extremes declined significantly, with frost days (−26.8 days/century), ice days (−13.0 days/century), and cold spell duration (−29.3 days/century) showing pronounced decreases. In contrast, warm extremes intensified, including increases in tropical nights (+3.6 days/century), hot days ≥30 °C (+10.2 days/century), warm spell duration (+7.2 days/century), and growing season length (+21.5 days/century). Percentile indices confirm a shift from cold to warm conditions, with warm nights (TN90p) increasing by +9.5%/century and cold nights (TN10p) decreasing by −7.6%/century, highlighting asymmetric nighttime warming. Absolute indices show significant warming of annual minimum temperatures, while changes in annual maxima are weaker. Teleconnection analysis indicates that NAO strongly modulates cold-season extremes, whereas summer heat extremes are less directly controlled by large-scale circulation. These findings provide robust observational evidence of long-term climate warming in southeastern Europe and offer essential scientific support for national adaptation strategies.
Northern Hemisphere extreme snowfall changes are studied using the Large ENsemble TIme Slice (LENTIS) model. Extreme snowfall frequency and amount are projected to increase more strongly than light snowfall in several regions. The sign and magnitude of change depend on the regions present-day seasonal mean climatological temperature. While light and extreme snowfall are projected to decline significantly, particularly in maritime and mid-latitude regions near the melting point, increases are expected in high-latitude maritime, select high-altitude and continental areas. Compared to light snowfall, extreme snowfall only begins to decline at seasonal mean temperatures roughly 7 °C higher. Consequently, increases in light snowfall are limited to the coldest regions, whereas increases in extremes already occur in relatively warmer, yet still freezing regions. In sufficiently cold regions, warming enhances extreme event frequency (up to 278%) and amount (up to 271%) more strongly than light snowfall (up to 101% and 152%). Regions near the melting point are thermodynamically controlled due to climate warming, while colder regions are likely influenced by both thermodynamic warming and dynamical circulation changes, evident for Greenland. Extreme Greenland snowfall is found to be associated with a sea level pressure anomaly dipole between Greenland and Northern Europe, promoting warm moist Atlantic air advection. Using the Greenland Oscillation Index (GOI) – which measures the strength of the dipole – it is found that increases in extreme snowfall events are linked to a higher frequency of favorable circulation patterns with above-median GOI, particularly over Eastern, Central and Northern Greenland.
Atmospheric moisture transport plays a fundamental role in regulating hydroclimatic variability, yet how moisture sink regions reorganize between wet and dry years remains insufficiently quantified. Using the Water Accounting Model–2 layers (WAM2layers) driven by ERA5 reanalysis and combined with empirical orthogonal function (EOF) analysis for 1980–2023, we examined the evolution of moisture sinks in the Yellow River Basin through a centroid-based diagnostic framework. Results reveal that mean sink precipitation is concentrated over northern China with a northeastward extension into downwind continental and oceanic regions, with maxima reaching 0.3–0.5 mm day−1. Sink precipitation has weakened over recent decades at a rate of −0.002 mm day−1 yr−1. Moisture sink distributions are more concentrated during wet years, whereas dry years exhibit greater spatial dispersion (15.67° vs. 16.30°) and a broader downstream redistribution of sink activity. EOF analyses identify a dominant spatial reorganization mode linking the upper reaches of the Yellow River Basin with downstream sink regions, with dry years exhibiting a stronger northeastward expansion of sink activity. Enhanced atmospheric humidity and upward motion during wet years favor moisture convergence within the basin, whereas subsidence during dry years promotes downstream redistribution of sink activity. This study reveals that circulation-driven reorganizations of moisture transport patterns are key to understanding and predicting regional hydroclimatic variability.
The increasing frequency of extreme heat naturally raises the question of whether, and to what extent, anthropogenic factors are responsible. However, attribution results can vary across modeling frameworks, and the widely used risk ratio (RR) becomes highly uncertain for extremely rare events. Here we focus on comparing two probabilistic attribution frameworks—unconditional (CMIP6) and conditional (GAMIL3.0, the first SST-driven large-ensemble model developed by Chinese institutes)—applied to the 2022 and 2023 extreme heat across China. Under both frameworks, we find that anthropogenic forcing substantially increases the likelihood of extreme heat nationwide, especially in Northwestern China, Northern China, and the Yangtze River Valley. While the magnitude of estimated influence differs between the two frameworks, the lower bounds of RR confidence intervals are largely consistent; for example, anthropogenic forcing increases the likelihood of the 2022 extreme heat in Eastern China by at least a factor of 11. For extremely rare events where RR becomes unstable, we additionally employ two complementary metrics—the non-risk ratio (NRR) and probability of sufficient causation (PS)—which remain bounded and interpretable. Our study demonstrates that integrating multiple frameworks, supplemented by targeted metrics for rare extremes, yields a more comprehensive attribution assessment, as no single approach is universally optimal.
Tropical cyclones (TCs) are associated with extreme winds that can cause damage to structures, infrastructure systems, and buildings, resulting in economic losses and fatalities. Results of the forecasted TC tracks and TC wind hazards can be used by emergency preparedness planners to protect lives, properties, and reduce damage costs. In the present study, a model based on the Unet++ architecture was trained to detect and track TCs. The training was focused on the TC or typhoon activities that occurred in the Northwestern Pacific Basin, and carried out using the ERA5 reanalysis dataset from the European Center for Medium-Range Weather Forecasts (ECMWF). The comparison of the identified TC tracks to the best TC track dataset from the China Meteorological Administration indicates that the trained model has very good precision and recall, which are 87% and 96%, respectively. Also, the statistical characteristics of the landfalling TCs for those detected and tracked by the developed model match well those of the best TC track dataset. The use of the newly developed model to detect and forecast TCs that occurred in 2024 was carried out based on the global weather forecasts given by the Integrated Forecast System (IFS) model from ECMWF and the Global Forecast System (GFS) model from the National Centres for Environmental Prediction. A comparison indicates that the forecasted locations of the TC centre by using the developed model in conjunction with the weather forecasts given by the IFS model are better than or comparable to many of the forecasted TC tracks reported by several different agencies. Moreover, a simple and practical framework for forecasting the TC wind hazards based on the forecasted TC tracks by using the developed model, a developed empirical relation for estimating the central pressure, and the gradient wind field model was presented and illustrated.
On 15 November 2024, a large-scale coastal flooding event hit China coastal areas, highlighting the urgency of understanding tide-driven flood risks. Yet this type of tide-driven flooding has received little research attention across the China coastline. Here we aim to raise awareness of high-tide flooding (HTF) along the China coastline by quantifying the future potential of frequent HTF occurrence. Using a likelihood-based proxy for HTF flooding threshold and modeled water levels, we quantify how the combined effects of sea level rise (SLR) and amplified astronomical tides drive increases in future HTF exceedance frequency. Our results indicate a marked increase in HTF exceedance frequency in the coming decades, driven by SLR and the ascending phase of the nodal tidal cycle. Specifically, under the Shared Socioeconomic Pathway 2-4.5 (SSP2-4.5) scenario, HTF exceedance events are projected to occur twice weekly on average along China northern coast by the 2050s, significantly more frequently than along the southern coast. Chronic HTF exceedance will first emerge in Bohai Bay under the SSP2-4.5 scenario in the mid-2040s. We also demonstrate that future tidal amplification will not only increase the number of HTF exceedance days but also advance the onset of chronic HTF. Our results highlight the substantial projected rise in HTF occurrence potential along China coastline, emphasizing the need for greater attention to HTF in this region.
Air pollution, heatwaves, droughts, and wildfires are increasingly interacting to create compound extremes that amplify risks to ecosystems, human health, and societal resilience. Yet, their global prevalence, intensity, and cascading impacts remain poorly quantified. Here, we provide a global assessment (2003–2022) of compound climate–pollution events, benchmarking outcomes against the World Health Organization (WHO) Air Quality Guideline (AQG) for fine particulate matter (PM2.5). Using a harmonized framework at 0.75° × 0.75° spatial resolution that integrates ERA5 reanalysis, CAMS PM2.5, and MODIS fire radiative power, we define compound events as the simultaneous co-occurrence of two or more hazards within the same grid-cell and day, map hotspots defined as grid-cells exceeding the 90th percentile of yearly compound event frequency, quantify their frequencies, and evaluate the effect of compound events on fire intensity and pollution. We find that the co-occurrence of pollution with heatwaves, droughts, and fires is widespread, particularly across South America, Africa, Asia, boreal regions, and parts of Europe and Australia. During fire seasons, daily PM2.5 concentrations on compound heat–drought–fire–pollution (PHDF) days exceed pollution-only levels by more than 200% on average and up to 800% at the upper tail. These extremes consistently push air quality far beyond the WHO AQG, revealing a pronounced climate burden on clean air. Our findings highlight the global reach and intensity of compound climate–pollution events, emphasizing the urgent need for integrated strategies that jointly address air quality, wildfire risk, and adaptation to safeguard public health in a warming climate.
