Intensified exposure to compound extreme heat and ozone pollution in summer across Chinese cities

Introduction

In the context of climate change, the increasing frequency of extreme temperature events and air pollution incidents pose significant threats to the sustainability of rapidly expanding urban areas¹. Extreme heat events are becoming a common summer occurrence, exposing urban populations worldwide unevenly and leading to increased health risks and significant economic losses^2,3,4,5. High-temperature environments are often closely associated with severe air pollution. Ozone has emerged as the primary pollutant in urban areas during the summer, with 12% of the population in 261 cities worldwide exposed to high levels of ozone pollution in 2020 (80–160 µg/m³), particularly in China and India⁶. The proportion of urban areas in the global ozone-attributable mortality burden continues to rise⁷.

Extreme heat events and ozone pollution are frequently simultaneous within environments characterized by intense solar radiation, reduced wind speeds, and low precipitation levels^8,9,10. The increased severity and frequent exposure to high temperatures and ozone pollution present significant health risks to the global population, especially as their combined effects are more harmful than individual events, disproportionately affecting vulnerable populations such as the elderly, young children, and low-income groups^{11,12,13,14,15,16}. Research indicates that elevated environmental temperatures can alter an individual’s physiological response and sensitivity to ozone pollution, with co-exposure to high temperatures and ozone levels potentially resulting in synergistic effects^17,18. Schnell and Prather¹⁹ systematically demonstrated that climate change contributes to the simultaneous occurrence of particulates, ozone, and extreme temperatures in eastern North America, where extreme events often cluster in time and space, presenting the highest pollution levels and hottest temperatures, thus posing significant potential health risks. In several regions of Europe, it has been observed that the daily excess mortality rates associated with compound heat-ozone events are 1.8 to 3.5 times higher than those resulting from individual heatwaves or ozone pollution events²⁰. A study of 250 counties in China showed that exposure to concurrent heatwaves and high O₃ pollution was associated with higher all-cause mortality risk, with annual average excess deaths attributed to concurrent events reaching 6,249 from 2017 to 2020, 5.7 times higher than from 2013 to 2016¹⁵.

The concurrent occurrence of extreme heat and ozone pollution constitutes one of the most influential compound climate events worldwide^19,21. Considering the spatial distribution changes of extremely high temperature and ozone pollution events and their adverse health impacts on residents, it is essential to investigate the risks and disparities associated with exposure to these compound events. This analysis facilitates the precise identification of vulnerable areas and delivers specific information for shaping future public health policies that aim to mitigate and adapt to the risks posed by climate change. Research on high temperatures and ozone pollution in China from 2013 to 2020 has found that the simultaneous occurrence of these extreme events is intensifying across the country²². Particularly, the North China Plain has become the region most severely threatened by the compound extreme events of high temperatures and ozone pollution in China²³. In the future, climate change is expected to exacerbate global exposure to compound extreme events of heatwaves and ozone pollution, with low-income countries facing the most adverse effects²⁴. However, the patterns of compound exposure to urban heatwaves and high ozone pollution remain to be elucidated, especially in densely populated East Asia, and previous studies have lacked quantitative analysis of the spatiotemporal changes in the compound exposure of these two events.

Here, we conducted high-resolution longitudinal estimates of the urban population in China exposed to compound extreme heat-ozone conditions from the summer of 2003 to 2020. To achieve this, we coordinated the estimates of daily maximum temperature and ozone pollution at a high resolution (1 km spatial resolution) in China with annual urban area and population data. For each pixel, extreme temperature and ozone pollution days were defined using absolute thresholds, and the number of days of compound events during the summer was calculated, as well as the exposure, annual compound event days multiplied by the total population. Subsequently, we estimated the exposure levels and annual growth rates nationwide and at the city level from 2003 to 2020. At the city level, we quantified the contributions of compound events and population growth to exposure to better understand the spatial and temporal patterns of differences in exposure to compound events. Meanwhile, we also compared the changes in exposure differences across urban areas.

Results

Temporal trend of compound events exposure

China extreme heat and ozone pollution compound event days increased by 180%, from 489 thousand days in 2003 to 1.37 million days in 2020, and an annual increase of 61 thousand days yr⁻¹ (Fig. 1a). The urban population exposure to extreme heat and ozone pollution increased by 67% in 18 years, from 3.8 billion person-days in 2003 to 6.3 billion person-days in 2020, growing by 0.2 billion person-days yr⁻¹ (Fig. 1d). Both the compound events exposure and days showed a significant growth trend (P < 0.05). Furthermore, extreme heat events show a fluctuating increase, while ozone pollution events continue to increase, both of which have risen significantly since 2015. The days and exposure person-days of extreme heat events in 2020 increased by 173 and 68%, respectively, compared to 2003 (Fig. 1b, e), while the days increased by up to 291% (Fig. 1c) and the exposure to ozone pollution increased by 121% (Fig. 1f).

Intensified exposure to compound extreme heat and ozone pollution in summer across Chinese cities — **Fig. 1: Frequency of extreme heat and ozone pollution events and population exposure in Chinese cities from 2003 to 2020.**

Spatial feature of compound event exposure

We calculated total compound exposure for each city (SI Appendix, Table S1) and estimated the annual increase in exposure from 2003 to 2020 (SI Appendix, Table S2). From 2003 to 2020, the number of days and exposure to compound events of high temperatures and ozone pollution in Chinese cities increased rapidly, with the growth trend exhibiting significant spatial heterogeneity (Fig. 2). The number of compound event days and exposure in Chinese cities in 2020 significantly increased compared to 2003, especially in eastern cities. Cities in the North China Plain, the Yangtze River Basin, and the southeastern coast have shown marked increases in compound event days and exposure levels. In 2020, these regions comprised most of the top 30 cities with the highest compound exposure and fastest exposure growth, indicating strong spatial clustering. In 2020, the 15 cities with populations over 5 million collectively added over 500 days of extreme heat-ozone events annually, with Chongqing and Chengdu increasing by nearly 3000 days. At the same time, just the top 15 cities contributed 50% of the national urban annual rate increase in compound event exposure. From 2003 to 2020, 38% (141) of cities in China had statistically significant exposure trajectories (P < 0.05). These cities are primarily concentrated in the North China Plain region. In 2020, these cities comprised 23% of China’s total population, or 330 million people. Beijing and Chengdu had the highest exposure increase rate, exceeding 15 million person-days yr⁻¹.

**Fig. 2: City-level compound extreme heat and ozone pollution event days and exposure spatial distribution.**

In Chinese cities, extreme heat and ozone exposure are mainly concentrated in urban core areas, and the difference with expansion areas is growing each year (Fig. 3). In 2020, 17% (62) of cities had an exposure difference of over 80% between their core and expansion areas. These cities were mainly concentrated in the North China Plain and southwestern regions of China (Fig. 3a). In 41% (150) of cities, the exposure difference ratio shows a significant increasing trend (P < 0.05), particularly in small and medium-sized cities (population < 1 million) in the Southwest, Central China, and Xinjiang, with annual growth rates exceeding 4% (Fig. 3b). These cities often have small urban core areas but are expanding rapidly. As the exposure levels increase in the core areas, while the exposure in the expansion areas has remained constant or declined, intensifying the exposure disparities. In 2020, the core areas of densely populated large cities contributed up to 80% to the exposure to compound events. This is evident in cities like Beijing (the capital), Wuhan (a major city in the central region), and Chengdu (a major city in the southwestern region), where the exposure disparity in central urban areas significantly increased, especially in Chengdu, which saw rapid growth (Fig. 3c, d, SI Appendix, Fig. S1). Additionally, as large cities expand, small high-exposure centers are emerging at the urban edges (Fig. 3c).

**Fig. 3: The spatial variation in exposure to compound extreme heat and ozone pollution across different urban regions.**

Contribution to compound exposure from population growth versus compound events

Urban population and cooccurrences of extreme heat and ozone pollution events exhibit evident spatial heterogeneity in their contributions to exposure trajectories at the city level (Fig. 4). Population growth remains the predominant factor driving compound extreme event exposure in Chinese cities. In 71% of cities (259 cities), the growth in exposure trajectories is primarily attributed to population growth (contributing more than 50%), particularly notable in cities located in the southeast region of China (Fig. 4a, e). Even in cities with similar trajectories, there are substantial differences in their contributions to the trajectory. Beijing and Chengdu, the two cities with the highest average annual growth rates of compound exposure, display similar exposure patterns. However, population growth accounts for 65% of the annual exposure growth rate in Beijing, while in Chengdu, it accounts for 85%. cities were categorized based on traditional geographical divisions and the spatial patterns of population growth and compound event contributions to exposure trajectories (Fig. 4c, d). The increase in the number of days with compound events has significantly contributed to the growth of urban compound exposure in Northeast China (NE), the North China Plain (NCP), Southwest China (SW), and Fujian (FJ). The median contribution of compound event growth to overall exposure growth was highest in NE (64%), followed by FJ (37%), NCP (35%), and SW (26%), with these regions exhibiting comparatively stronger contributions than others. In contrast, the contribution of compound events to exposure in rapidly urbanizing southern China has diminished, with population growth playing a significant role in the increased exposure.

**Fig. 4: The share of population versus compound event in the rate of increase of co-exposure to extreme heat and ozone pollution.**

Discussion

From 2003 to 2020, the risk of urban residents in China being exposed to more frequent extreme high temperatures and ozone pollution during the summer has continuously increased. Our assessment of urban residents in China exposure to extreme heat and ozone pollution over 18 years at a 1-km scale indicates that the compound exposure in urban areas nationwide has doubled. During this period, the annual average increase in compound event exposure for the urban population in China exceeded 200 million person-days. In particular, both individual occurrences and simultaneous events of extreme heat-ozone in China have significantly increased after 2015. However, the trends and impacts of compound high temperature and ozone events have not yet received widespread attention. In the future, with the increasing frequency of simultaneous high temperature and ozone occurrences, as well as the continuously growing urban population, exposure to compound extreme events will pose a significant threat to public health.

The North China Plain (NCP) has emerged as a primary high-risk region for extreme heat and ozone pollution in China, with a marked increase in the frequency of compound events and exposure across most cities (SI Appendix, Fig. S2). The heightened occurrence of these compound events has significantly contributed to the accelerated growth of urban exposure trajectories within the region. Previously, it has been observed that the trend of rising temperatures in Northern China surpasses that of the South²⁵. High temperatures have been identified as the leading meteorological factor contributing to the worsening ozone pollution on the North China Plain²⁶. The simultaneous increases in ozone levels and temperature on the North China Plain have led to the frequent occurrence of compound extreme events^23,27. The frequent occurrence of extreme heat and ozone pollution events poses greater health risks to urban residents in the North China Plain. Additionally, this phenomenon has also been observed in Fujian, the Yangtze Plain, and the Southwest regions of China (SI Appendix, Fig. S2). In addition to human activities exacerbating the growth of compound exposure, meteorological factors have also contributed to the occurrence of compound extreme events. Summer typhoon activity facilitates cross-regional ozone transport from the Yangtze River Delta, while extreme heat increases ozone concentrations in Fujian^28,29. Together, these factors have contributed to the rise in compound exposure in the region. The westward shift of the Western Pacific Subtropical High during summer has played a dominant role in the co-occurrence of extreme heat, intense surface ultraviolet radiation, and severe ozone pollution in the Yangtze Plain³⁰. The basin topography of the Southwest region, along with high summer temperatures, intensifies ozone pollution, resulting in the frequent occurrence of compound events^31,32. Therefore, these areas must take measures to mitigate compound events of high temperatures and ozone pollution to reduce the risks of compound exposure.

In the internal spaces of the city, we find that the compound exposure in the urban core is significantly higher than in the peripheral expansion areas, and this trend continues to intensify. This may be caused by the high-density human activity emissions in urban centers, especially from transportation pollution sources³³. Model predictions suggest that the simultaneous occurrence of extreme heat and ozone events in China may intensify under high emission scenarios³⁴. The welfare of urban residents in China, especially in densely populated city centers, faces significant challenges. This highlights the importance of proactive emission reduction climate policies for sustainable urban development. Considering the relationship between high temperatures and ozone pollution, addressing rising temperatures can bring co-benefits in reducing air pollution. According to reports, China’s actions to mitigate carbon dioxide emissions may lead to a significant reduction in ozone concentrations by 2050³⁵. Such emission reduction strategies help to further mitigate the exposure impacts of extreme compound events³⁶.

Furthermore, our results indicate that the trajectory of compound exposure in Chinese cities primarily stems from contributions of population growth, particularly in rapidly growing southern cities. This underscores the necessity of implementing adaptive measures, such as increasing blue-green infrastructure and urban shading, in addition to emission reduction policies, especially in the context of rapid urban population growth. Such measures are particularly important for protecting vulnerable groups, including the elderly, children, and outdoor workers. The spatial differences in the contributions of population growth to exposure trajectories in various cities reflect the urbanization trends in China. However, even among cities of similar population sizes, there are significant differences in the contributions of population growth and the occurrence of compound events to exposure trajectories. Overall, we have described in detail the spatiotemporal variations of urban compound events and exposure trajectories at the city level, as well as the spatial heterogeneity of contributions from population growth and extreme events to the exposure trajectories in different cities. This provides essential information for managers to plan future climate adaptation policies. We also recognize that using thresholds to directly define exposure to extreme events does not take into account the risk differences among populations with varying socioeconomic backgrounds or health conditions, particularly as the elderly have higher vulnerability in the context of climate change³⁷. Subsequently, the impacts of exposure on different populations could be considered.

Our analysis provides evidence of spatiotemporal disparities in exposure to compound extreme heat and ozone pollution events at national, city, and grid scales. The occurrence and exposure of compound events in most cities in China show a significant increase. Our results identify high-risk cities and regions in urgent need of mitigation measures, such as those in the North China Plain. Detailed information on the differences in contributions from compound event days and population growth to exposure trajectories will help in devising more targeted measures and allocating resources. Additionally, integrating high-resolution data on the occurrence of compound events in urban areas can provide opportunities for effective early warning of extreme events to mitigate the health burden of exposure.

Methods

Urban population data

We used the 1 km global dataset of annual urban extents (2003–2020)³⁸, in which urban spatial extents related to local high-intensity human activities were mapped from nighttime light images using a new stepwise partitioning framework. Meanwhile, we extracted annual grid data of China’s urban population (2003–2020) from the Worldpop (https://www.worldpop.org/) 1 km population dataset based on the above urban extent data.

Daily temperature and daily ozone

We adopted a global dataset of daily maximum and minimum near-surface air temperature at 1 km resolution over land across 50 °S–79 °N from 2003 to 2020³⁹. This dataset is based on ground station near-surface temperature measurement data and satellite observations, including digital elevation models and surface temperature products. ChinaHighO3 is the seamless daily, monthly, and yearly 1 km ground level maximum 8-h average Ozone dataset in China from 2000 to present^40,41. This dataset is one of ChinaHighAirPollutants (CHAP), which provides multiple air pollutant datasets by combining artificial intelligence technology and multi-source data. We extracted daily maximum temperature and ozone concentration data for the summer (June, July, and August) in urban areas of China from 2003 to 2020 at 1 km resolution. This data was utilized to investigate the joint exposure of urban populations in China to extreme heat and ozone.

Urban population exposure to extreme heat and ozone

We followed the definition provided by the China Meteorological Administration, selecting a threshold of 35 °C for daily maximum temperature to define extreme heat events. Then, according to the Air Quality Guidelines (AQG) by the World Health Organization, we’ve set a threshold for ozone pollution concentration, daily maximum 8-h average ozone concentrations exceeding 100 µg/m³ ⁴². Finally, we measured the number of days of compound events in each grid cell, where extreme heat and ozone pollution co-occurred. In addition, we also separately calculated the number of days for extreme heat and ozone pollution events. We employed absolute thresholds to define extreme events rather than relying on percentile-based criteria. This approach ensures the comparability of exposure trajectories for compound events across different cities.

We quantified the exposure of compound extreme heat and ozone events for each city grid cell from 2003 to 2020, measured in person-days. Person-days is widely used to compare and contrast exposure levels across different regions and periods. We also calculated the number of extreme heat days and ozone pollution days, as well as the levels of heat exposure and ozone exposure. The heat-ozone events, heat events, and ozone events are calculated only within urban area. For a given grid cell, we multiplied the population of that grid cell in year by the number of compound event days (or ozone pollution days, extreme heat days) in that year, yielding its exposure, as shown in Eq. (1):

$$Ex{p}_{ij}={N}_{ij}times Day{s}_{ij}$$

(1)

where Exp_ij, N_ij and Days_ij are human exposure to compound heat-ozone events (or heat events, ozone events), population and number of events days in the i-th year and j-th grid, respectively.

To describe the spatial pattern differences of exposure within a city, we defined the urban boundary as of 2003 as the Urban Core Area (UCA), and the newly expanded areas from 2003 to 2020 as the Urban Expansion Area (UEA). We then calculated the ratio of the average compound events exposure difference (EDR) between the UCA and the UEA for each city, as shown in Eq. (2):

$$EDR=frac{overline{{E}_{UCA}}-overline{{E}_{UEA}}}{overline{{E}_{UCA}}}times 100 %$$

(2)

where (overline{{E}_{UCA}}) and (overline{{E}_{UEA}}) represent the average compound exposure levels for UCA (Urban Core Area) and UEA (Urban Expansion Area), respectively. The higher the ratio, the higher the compound extreme heat and ozone exposure pressure in the urban core area.

Urban population exposure trends

After summing exposure for each year at city and national scales, we evaluated the annual growth rates (in person-days yr⁻¹) of exposure and the annual growth rates (in days yr⁻¹) of compound event occurrences from 2003 to 2020 by fitting simple ordinary least squares linear regression models (OLS). The study included a total of 367 cities in China. For instance, at the city level, we estimated the rates of change for compound event occurrences and exposure from 2003 to 2020, as shown in Eqs. (3) and (4):

$$Ex{p}_{ij}={beta }_{0}+{beta }_{exp }{Y}_{i}+varepsilon$$

(3)

$$Day{s}_{ij}={beta }_{0}+{beta }_{days}{Y}_{i}+varepsilon$$

(4)

where β_exp and β_days are the annual change rates of compound exposure and events (person-days yr⁻¹ and days yr⁻¹), and the regression was tested at a significance level of P < 0.05.

Contribution to exposure from population growth versus compound event

At the city level, we drew on the methodology developed by Tuholske, et al.² to quantify the contributions of population growth and the increase in compound events to the exposure trajectories. For a given year and city, the contribution share of climate change to the exposure increase was calculated by multiplying the fixed population of urban settlements in 2003 by the increase in days of extreme heat and ozone compound events annually (Eq. 5). The contribution share of population growth to the exposure increase was calculated by multiplying the population growth since 2003 by the number of compound event days (Eq. 6).

We applied simple OLS regression to measure the annual change rates in the proportion of exposure from population growth and compound events. The resulting coefficients represent the average rates of change for compound event occurrence and population growth (in person-days yr⁻¹), respectively. These coefficients assessed the proportional contributions of exposure increase induced by population growth and climate change from 2003 to 2020 at the city level. To achieve this, we divided the exposure growth rates induced by population growth and compound extreme heat and ozone by the total exposure growth coefficient, resulting in the shares of population growth and compound event to the total exposure growth (Eqs. 7, 8).

$${rm{Heat}}-{{rm{Ozone}}}_{ij}={N}_{j}^{0}times (Day{s}_{ij}-Day{s}_{j}^{0})$$

(5)

$${{rm{Pop}}}_{ij}=({N}_{ij}-{N}_{j}^{0})times Day{s}_{ij}$$

(6)

$${{rm{C}}}_{pop}={beta }_{pop}/{beta }_{exp }times 100 %$$

(7)

$${{rm{C}}}_{heat-ozone}={beta }_{heat-ozone}/{beta }_{exp }times 100 %$$

(8)

where Heat-Ozone_ij and Pop_ij are human exposure to heat-ozone contributed by the compound event and the population in i-th year, and ({N}_{j}^{0}) and (Day{s}_{j}^{0}) are population and the number of compound event days in 2003 (i.e., initial year) of the j-th city. ({{rm{C}}}_{heat-ozone}) and C_pop are the contribution of compound event and increasing population to the exposure. ({beta }_{heat-ozone}) and β_pop are the annual change rates of compound exposure contributed by compound event and increasing population, respectively.