Introduction

The heterogeneous chemistry in the lower atmosphere is significant in areas with high aerosol loading and gas-phase pollutants1. For example, the heterogeneous interaction of N2O5 with aerosols significantly influences the NOx-O3 cycle in the atmosphere, depleting atmospheric NOx and leading to reduced surface ozone levels in the lower atmosphere (800–1000 hPa)2,3. However, the termination of these reactions is also important in determining the chemistry of the lower atmosphere. For instance, surface ozone (O3) is formed through radical chain reactions involving the oxidation of volatile organic compounds (VOCs) and nitrogen oxides (NOx), initiated by the photolysis of compounds like O3 and HCHO4,5,6,7. Termination occurs via peroxyl radical self-reactions or the reaction of hydroxyl radicals with NO2, influencing whether reductions in NOx or VOC emissions are more effective for controlling O3 pollution8. In ozone mitigation efforts, these reactions are typically considered within two regimes called ‘NOx-limited’ and ‘VOC-limited’9,10. The NOx-limited chemical regime describes conditions in which NOx emissions reductions are most beneficial to reduce ozone, while the VOC-limited regime represents situations where lowering emissions of organic compounds would be most effective.

Wang et al.11 noted that the trade-off between ozone (O3) and particulate matter (PM2.5) has arisen as an unexpected result of China’s Clean Air Action Plan, which sought to mitigate air pollution. Subsequent to the plan’s execution, ozone concentrations elevated during summer in the North China Plain, attributed to diminished NOx emissions and stable or escalating VOC emissions, along with substantial declines in PM2.5 levels. This suggests that tackling ozone pollution necessitates a more sophisticated comprehension than merely classifying it into NOx-limited and VOC-limited categories, wherein the impact of PM2.5 and aerosols are important12. Consequently, forthcoming air quality control systems globally must account for the complicated links among ozone, particulate matter and precursor emissions to successfully mitigate both pollutants concurrently.

Ivatt et al.13 presented a third regime, termed the Aerosol-Inhibited Regime (AIR), in which hydroperoxyl radicals (HO2) on particulate surfaces influence ozone formation. The experiment conducted with GEOS-Chem, a global chemical transport model (CTM), demonstrated that between 1970 and 2014, the percentage of the Northern Hemisphere’s population residing in this regime increased from 2 to 21%. At first, in 1750, regions dependent on biomass burning experienced the most significant effects, but industrialised areas such as North America and Europe had become the most affected by 1970. By 2014, the regions of South and East Asia, with a particular emphasis on South Asia, emerged as the most impacted areas. To address the rise in surface ozone due to a 50% reduction in PM2.5 precursors, a 40% decrease in NOx emissions is essential, highlighting the complicated interactions among various pollutants in the pursuit of effective air quality management. Therefore, it is essential to take into consideration not only the trade-offs between ozone and particulate matter but also the influence of aerosol particles on ozone formation. This emphasises the necessity for more detailed and regionally tailored approaches to effectively tackle both ozone and particulate pollution in the present and future emission contexts14,15.

Shared Socioeconomic Pathways (SSPs) represent scenarios designed to investigate various potential futures related to climate change and its effects16. This framework discusses the connection between different levels of socioeconomic development, technological progress and demographic shifts and their effects on greenhouse gas emissions and susceptibility to climate-related impacts. Five scenarios exist, starting with SSP1, which is defined by sustainable development and low emissions and extending to SSP5, which envisages rapid economic growth and high fossil fuel consumption. Each pathway presents unique challenges for climate mitigation and adaptation, assisting experts and decision-makers in assessing potential climate strategies. For example, SSP2 illustrates a balanced scenario that underscores the complexity and unpredictability in reaching climate objectives, whereas SSP3 emphasises a divided world characterised by elevated emissions stemming from insufficient collaboration. The SSP framework allows for the evaluation of climate impacts, adaptation requirements and mitigation approaches across different scenarios, and thereby supporting informed decision-making17. Additionally, the SSPs are frequently integrated with Representative Concentration Pathways (RCPs) to simulate various climate scenarios influenced by differing levels of greenhouse gas concentrations18,19.

This study aims at estimating the surface ozone concentrations using the GEOS-Chem framework under different SSP-RCP scenarios, addressing both near-future (2046) and distant-future (2096) scenarios, inclusive and exclusive of aerosol uptake effects. Previous studies have shown that the chemical effect of aerosols on oxidation, specifically by reactive uptake, exceeds their radiative effects, like photolysis rates, in influencing the formation of ozone near the surface20. Thus, the growing complexity of ozone dynamics, especially considering the unforeseen effects of air quality policies—exemplified by those in China—underscores the imperative to integrate aerosol interactions into atmospheric models. The relationship between reduced PM2.5 emissions and increased ozone levels highlights the necessity of a holistic strategy for air quality management21. This study utilises the SSP framework for clarifying the impact of diverse socioeconomic trajectories on ozone pollution across distinct aerosol conditions. Understanding these interactions is essential for developing effective climate and air quality measures, as the anticipated rise in populations residing in aerosol-inhibited areas presents considerable hazards to public health and environmental sustainability. This could further benefit policy makers in determining effective strategies for decreasing ozone levels while maintaining improvements in particulate matter and aerosol reduction22. As international initiatives to combat climate change escalate, it is essential to understand the ozone generation processes to formulate adaptive policies that address both local and global air quality issues.

Results and discussion

Sensitivity of aerosol uptake coefficients on photochemical ozone regimes

The aerosol uptake coefficient notably affects chemistry by modifying the interactions between aerosols and trace gases, including O₃, NO₂ and VOCs. Aerosols, consisting of fine particulate matter suspended in the atmosphere, serve as surfaces for heterogeneous reactions with gases. With an increase in the aerosol uptake coefficient, the likelihood of aerosols to react with radical species such as HO2 improves, resulting in lower surface ozone formation. Here, we assess the effect of varying aerosol uptake coefficients (γHO2)—0, 0.1, 0.2 and 0.4—on the distribution of AIR and their relationship with NOx-limited and VOC-limited regimes across the tropical and extratropical regions and seasons. The aerosol uptake coefficient is essential in regulating the interaction between aerosols and atmospheric gas-phase chemistry, specifically in assessing the degree to which aerosols impede or facilitate chemical reactions in the atmosphere23,24. Our analysis shows that an increase in the γHO2 results in a rise in the percentage of pixels categorised as the regions within the aerosol-inhibited regime. This effect is especially evident in areas where NOx-limited regimes typically dominate. In the NH winter season, an increase in the aerosol absorption coefficient from 0.0 to 0.2 results in a substantial reduction of around 13% of the NOx-limited regime, whereas the AIR expands by almost 15% (compare Fig. 1 with Supplementary Fig. 1). Even though aerosol loading is generally lower in high latitudes, conditions in these regions still favour aerosol-inhibition, primarily due to the longer atmospheric lifetimes and limited sunlight, particularly during winter, which reduce photochemical reactions that typically generate HO2, such as the photolysis of peroxides25. Also, the lower temperatures in high-latitude regions influence the chemical properties of aerosols. Cold temperatures can enhance the efficiency with which HO2 is uptaken onto aerosol surfaces. Certain types of aerosols, like sulphate aerosols, become more reactive in colder conditions, which further promote the uptake of HO2 and contribute to the inhibition of surface ozone26,27. Supplementary Figs. 1–3 show the spatial variation of the extent of AIR with different γHO2. This transition indicates that increasing aerosol reactive uptake reduces the availability of reactive gases such as HOx, and thus, amplifying the AIR at the expense largely of the NOx-limited regime. A similar pattern exists in other seasons, characterised by a steady decline of 5–10% in NOx-limited regions and a commensurate rise in aerosol-inhibited areas.

Fig. 1: Photochemical Ozone formation regimes in 2018.
figure 1

a–d The regional distribution of various ozone generation photochemical regimes modelled with γHO2 of 0.2 in 2018. Here, DJF is December-January-February, MAM is March-April-May, JJA is June-July-August, SON is September-October-November, USA is United States of America, EU is Europe, NNA is Northern North America, NSA is Northern South America and SSA is Southern South America.

Full size image

For instance, India is characterised consistently by an AIR across all seasons, but with occasional seasonal variations. In December-January-February (DJF), the Indo-Gangetic Plain (IGP) region, noted for its lush vegetation28, has higher VOC emissions, resulting in the formation of VOC-limited regimes. The elevated VOC emissions in these seasons may account for a shift in chemical regimes, since VOCs serve as precursors for atmospheric ozone generation, affecting the equilibrium between NOx-limited and VOC-limited regimes4. Also, during winter (DJF), lower temperatures and stagnant atmospheric conditions reduce the dispersion of pollutants, leading to high concentrations of VOCs29,30. However, these regions are in a VOC-limited regime, where ozone formation is constrained by the lack of available VOCs to interact with excess NOx available due to a combination of meteorological factors and increased emissions31. Nevertheless, the AIR persists throughout the rest of the year (March-April-May (MAM), June-July-August (JJA), September-October-November (SON)). East Asia also shows a consistent AIR for most of the year, except in SON. The shift from aerosol-inhibited to VOC-limited regimes in this area could be affected by seasonal fluctuations in emissions and atmospheric dynamics, particularly in autumn, owing to changes in vegetation and air mass transport32,33.

The Sahara region in Africa also shows a year-round AIR. Nevertheless, during JJA, the extent of the area under AIR markedly diminishes compared to DJF and MAM. This decrease can be attributed to the decreasing dust concentration in the atmosphere during this period34,35. The Sahara, a substantial producer of dust, has its aerosol concentrations affected by the intensity and direction of trade winds, which can reduce dust particles over extensive regions during JJA, and thus reduce the impact of aerosol absorption in the area36. Also, weaker winds during the summer monsoon are less effective at picking up dust from the desert surface compared to the stronger westerly winds in winter37. The seasonal distribution of dust in the Sahara could influence the AIR, as dust aerosols interact with other atmospheric substances, affecting the heterogeneous chemical equilibrium. Furthermore, the extratropical regions of Northern North America and Southern South America show a pronounced seasonal change in AIR, characterised by widespread aerosol-inhibited regions during the winter months. The seasonal variations can be attributed to variations in atmospheric circulation patterns, temperature and aerosol concentrations, which fluctuate with the seasons and affect the degree of the AIR. In these regions, the winter months generally result in stable atmospheres with temperature inversions and augmented aerosol accumulation due to high heating demand, which could foster the formation of AIR38.

Surface O3 sensitivity to aerosol uptake coefficients

Recent-past (2018)

Figure 2 shows the difference in surface ozone concentrations between a simulation with γHO2 of 0 and 0.2, representing the impact of a reduction in the aerosol heterogeneous uptake sink on surface O3. (Supplementary Figs. 4–6 show the percentage change in surface ozone concentrations for γHO2 of 0 with 0.1, 0.2 and 0.4, respectively). An uptake coefficient (γHO2) of 0.2 is used as the baseline for analysing surface ozone variations in 201839,40. Changes are most pronounced in the terrestrial regions, with negligible variation over the marine atmosphere. The most marked changes in surface ozone levels are observed in South and East Asia, where aerosols are most abundant globally currently and in 201841,42, especially during the Northern Hemisphere winter, where ozone concentrations increase by about 8–10 ppb (30–40%) due to the weakening of the radical sink. Although surface ozone variations are considerable in these regions during different seasons, the percentage change generally lies between 20 and 30%. Conversely, central Africa exhibits minimal surface ozone variations, between 5 and 10%, with a modest reduction between the Southern Hemisphere summer (JJA) and autumn (SON). Eastern North America also shows an increase in surface ozone, but only during the Northern Hemisphere winter, which is generally within 10–15% in 2018. The regional distribution of ozone variations is uniform, but the extent of these changes is more substantial with elevated aerosol uptake coefficients. With a γHO2 of 0.4, surface ozone increases by around 40–50% in East and South Asia, but this is limited to 15–25% when the HO2 uptake coefficient is 0.1. This change indicates that the γHO2 profoundly influences surface ozone concentrations, with more uptake resulting in more pronounced change in ozone levels, particularly in densely inhabited and industrialised regions such as East and South Asia.

Fig. 2: Ozone changes with and without γHO2 in 2018.
figure 2

a–d The change in surface ozone between when γHO2 is 0 and 0.2 in 2018. Here, DJF is December-January-February, MAM is March-April-May, JJA is June-July-August and SON is September-October-November.

Full size image

Near-future (2046)

Using GCAP 2.0, we modelled the future development of AIR globally under the SSP1–2.6, SSP2–4.5 and SSP5–8.5 scenarios to analyse the impact of aerosols on surface ozone levels by 2046. Figure 3 shows the photochemical ozone formation regimes under SSP2–4.5 and Supplementary Fig. 7 and 8 show the same under the SSP1–2.6 and SSP5–8.5 scenarios, respectively. AIR regions show a notable decrease in extent in future scenarios relative to current levels, as aerosols impede ozone formation via heterogeneous reactions39. Regions such as India and Sahara, which currently experience year-round AIR, show a reduction in their extent. In the SSP2–4.5 scenario, the global extent of AIR shrinks to 10.8% in December–February (DJF), 3.6% in March–May (MAM), 1.2% in June–August (JJA) and 4.2% in September–November (SON), due to a mid-century decline in aerosol and precursor emissions in most world regions. The SSP1–2.6 scenario, characterised by even more stringent mitigation efforts and reduced emissions, indicates a further decline in the extent of AIR by about 8.5%, 1.2%, 0.9% and 4.4% in DJF, MAM, JJA and SON, respectively. The decline can be attributed to lower aerosol emissions and concentrations due to cleaner air policies, reduced emissions and alterations in the availability of aerosol precursors, resulting in diminished aerosol-induced inhibition of ozone formation43. In the SSP2-4.5 scenario, by 2046, we observe a substantial reduction in the total aerosol surface area by about 50–100% (0.5–0.75 × 10−⁶ cm² cm⁻³) throughout much of East and South Asia and the Middle East, where the AIR regions dominated in the recent past (Fig. 4). This decrease could be attributed to the changes in aerosol emissions and atmospheric conditions influenced by socio-economic and climate trajectories. Furthermore, Black Carbon (BC) emissions in the near-future are projected to rise by about 25–50% (0.25–0.5 × 10−⁶ cm² cm⁻³) across East Asia, signifying an escalating contribution to regional air pollution and possible effects on climate forcing in the SSP 2–4.5 emission scenario (Supplementary Fig. 9). Sulphate aerosols will likely rise by 20–40% (2–3 × 10−⁶ cm² cm⁻³), especially during the Northern Hemisphere winter, across East Asia, North America and Northern Asia44, which could be due to the changes in industrial emissions and atmospheric transport patterns (Supplementary Fig. 10). Despite the overall reduction in the extent of AIR, their geographic distribution shifts, becoming more pronounced in areas such as the United States and Europe. This shift can be due to the near-zeroing of NOx emissions in these countries, which outpaces aerosol emissions reductions and indicates that aerosol-induced ozone inhibition may become considerable in densely populated areas there, intensifying air quality challenges in regions already facing health impacts from air pollution. The rise in sulphate aerosol surface area may contribute to the expansion of AIR in the US, notably in NH winter, as the spatial distribution of sulphate aerosols correlates highly with the extension of AIR there compared to that in 2018 (Supplementary Fig. 10). Furthermore, by 2046, it is projected that nearly 90% of the globe will be in a NOx-limited regime across all SSP scenarios, suggesting that NOx will emerge as the primary limiting factor for ozone production.

Fig. 3: Future ozone formation regimes.
figure 3

a–d The regional distribution of ozone generation photochemical regimes modelled with γHO2 of 0.2 in the SSP 2-4.5 scenario for the year 2046 (e–h) and 2096. Here, DJF is December-January-February, MAM is March-April-May, JJA is June-July-August and SON is September-October-November.

Full size image
Fig. 4: Future changes in aerosol surface area.
figure 4

The change in Aerosol surface area density in SSP 2-4.5 for a–d 2046 and e–h 2096 when compared to that of 2018.

Full size image

We also compare the effect of aerosol uptake on surface ozone levels under various SSP scenarios by comparing simulations that include aerosol uptake to those that do not, using a γHO2 of 0.2. Figure 5 shows the change in surface ozone for the SSP2–4.5 scenario and Supplementary Figs. 11–16 and Fig. 6 show the percentage change in surface ozone with and without aerosol uptake for the different SSP scenarios compared to that of the recent past. We observe a consistent increase in ozone, especially in areas with high aerosols, including South and East Asia, where a reduction in aerosol surface area leads to an increase in the radical species like HO2. In the SSP1–2.6 scenario, ozone concentrations increase by ~5–10% in these regions throughout the seasons (Fig. 6). However, the increase in ozone is less during MAM and JJA, which is within 5%. This increase in ozone levels can be attributed to a lower aerosol uptake, which facilitates the formation of ozone through heterogeneous reactions. There is also a notable increase in surface ozone in the Southern Hemisphere during the JJA and SON. In areas characterised by elevated aerosol concentrations, including South and East Asia, the SSP2–4.5 and SSP5–8.5 scenarios indicate a more pronounced increase in ozone levels, by ~10–20%, in contrast to the rise of 5–10% observed in the SSP1-2.6 scenario. This is much lower than recent past (2018) sensitivity of O3 concentration to γHO2 (30–40% in Asia), which further confirms that the extent of AIR will further reduce in the future.

Fig. 5: Future ozone changes with and without γHO2.
figure 5

The change in surface ozone between when γHO2 is 0 and 0.2 in a–d 2046 and e–h 2096. Here, DJF is December-January-February, MAM is March-April-May, JJA is June-July-August and SON is September-October-November.

Full size image
Fig. 6: Regional difference in aerosol surface area and surface ozone.
figure 6

The difference in a–e aerosol surface area and f–j surface ozone between γHO2 values of 0 and 0.2 averaged over 5 days across different regions for the year 2018, and for various SSP scenarios in 2046 and 2096. Here, NSA is Northern South America, USA is United States of America, EU in Europe and EA is East Asia, as shown in Fig. 1.

Full size image

Distant-future (2096)

A comparison of these projected results for 2046 and 2096 reveals a decrease in the extent of AIR in 2096 under the SSP1-2.6 and SSP2-4.5 scenarios (Fig. 3 and Supplementary Fig. 14). The global extent of AIR diminishes from 10.8% to 7.4% in DJF, 3.6% to 1.3% in MAM, 1.2% to 0.4% in JJA and 4.2% to 3.1% in SON. The shrink is particularly evident in the United States in the winter months (DJF). India continues to be primarily within a VOC-limited regime in both 2046 and 2096, with AIR decreasing to fewer than 10% globally by 2096. In Northern Europe, most regions continue to be in AIR in both 2046 and 2096. The observed temporal patterns are similar in both the SSP1–2.6 and SSP2–4.5 scenarios (Fig. 6). Nevertheless, the aerosol surface area is higher when γHO2 is 0 because a lower uptake coefficient implies no uptake, leaving more aerosol surface area available in the atmosphere. In contrast, a higher uptake coefficient results in more gas-phase particles being absorbed by aerosols, thereby decreasing available aerosol surface area. For instance, in India, the aerosol surface area is ~300 × 10¹⁰ cm² cm⁻³, which could have taken up around 6 ppb of ozone during winter in 2018. However, under the extreme SSP5–8.5 scenario, a notable transformation takes place in North America, where northern areas of the US and Canada fall entirely in AIR, while southern regions of the US shift to a VOC-limited regime (Supplementary Fig. 15). Furthermore, Southeast Asia is in AIR in 2046, but it turns out to be predominantly NOx-limited by 2096.

The γHO2 further impacts surface ozone concentrations, with regional variations observed during 2096 as well (Figs. 5, 6 and Supplementary Figs. 16–18). For instance, in the SSP 2-4.5 scenario, the comparison of γHO2 ranging from 0 to 0.2 indicates an increase in surface ozone levels by ~10–15% in the regions including South and East Asia during the DJF season. The northern extratropics, encompassing the US, Canada, China, Russia and Europe, shows a notable reduction in ozone of 5–10%, except in JJA, during which a decrease of less than 5% is found. These changes are more pronounced in the SSP 5–8.5 scenario and less evident in SSP 1–2.6, analogous to that observed in 2046 (as shown in Fig. 6). An additional decrease in total aerosol levels by the end of the century compared to that of current levels, along with reductions in sulphate and BC emissions, is anticipated to reduce surface ozone, as shown in Fig. 4 and Supplementary Figs. 9, 10. Aerosols, especially sulphates and BC, influence ozone production, impact the availability of hydroxyl radicals (OH) and thus influence the extent of AIR.

Future projections of surface ozone

Near-future (2046)

We also estimate the total surface O3 change from future to present. Surface ozone levels in the Northern Hemisphere tropics are projected to rise by ~50–75% by 2046 when compared to the current levels in 2018, with an increase of about 150–200% in areas such as South America and East Asia, although these variations are seasonal in the SSP 2-4.5 scenario (Fig. 7 and Supplementary Fig. 20). Using our γHO2 sensitivity simulations, we attribute that about 40–60% of the total surface ozone enhancement is due to aerosol reductions in the regions dominated by AIR. Northern South America is expected to see a continuous increase in surface ozone of ~200%, irrespective of the seasons, whereas mid-Africa is anticipated to experience a steady rise of 50–100% year-round, with an exception in DJF. India also has a seasonal rise of about 30 ppb in surface ozone levels, especially during JJA. The observed spatial patterns are uniform across all SSP scenarios (SSP1–2.6, SSP2–4.5 and SSP5–8.5), with more increase in ozone levels noted in the SSP2–4.5 and SSP5–8.5 scenarios, which is about 20–30% higher when compared to current levels (Supplementary Figs. 19–21). The distribution of ozone in higher emission scenarios shows a more widespread pattern, mainly in the Northern Hemisphere extratropics, where an increase of 50–75% is noted throughout the globe. Nevertheless, regions within AIR regions in the near-future mainly affects North America, Northern Europe and East Asia, especially during the Northern Hemisphere winter, with an approximate increase of 50–60% surface ozone over North America and Northern Europe and about 20–30% over East Asia.

Fig. 7: Projected surface ozone changes for 2046 and 2096.
figure 7

a–d The change in surface ozone concentrations for the SSP 2-4.5 scenario for the year 2046 and e–h 2096 when compared to that of 2018. The model simulation uses an aerosol uptake coefficient (γHO2) of 0.2. Here, DJF is December-January-February, MAM is March-April-May, JJA is June-July-August and SON is September-October-November.

Full size image

Biogenic emissions strongly impact ozone formation, which is also impacted by atmospheric NO2 concentrations45. Figure 8 shows the change in HCHO in near and distant-future from GCAP 2.0 simulations. We observe an increase in HCHO (a proxy for VOCs), primarily of biogenic origin, in both near-future and distant-future simulations. The oxidation of VOCs contributes to the generation of secondary organic aerosols, with the degree of contribution influenced by temperature, vegetation type and foliar mass46. These characteristics are acutely responsive to climate change and alterations in vegetation distribution47. For instance, Smith et al.48 shows the projected global rise in agricultural land and forest cover, which will subsequently increase biogenic VOC emissions. Furthermore, agricultural waste burning may increase by 1.5 to 2 times across different emission scenarios by the end of the century, resulting in elevated NOx and VOC emissions, enhancing ozone formation from these precursors.

Fig. 8: Future changes in HCHO.
figure 8

The change in HCHO for the years a–d 2046 and e–h 2096 when compared to that of 2018 for the SSP 2-4.5 scenario. Here, DJF is December-January-February, MAM is March-April-May, JJA is June-July-August and SON is September-October-November.

Full size image

Urbanisation, energy use and air pollution are closely related. Urban Heat Islands in densely populated cities raise temperatures, and thus, increase the demand for cooling energy49,50. This results in increased residential emissions, but it varies by region. We observe that VOC emissions in Central Africa, Northern South America and India will increase by about 2–3 ppb in the future, and exacerbate the ozone formation (Fig. 8). Similarly, in these regions, residential emissions are projected to rise until the mid-century, before reverting to historical levels by 210051. We also observe a notable increase in NO2 emissions, particularly in India (about 5 ppb), South America and Africa (about 1–2 ppb), in the near-future. However, under the SSP 2-4.5 scenario, these emissions are projected to decrease substantially by about 1–2 ppb when compared to that of the recent past by the end of the century, as shown in Fig. 9. This reflects potential improvements in emission control measures, shifts in industrial activity and changes in energy consumption patterns, leading to a reduction in NO2 levels over time. A cleaner environment with less NOx and still high aerosols can further reduce the aerosol uptake and thus increase the ozone concentration there52. Thus, these findings emphasise the necessity for efficient control methods to guarantee improved air quality in the future. Tian et al.53 studied energy consumption and air pollution trends in China since 1990, providing recommendations for sustainable energy utilisation, air pollution management and CO2 reduction. By 2003, China’s total primary energy consumption had attained 1678 million tonnes of standard coal equivalent. Consequently, emissions of SO2 and NOx increased to 21.58 million tonnes and 16.13 million tonnes, respectively. This suggests that more rigorous laws, standards and efficient economic strategies are needed for resolving those challenges. Thus, the increasing demand of energy would lead to more emissions, but can be tackled using a clean and sustainable practice for energy production54.

Fig. 9: Future changes in NO2.
figure 9

The change in NO2 for the years a–d 2046 and e–h 2096 when compared to that of 2018 for the SSP 2-4.5 scenario. Here, DJF is December-January-February, MAM is March-April-May, JJA is June-July-August and SON is September-October-November.

Full size image

Also, climate change is anticipated to modify atmospheric circulation and precipitation patterns, impacting meteorological conditions that deteriorate air quality. In a warming climate, the buildup of particulate matter in the atmosphere, especially during winter, is expected to rise due to stagnation. This stagnation is characterised by lack of precipitation, weak surface winds and poor vertical mixing, all of which facilitate the accumulation of pollution at the surface. Stagnation events can exacerbate PM2.5 pollution even in the absence of substantial emissions. Zhou et al.55 found that changes in stagnation patterns are influenced by both global CO2-induced circulation modifications and local aerosol-driven meteorological reactions. By 2100, winter stagnation in IGP is projected to extend by 7 ± 3 days, resulting in an increase in PM2.5 concentrations by roughly 7 µg/m³ under high-warming and high-aerosol scenarios. Nevertheless, annual stagnation instances are anticipated to diminish in most of India. Horton et al.56 used an air stagnation index and a series of bias-corrected climate model simulations to evaluate potential alterations in stagnation frequency and duration due to global warming. Their study indicates that stagnation events may rise in regions encompassing 55% of the global population, with the affected population in these areas being ten-fold greater than in regions experiencing a decline in stagnation. By the late twenty-first century, substantial increases—up to 40 additional days a year—are predicted for many areas in the tropics, subtropics and certain mid-latitude regions. These alterations will probably exert most notable impacts on countries such as India, Mexico and the western United States, where the large populations coincide with prolonged stagnation episodes. This stagnation may result in the prolonged existence of trace gases in the atmosphere, and thus, increase ozone exposure in the densely populated areas.

Distant-future (2096)

In 2096, the change in surface ozone levels is less than those in 2046 across all SSP scenarios when compared to that of the recent past, as shown in Fig. 5 and Supplementary Figs. 19–21. The reduction is attributable to several factors, including stricter emission controls, enhanced air quality regulations and possible alterations in atmospheric dynamics that may decrease ozone formation57. The dispersion of ozone from hotspot regions diminishes by roughly 25–50% across regions. The reduction is noticeable particularly in the tropical marine atmosphere. The spatial extent of hotspot regions in 2096 is similar to that of 2046, with consistently higher ozone levels (100–150%) in regions such as northern South America and Central Africa, irrespective of the seasons. These regions may persistently experience increasing ozone attributable to local sources of ozone precursors (as shown in Figs. 8 and 9), including biomass burning or industrial emissions, together with climate conditions favourable to ozone formation58. East and South Asia show an increase in ozone during DJF and JJA, likely influenced by seasonal emission variations, including elevated industrial activity in winter and agricultural burning in summer59,60. Furthermore, surface ozone levels in North America are projected to be higher by 20–30 ppb in 2096 compared to that of 2018, potentially attributable to changes in transportation emissions, urbanisation, or a lag in atmospheric chemistry’s response to reductions in precursor pollutants61,62.

Discussion

A comparison of Fig. 5 with Fig. 7 shows the impact of γHO2 on surface ozone concentrations in both the near and distant-future, compared to the recent past. In comparing the results from 2018 to those from 2046, East and South Asia have the most dominant effect of γHO2 on ozone concentrations, with an increase of around 5–10 ppb, which is nearly two times less than the contribution noted in the recent past (as shown in Fig. 2). Future projections indicate that surface ozone levels in India will significantly rise during summer monsoon (JJA), with an overall increase of 10–30 ppb, of which 5–10 ppb is ascribed to the impact of γHO2, or around 20–40% of the total increase. In East Asia, the most significant increase is anticipated in winter, when aerosol effects are more evident, contributing 5–7.5 ppb to the total 30 ppb rise, corresponding to ~25% of the ozone increase. In West Africa, γHO2 is expected to contribute roughly 5 ppb during the dry season (DJF), whereas Central Africa will have a contribution of about 5 ppb in the fall (SON). The anticipated total increase in surface ozone in West and Central Africa is around 15–20 ppb, with γHO2 contributing 25–30% to this rise. By 2096, the influence of γHO2 on surface ozone is expected to continue to decline, with the most significant effect prevailing in India, where the contribution will be less than 5 ppb, which represents ~10–20% of the overall rise. Conversely, by the end of the century, Central and West Africa will continue to have an increased impact than East Asia, although this will account for merely 10% of the total ozone enhancement attributed to higher aerosol uptake by HO2.

Additionally, Fig. 7 also shows that regions that have lower surface ozone show more variation in the future, ~100–200% (20–40 ppb). Ozone precursors are, in general, more difficult to control and ozone levels have a larger impact from remote sources as well as increasing methane concentrations63. We found that, in the SSP scenarios, regional ozone levels do show clear increase in the future. For instance, Northern South America currently has an ozone level of 10 ppb, which is projected to increase by around 200%, reaching about 30–40 ppb by the end of the century in the SSP2–4.5 scenario. This is also analogous to several other regions, including Northern Europe and East Asia. Current hotspot regions, such as Central Africa, show an increase of 50–100%, resulting in levels around 20–30 ppb by 2100. India currently has higher ozone levels (about 25–35 ppb), which are projected to remain stable by 210064. This indicates that the majority of worldwide regions would experience ozone levels of 40–80 ppb in the future, which is above the World Health Organization’s recommended guideline, increasing the population’s susceptibility to pre-existing health problems associated with air pollution.

The results show a substantial influence of γHO2 on the distribution of aerosol-inhibited regimes and surface ozone concentrations in different global regions. The increase in γHO2 correlates with an expansion in the AIR, especially in otherwise NOx-limited regions like South and East Asia, particularly evident in colder months when lower temperatures and stagnant atmospheric conditions intensify aerosol effects. Future projections across various SSP scenarios indicate a global reduction in AIR extent, linked to decreased aerosol concentrations and more stringent emission regulations. The incorporation of these iterative impacts could modify the concentrations of these contaminants63. However, previous studies suggest that the stagnation effect is expected to intensify in the context of future climate change, leading to an increased accumulation of PM and, consequently, more aerosols in the atmosphere65. Therefore, incorporating HO2 uptake in future studies is essential for a precise understanding of atmospheric chemistry, as it directly affects the atmosphere’s oxidative capacity, ozone production and the behaviour of key reactive species. This is particularly essential in regions with lower NOx levels but aerosol surface area is still high, lower levels of γHO2 can greatly influence HO2 concentration and, in turn, ozone production regimes52.

The future trajectory of air quality, especially with higher ozone and lower aerosol levels, poses considerable concerns and health dangers worldwide. Projections based on the SSP-RCP scenarios indicate geographical variances in ozone concentrations, with regions such as Northern South America, Northern Europe and East Asia anticipated to experience notable increases by the end of the century. The rise in ozone is primarily influenced by many elements, such as aerosol reactive uptake, agricultural emissions, forest fires and the impact of climate change on vegetation and air circulation. The increase in biogenic VOC emissions, coupled with improved agricultural techniques and garbage incineration, will intensify ozone formation, particularly in areas such as Africa and Asia66,67. The interaction between urbanisation, energy consumption and air pollution highlights the pressing necessity for greener energy alternatives and efficient pollution mitigation strategies. Climate-induced alterations in weather patterns, including stagnation episodes, would exacerbate particulate matter concentrations and extend exposure to detrimental pollutants, especially in densely populated regions. Therefore, efforts to reduce aerosol concentrations must be integrated with policies that tackle the broader impacts of climate change, including reducing the possibility of wildfires and agricultural emissions, as these elements will intensify ozone formation. Furthermore, policies that advocate for cleaner energy choices, urban air quality control and sustainable agricultural methods will be essential in alleviating the total environmental burden. Considering the health hazards linked to increased ozone concentrations, it is imperative for states to collaborate in establishing and enforcing international norms in accordance with the WHO’s recommendations. These initiatives must encompass investment in research and technology to assess air quality and evaluate the efficacy of executed actions.

Methods

Ground-based measurements

The Southern Hemisphere ADditional OZonesondes (SHADOZ) project was established in 1998 to monitor ozone profiles at various tropical stations using ozonesondes. These measurements utilise electrochemical concentration cell sondes, which are equipped with a gas-sampling pump connected to an ozone sensor and a radiosonde for transmitting data68. The accuracy of the ozonesonde readings is within 5%. A comprehensive description of the data can be found in Thompson et al.69. We also use the ground-based measurements of ozone from the Central Pollution Control Board (CPCB) monitoring stations for Asansol, India. The inherent precision of ozone measurements from CPCB analysers is about ±1 ppb. Further details regarding these data are available in CPCB70.

GEOS-Chem model

GEOS-Chem is a CTM developed for modelling complex oxidant-aerosol chemistry in the troposphere and stratosphere71,72. Employing the Kinetic PreProcessor (KPP) 3.0 as its chemical solver, GEOS-Chem integrates sophisticated functionalities via the FlexChem interface, facilitating a flexible methodology for chemical kinetics73. The model complies with the most recent JPL/IUPAC guidelines for chemical mechanisms, with substantial modifications that improve the depiction of diverse chemical processes, including those related to isoprene, aromatics and nitrates74. Recent advancements have enhanced the treatment of intricate reactions, including methanol synthesis and mercury redox chemistry, facilitating more precise atmospheric forecasts. Additionally, the model incorporates the reactive absorption of nitrogen oxides by aerosols and computes aerosol hygroscopicity, which is essential for comprehending aerosol-cloud interactions. GEOS-Chem offers a comprehensive framework for investigating atmospheric chemistry and its effects on air quality and climate change by accommodating various chemical species and reactions, including halogens and hydroxymethanesulfonate. This model provides thorough understanding into ozone generation mechanisms and pollutant interactions, serving as an essential resource for addressing air quality challenges and comprehending the wider implications of atmospheric dynamics. GEOS-Chem has been rigorously validated by researchers globally. For instance, Travis et al.75 showed that, despite the successful simulation of ozone and its precursors in the SEAC4RS aircraft data below 1 km of altitude, Maximum Daily 8-h average surface ozone (MDA-8) was biased high in the model by +6 ppb on average.

David et al.76 used the GEOS-Chem transport model over India and observed that the model reasonably simulated the tropospheric O3 abundances and vertical profiles, with a mean bias of 1–3 DU compared to observations for the period 2000–2015. Christiansen et al.77 observed that the GEOS-Chem model, when validated against 25 ozonesonde sites, aircraft and satellite observations globally, has a reasonable agreement with a mean bias of 1–3 DU for the period from 1990 to 2017. Mao et al. 78 reported that the GEOS-Chem model, validated against MDA8 O3 observations in central and eastern China from May to July 2017, had a strong correlation of 0.77 (95% confidence level). In Nanjing, the simulated MDA8 O3 concentrations converged with the observed trend, with a correlation coefficient of 0.65, a normalised mean bias of 5% and a normalised mean error of 21%. We compared the modelled ozone to the surface ozone measurements from the SHADOZ ozonesonde network and ground-based measurements from CPCB and OpenAQ datasets, as shown in Supplementary Fig. S22 and S23, respectively. Although the modelled and in-situ observations are generally consistent, the model tends to overestimate surface ozone at the majority of the stations. This overestimation can be attributed to factors like uncertainties in the emissions of both anthropogenic and biogenic precursors, as well as inaccuracies in ozone sinks such as dry deposition. Additionally, the model’s coarse resolution limits its ability to capture localised chemical processes like ozone titration9.

GCAP 2.0 model

The Global Change and Air Pollution (GCAP 2.0) framework signifies a notable technological enhancement compared to the original GCAP model, first articulated by Wu et al.79 and subsequently refined by Murray et al.80. This revised model utilises meteorological data sourced from version E2.1 of the NASA Goddard Institute for Space Studies General Circulation Model, allowing the GEOS-Chem to conduct simulations across diverse scenarios, including pre-industrial, recent historical and multiple future conditions in accordance with the CMIP6 experiments81,82. GCAP 2.0 resolves significant difficulties noted in previous iterations, notably the excessive mass transfer from the stratosphere to the troposphere that complicated the tropospheric ozone budget. This problem frequently obstructed the analysis of polar ice-core data, particularly with the integration of online interactive stratospheric chemistry in GEOS-Chem. To address these complexities, GCAP 2.0 employs a unidirectional offline coupling technique, enabling the GEOS-Chem CTM to utilise stored meteorological data from any historical or forthcoming time. Essential attributes encompass the incorporation of a cohesive chemical mechanism spanning from the surface to the mesopause, facilitating enhanced atmospheric representations, alongside a versatile emissions pre-processor to accommodate diverse emissions situations. GCAP 2.0 facilitates global simulations encompassing 72 vertical layers down from 1000 hPa to 0.01 hPa. The model integrates innovations such as one- and two-way linked nested regional simulations, enabling localised air quality investigations in conjunction with global evaluations and an adjoint for inverse modelling that improves its functionality.

THe GEOS-Chem and GCAP 2.0 are run on a 2.5° × 2.5° grid resolution globally. GEOS-Chem classic version has 72 vertical levels from the surface (1000 hPa) to the top of the atmosphere (0.01 hPa), whereas GCAP 2.0 has a 41-level vertical gridding. The GEOS-Chem model utilises Goddard Earth Observing System Forward Processing simulations for meteorological input and Harmonized Emissions Component for emissions. One year spin up simulations with different aerosol uptake coefficients are used to generate the model restart files. Termination rates of chain reactions were determined using archived species concentrations and physical parameters such as temperature, pressure and humidity, in addition to aerosol properties, as per Ivatt et al.13. Although we do not classify radical product generation in peroxyl-radical self-reactions as termination stages, we regard non-radical products as ongoing termination processes. For our simulations, we employed a baseline HO2 reactive uptake coefficient (γHO2) of 0.2. Impact of HO2 uptake on tropospheric composition (γHO2) value of 0.2 was recommended for use in models of tropospheric chemistry based on a combination of laboratory data (yielding values of 0.05–0.2)83,84,85 together with results from model-observation comparisons86,87. However, laboratory studies of pure synthetic aerosols indicate lower uptake coefficients (γHO2 < 0.2), whereas real-world aerosol studies reveal values between 0.08 to 0.40, implying that elements such as transition metals may increase aerosol uptake65,88,89. To assess the influence of HO2 uptake, we performed simulations with γHO2 values of 0, 0.1 (half of the baseline), 0.2 (baseline for future projections in this study) and 0.4 (twice the baseline). Taketani et al.90 measured the HO2 uptake coefficients for ambient aerosol particles collected on quartz fibre filters using a high-volume air sampler in China. The measurements were conducted with an aerosol flow tube coupled to a chemical conversion/laser-induced fluorescence technique at 760 Torr and 298 K, with a relative humidity of 75%. The aerosol particles were regenerated with an atomiser using water extracts from the particles. Over 10 samples, the HO2 uptake coefficients at the Mt. Tai sites ranged from 0.13 to 0.34, while those at the Mt. Mang sites ranged from 0.09 to 0.40. Zhou et al.91 studied the first online measurement of the total HO2 reactivity caused by the ambient gas phase (kg’) and aerosol phase (ka’) in the summer of 2018 in Kyoto, Japan, using a combined laser-flash photolysis and laser-induced fluorescence (LFP–LIF) technique coupled with a versatile aerosol concentration enrichment system that enriches ambient aerosols. The related γ ranged from 0.08 (25th percentile) to 0.36 (75th percentile), with an average value of 0.24, which is comparable with the values used in previous modelling studies (0.2); however, it presents a large variation of ±0.20 within the measurement time. When γHO2 exceeds 1, HO2 concentrations can be notably affected, especially in regions with low NO concentrations and/or high aerosol concentrations92,93.

Though we presumed H2O to be the exclusive product of HO2 absorption, our results remain valid when H2O2 is taken into account. Employing a singular γHO2 value likely oversimplifies the variability, as existing models fail to account for its temporal oscillations94,95. The heterogeneous radical loss rate of HO2 is then determined by iterating through the radius and surface area of various aerosol types, including organic carbon (OC), BC, dust, sea salt, nitrate–sulphate–ammonium and secondary organic aerosol (SOA). These factors, along with temperature and air density, were used to calculate the first-order loss rate of HO2. However, this study has not accounted for the variation in photolysis rates of aerosols and its effect on surface ozone. The seasons are defined as DJF (December-January-February), MAM (March-April-May), JJA (June-July-August) and SON (September-October-November) in this study.