Introduction

In recent years, with the implementation of a series of emission reduction measures in China, primary air pollutants, notably fine particulate matter (PM2.5), nitrogen dioxide (NO2), sulfur dioxide (SO2), and carbon monoxide (CO), have shown significant decreasing trends1,2. However, the secondary air pollution represented by ozone (O3) has not been effectively controlled3,4,5. Particularly in summer, O3 has replaced PM2.5 as the primary air pollutant in eastern China and may persist in the country for a long time to come. Near–surface O3 is formed from volatile organic compounds (VOCs) and nitrogen oxides (NOx) through a series of complex chemical reactions under sunlight6,7. Complex non–linear relationships exist between VOCs, NOx, and O3. In the NOx–limited regime, when the VOC concentration remains constant, decreasing NOx concentrations result in decreasing O3 concentrations. In the VOC–limited regime, when NOx remains unchanged, lowering the VOC concentration results in lower O3 concentrations. However, in the VOC–limited regime, if the VOC concentration remains constant, reducing the NOx concentration may lead to an increase in the O3 concentration6,8. Therefore, fully understanding the mechanism of O3 generation and clarifying limiting regimes for O3 formation are the basis for the development of O3 pollution control measures.

Empirical kinetics modeling approach (EKMA) curves based on offline sampling data at ground stations are the main method currently used to investigate O3 formation sensitivity (OFS)9,10. However, owing to the wide variety of VOCs and their different activities, the heavy workload of offline analysis and long analysis cycles affect the timeliness of O3 pollution control. Ambient formaldehyde (HCHO), which is mainly generated by the oxidation of VOCs, has a short atmospheric lifetime and is positively correlated with the peroxyl radical (RO2). With limited information, HCHO can be used to represent the total reactivity of VOCs to analyze the OFS11,12. In addition, because of the short lifetime of NOx, NO2 is usually considered as a substitute for NOx concentration13. Based on these assumptions, many studies have used the ratio of HCHO (a proxy for VOCs) to NO2 (a proxy for NOx) (FNR) to characterize the OFS14,15,16,17, namely, FNR = HCHO/NO2. In particular, satellite observations can simultaneously identify these two compounds, which has prompted researchers to study the spatial heterogeneity of the O3 formation mechanism through FNR. For example, Ren et al.14 found that the VOC–limited regime was widespread in Chinese mega–city clusters, such as the North China Plain (NCP), Yangtze River Delta (YRD), and Pearl River Delta (PRD), whereas the NOx–limited regime dominated most of the remaining areas. However, satellite observations provide only one FNR observation per day, which may not be fully representative of the daytime OFS. In addition, satellite measurements can only provide tropospheric column concentrations and cannot identify the OFS at different heights. Therefore, high temporal resolution HCHO and NO2 data are required to explore the daily evolution of the OFS. Moreover, temporal and regional differences will lead to different spatio–temporal characteristics of O3 precursors (NOx and VOCs) in different regions. Therefore, a comprehensive understanding of the spatial and temporal characteristics of O3 and its precursors is essential for regional control of atmospheric photochemical pollution.

Multi–axis differential optical absorption spectroscopy (MAX–DOAS) enables long–term rapid simultaneous vertical profile measurements of atmospheric trace gases; therefore, it is widely used in the diagnosis of atmospheric OFS. For example, Hong et al.18 and Lin et al.19 employed MAX–DOAS observations in Guangzhou to determine that diurnally, the near surface was basically in the VOC–limited regime in the morning, and in a transition limited regime in the noon and afternoon. Vertically, the OFS varied with height, from a VOC–limited regime (0.02–0.22 km) to a transition limited regime (0.22–0.42 km) and an NOx–limited regime (0.42–2.02 km). Hu et al.20 observed O3 and its precursor profiles (O3, HCHO, NO2) using MAX–DOAS in the NCP, YRD, and PRD simultaneously and found that the VOC–limited regime dominated near the surface, while the contribution of the NOx–limited regime increased with increasing altitude. Peng et al.21 combined the Geostationary Environmental Monitoring Spectrometer (GEMS) with MAX–DOAS measurements to analyze the OFS in the NCP region and noted that factors such as spatio–temporal characteristics and seasonality must be considered when formulating O3 control policies. Therefore, satellites combined with ground–based MAX–DOAS observations can provide a feasible method for investigating the spatio–temporal distribution of key O3 precursors and determining the OFS in the lower troposphere.

Previous studies based on satellites and MAX–DOAS have mainly focused on the spatio–temporal characteristics of the OFS over the entire observation period or O3 pollution period. The differences in meteorological conditions between O3 pollution days and clean days have been revealed by a large number of studies3,5,22. However, the differences in the spatio–temporal evolution characteristics of O3 precursors and the OFS between O3 pollution days and clean days have received less research attention18. Clarifying these differences is the key to accurately preventing and controlling O3 pollution and reducing the number of O3 exceedance days, because it is difficult to accurately understand the detailed driving force of ozone pollution by merely investigating the entire observation period or O3 pollution period, especially in the context of the high ozone precursor concentrations in eastern Chinese cities. In this study, long–term ground–based MAX–DOAS and satellite observations were combined to investigate the vertical structural features and differences, as well as the regional distribution features and differences, in the OFS in eastern Chinese cities under different ambient O3 levels. This study aims to provide a better understanding of the driving force and prevention methods for O3 pollution in eastern Chinese cities from the perspective of its vertical and spatial evolution. Note that all of the times mentioned in this study refer to the local time.

Results

Spatial and vertical variations in O3 and its precursors in eastern China

The concentration of industry, dense population, and heavy traffic in eastern China lead to high emissions of NOx and VOCs, which are the main precursors for O3 formation2,22. Influenced by anthropogenic emissions, vegetation activity, and heat source emissions23,24,25, the Taihang Mountains form a geographical barrier that prevents the transport of high concentrations of NOx and VOCs from the south to the north (Fig. 1). Thus, NO2 is mainly concentrated in urban agglomerations (0.6–1.3 \(\times {10}^{16}molecules/c{m}^{2}\)) in the NCP (Beijing, Tianjin, Hebei, Shandong, northern Henan, northern Anhui, and northern Jiangsu) and the Middle–Lower Yangtze River plain (MLYRP: Shanghai, southern Jiangsu, and northern Zhejiang). In contrast, the southeastern hilly areas (western Hubei, Hunan, Jiangxi, southern Anhui, and southern Zhejiang), which have high vegetation cover, have relatively low NO2 concentrations (less than 0.6\(\times {10}^{16}molecules/c{m}^{2}\)). High HCHO concentrations (1.1–1.9\(\times {10}^{16}molecules/c{m}^{2}\)) are distributed over a much wider area, occupying almost the entirety of eastern China, and the HCHO levels are mainly affected by industrial emissions, biogenic VOC in vegetation, and thermogenic emissions from biomass burning24,25. As a result, high HCHO concentrations exceeding 1.5\(\times {10}^{16}molecules/c{m}^{2}\) during April–October 2020–2023 are mainly concentrated on the southern side of the Taihang Mountains, and the HCHO concentrations north of the Taihang Mountains are below 1.1 \(\times {10}^{16}molecules/c{m}^{2}\).

Fig. 1: Spatial distribution of ozone and its precursors.
figure 1

Spatial distributions of tropospheric mean a NO2 VCD, b HCHO VCD, c surface mean MDA8O3, and d mean FNR derived from TROPOMI observations in eastern China during April–October, 2020–2023. The white stars in (a) represent the cities equipped with ground-based MAX–DOAS observations.

Full size image

High maximum daily 8–h average O3 (MDA8O3) concentrations (>85 ppb) are primarily distributed in the NCP (Fig. 1c), particularly in Hebei, Henan, Shandong, Jiangsu, and northern Anhui, with relatively low O3 concentrations in the southeastern hilly areas. Overall, O3 in eastern China shows an increasing spatial distribution from south to north. Differences in the regional distributions of O3 precursors (HCHO and NO2) inevitably lead to differences in the OFS. FNR derived by Tropospheric Monitoring Instrument (TROPOMI) shows that most cities in the NCP and the MLYRP have FNR values ranging from 1.2–2.6 (mainly in the VOC–limited regime, details refer to METHODS M4), which is considerably lower than that of the southeastern hilly area (>3, mainly in the transition limited regime or NOx–limited regime, details refer to METHODS M4). The spatial distribution of FNR exhibits a diametrically opposite trend to that of the O3 concentration (Supplementary Fig. 1), showing a significant negative correlation, with a Pearson’s correlation coefficient of –0.46 (p < 0.01). In other words, regions with high O3 concentrations have low FNR values and are more sensitive to VOC concentrations in eastern China.

Figure 2 shows the mean diurnal changes in the vertical distributions of NO2, HCHO, and FNR in Hefei, Huaibei, and Taian. The surface O3 concentrations in Hefei, Huaibei, and Taian increase rapidly from 08:00 to 13:00, with a significant slowdown after 13:00. The diurnal variation of the NO2 vertical distribution shows that NO2 is mainly confined below 0.3 km, with average concentrations of approximately 5.2 ppb in Hefei, 4.1 ppb in Huaibei, and 5.0 ppb in Tai’an. This is primarily attributed to ground–based emission sources26, such as residential areas, industrial facilities, and power plants. It is worth noting that the peak NO2 concentration occurs between 8:00 and 10:00, and the mean concentration exceeds 9.0 ppb in Hefei, 7.7 ppb in Huaibei, and 10.2 ppb in Taian, corresponding to the morning rush hour27. The decreased NO2 concentrations at noon and in the afternoon are attributed to the increase in light intensity during the warm season, which accelerates NO2 photolysis. The diurnal variation in the vertical distribution of HCHO is significantly different from that of NO2. It can be clearly seen that HCHO is mainly distributed below 0.6 km. The average HCHO vertical profiles in Hefei and Huaibei peak at the heights of 0.1–0.3 km (5.3 ppb in Hefei and 4.4 ppb in Huaibei) rather than at the bottom layer. The peak HCHO concentration in Taian after 11:00 also appears at 0.1–0.3 km. The most probable reason for the elevation of the HCHO layer during the period of 10:00–16:00 is that more active photochemical reactions occur in the elevated layer (0.1–0.3 km)28. For example, owing to the upward transport of its precursors (such as isoprene) to the elevated layer, it is oxidized to secondary HCHO under sunlight, which is consistent with the results of previous studies on atmospheric pollution in China29. Overall, the diurnal variation in HCHO during the warm season in eastern China shows a continuous upward trend in the morning, peaking at around 11:00, and the trend then remains relatively stable until the afternoon. This pattern may be attributed to the intense and persistent sunlight and favorable weather conditions during the warm season in eastern China, where vegetation releases large amounts of biogenic precursors30, and the intense solar radiation from midday to afternoon enhances the photochemical oxidation process, leading to the accumulation of HCHO near the surface.

Fig. 2: Diurnal variations of O3 and its precursors.
figure 2

Mean diurnal variations in the surface O3 concentrations in a Hefei, b Huaibei, and c Tai’an during April–October, 2020–2023. Mean diurnal variations in the d–f NO2 vertical distributions, g–i HCHO vertical distributions, and j–l FNR vertical distributions in Hefei, Huaibei, and Tai’an during April–October, 2020–2023.

Full size image

The OFS vertical distributions also show significant diurnal variations, influenced by the differences in the diurnal variations of the NO2 and HCHO vertical distributions. Figure 2j–l shows that the diurnal variations in the FNR vertical structures in Hefei, Huaibei, and Taian have almost the same trends. The FNR value below 0.5 km is less than 1.5 at 08:00–10:00, the thresholds for the transition limited regime of Hefei, Huaibei, and Tai’an were 1.94–2.43, 2.01–2.59, and 1.98–2.56, respectively (details refer to METHODS M4). Thus, the OFS is in the VOC–limited regime. With the enhancement of solar radiation, the intense photochemical reaction causes the HCHO concentration to increase, the NO2 concentration to decrease, and the FNR value to increase continuously, and it usually reaches a maximum value at 12:00–14:00 (1.8–2.1, mainly in the transition limited regime). After 14:00, the FNR value continues to decrease, and the OFS shifts to the VOC–limited regime again. In terms of the vertical distribution, the FNR values below 1.5 km continue to increase with height, which is caused by the difference in the vertical distribution of HCHO and NO2; NO2 is more concentrated near the surface, whereas high concentrations of HCHO can extend to higher altitudes. Based on the surface fitted FNR threshold, the OFS vertical distributions in eastern China can be summarised as follows: (1) In the early morning (8:00–10:00), the VOC–limited regime usually occurs below 0.5 km, the transition limited regime usually occurs between 0.5 and 0.7 km, and the NOx–limited regime usually occurs above 0.7 km. (2) At noon (12:00–14:00), a transition limited regime usually occurs below 0.3 km, and a NOx–limited regime usually occurs above 0.3 km.

Vertical difference of ozone precursors under different ozone pollution levels

Many previous studies have investigated the relationship between O3 and its precursors during O3 exceedance days10,31,32,33. In this study, we focus on investigating the differences in the spatial and temporal evolution of vertical distribution of O3 precursors (HCHO, NO2) under different O3 pollution levels. The entire observation period was classified into O3 level 0 days (clean days, MDA8O3 < 100 µg/m3), O3 level 1 days (light pollution days, 100 µg/m3 < MDA8O3 < 160 µg/m3), and O3 exceedance days (heavy pollution days, MDA8O3 > 160 µg/m3), details refer to METHODS M2. Supplementary Fig. 2 shows the mean diurnal variations in the surface O3 and∆O3 concentrations in Hefei, Huaibei, and Tai’an under different O3 pollution levels during April–October 2020–2023;∆O3 is defined as the difference in O3 concentration between two adjacent hours. Significant differences are observed in the diurnal variations in surface O3 concentrations in Hefei, Huaibei, and Tai’an under different O3 pollution levels. The cumulative O3 concentration is higher in the early morning on the O3 exceedance days. The mean O3 concentration at 7:00 on O3 level 0 days are 19.0, 20.4, and 20.4 ppb in Huaibei, Hefei, and Tai’an, respectively. The mean O3 concentration at 7:00 on O3 exceedance days in Huaibei, Hefei, and Tai’an are 24.3, 33.7, and 34.2 ppb, respectively. More importantly, the differences in the O3 growth rate between 8:00 and 14:00 under different O3 pollution levels in Hefei, Huaibei, and Tai’an are even larger, with the mean O3 growth rate on O3 level 0 days being 2.4, 2.1, and 2.4 ppb/h, respectively, while the mean O3 growth rate between 8:00 and 14:00 on O3 exceedance days is 10.7, 9.6, and 10.1 ppb/h, respectively. The maximum O3 growth rate in Hefei (14.8 ppb/h) and Huaibei (13.2 ppb/h) occurs at 10:00, and the maximum O3 growth rate in Tai’an (13.2 ppb/h) occurs at 11:00. Therefore, Accumulation of the previous night’s O3 concentration and the rapid increase in O3 concentration between 8:00 and 14:00 is the main reason for the O3 exceedance.

Figure 3 shows the mean diurnal variations in the NO2 vertical distributions at different O3 pollution levels in Hefei, Huaibei, and Tai’an during April–October, 2020–2023. The diurnal variation in the NO2 vertical distributions is essentially the same under different O3 pollution levels. The NO2 is mainly concentrated below 0.3 km, with high NO2 concentrations in the morning. NO2 photolysis occurs rapidly with an increase in sunlight intensity, and thus the NO2 concentration usually reaches its lowest value in the afternoon (13:00–15:00) and then increases slightly in the early evening. Further investigation of the difference in the diurnal variation of NO2 below 0.3 km on O3 exceedance days and O3 level 1 days reveals that the mean NO2 concentrations in Hefei, Huaibei, and Tai’an at 8:00–9:00 on O3 exceedance days are 1.22, 0.23, and 0.97 ppb higher than that on O3 level 0 days, respectively. Moreover, the mean NO2 concentrations below 0.3 km in Hefei, Huaibei, and Tai’an at 10:00–16:00 on O3 exceedance days are 1.18, 1.61, and 1.60 ppb lower than that on O3 level 0 days, respectively. This suggests that there is a greater accumulation of NO2 as a raw material for photochemical reactions in the early morning on O3 exceedance days. In addition, more rapid photolysis of NO2 occurs on O3 exceedance days, resulting in a large amount of O3 production. Meteorological conditions also support this conclusion, with higher downward ultraviolet radiation and temperature and lower relative humidity on O3 exceedance days (Supplementary Figs. 3–5). These meteorological conditions are conducive to the photolysis of NO2, resulting in lower NO2 concentrations at noon and in the afternoon on O3 exceedance days34,35,36,37. The NO2 concentrations below 0.3 km in Hefei (0.43 ppb) and Tai’an (1.99 ppb) at 8:00–9:00 on O3 level 1 days are lower than those on O3 level 0 days, whereas the NO2 concentration in Huaibei (0.03 ppb) is slightly increased. The NO2 concentrations in Hefei, Huaibei, and Tai’an at 10:00–16:00 on O3 level 1 days are significantly lower than those on O3 level 0 days, which may also be related to meteorological conditions (higher downward ultraviolet radiation and temperature).

Fig. 3: Diurnal variations in the NO2 vertical distributions.
figure 3

Mean diurnal variations in the NO2 vertical distributions on O3 a–c level 0, d–f level 1, and g–i exceedance days in Hefei, Huaibei, and Tai’an during April–October, 2020–2023. Mean diurnal variations in the difference in NO2 vertical distributions between O3 level 1 days and O3 level 0 days in j Hefei, k Huaibei, and l Tai’an during April–October, 2020–2023. Mean diurnal variations in the difference in NO2 vertical distributions between O3 exceedance days and O3 level 0 days in m Hefei, n Huaibei, and o Tai’an during April–October, 2020–2023.

Full size image

Significant increases in HCHO concentrations on O3 exceedance days and O3 level 1 days relative to those on O3 level 0 days occur throughout the day (Fig. 4). The mean HCHO vertical profile on O3 exceedance days is Gaussian in shape, with peaks typically occurring at altitudes of 0.1–0.3 km, which is caused by a combination of primary emissions of surface HCHO and photochemical oxidation processes at high altitudes. On average, the HCHO concentrations below 1.0 km on O3 exceedance days in Hefei, Huaibei, and Tai’an are approximately 1.3–2.1, 1.1–2.0, and 1.6–2.6 times higher than those on O3 level 0 days, respectively. There may be two reasons for the large difference between the HCHO concentrations on O3 exceedance days and O3 level 0 days. First, there is more secondary generation of HCHO in an O3–rich environment18, i.e., under favorable meteorological conditions (higher downward ultraviolet radiation and temperature, lower relative humidity), photooxidation of VOCs is promoted, thus increasing the HCHO concentration. Second, hot and dry conditions also promote natural emissions, e.g. vegetation emissions of VOCs38, which further contribute to the increase in HCHO concentration.

Fig. 4: Diurnal variations in the HCHO vertical distributions.
figure 4

Mean diurnal variations in the HCHO vertical distributions on O3 a–c level 0, d–f level 1, and g–i exceedance days in Hefei, Huaibei, and Tai’an during April–October, 2020–2023. Mean diurnal variations in the difference in HCHO vertical distributions between O3 level 1 days and O3 level 0 days in j Hefei, k Huaibei, and l Tai’an during April–October, 2020–2023. Mean diurnal variations in the difference in HCHO vertical distributions between O3 exceedance days and O3 level 0 days in m Hefei, n Huaibei, and o Tai’an during April–October, 2020–2023.

Full size image

Therefore, abundant O3 precursors (higher NO2 and HCHO concentrations) are generally present in the early morning on O3 exceedance days, and under favorable meteorological conditions, the rapid photolysis of NO2 produces a large amount of O3, causing the NO2 concentration to decline rapidly and reach its lowest value near the surface in the afternoon. The secondary formation of HCHO in O3–rich environments is more intense and reaches its maximum value at 0.1–0.3 km at noon. This result also indicates that O3–rich environments may be more favorable for photochemical reactions at high altitudes. In addition, the diurnal variations in the NO2 and HCHO concentrations are more drastic on O3 exceedance days than on O3 level 0 days.

Vertical differences in ozone formation sensitivities under different ozone pollution levels

Figure 2 shows that the mean FNR vertical profile exhibits a Gaussian shape, with a peak occurring around 1.0–1.8 km. At lower altitudes (0–0.5 km), the estimated mean FNRs are lower than 2.5, indicating that the surface OFS is controlled by the VOC–limited regime. However, FNR is a diagnostic indicator of OFS that varies with time and height18,19, and changes in long–term mean values usually cannot accurately reflect the transition of OFS with time and height. Therefore, based on the HCHO and NO2 vertical profiles under different O3 pollution levels, the occurrence probabilities of OFS at different heights in Hefei (Fig. 5), Huaibei (Supplementary Fig. 6), and Tai’an (Supplementary Fig. 7) during April–October 2020–2023 are classified and statistically analyzed. It is worth noting that the FNR here is derived using the hourly average NO2 and HCHO, and the determination threshold (details refer to METHODS M4) is applied to determine the VOC–limited regime, transition limited regime, and NOx–limited regime. Finally, the occurrence probabilities of the VOC–limited regime, transition limited regime, and NOx–limited regime are determined at different altitudes and different periods.

Fig. 5: Diurnal variations in the O3 formation sensitivity.
figure 5

Occurrence probability of diurnal variations in the a VOC-limited regime, b transition-limited regime, and c NOx-limited regime vertical distributions on O3 level 0 days; d VOC-limited regime, e transition-limited regime, and f NOx-limited regime vertical distributions on O3 level 1 days; and g VOC-limited regime, h transition-limited regime, and i NOx-limited regime vertical distributions on O3 exceedance days in Hefei during April–October, 2020–2023.

Full size image

On O3 level 0 days, the VOC–limited regime contributes most significantly to the surface OFS in Hefei (>79%), Huaibei (>67%), and Tai’an (>87%), and the contributions of the transition limited regime and NOx–limited regime are less than 20% (Fig. 5, Supplementary Figs. 6, 7). With increasing altitude, the contribution of the VOC–limited regime gradually decreases, the contribution of the NOx–limited regime increases, and the contribution of the transition limited regime remains almost constant. Above 1 km, the contribution of the VOC–limited regime reaches a minimum (approximately 28%, 27%, and 10% in Hefei, Huaibei, and Tai’an, respectively), and the contribution of the NOx–limited regime reaches a maximum (approximately 64%, 67%, and 84% in Hefei, Huaibei, and Tai’an, respectively). Diurnally, the VOC–limited regime has the greatest contribution at 08:00 and 16:00 (more than 85% below 0.3 km in Hefei, Huaibei, and Tai’an), and the contributions of the transition limited regime (approximately 10%, 12%, and 5% below 0.3 km in Hefei, Huaibei, and Tai’an, respectively) and NOx–limited regime (approximately 17%, 22%, and 10% below 0.3 km in Hefei, Huaibei, and Tai’an, respectively) increase at midday (12:00–14:00). Similarly, diurnal and vertical variations in the OFS have been observed in other cities in China19,20.

The height–temporal evolution of the OFS occurrence probability on O3 exceedance days and O3 level 1 days is generally similar to that on O3 level 0 days, but there are three differences (Fig. 5, Supplementary Figs. 6, 7). First, the contribution of the NOx–limited regime at high altitudes increases significantly, particularly on O3 exceedance days. The average contributions of the NOx–limited regime within 0.5–2 km on O3 level 1 days in Hefei, Huaibei, and Tai’an are 66%, 55%, and 78%, respectively. The average contributions of the NOx–limited regime within 0.5–2 km on O3 exceedance days in Hefei, Huaibei, and Tai’an are 77%, 78%, and 86%, respectively. Second, the average contributions of the surface transition limited regime and NOx–limited regime increase at midday (12:00–14:00). The average contributions of the NOx–limited regime (transition limited regime) on O3 level 1 days in Hefei, Huaibei, and Tai’an are 27% (16%), 28% (21%), and 21% (21%), respectively, and the average contributions of the NOx–limited regime (transition limited regime) on O3 exceedance days in Hefei, Huaibei, and Tai’an are 17% (30%), 22% (16%), and 8% (16%), respectively. Third, the contribution of the transition limited regime increased significantly on O3 exceedance days, particularly at 0.1–0.4 km at midday (12:00–14:00), with average contributions of the transition limited regime in Hefei, Huaibei, and Tai’an of 22%, 29%, and 30%, respectively. Overall, the occurrence probability of the VOC–limited regime decreases on O3 exceedance days, while the occurrence probabilities of the transition limited regime and NOx–limited regime increase, especially at high altitudes on O3 exceedance days.

Furthermore, the OFS is generally in the VOC–limited regime on O3 exceedance days during the period of a sharp rise in O3 concentrations (8:00–12:00), and the contributions of the transition limited regime or NOx–limited regime increases when the O3 concentration peaks (12:00–14:00). Therefore, the reduction in the VOC concentrations is the key to reversing the current high O3 levels, while control of the NOx concentration cannot be neglected to further reduce the peak O3 concentration and number of O3 exceedance days. When formulating O3 control policies, it is important to consider diurnal and vertical variations in the OFS while addressing O3 precursor emissions at different times. Measures such as vehicle restrictions and the adoption of renewable energy sources can be implemented at midday, and more stringent regulations of chemical plants are needed in the early morning and mid–morning to reduce the release of VOCs, e.g., by delaying the start–up time of chemical plants. The goal is to reduce VOC emissions during the period of a sharp rise in O3 concentrations and reduce NOx emissions at midday and afternoon based on the primary OFS regime during these times.

Spatial differences in ozone formation sensitivities under different ozone pollution levels

Figure 6 shows the occurrence probabilities of the VOC–limited regime, transition limited regime, and NOx–limited regime spatial distributions derived from TROPOMI observations under different O3 pollution levels in eastern China during April–October, 2020–2023. The OFS of most cities in the NCP and MLYRP is mainly in the VOC–limited regime on O3 level 0 days, whereas the OFS of most cities in the southeast hills is mainly in the NOx–limited regime on O3 level 0 days. On O3 level 1 days, the occurrence probability of the VOC–limited regime decreases for most cities in the NCP and MLYRP, whereas the contribution of the NOx–limited regime decreases in most cities in the southeastern hills, and this trend is more significant on O3 exceedance days. The regional mean contributions of the VOC–limited regime, transition limited regime, and NOx–limited regime on O3 level 0 days in the NCP (MLYRP) are 83% (72%), 10% (15%), and 7% (14%), respectively. The regional mean contribution of the VOC–limited regime in the NCP (MLYRP) decrease to 39% (63%) on O3 exceedance days, and the regional mean contribution of the transition limited regime in the NCP (MLYRP) increases to 38% (26%) on O3 exceedance days. On O3 level 0 days, the regional mean contributions of the VOC–limited regime, transition limited regime, and NOx–limited regime in the southeastern hilly region are 18%, 24%, and 58%, respectively. On O3 exceedance days, the regional mean contribution of the NOx–limited regime in the southeastern hilly region decreases to 34%, and the regional mean contribution of the transition limited regime increases to 35%. Therefore, targeted O3 control measures are necessary for different regions. VOC emissions control should mainly focus on most cities in the NCP and MLYRP, and a possible recommendation is to avoid planting trees with high VOC emissions21, such as poplars, pines, and abies, in areas with high NOx emissions, such as roadsides and power plants. However, NOx emissions control should be mainly concentrated in the cities in the southeastern hills.

Fig. 6: Spatial distribution of O3 formation sensitivity.
figure 6

Occurrence probabilities of the a VOC-limited regime, b transition-limited regime, and c NOx-limited regime spatial distributions on O3 level 0 days; d VOC-limited regime, e transition-limited regime, and f NOx-limited regime spatial distributions on O3 level 1 days; and g VOC-limited regime, h transition-limited regime, and i NOx-limited regime spatial distributions on O3 exceedance days in eastern Chinese cities derived by TROPOMI observations during April–October, 2020–2023.

Full size image

The shift in OFS at different O3 pollution levels may depend mainly on the photochemical conditions and emission characteristics of O3 precursors. Figure 7a, d shows the differences in NO2 concentrations observed by TROPOMI. NO2 is lower in most cities in the NCP on O3 level 1 days and O3 exceedance days relative to O3 level 0 days (Supplementary Fig. 8), because high temperatures and strong ultraviolet radiation are necessary for high O3 concentrations, and these favorable meteorological conditions are more conducive to the photolysis reaction of NO2. In contrast, NO2 concentrations in most cities in the southeastern hills increase on O3 level 1 days and O3 exceedance days compared with the NO2 concentrations on O3 level 0 days. Most cities in the southeastern hills are under the NOx–limited regime; thus, a high NO2 concentration is a necessary condition for inducing high O3 concentrations. A significant limitation of the TROPOMI observations is its inability to capture diurnal variations in air pollutant concentrations. Here, we further analyze the spatial distribution of surface NO2 concentrations measured by Ministry of Ecology and Environment of China (MEE) in eastern China (Supplementary Figs. 9, 10). Consistent with the MAX–DOAS measurements, the NO2 concentration is considerably higher in most cities in eastern China in the early morning on O3 level 1 days and O3 exceedance days than the NO2 concentrations on O3 level 0 days. Under favorable meteorological conditions (higher downward ultraviolet radiation and temperature and lower relative humidity), the rapid photolysis of NO2 in the morning leads to a rapid decrease in NO2 concentrations and a rapid increase in O3 concentrations in the late afternoon.

Fig. 7: Spatial distributions of the difference in O3 precursors.
figure 7

Spatial distributions of the difference in tropospheric a NO2 VCD, b HCHO VCD, and c FNR between O3 level 1 days and O3 level 0 days, and the difference in tropospheric d NO2 VCD, e HCHO VCD, and f FNR between O3 exceedance days and O3 level 0 days in eastern China during April–October, 2020–2023.

Full size image

The differences in HCHO concentrations under different O3 pollution levels are also consistent with the MAX–DOAS measurements, where favorable meteorological conditions on O3 level 1 days and O3 exceedance days enhance the photochemical reactions and increase the secondary HCHO concentrations as well as the emission of biogenic VOCs from vegetation. It is worth noting that the FNR values of most cities in the NCP and MLYRP on O3 level 1 days and O3 exceedance days increase compared with those on O3 level 0 days, whereas the FNR values of most cities in the southeastern hills on O3 level 1 days and O3 exceedance days decrease compared with those on O3 level 0 days. Therefore, the difference in FNR values between northern and southern cities in eastern China on O3 exceedance days narrows, and the OFS tends to be higher in the transition limited regime (Fig. 6h). This also implies that although there is a difference in the importance of O3 precursor emission control targets (NOx in the south and VOCs in the north) between northern and southern cities in eastern China, this does not mean that controlling only one of the precursors can achieve the best effect of O3 prevention, and the most appropriate ratio of O3 precursor emission reduction should be designed according to different regions.

Discussion

Owing to the limitations of the observational data, the analysis of vertical differences in O3 precursors and OFS under different O3 pollution levels is limited to three cities in eastern China. Although other cities in eastern China were further investigated using satellite observations, TROPOMI only provides observation results for column concentrations at approximately 13:30 each day, which did not allow us to obtain diurnal and vertical variations in the OFS. Further observations must be extended to southern and coastal cities to investigate the relationship between O3 and its precursors more comprehensively. For example, Wang et al.39 found that, driven by weather systems and mesoscale recirculation, the interaction between continental and marine air masses profoundly altered the atmospheric composition, which further affected the formation and redistribution of O3 in coastal areas, resulting in a significant difference in the photochemical processes of O3 between coastal and inland areas. When continental air intrudes into the marine atmosphere, O3 pollution at sea is amplified, and the elevated O3 will be recirculated to the coast near the shore by the sea breeze, resulting in increased O3 pollution in coastal cities. Another limitation of the study is that we used fixed FNR thresholds at different heights, a fixed threshold may underestimate the contribution of upper layer VOC–limited regime and overestimate the contribution of upper layer NOx–limited regime according to previous investigation20. In the future, OFS can be further diagnosed using height–dependent FNR thresholds to obtain more accurate results.

By combining ground–and satellite–based hyperspectral remote sensing with ground station observations, this study investigated the vertical and spatial features and differences in the OFS in eastern Chinese cities under different ambient O3 levels. The diurnal trends in the NO2 and HCHO vertical distributions were similar under different O3 pollution levels. The NO2 concentration was mainly concentrated below 0.3 km, with high NO2 concentrations in the morning. Compared with the O3 level 0 days, there was more accumulation of NO2 as a raw material for photochemical reactions in the early morning on O3 exceedance days, and more rapid photolysis of NO2 occurred on O3 exceedance days, resulting in a large amount of O3 production. The mean HCHO vertical profile under different O3 pollution levels was Gaussian in shape, with peaks typically occurring at altitudes of 0.1–0.3 km; compared with O3 level 0 days, the HCHO concentration was higher throughout the day on O3 exceedance days.

There were also significant differences in the OFS under different O3 pollution levels. The contribution of the NOx–limited regime at high altitudes increased significantly on O3 exceedance days compared with that on O3 level 0 days. In addition, the contribution of the transition limited regime also increased significantly on O3 exceedance days, particularly at 0.1–0.4 km at midday (12:00–14:00). Furthermore, the OFS was generally in the VOC–limited regime on O3 exceedance days during the period of a sharp rise in O3 concentrations (8:00–12:00), and the contributions of the transition limited regime or NOx–limited regime increased when the O3 concentration peaked (12:00–14:00). Therefore, reducing the VOC concentrations is the key to reversing the current high O3 levels, while controlling NOx concentrations cannot be neglected to further reduce the peak O3 concentration and the number of O3 exceedance days. Spatially, the OFS of most cities in the NCP and MLYRP was mainly in the VOC–limited regime on O3 level 0 days, whereas the OFS of most cities in the southeastern hills was mainly in NOx–limited regime on O3 level 0 days. On O3 level 1 days and O3 exceedance days, the difference in FNR values between northern and southern cities in eastern China on O3 exceedance days narrowed, the occurrence probability of the VOC–limited regime decreased for most cities in the NCP and MLYRP, the contribution of the NOx–limited regime of most cities in the southeastern hills decreased, and the OFS in eastern China tended to be under a transition limited regime.

Methods

M1 MAX–DOAS measurements

The ground–based MAX–DOAS stereo remote sensing network was continuously measured from December 22, 2020, to May 15, 2023; April 12, 2019, to May 27, 2022; and July 15, 2021, to May 15, 2023, in Hefei, Huaibei, and Tai’an, respectively. The three cities transition sequentially from south to north (Fig. 1a) to obtain the vertical profiles of O3 precursors in eastern China. The MAX–DOAS instrument consists of a telescope, two spectrometers, and a controlling computer. MAX–DOAS obtains trace gas vertical profiles from scattered sunlight observed at different elevation angles, with a complete scan lasting approximately 12 min. The system operates only during the daytime (08:00–16:00 local time) with a temporal resolution of 15 min, a spatial resolution of 100 m for altitudes up to 1 km, and a spatial resolution of 200 m for altitudes of 1–2 km. Detailed information regarding the NO2 and HCHO vertical profile retrieval algorithms and their validation can be found in our previous studies40,41,42,43.

In addition, we compared the surface NO2 concentrations observed by the MAX–DOAS with the urban surface NO2 concentrations measured by the MEE (Supplementary Fig. 11), and the results were comparable to those reported in previous studies19,44, with Pearson’s correlation coefficients for Hefei, Huaibei, and Tai’an of 0.74 (P < 0.01), 0.66 (P < 0.01), and 0.73 (P < 0.01), respectively. Therefore, the MAX–DOAS retrieval is considered reliable. The differences between the MAX–DOAS and MEE observations arise from two components. First, the detection modes differ: the urban NO2 concentration observed by MAX–DOAS is the result of scanning along a fixed direction, whereas the urban NO2 concentration observed by MEE is the result of in situ sampling. Second, there are some differences in the observation locations, as the urban NO2 concentration from the MEE is the average of several in situ stations, whereas only one MAX–DOAS observation site is equipped in each city.

M2 Surface measurements

In this study, we investigated the OFS in 156 cities in eastern China. The spatial distribution of these 156 cities is shown in Fig. 1c. The data for the MDA8O3, hourly O3 concentration, and hourly NO2 concentration in 156 cities from April to October in 2020–2023 were obtained from the MEE open access website, https://www.mee.gov.cn (last accessed: July 7, 2024) and archived at https://quotsoft.net/air/ (last accessed: July 7, 2024). After August 31, 2018, the MEE reported a concentration unit of µg/m3 at the reference state (298 K, 1013 hPa), and the mass concentrations (µg/m3) at each site were converted to volume mixing ratios (VMRs, ppb). To explore the O3 formation sensitivities under different O3 pollution levels, the entire observation period was classified into O3 level 0 days (clean days, MDA8O3 < 100 µg/m3), O3 level 1 days (light pollution days, 100 µg/m3 < MDA8O3 < 160 µg/m3), and O3 exceedance days (heavy pollution days, MDA8O3 > 160 µg/m3) according to the “China Ambient Air Quality Index Technical Regulations (Trial)” (HJ633–2012) standard (Supplementary Table 1). To discuss the differences in the O3 formation sensitivities under different O3 pollution levels, the O3 precursor concentration on O3 level 0 days was used as a reference, and the differences in the O3 precursor concentrations under different O3 pollution levels were defined as the O3 precursor concentration on O3 level 1 days minus the O3 precursor concentration on O3 level 0 days, and the O3 precursor concentration on O3 exceedance days minus the O3 precursor concentration on O3 level 0 days.

M3 Satellite observations

The TROPOMI onboard the European Space Agency’s Copernicus Sentinel 5 precursor satellite can provide daily global coverage of HCHO and NO2 vertical column density (VCD) observations45,46,47. TROPOMI has a high spatial resolution of 3.5 × 7.0 km and an overpass of approximately 13:30. In this study, TROPOMI NO2 (“S5P_OFFL_L2__NO2…”) and HCHO (“S5P_OFFL_L2__HCHO…”) tropospheric VCD products from April to October 2020–2023 were employed48 (downloaded from https://search.earthdata.nasa.gov/search; Last visited: January 7, 2024). The recommended quality control (QC, in the 0–1 range) filter was used, and HCHO retrieval values with QC values less than 0.5 and NO2 retrieval values with QC values less than 0.75 were excluded. To match the O3 concentrations measured at the MEE sites, we averaged the TROPOMI observations (HCHO and NO2 tropospheric VCDs) over a range of 0.2° at the surface MEE urban stations.

The NO2 and HCHO tropospheric VCDs measured using MAX–DOAS and TROPOMI were also compared. Considering the TROPOMI satellite transit time of around 13:30, the mean MAX–DOAS values from 13:00–14:00 were used for comparison. The TROPOMI observations were averaged over a range of 0.2° at the ground–based MAX–DOAS station. Comparisons of the tropospheric VCDs of NO2 and HCHO observed by MAX–DOAS and TROPOMI are shown in Supplementary Figs. 12 and 13. In general, the MAX–DOAS and TROPOMI observations are relatively consistent, with Pearson correlation coefficients of 0.99 (0.74), 0.96 (0.66), and 0.96 (0.73) for NO2 (HCHO) in Hefei, Huaibei, and Tai’an, respectively. The difference in comparisons between MAX–DOAS and TROPOMI observations can be caused by many factors, such as different observation modes, pixel size, fitting errors, a priori model bias, cloud and aerosols, and spatio–temporal resolution48.

M4 Diagnosing the ozone formation sensitivity

Based on the monthly mean surface midday (13:00–14:00) O3 concentration observed by MEE in 156 cities in eastern China from April to October 2020–2023, and the monthly mean HCHO and NO2 tropospheric VCDs observed by TROPOMI during the same period, the OFS was determined using the method proposed by Jin et al.49 and Ren et al.14 to derive FNR thresholds for transition limited regimes. The data were arranged according to the order of FNR values from small to large, forming bins at intervals of 0.1. Then, the O3 concentration of the grouping was plotted as a function of the grouping FNR. O3 initially increased and then decreased with an increase in FNR. Therefore, a polynomial rather than a linear function was used to fit the FNR curve. We used the third–order polynomial model to derive the maximum average O3 concentration (peak of the curve in Fig. 8a). Assuming that the peak of the curve (where the curve slope is 0) marks the transition from a VOC–limited regime to a NOx–limited regime, the transition zone (transition limited regime) was defined as the range of the curve with slopes between -3 and +3. Figure 8a summarizes all of the available observations and examines their empirical relationships with the observations of all cities in eastern China. It is found that the O3 peaks at FNR = 3.31 and the transition limited regime ranges from 2.67 to 4.04. Therefore, based on the TROPOMI observations, we roughly defined the following three states: FNR < 2.67 is the VOC–limited regime; FNR > 4.04, is the NOx–limited regime; 2.67 < FNR < 4.04 represents a transition limited regime. This result is similar to that of a recent study reporting FNR thresholds for major cities in China, which recommended FNR values between 2.3 and 4.2 as a transition limited regime50.

Fig. 8: Diagnosing the O3 formation sensitivity.
figure 8

Fitting of a MEE-observed surface monthly mean O3 concentrations as a function of the binned FNR measured by TROPOMI during April–October, 2020–2023. Fitting of the MEE-observed surface hourly mean O3 concentrations as a function of the binned FNR measured by MAX–DOAS in b Hefei, c Huaibei, and d Tai’an. The solid blue line represents the third-order polynomial fitting, and the blue shadow represents the 95% confidence band. The correlation coefficients are shown at the top right of each panel, and the superscript asterisk indicates P < 0.01. The vertical red dotted line represents the maximum value of the fitted curve, and the vertical red shadow represents the range of the curve slope from −3 to +3 (transition regime). The area to the left of the red shading is the VOC-limited regime and the area to the right is the NOx-limited regime. The labels at the top right of each panel represent the threshold values for the NOx-limited regime (high) and VOC-limited regime (low).

Full size image

Based on MEE–observed surface hourly O3 concentrations, MAX–DOAS retrieved the HCHO and NO2 profiles, and a similar approach was used to determine the FNR thresholds. First, the hourly averages of the surface HCHO and NO2 concentrations observed by MAX–DOAS were used to match the O3 concentrations observed by the MEE. The available observational values in Hefei, Huaibei, and Tai’an from April to October 2020–2023 were then summarized, and bins were formed at intervals of 0.1 according to the FNR values arranged in order from small to large. Finally, the grouping O3 concentration was plotted as a function of the grouping FNR, and the third–order polynomial fitting is shown in Fig. 8b–d. The thresholds for the transition limited regime of Hefei, Huaibei, and Tai’an were 1.94–2.43, 2.01–2.59, and 1.98–2.56, respectively. In addition, we hypothesised that the FNR thresholds for determining the OFS would not change markedly with height by referring to recent studies by Liu et al.51, Lin et al.19, and Hong et al.18, because the limitations of our research, the retrieval of the ozone profile cannot be realized52.