Main

Climate change has been escalating the intensity and frequency of heatwaves, with detrimental effects on human health, agriculture and ecosystems1,2. Heat extremes, defined as single days with daily maximum temperature (Tmax) exceeding the 90th percentile in this study, often coincide with severe ozone (O3) pollution, as strong sunlight, little rainfall and stagnant air under high-pressure systems facilitate the accumulation of both air pollutants and heat3. Exposure to concurrent heatwaves and O3 extremes poses threats to human health, particularly for cities due to overlapping socioeconomic, demographic and climatic factors4.

O3 is predominantly formed via complex and nonlinear photochemical reactions of nitrogen oxides (NOx) and volatile organic compounds (VOCs)5. Heatwaves and O3 extremes are intricately linked and have been extensively studied based on surface observations6,7,8. During heatwaves, soaring temperatures and intense sunlight accelerate the chemical kinetic rates of photochemical O3 production9. High temperatures also exacerbate ground-level O3 by triggering a surge of biogenic volatile organic compound (BVOC) emissions from vegetation, particularly from highly temperature-sensitive herbaceous species in urban landscapes10. Furthermore, heatwaves drive soil drying and decrease the stomatal conductance of vegetation, thereby weakening dry deposition and worsening O3 pollution near the surface11.

Beyond the interconnections near the surface, long-lasting and widespread heatwave-associated O3 pollution usually extends into the upper air12. Existing studies demonstrate that O3 exhibits heterogeneous vertical profiles largely determined by multiple physical and chemical processes, such as stratospheric intrusion13, synoptic situations14 and planetary boundary layer (PBL) dynamics15, as well as nonlinear photochemistry16. Meanwhile, heatwaves feature a positive temperature anomaly spanning the troposphere with intensity varying greatly across altitudes17, which affects the magnitude of near-surface heat18 and, consequently, O3 levels. However, vertical insights into the photochemical stratification and intricate connection between heatwaves and O3 are relatively limited owing to sparse vertical observations, and a comprehensive mechanism understanding is imperative to cope effectively with such compound extremes.

Rapid urbanization and industrialization have led to substantially increased warming and O3 pollution across the Northern Hemisphere over recent decades19. With one-sixth of the world’s population and the fastest urbanization process globally20, China is among the regions most severely affected by compound heat–O3 extremes, especially in highly populated megacities. Unlike Europe and the United States where O3 levels have declined due to emission controls, O3 pollution in China deteriorates continuously, with warm-season daily maximum 8 h average (MDA8) O3 increasing by 1.4 ± 1.2 ppb per year during 2013–2022 (ref. 21). With a warming rate nearly twice the global average22 and expanding urban growth, China is exceptionally sensitive to climate extremes. As urbanization and climate warming intensify heatwaves and further exacerbate O3 pollution23, tackling these dual challenges of climatic extremes and air pollution is an urgent need.

In this study, we integrate satellite and ground observations, a specially designed airship campaign, and meteorology–chemistry coupled modeling to disentangle the underlying mechanisms of heatwave-associated O3 pollution from a vertical perspective. Furthermore, we assess the potential of targeted emission reductions to mitigate urban photochemical pollution in a warming climate, offering critical insights for future climate and air quality strategies tailored for cities in China and other regions facing similar dual challenges.

Heatwave-associated ozone pollution and vertically heterogeneous anomalies

In China, O3 pollution events with MDA8 O3 exceeding China’s Ambient Air Quality Standards Grade II of 160 μg m−3 occur frequently during the warm season (May–August), and most of them are linked to heat extremes. Observations from air quality monitoring networks over the past decade (2014–2023) illustrate a widespread O3 enhancement during heatwaves (Fig. 1a). The average MDA8 O3 concentration increases from 106.9 μg m−3 on normal days to 131.8 μg m−3 during heatwaves. The proportion of sites with O3 exceedances more than doubled, rising from 11.5% to 26.6% during heatwaves. These positive O3 anomalies are particularly evident in densely populated and industrialized city clusters such as the North China Plain (NCP), the Yangtze River Delta (YRD), the Pearl River Delta (PRD) and the Sichuan Basin (SCB) (Fig. 1a). Among these regions, YRD experienced the most prominent O3 increment, reaching 37.6 μg m−3. The co-occurrence frequency of warm-season O3 and heatwaves (defined as the days with both Tmax and MDA8 O3 exceedances) in these highly populated city clusters has surpassed 4.2 days per year over the past decade (Fig. 1a), increasingly endangering city dwellers and the urban environment.

Fig. 1: Heatwave-associated O3 and NO2 anomalies in main city clusters of China.
Fig. 1: Heatwave-associated O3 and NO2 anomalies in main city clusters of China.The alternative text for this image may have been generated using AI.
Full size image

a, The difference in MDA8 O3 concentrations between heatwave days and normal days (red dots, sized by the frequency of concurrent heatwave and O3 pollution) during May–August from 2014 to 2023. The shaded gray area shows the gridded population density and three types of grid cells classified by the degree of urbanization50. b, The difference in tropospheric NO2 column densities from satellite retrievals (shaded contour) and surface NO2 concentrations in the afternoon (12:00–16:00 LT) at sites with O3 anomalies greater than 30 µg m−3 (dots) between heatwave days and normal days during May–August from 2014 to 2023.

Source data

NOx is a vital precursor of urban O3 formation and originates mainly from industry, transportation and power generation in cities24. During heatwaves, NOx also exhibits notable anomalies characterized by substantial vertical contrasts (Fig. 1b). The satellite-retrieved tropospheric NO2 column densities show an obvious increase, possibly due to elevated NOₓ emissions from soil and industrial sources25,26. In contrast, surface NO2 concentrations show a negative anomaly, resulting in a seemingly conflicting shift of NOx that is particularly evident in the YRD megacity cluster. Such vertical disparities provide new insights into the linkages between heatwaves and vertical stratification of photochemical ozone pollution. However, current measurements on photochemically active species are predominantly ground-based rather than for the urban boundary layer, which fail to grasp the full scope of the underlying processes involved in compound heat and O3 extremes.

To comprehensively understand the physical and chemical mechanisms of compound heat and O3 extremes, we conducted an intensive airship field campaign near the Station for Observing Regional Processes of the Earth System (SORPES) in Nanjing, a megacity in YRD with frequent heatwaves and O3 pollution, in the early summer (May–June) of 2023 (Extended Data Fig. 1a). We deployed in situ instruments mounted on a rack underneath the airship to provide highly time-resolved observations of O3 and its precursors (NOx and VOCs), as well as key meteorological parameters (Methods and Extended Data Fig. 1b). At the same time, real-time surface measurements with a wide range of photochemistry-related species and meteorological parameters were also conducted at SORPES. In total, 197 vertical profiles were recorded from the surface to a maximum altitude of 1,200 m during the 2 month campaign experiment. On the basis of these measurements, O3 pollution did show a tight link with high temperatures (Extended Data Fig. 1c). This experiment thus provides vertical insights into the photochemical and boundary layer processes involved in O3 pollution in a typical urban environment.

As there is less cloud cover and soil evapotranspiration during heatwave days, more heat is released into the air, reinforcing the high air temperature and convective turbulence27,28,29 (Extended Data Fig. 2). Therefore, the urban boundary layer usually fully develops during heatwaves, which is well reflected by anomalously high planetary boundary layer height (PBLH) and strong turbulent mixing (Fig. 2a). Accordingly, during heatwave days, airship-measured NOx vertical profiles show quite distinct daytime evolution, which feature a notable increase (2.91 ppb) aloft, while a distinct decline (–3.89 ppb) near the surface (Fig. 2b). This vertical heterogeneity of NOx agrees well with the long-term records in Fig. 1b. Such pronounced vertical changes in NOx concentration result from the fact that heat-enhanced convection fuels turbulent mixing and the uplift of anthropogenic emissions, which even outweigh NOx emission increases from industrial sources or soil volatilization owing to high temperatures25,26. This is further supported by a relatively lower surface NOₓ concentration observed during heatwave afternoons (that is, 4.17 ppb), compared with 7.13 ppb on warm-season normal days (Extended Data Fig. 3).

Fig. 2: Responses of urban boundary layer, O3 and NOx to heatwaves.
Fig. 2: Responses of urban boundary layer, O3 and NOx to heatwaves.The alternative text for this image may have been generated using AI.
Full size image

a, The averaged backscattering coefficient (Bscat. coeff.) and PBLH from the ceilometer on normal days and heatwave days at the SORPES station in May–June 2023. b, Daytime evolution in O3 and NOx profiles from the morning (7:00–9:00 LT) to afternoon (12:00–16:00 LT). The solid lines represent mean values and shaded areas show the 25–75th percentiles. c, An isopleth diagram of daily maximum O3 against daytime NOx concentration and VOC reactivity. The circles indicate the mean value of NOx concentration and VOC reactivity at the surface on normal days (blue circle) and heatwave days (red circle). The squares represent the mean value of NOx concentration and VOC reactivity near 1,000 m and the error bars show ± 50% s.d. of VOC reactivity. There are 27 observation points for VOC observation near 1,000 m.

Source data

On the other hand, the O3 concentration rises at both the surface and aloft during heatwaves (Fig. 2b). That is, O3 shows a synchronous increase with NOx in the upper air but changes oppositely near the surface. Here, the altitude-dependent O3 response to NOx was mainly attributed to the vertical heterogeneity of VOC concentrations relative to NOx (Extended Data Fig. 4a,b), which governs the regime of O3 photochemical formation. As shown in Fig. 2c, the photochemical O3 regime gradually shifts from VOC-sensitive to NOx-sensitive with altitude owing to the vertical stratification of NOx and VOCs16,30. In the NOx-sensitive upper boundary layer, heatwaves supplement NOx by stronger convection and fuel O3 photochemical production. Near the surface where photochemical production of O3 is predominantly VOC-sensitive, heatwaves elevate BVOC emissions and the ambient isoprene level (from 1.4 to 2.9 ppb; Extended Data Fig. 4c), while also stimulating the vertical dispersion of near-surface NOx (from –5.9 ppb to –12.0 ppb; Extended Data Fig. 3a), contributing to a notable increment in daytime O3 chemical production (from 5.5 to 10.6 ppb h−1). Overall, the airship measurements indicate that, by reallocating precursors vertically, heatwaves shift the stratification of photochemical regime and facilitate O3 formation both near the surface and in the upper boundary layer, jointly exacerbating O3 pollution.

Vertical stratification of photochemical responses to heatwaves

To quantitatively understand the key physical and chemical mechanisms in heatwave-enhanced O3 pollution, coupled meteorology–chemistry simulations with process diagnosis were performed using a single-column model (SCM) version of Weather Research and Forecasting model coupled with Chemistry (WRF–Chem) (Methods). Two groups of simulations (each including three parallel experiments) were conducted, driven by heatwave and normal meteorological conditions, respectively. As shown in Extended Data Fig. 5, the simulations are capable of capturing the meteorological differences between heatwave and normal days. The modeling results indicate that heatwaves elevate the vertical O3 levels by reinforcing photochemical production, which can further be disentangled into three key pathways. Specifically, heatwaves favor O3 formation directly through accelerating photochemical reaction rates under higher temperatures and radiation (PhotChem_Rate) and indirectly by boosting biogenic VOC emissions (Emis_BVOC). In addition, consistent with the airship observation, heatwaves tend to vertically redistribute the precursor profiles, particularly NOx (Met_NOxVmix), which modulates O3 photochemical production. The magnitude and relative contribution of three chemical pathways vary greatly with altitude owing to the vertical stratification of O3 photochemical regime (Fig. 3a). Near the surface (<100 m), precursor redistribution and biogenic emissions have comparable contributions, accounting for 46.7% and 48.8% of enhanced O3 production, respectively. However, in the middle boundary layer, the accelerated photochemical reaction becomes increasingly pronounced, peaking at nearly 31.9% around 600 m. At higher altitudes (1,000–1,500 m), during heatwaves, supplemental NOx from the uplift of near-surface emissions dominates the enhancement in O3 chemical formation (60.5%). The enhanced near-surface O3 chemical production, together with vertical transport, jointly drives O3 elevation at higher altitudes where precursors are limited. Such elevated O3 aloft could be kept in the residual layer overnight with negligible chemical and deposition loss, exacerbating surface O3 pollution after the residual layer breaks up the following morning15.

Fig. 3: Heatwave-induced O3 anomalies through different pathways and the responses to anthropogenic NOx emission reductions.
Fig. 3: Heatwave-induced O3 anomalies through different pathways and the responses to anthropogenic NOx emission reductions.The alternative text for this image may have been generated using AI.
Full size image

a, Diagnostic process contributions to heatwave-induced daytime (10:00–16:00 LT) O3 changes in Nanjing simulated by SCM. Met_O3Vmix represents the reinforced turbulence mixing of O3 owing to heatwaves; PhotChem_Rate, Emis_BVOC and Met_NOxVmix represent heatwave-enhanced chemical production of O3 from accelerating photochemical reaction rates, boosting biogenic emissions and redistributing anthropogenic precursors, respectively. b, The response of surface daytime O3 enhancement (solid lines) and chemical formation contributions (Chem_O3, dotted lines) (<100 m) to reductions in anthropogenic NOx emission. The solid and hollow circles indicate the tipping points where O3 mitigation and photochemical regime transition, respectively. The red and blue lines represent the heatwave and normal scenarios, respectively. The gray bars show the difference between the two scenarios, representing heatwave-induced O3 changes, with positive values indicating heatwave penalties and negative values signifying heatwave rewards.

Source data

Given the vital role of NOx precursors in both the surface and upper air, anthropogenic emissions could influence urban O3 pollution by altering photochemical regimes. Under high ambient NOx levels in megacities of China (for example, Nanjing), initial NOx reductions would enhance O3 chemical production near the surface and decrease it aloft owing to the aforementioned vertical disparities of the photochemical sensitivity regime. As emission reductions increase, both O3 production at the surface and aloft would be NOx-sensitive, where limited NOx availability will sharply reduce O3 levels (Fig. 3b and Extended Data Fig. 6).

Heatwaves help remove surface NOx and supplement upper NOx, thereby advancing the tipping point of surface photochemical regime transition and O3 mitigation by 10% in terms of NOx reductions (Fig. 3b). This suggests that while heatwaves exacerbate O3 pollution (heatwave penalty) due to unfavorable meteorology, short-term emission control according to the weather forecast would be expeditious to mitigate O3 extremes. In the long term, with the surface transitioning toward NOx-sensitive, this penalty is expected to even reverse into O3 reduction as heatwave-enhanced dispersion outweighs its adverse effect on O3 formation. When NOx emissions decrease by 80%, heatwaves lower surface O3 by 12.9% compared with normal days. In the upper urban boundary layer, the heatwave penalty also diminishes as reduced upward mixing of O3 and NOx limits O3 production and accumulation (Extended Data Fig. 6c). Consequently, O3 pollution gets progressively alleviated, indicating that urban emissions control can reduce the likelihood that heatwaves develop into compound climate and air pollution extremes.

Mitigating compound heat–ozone extremes in a warming climate

In the past decade, China has steadily reduced anthropogenic NOx emissions by enforcing two phases (2013–2017 and 2018–2020) of clean air actions31. Although it has partly altered the surface O3 photochemical regime, the VOC-sensitive regime and photochemical stratification remain widespread in megacities32,33 (Extended Data Fig. 7a). The health and environmental burdens triggered by heatwave–O3 associations thus threaten the climate resilience of city clusters, particularly amid accelerated climate change, urbanization and population aging.

Under the medium greenhouse gas emission scenario (Shared Socioeconomic Pathway (SSP) 2-4.5), which approximates current climate commitments, the surface daily maximum temperatures in China are projected to increase by over 1.5 °C and urban areas to expand by 1.4 × 104 km2 (12.1%) in the coming decades (Fig. 4a), causing an escalation of urban heatwave extremes and the associated O3 pollution. If applying lax emission policies without additional climate or air quality target constraints (2060 Baseline), China’s major city clusters (that is, NCP, YRD, SCB and PRD) remain hotspots subject to concurrent heatwave and O3 pollution, with the co-occurrence frequency during warm season rising to 9.5 days per year, 15.3 days per year, 8.4 days per year and 3.5 days per year by 2060, respectively, highlighting an urgent need for mitigating such compound risks (Fig. 4c). To combat climate change, China has committed to peak carbon dioxide emissions before 2030 and achieve carbon neutrality before 206034. As cities account for more than 70% of carbon emissions and energy consumption, a transition to low-carbon energy and sustainable cities is a key solution35, which will gain synergistic reductions in urban pollutant emissions given the shared emission sources36 (Extended Data Fig. 8). Compared with the Baseline scenario, China’s NOx emissions are estimated to reduce by around 87.1% by 2060 owing to carbon neutrality and air pollution control policies (2060 on-time peak-net zero-clean air) (Fig. 4a).

Fig. 4: Future co-occurrence of heatwave and O3 extremes during the warm season.
Fig. 4: Future co-occurrence of heatwave and O3 extremes during the warm season.The alternative text for this image may have been generated using AI.
Full size image

a, Trends in China’s NOx emissions under the baseline scenario and on-time peak-net zero-clean air scenario, along with the daily maximum temperature (red line) and urban areas (blue line) under the SSP 2-4.5 scenario from 2020 to 2060. The daily maximum temperature is presented as mean ± 50% s.d. of 30 global climate models. b, Warm-season NO2 concentrations and MDA8 O3 levels under heatwaves across four city clusters (NCP, YRD, SCB and PRD) under the 2060 baseline and on-time peak-net zero-clean air scenarios. The bars show the mean concentrations across city clusters from a 4 month (123 days) simulation. The circles show the median values and the error bars present the 10th and 90th percentiles. The dashed gray lines represent the WHO Interim Target 1 and AQG. c, The spatial distribution of heatwaves and O3 pollution co-occurrence frequency (days per year) during May–August under the 2060 baseline scenario in China and the comparison of co-occurrence frequencies between the recent decade and 2060. The gray and purple bars indicate the impacts of climate change and emission control, respectively.

Source data

On the basis of meteorology–chemistry modeling, it is estimated that the steep emission reductions owing to carbon neutrality policies could lower the warm-season NO2 concentrations by 82.6–85.9% compared with the baseline scenario in 2060, falling below the World Health Organization (WHO) air quality guideline (AQG) level of 10 µg m−3 (Fig. 4b). As mentioned previously, this anthropogenic emissions reduction will shift O3 photochemical regimes in megacities toward NOx sensitivity and narrow its vertical disparities, enabling the further alleviation of heatwave-associated O3 pollution by reducing photochemical production (Extended Data Fig. 7b–d). Quantitatively, the MDA8 O3 levels under heatwaves could decrease by 50.3 µg m−3 (40.9%), 59.2 µg m−3 (47.4%), 53.4 µg m−3 (43.3%) and 33.2 µg m−3 (43.2%) in the four city clusters (NCP, YRD, SCB and PRD), respectively, where the NCP, YRD and SCB regions will meet the WHO Interim Target 1 of 100 µg m−3 and PRD will achieve stricter AQG targets of 60 µg m−3. Although anthropogenic VOC (AVOC) emissions may also decrease via future end-of-pipe technologies and energy transitions, further analysis shows that NOₓ reductions play a dominant role in O3 pollution mitigation, accounting for 74.2–89.9% of the total reduction (Extended Data Fig. 8e,f). Consequently, the frequency of heatwaves and O3 pollution co-occurrence during the warm season is expected to drop sharply to less than 0.1 days per year by 2060, despite plausible increase in heatwave frequency by more than 19.0 days per year. These results highlight the importance of urban emission control in mitigating compound heat–air pollution extremes.

Discussion

Our findings reveal that heatwaves tend to alter the vertical heterogeneity of precursors (that is, NOx) as well as the photochemical regime, thus exacerbating O3 pollution through more effective chemical production both at the surface and aloft. Despite intensifying heatwaves, anthropogenic emissions can influence whether a meteorological extreme evolves into a compound air pollution hazard, especially in populated cities. In the short term, strategic emission reductions based on near real-time weather and air quality forecasts (for example, targeting more reduction under heatwave extremes) could alleviate O3 pollution in a cost-effective way. In the long run, emission reductions entailed in China’s carbon neutrality pledge could substantially mitigate heatwaves and O3 pollution co-occurrence, even under a warming climate.

Our results are primarily based on in situ measurements and meteorology–chemistry model simulations, yet several heat-related aspects warrant further investigation. First, the single-day heatwave definition adopted here serves as a practical compromise between study motivation and the limitations of vertical observations. It reasonably captures daytime heat anomalies that drive turbulence and photochemistry, aligning with our focus on the subdiurnal meteorology–chemistry coupling processes. Second, AVOCs are also sensitive to temperature37 and may contribute to O3 formation38. On the basis of the observation analysis, we found that AVOCs accumulate more at night but do not exhibit a noticeable daytime increase (Extended Data Fig. 4d). However, vertical in situ measurements of VOCs are challenging and their role in heatwave and O3 extremes still needs further exploration. In addition, beyond the inherent complexity of chemical and physical processes in compound heat–O3 extremes, their intricate feedback with climate policies and socioeconomic activities adds further complexity24. For instance, dynamic emissions owing to heat-induced anthropogenic activity changes (that is, restricted outdoor labor and increasing energy demand) are not represented in the static inventories, possibly introducing uncertainties in this assessment.

Overall, the intricate interplay between meteorological, chemical and social processes underscores the imperative of understanding the extent to which humans can determine the escalation of weather extremes into compound extremes. While our findings are rooted in China’s megacities, similar challenges confront other rapidly urbanizing regions such as the global south, where reliance on fossil fuels and inadequate energy transition technologies may exacerbate future NOx emissions and compound risks39. This study therefore provides a valuable preview and model for emerging cities to recognize and address the foreseeable challenges of climate change and air pollution, enhancing their climate resilience and sustainability in a warming future.

Methods

Observational and reanalysis data

Ground-based observations of NO2 and O3 at more than 1,500 stations for the past 10 years (2014–2023) are archived at the air monitoring data center of the Ministry of Ecology and Environment of China, which were collected to analyze the spatial distribution characteristics. Tropospheric NO2 and formaldehyde (HCHO) column density with a pixel size of approximately 13 × 24 km2 were retrieved from the ozone monitoring instrument mounted on the Aura satellite40.

We used hourly 2 m air temperature with a spatial resolution of 0.25° × 0.25° from the fifth-generation European Centre for Medium-Range Weather Forecasts reanalysis dataset (ERA5) to calculate the daily maximum temperature (Tmax)41. The data for 1969–2019 were used to calculate the present heatwave threshold, and data for 2014–2023 were paired with station observations of O3 to determine heatwave–O3 co-occurrences. Surface downward shortwave radiation flux, surface latent heat flux, 2 m air temperature and PBLH from the ERA5 were also used in the analysis.

Maximum near-surface air temperatures for 2020–2060 under scenario SSP 2-4.5 were obtained from the NASA Earth Exchange Global Daily Downscaled Projections (NEX-GDDP CMIP6) dataset42 and used to identify future heatwaves. This dataset provides bias-corrected statistically downscaled data at 0.25° horizontal resolution from Coupled Model Intercomparison Project Phase 6 (CMIP6) as procurable from 35 global climate models. Considering the availability of the variable (maximum near-surface air temperatures) and scenario (SSP 2-4.5), 30 global climate models (Supplementary Table 1) were selected.

There is no universal definition for heatwaves yet. In this study, we defined a heatwave day using the 90th percentile of Tmax, which is calculated using a 15 day moving window centered on each calendar day, based on 1969–2019 ERA5 data for the present and 2020–2060 CMIP6 projections for the future. This relative standard accounts for high spatiotemporal temperature variability and better reflects extreme heatwaves under future climatic conditions. It should be noted that the threshold criterion is a single day, rather than consecutive days, as we want a reasonable sample size with limited field observations to ensure a more robust analysis, as shown in Supplementary Fig. 1. Therefore, heatwave–O3 co-occurrence is a day in which Tmax exceeds its 90th threshold and MDA8 O3 concentration exceeds 160 µg m−3.

Airship field campaign

The airship field campaign was part of a multiplatform observation study of complex air pollution in the megacity cluster of eastern China, aiming to investigate the underlying causes of urban O3 pollution during the warm season. The experiment was conducted at the Xianlin Campus of Nanjing University (32° 06′ N, 118° 57′ E) from 7 May 2023 to 28 June 2023. The airship platform consists of a tethered mega-balloon and customized rack for instrument deployment. It can record a vertical profile from surface to ceiling altitude of 1,200 m within 20–30 min as well as hover at a specified altitude, which allows for flexible vertical measurements and high temporal resolution. Owing to limitations such as precipitation, fierce wind and aviation control, we were unable to perform the vertical measurements at all times and ended up measuring 197 vertical profiles.

The airship deployed multiple instruments to record O3, VOCs, reactive nitrogen species and other key pollutants. Meteorological parameters and gas species (O3, NO and NO2) were measured with a temporal resolution of 1 min. Ambient NO and NO2 were measured by an NOx analyzer (Thermo Scientific, Model 42i Trace Level) via the gas-phase chemiluminescence method. O3 was measured by a 2B Technologies Personal Ozone Monitor and O3 analyzer (Thermo Scientific, Model 49i) at different periods. Meteorological parameters, including relative humidity, wind speed, wind direction, pressure and air temperature were recorded by an Integrated CO2 and H2O Open-Path Gas Analyzer and 3-D Sonic Anemometer (IRGASON). In addition, high-frequency measurements (time resolution of 5 min) of VOCs and their oxidation products were conducted from 12 June to 16 June using the Vocus proton-transfer reaction time-of-flight mass spectrometer (Vocus PTR-ToF-MS, 2R), which improves the detection efficiency of product ions compared with conventional PTR instruments and is sensitive to a large range of oxygenated VOCs.

Surface measurements of aerosols, trace gases, VOCs and meteorological parameters were conducted synchronously at the SORPES station, which is located near the airship platform and is a representative urban site in the YRD region. Trace gases (SO2, O3, NOx and CO) were continuously measured using a Thermo TEI 43i, TEI 49i, TEI 42i or TEI 48i, respectively. Meteorological parameters were measured at a height of 4 m from the meteorological tower and a ceilometer (CL51, Vaisala) was used to provide backscattered signals and PBLH. Nonmethane hydrocarbons with 56 species and C2–C6 oxygenated VOCs were measured by a gas chromatograph–mass spectrometer/flame ionization detector (GC–MS/FID) system coupled with a cryogen-free preconcentration device. The time resolution was 1 h, and ambient air was sampled during the first 5 min of each hour for both channels. The uncertainties for VOCs measurements by the GC–MS/FID system are estimated to be 15–20%.

Observation-based model

A zero-dimensional observation-based box model (the framework for 0-D atmospheric modeling; F0AM) was used to simulate O3 sensitivity43. This model adopts the gas-phase chemical mechanism Master Chemical Mechanism (MCM v3.3.1) that explicitly describes the degradation pathways of 143 primary VOCs, more than 5,900 species and 17,000 reactions, and also incorporates physical processes, including solar evolution, planetary boundary layer evolution, dry deposition and dilution44. In this study, averaged surface measurements of meteorological parameters (air temperature, relative humidity, pressure and boundary layer height) and chemical species (O3, CO, SO2, NO, NO2, HONO and VOCs) obtained at the SORPES station during May–June 2023 were used to constrain the simulation and serve as the base scenario. The major VOCs observed by GC–MS/FID are presented in Supplementary Table 2. CH4 and H2 mixing ratios were assumed to be 1.9 ppm and 550 ppb, respectively. The photolysis frequencies (J values) were calculated by the National Center for Atmospheric Research tropospheric ultraviolet and visible radiation transfer model, which considers the influence of solar zenith angle, elevation, ground albedo and vertical O3 column. We further scaled the J values according to the observed cloud cover and the observed NO/NO2 ratio. Before running the base scenario, we conducted a 3 day model spin-up to initialize the unmeasured compounds. Then we proportionately increased or decreased the VOCs and NOx concentrations from the base scenario to cover the practical VOCs and NOx levels. Meanwhile, other conditions, including photolysis frequencies, temperature and relative humidity, were kept the same as the base scenario. The model results were interpolated into an isopleth for illustrating the impact of changes in NOx and VOCs on the daily maximum O3 concentrations, also known as the empirical kinetic modeling approach isopleth diagram. The performance of the base scenario is shown in Supplementary Fig. 2.

The O3 photochemical regimes on normal days at the surface and around 1,000 m altitude were determined using averaged NOx and the VOC observations from the SORPES station and airship platform. Owing to the limited availability of vertical VOC measurements, the direct calculation of the OH reactivity of VOCs, which is defined as the concentrations of VOCs multiplied by their respective reaction rate coefficients with OH, was not robust. Therefore, we utilized all VOC observations (excluding data with wind speeds greater than 5 m s−1) to characterize the vertical distribution of VOCs and derived the surface OH reactivity of VOCs (RVOCs_surface) from SORPES observations. The OH reactivity of VOCs near 1 km (\({R}_{\mathrm{VOCs}}{\rm{\_}}1\,{\rm{km}}\)) was then extrapolated as follows:

$$\frac{{R}_{\mathrm{VOCs}}{{\_}}\,1{\rm{km}}}{{R}_{\mathrm{VOCs}}\_{\rm{surface}}}=\frac{{C}_{\mathrm{VOCs}}{{\_}}\,1{\rm{km}}\_{\rm{Vocus}}}{{C}_{\mathrm{VOCs}}\_{\rm{surface}}\_{\rm{Vocus}}}$$

Here CVOCs_surface_Vocus and CVOCs_1km_Vocus represent the sum of BVOC (isoprene) and oxygenated VOC (that is, MVK, MACR and methylglyoxal) concentrations at the surface and around the altitude of 1,000 m measured by Vocus, respectively. Alkanes were not included here because they are a comparably minor and relatively unreactive fraction of total VOC, and the PTR method used by Vocus is not sensitive to alkanes.

The VOC reactivity and NOx concentrations on heatwave days were obtained by adding heatwave-induced changes to the results for normal days. Heatwave-induced changes in VOCs and NOx were estimated from the difference between the sensitivity experiments of the single-column model (see ‘Meteorology–chemistry coupled simulations’ section). This approach eliminates the uncertainties caused by daily variations in airship sampling heights and times between heatwave days and normal days. Although this derivation carries some uncertainties, it provides a feasible way to qualitatively assess O3 sensitivity in the absence of sufficient observational data.

Meteorology–chemistry coupled simulations

The WRF–Chem model (version 3.9.1) was applied to identify the major pathways of heatwave-induced O3 exacerbation and assess the change of O3 and heatwave co-occurrence under future scenarios. This model considers a variety of physical and chemical processes, such as the emission and deposition of pollutants, advection and diffusion, gaseous and aqueous chemical transformation, aerosol chemistry and dynamics45. In addition, it embeds a diagnostic analysis tool to calculate the variations of O3 concentrations induced by individual physical and chemical processes at each integration step, which have been successfully adopted in our previous work46.

The major pathways of heatwave-induced O3 changes were quantified using the SCM, which is a one-dimensional, idealized model including all physical and chemical processes in the regional version of WRF–Chem except for horizontal transport. Without the interference of the atmospheric three-dimensional dynamics, it is a key tool for understanding vertical physical and dynamic processes (land–atmospheric interaction, turbulence in PBL and so on) that drive heatwave events47. It provides a computationally efficient way to perform numerous numerical experiments and can be initialized by long-term observations at specific locations. Here, we designed six parallel numerical experiments to quantify the contribution of different processes on heatwave-induced O3 exacerbation, with the first three affiliated with the HW (heatwave) group and the last three affiliated with the NM (normal) group (Supplementary Table 3).

For the HW group of simulations, averaged radiosonde measurements at 08:00 LT during the heatwave days in Nanjing were used as the initial meteorological fields to drive the model. HW_CTRL is the control simulation of this group, in which the impact of vertical turbulent mixing and biogenic emissions was excluded. On this basis, HW_Vmix incorporated the vertical mixing process and HW_Vmix_BVOC went a step further by considering biogenic emissions. The experiment designs of the NM group correspond individually to the HW group, but use averaged soundings from normal days to initialize the model. All experiments included anthropogenic emissions from the Multi-resolution Inventory for China, which were assumed constant without considering the heatwave-induced emission enhancement (for example, power plant emissions). More details about model configurations are provided in Supplementary Table 4.

By comparing the HW and NM group simulations with the aid of the diagnostic analysis tool, we calculated the O3 change driven by vertical mixing (Met_O3Vmix = Vmix_HW_Vmix_BVOC − Vmix_NM_Vmix_BVOC) and chemical production processes (Met_O3Chem = Chem_HW_Vmix_BVOC − Chem_NM_Vmix_BVOC) in response to heatwave-induced meteorological anomalies. Enhanced chemical formation was subdivided into three crucial pathways: accelerated photochemical reaction rate due to high temperature and strong radiation (PhotChem_Rate = Chem_HW_CTRL − Chem_NM_CTRL), vertical redistribution of anthropogenic precursors by reinforced turbulence (Met_NOxVmix = Chem_HW_Vmix − Chem_NM_Vmix − PhotChem_Rate) and boosted BVOC emissions (Emis_BVOC = Chem_HW_Vmix_BVOC − Chem_NM_Vmix_BVOC − Met_NOxVmix).

We utilized WRF–Chem modeling to perform a set of simulations to evaluate the heatwave and O3 extremes and their co-occurrences in 2060 under different emission scenarios. The model domain covered all of China and its surrounding areas with a grid resolution of 27 km and 30 vertical layers. The simulations spanned the warm season (May–August) of 2060, with each run covering 5 days, and the chemical outputs from the preceding run were used as the initial conditions for the following one. The initial and boundary meteorological conditions were derived from the bias-corrected large-scale forcing climate model simulations of 2060 under scenarios from CMIP6 SSP 2-4.5, with a horizontal grid spacing of (1.25° × 1.25°) at 6 h intervals48. Key physical parameterizations included the Noah land surface scheme, the RRTMG radiation scheme and the Yonsei University PBL scheme, whose combination allow more reliable and plausible descriptions of land–atmosphere interactions and vertical turbulent mixing processes in the PBL, enabling better simulation of surface temperatures and pollutants dispersion. For the chemical mechanisms, the carbon-bond photochemical mechanism combined with the Model for Simulating Aerosol Interactions and Chemistry aerosol module were utilized.

Natural emissions in both the single-column model and regional model were calculated by the Model of Emissions of Gases and Aerosols from Nature (MEGAN)49, which estimates biogenic emissions online based on static parameters (for example, fraction of plant function type and leaf area index) provided by MODIS observations and real-time meteorological conditions (for example, temperature, solar radiation and soil moisture). This tight coupling between emission calculations and meteorological drivers makes the MEGAN well capable of capturing the temperature dependence of BVOC emissions7. It also calculates NOx emissions from the soil nitrogen pool using a temperature-dependent rate. Given that soil NOx emission rates depend not only on temperature but also other variables influencing nitrification/denitrification that vary with temperature (especially soil moisture), such simplification may underestimate the soil NOx emissions.

Future anthropogenic emissions in 2060 were derived from an integrated assessment model GCAM (China-focused Global Change Assessment Model) and a technology-based emission projection model DPEC (the dynamic projection model for emissions in China)36. This model provides five scenario-specific air pollutants and CO2 emissions for China, which considers the growing energy demand and low-carbon policy in power and industry sectors. We selected the baseline scenario and the on-time peak-net zero-clean air emissions scenario for further analysis. The two scenarios share the same socioeconomic development roadmap, but the 2060 baseline scenario does not consider additional climate or air quality target constraints. By contrast, the on-time peak-net zero-clean air scenario deploys a range of carbon reduction measures during 2020–2030, a more ambitious low-carbon transition after 2030 and the optimal end-of-pipe pollution control during 2020–2060 to achieve carbon neutrality and air quality improvement. Such strict measures could reduce anthropogenic NOx and non-methane volatile organic compound emissions by 64% and 87%, respectively. Therefore, the difference between the on-time peak-net zero-clean air and baseline represents the impact of anthropogenic endeavors on heatwave–O3 co-occurrence extremes. On this basis, the only NOx cut scenario reduces NOx emissions to the on-time peak-net zero-clean air level while maintaining other emissions consistent with the baseline to isolate the contribution of anthropogenic NOx control.

We validated the performance of the WRF–Chem model in capturing heat and O3 pollution extremes using available meteorology and air quality observations. As illustrated in Supplementary Figs. 3 and 4, the baseline simulation during the 2023 airship campaign shows good performance in capturing the magnitude and hourly variation of meteorological parameters and pollutants at the campaign site, as well as the temperature and O3 anomalies during heatwaves across major urban clusters. In addition, we further evaluated the impact of different chemical mechanisms in the model on future O3 predictions. We conducted an additional simulation using the common reactive intermediates gas-phase chemical mechanism. As shown in Supplementary Fig. 5, carbon-bond photochemical mechanism and common reactive intermediate simulations show reasonably consistent O3 levels and temporal variations across major city clusters, with modest differences in MDA8 O3 especially at high temperatures, indicating the robustness of future O3 predictions under a changing climate and precursor levels.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.