**2. Photochemical Formation Mechanism of Tropospheric O3**

Most of the tropospheric O3 is generated due to the photochemical reactions of some primary pollutants, such as NOx and VOCs, under the strong sunlight in the troposphere. Some main reactions of the formation and loss mechanisms of tropospheric ozone are summarized in Table 1.

In the reactions in Table 1, R and M stand for organic group and other matters in the atmosphere, respectively. Tropospheric O3 is formed by the photolysis of NO2, with the reactions R1 and R2 [16]. The three reactions from R1 to R3 constitute a rapid cyclic process, which can reach a dynamic equilibrium under certain conditions without causing an increase in the total amount of O3 when no other chemical species are involved. However, in the atmosphere polluted by organic matter, peroxy radicals (such as RO2· and HO2·) can replace the O3 in reaction R3, so the conversion of NO to NO2 does not need to consume O3, but the continuous reactions of R1 and R2 occur subsequently, thereby destroying the photochemical reaction cycle of NO2-NO-O3, resulting in the accumulation of O3. The rate of photochemical O3 production is primarily determined by the reaction of NO with peroxy radicals such as RO2· and HO2·, with the reactions of R4 and R5. Peroxy radicals

RO2 and HO2· can be produced by the reactions of ·OH with hydrocarbon (abbreviated as RH) and CO, with the reactions from R6 to R9.


**Table 1.** Main reactions of the formation and loss mechanisms of tropospheric ozone.

There is a series of chain reactions centered on various free radicals, resulting in the accumulation of O3 [17]. In the clean troposphere, the ·OH radicals are mainly derived from the reaction of water vapor with O (1D) atoms, which are usually produced by the photolysis of O3, with the reactions R10–R11. In the polluted troposphere, the OH radicals are mainly formed from the photolysis of HONO, with reaction R12. At the same time, O3 can be removed from the atmosphere by some reactions such as R3, R10 and R13–R15. Hence, the net generation rate of O3 is equal to the total generation rate minus the removal rate. It was reported that the destruction of O3 could occur in many ways, and the most important pathway is the surface deposition [18]. For example, O3 consumption pathways can be achieved by oxidation of SO2 in the liquid phase reaction. The rates of these reactions vary greatly depending on the meteorological and photolysis conditions, in addition to the rate of competitive transport and removal processes.

#### **3. Spatiotemporal Distribution of Tropospheric Ozone in China**

Tropospheric O3 exhibits different characteristics in different regions. Understanding the spatiotemporal characteristics of O3 concentration is essential for controlling atmospheric O3 pollution. Since 2012, the Chinese government has included atmospheric O3 as a regular pollutant monitoring indicator, and the national monitoring network has brought convenience to the study of the spatial and temporal characteristics of atmospheric O3.

Most of Chinese population lives in the east of China, especially in the three most developed regions of Jing-Jin-Ji (JJJ, including Beijing, Tianjin, and Hebei province), Yangtze River Delta (YRD, including Shanghai, Zhejiang, Jiangsu, and Anhui provinces), and Pearl River Delta (PRD, including nine cities in south-central of Guangdong province). These regions are also the areas with the highest emissions of anthropogenic NOx and VOCs, thus leading to serious regional atmospheric ozone pollution. Therefore, these regions are the key areas for preventing and controlling air pollution. Figure 1 shows the spatial distribution of annual average O3-max-8 h in China from 2013 to 2018 [19]. The overall O3 concentration presented a spatial distribution pattern of higher in the east and lower in the west. The high-value areas of O3-max-8 h were mainly concentrated in the North China Plain in the east, such as Hebei province and Shandong province, where O3-max-8 h was higher than 180 μg·m<sup>−</sup>3; followed by the Yangtze River Delta and its nearby areas with an O3-max-8 h ranging from 120 to 160 μg·m<sup>−</sup>3. The O3-max-8 h in the southern Pearl River Delta region was also in the range of 120 to 160 μg·m<sup>−</sup>3, but the high-value area was smaller than the Yangtze River Delta area. The O3-max-8 h was lower in the western

region, ranging from 70 to 100 μg·m<sup>−</sup>3, and reaching as low as 62 μg·m−<sup>3</sup> in the Hami area. The spatial distribution of O3-max-8 h was consistent with the distribution pattern of its precursor emissions. The NOx emission intensity was higher in the east than that in the west, with the highest values distributing in the Beijing–Tianjin–Hebei region, Yangtze River Delta and Pearl River Delta. From 2013 to 2018, the 90th percentile of O3-max-8 h concentration in China gradually increased with an annual growth rate of 2.6 μg·m−<sup>3</sup> per year. The highest O3-max-8 h (≥180 μg·m<sup>−</sup>3) zone mainly occurred in the North China and Yangtze River Plains, which gradually expanded in the North China Plain (NCP) while shrinking in the YRD and PRD.

**Figure 1.** Spatial distribution of annual average O3-max-8 h in China from 2013 to 2018. Reprinted with a permission from ref. [19]. Copyright 2021 Li Ze Yuan.

Based on the data of ozone monitoring instruments (OMI) from 2005 to 2014, the tropospheric ozone trend in mid-eastern China (including 10 major cities) was studied [20]. The results showed that the mixing ratios of tropospheric ozone column were fairly stable, but those of ground-level clearly increased, by 12.38%. The concentration of ground-level ozone reached the maximum value from May to June, while the minimum value was from November to December. The concentrations of ground-level ozone increased with the cumulative increments of 6.3, 6.6, and 10.2 ppbv (parts per billion by volume) in Beijing, Shijiazhuang and Tianjin, respectively, from 2005 to 2014. Additionally, the concentration of ground-level ozone increased rapidly in Tianjin during 2012-2014, showing an increase of 13.25% compared with 2010–2011, which might be due to the more rubber and chemical companies around Tianjin. In contrast, the concentration of ground-surface O3 in the Beijing area showed a slower rising trend from 2005 to 2014. According to previous studies, atmospheric O3 pollution often appeared in the region of Beijing–Tianjin–Hebei, among which Beijing and Baoding were more polluted [21,22].

The temporal and spatial distribution characteristics of atmospheric O3 in the Beijing– Tianjin–Hebei region during 2013–2015 indicated that O3 concentration presented obvious seasonal variation, with the highest concentration in late spring and summer, and showed a single peak distribution during daytime, with the maximum value appearing around 15:00. In contrast, the concentration was lower and had little fluctuation throughout the

day in autumn and winter. The higher values of O3-max-8 h were mainly distributed in north-central Beijing, Chengde and Hengshui [23]. The seasonal variations of tropospheric O3 concentration distribution in Beijing, Shanghai, Guangzhou and Chengdu were similar, with the highest value generally occurring in summer and the lowest value generally appearing in winter [24]. Table 2 summarizes the tropospheric ozone concentrations in some regions of China.


**Table 2.** Summary of tropospheric ozone concentrations in some regions of China.

1. O3-1 h: Maximum 1 h average; O3-8 h: Maximum 8 h average. 2. Note: \* For rough estimates from the literature.

The summer–winter differences are due to the general meteorological conditions including the variability of irradiation levels affecting free-tropospheric and boundarylayer photochemistry, which is also one of the main sources of the high background O3 on the surface [9]. From 2013 to 2019, the weather in the North China Plain (NCP) drove an increase in surface O3 [34]. The hot weather in the NCP in summer is usually driven by a wide range of anticyclone conditions, which is regarded as a typical climate pattern for the number of days of O3 pollution [33]. The influence of the boundary layer on ozone in the summer afternoon cannot be ignored. Under the conditions of free convection, the stronger the ultraviolet radiation (UV), the higher the temperature, the lower the relative humidity (RH) and the higher the boundary-layer height (BLH), the more serious the ozone pollution was in Shijiazhuang in summer of 2018–2019 [35]. The increase in radiation during the day may cause the boundary layer to rise, and the accumulated O3 may mix down to the boundary layer, affecting the near-surface ozone concentration. The history of the air mass is an important factor in determining the magnitude and potential signs of the impact of entrainment on surface O3 through atmospheric boundary-layer growth [36]. As the height of the boundary layer increases, the O3 in the residual layer (RL) is transported to the boundary layer. Some studies have found that the mixed ozone from the RL contributes 50–70% of the maximum concentration near the surface for the next day, and the rest comes from chemical production and possible advection [37,38].

Based on the tropospheric O3 concentration data of 16 urban monitoring stations from June 2013 to May 2014, the spatiotemporal distribution characteristics of atmospheric O3 in the Yangtze River Delta region were studied [27]. The results showed that the annual

average O3 concentration was higher in the cities near the sea and lower in the cities that are inland. The concentration of atmospheric O3 showed a seasonal variation, with higher concentration in summer and lower concentration in winter. The higher O3 pollution area was located in the north of Hangzhou Bay in summer, while the higher O3 pollution area was located in the eastern coastal zone in winter. The diurnal variation of O3 concentration in the Yangtze River Delta was unimodal throughout the four seasons. The daily minimum O3 concentration appeared around 06:00 in summer, and was delayed by about one hour in the other seasons, and the daily maximum O3 concentration appeared around 15:00 in all seasons. Based on the OMI data, the spatiotemporal distribution of the tropospheric O3 in the Yangtze River Delta region showed a significant zonal difference, increasing with latitude [39]. According to the monitoring data of 72 state-controlled stations in Jiangsu province from 2013 to 2015, the spatiotemporal distribution characteristics of tropospheric O3 were studied [26]. The results indicated that the annual mean value of atmospheric O3 in Jiangsu province showed a significant spatial difference, with the concentration gradually decreasing from coast to inland. Tropospheric O3 showed the highest concentration in Yancheng city, while it was lower in Changzhou, eastern Wuxi and Xuzhou cities. The 90th percentile concentrations of O3-8 h were significantly different from north to south. The atmospheric O3 concentration was relatively higher in the cities of Nanjing, Yangzhou and Zhenjiang, while the lower concentrations were found in the cities of Xuzhou and Suqian. It was reported that the atmospheric O3 peak occurred in the afternoon in Shanghai from 2006 to 2016 [2]. The areas with O3 concentration exceeding the limit of Chinese national ambient air quality standards were mainly in the southwest suburbs of Shanghai, and the atmospheric O3 concentration decreased from the southwest suburb to the northeast urban areas. It was reported that the tropospheric O3 pollution in the Yangtze River Delta region was more serious in Shanghai, Ningbo and other cities [28]. It should be noted that the concentrations of near-ground-surface O3 in Shanghai, Hangzhou, Hefei and Nanjing in the Yangtze River Delta region have increased slightly during the past 10 years (from 2005 to 2014), but the increase degree was smaller than that in the Beijing–Tianjin–Hebei region. The distribution of atmospheric O3 pollution in the Yangtze River Delta showed relatively obvious flaky distribution characteristics, and the higher emissions from motor vehicles in the Yangtze River Delta urban cities were the main sources of atmospheric O3 in this region [20].

The average near-ground-surface O3 concentration in the Pearl River Delta region was slightly lower than those in the Beijing–Tianjin–Hebei region and the Yangtze River Delta region during 2013–2018 [19]. The characteristics of atmospheric O3 pollution in the Pearl River Delta region and Guangdong province were reported based on the large-scale and long-term continuous O3 monitoring data of recent years [40]. The results showed that the atmospheric O3 concentration in the Pearl River Delta region was higher than that in the northwest of Guangdong province. Outside of the Pearl River Delta region, the eastern area of Guangdong province has the highest atmospheric O3 level. The O3 concentration was higher in the central southern part of the Pearl River Delta and the eastern part of Guangdong, while it was lower in the west. The concentration of atmospheric O3 was higher in summer and autumn, and lower in winter and spring. Due to the large differences of the climate between the Pearl River Delta region and the Beijing–Tianjin–Hebei region and the Yangtze River Delta region, the better atmospheric diffusion conditions made it difficult for atmospheric O3 to accumulate in the Pearl River Delta region.

In addition to the regions of Beijing–Tianjin–Hebei, the Yangtze River Delta and the Pearl River Delta, other regions in China have also been conducted research on local atmospheric O3 pollution. The spatial and temporal distribution of atmospheric O3 pollution in the Bohai Rim region of Liaoning province was reported [41]. The results showed that the atmospheric O3 pollution presented obvious seasonal variation characteristics, and the main months in which the O3 concentration exceeded the limit of Chinese national ambient air quality standards were from May to August. The diurnal variation of atmospheric O3 was unimodal, and the peak concentration appeared in the afternoon. The higher O3

concentration areas were mainly located in Yingkou in the central Bohai Sea Economic Rim of Liaoning, while the O3 level was relatively lower in Dalian and Huludao. The investigation of atmospheric O3 pollution in Shenyang area from 2013 to 2015 showed that the concentration of O3 in the periphery of the city was higher than that in the center of the city [29]. Compared with the periphery of the city, the concentration of NO emissions is higher in urban centers. The increase in NO emissions leads to an increase in the titration of O3, which inhibits the accumulation of O3. The variation of O3 concentration showed obvious seasonal characteristics, with the highest being in summer and the lowest in winter. The diurnal variation showed a unimodal distribution, with the trough value at 06:00 and the peak value at 14:00. Over continental sites, important nocturnal ozone destruction is observed due to dry deposition and NO titration [42]. The tropospheric O3 concentrations showed significant "weekend effects", with higher O3 concentrations in weekends than in weekdays during the daytime while little difference at night. The spatiotemporal distribution characteristics of surface O3 concentrations in Fujian province in 2016 was studied [43]. The results showed that the O3 concentration was higher in spring and autumn, whereas it was lower in winter. The O3 concentrations in the coastal cities were higher than those in the inland cities. The monthly changes in O3 concentration presented a bimodal pattern, with peaks generally appearing in May and September. The diurnal variation curve of O3 concentration was a single peak, which usually appeared at about 14:00. The spatiotemporal distribution characteristics of the atmospheric O3 concentrations in Jiangxi province during 2015–2017 showed that the higher values of atmospheric O3 were mainly distributed in the northeast areas such as Nanchang city and Jiujiang city, while the lower values were mainly distributed in the western areas such as Xinyu city and Yichun city [30]. The monthly variation of atmospheric O3 concentration showed a double-peak pattern with higher values in May and September, while the daily variation showed a single-peak pattern with higher values at 14:00–16:00. The temporal characteristics of atmospheric O3 pollution and the meteorological factors in Chengdu during 2014–2016 were reported, and the results showed that the situation of atmospheric O3 pollution in Chengdu became worse in recent years [31]. The concentrations of atmospheric O3 showed obvious seasonal variation characteristics—higher in summer and spring, while lower in winter and autumn. The diurnal variation of O3 concentration showed a unimodal distribution, with the peak appearing at around 15:00, which was consistent with the diurnal variation of air temperature and solar irradiance. The distribution of surface O3 in Chongqing city in 2018 showed that the O3 concentration in spring to autumn exceeded the limit of Chinese national ambient air quality standards [44]. It was pointed out that the concentration of atmospheric O3 was the highest and the pollution lasted for a long time in summer. Severe O3 pollution in the Sichuan basin in summer was also reported [32].

For most urban stations, the potential ozone (Ox = O3 + NO2) is a conservative amount over a short time scale. When the freshly emitted NO reacts with O3, NO2 is formed in a few minutes, so some local NO2 in the troposphere is produced at the expense of O3 [45,46]. Generally, the surface ozone production is controlled by NOx. The diurnal patterns of O3 and nitrogen dioxide were opposite in Chengdu, indicating that the O3 sensitivity was VOC-limited [32]. The relationship between atmospheric O3 with nonmethane hydrocarbons (NMHCs) and NOx in Guangzhou in 2011 was discussed, and the results showed that controlling highly reactive NMHCs and NOx could effectively reduce O3 concentration [47]. It should be noted that the reduction in NOx may have positive or negative impact on local ozone production. Ozone sensitivity was different at different stages, and reducing NOx emissions had a negative impact on Shenzhen's ozone pollution control from 2015 to 2018 [48].

#### **4. Relationship between Tropospheric Ozone and Its Precursors**

As mentioned before, tropospheric O3 can be produced by photochemical reactions of VOCs, NOx and other primary pollutants under solar radiation. Theoretically, the content of O3 in the troposphere can be controlled by controlling the emission of VOCs

and NOx. However, the execution difficulty is that the relationship between the generation of O3 with VOCs and NOx is nonlinear. An investigation into the relationship between the atmospheric O3 with NOx and VOCs showed that the formation of O3 depended on NOx in rural areas, while it depended on both NOx and VOCs in urban areas [49]. A study on the formation of tropospheric O3 and the effect of VOCs in Shanghai found that alkanes and aromatic hydrocarbons were the dominant VOCs, and aromatic hydrocarbons contributed most to the chemical production of atmospheric O3 [50]. Similar results were found in Guangzhou, where aromatics accounted for 70% of the atmospheric O3 formation potential (OFP) [51].

A numerical simulation control of atmospheric O3 pollution was carried out in Shenzhen city based on the two-dimensional air-quality model [52]. The results showed that the generation of atmospheric O3 was the product of the interaction between NOx and VOCs, and the emission of VOCs was more important. The co-emission reduction in the precursors might effectively reduce the atmospheric O3 pollution. As one of the major species of VOCs emitted from biogenic sources, isoprene is highly reactive and plays an important role in the generation of oxidants for a range of photochemical reactions. A study on the contribution of isoprene emissions to the ground-level O3 formation in Beijing showed that isoprene emissions accounted for almost half (49.5%) of OFP at 13:00 in August of 2010, suggesting that isoprene played an important role in the ozone formation [53]. According to the results of field sampling, the most influential substances related to OFP in Zhengzhou urban area were ethanol, 2-hexanone, o-trimethylbenzene, and the industrial VOCs were a source of O3 pollution in Zhengzhou [54]. It is reported that fire can affect NOx, CO and VOCs, which will significantly affect the background value of O3 [55].

Therefore, the prevention and control of atmospheric O3 pollution cannot be simply through a programmed control of primary pollutants. The influence of VOCs and NOx on atmospheric O3 production can be characterized by a VOCs-sensitive zone and NOxsensitive zone [56]. In general, the oxidation of VOCs with high concentrations of VOCs can produce higher concentrations of RO2·, and the emission of NO can lead to reaction R4 enhancement. Therefore, the amount of O3 production increases with the increase in NOx, and this type of O3 generation mechanism is described as the NOx-sensitive (limiting) type. When the concentration of NOx is high and the concentration of VOCs is low, the reaction rate of NO + O3 is faster than that of NO + RO2·. In this case, the cumulative amount of O3 may decrease with the increase in NOx, and may increase with the increase in VOCs, hence this mechanism is described as VOC-sensitive (limiting) or NOx saturation. When the generation of O3 is restricted by VOCs, the O3 generation can be controlled by reducing the emission of VOCs. Similarly, when the formation mechanism of O3 is NOx-limiting type, the O3 content can be controlled by reducing the emission of NOx. The sensitivity of summer O3 in Beijing during 2010–2015 was studied [57]. The results showed that when VOCs/NOx was 2.0, the urban areas were more sensitive to VOCs and high concentrations of VOCs persisted in western and northern rural areas. When VOCs/NOx was 3.0~5.0, O3 precursors aged, and lower VOCs concentrations appeared in the northern and southern suburbs. A comprehensive investigation into O3 and its precursors and low tropospheric aerosols over a survey site located at the University of Chinese Academy of Sciences in Beijing showed that the photochemical generation of O3 in the boundary layer was restricted by VOCs in hazy weather, while the photochemical reaction of O3 became VOCs–NOx-limiting in the clean weather [58]. According to the sensitivity analysis, the atmospheric O3 generation was largely determined by VOCs when air masses came from the polluted areas in the south. Therefore, reducing VOCs emissions from the industrial areas and urbanized areas could help to reduce the ozone pollution at this site.

Currently, there are a variety of methods that can be used to study the sensitivity of atmospheric ozone generation. Some commonly used methods are as follows.

(1) Ozone production efficiency (OPE, defined as the number of ozone molecules produced for each NOx molecule oxidized). A lower OPE value (<4) indicates that the free radical cycling efficiency is lower, so VOCs are the limiting factor, and the

formation of O3 is controlled by VOCs. Conversely, a higher OPE value (>7) indicates that the free radical cycling is efficient and the formation of O3 is limited by NOx. When the OPE value is medium (4–7), O3 generation is controlled by both VOCs and NOx. The OPE values in rural and suburban areas of Beijing were measured during the 2008 Olympics [59]. The results showed that higher OPE values corresponded to NOx limiting under low NOx conditions, whereas OPE values were lower under high NOx conditions.

(2) Relative incremental reactivity (RIR, defined as the ratio of the decrease in O3 production rate to a given reduction in the precursor concentration) is a measure of the sensitivity of a single precursor. Cardelino et al. [60] first used a scenario test calculated by a box model to simulate the response of ozone to changes in precursors. The calculation result can be expressed by the following formula.

$$\text{RIR}(\text{X}) = \frac{\Delta \text{O}\_3(\text{X})/\text{O}\_3}{\Delta \text{C}(\text{X})/\text{C}(\text{X})}$$

where X represents a group of major pollutants, and O3 represents the modelled O3 concentration. ΔC(X)/C(X) gives the relative change in the primary pollutants in one of the sensitivity tests, and the relative change in modelled ozone concentration is given by ΔO3(X)/O 3. In the study on atmospheric ozone pollution conducted in Chengdu in September 2016, the anthropogenic variation of the main pollutant in the sensitivity test was chosen as 20% in the RIR analysis, because when the variation value was greater than 20%, the RIR value deviated due to the significant change in the simulated free radical concentration [61]. The RIR results demonstrated that anthropogenic VOCs reduction is the most efficient way to mitigate ozone pollution, of which alkenes dominated more than 50% of the ozone production [61].


According to current research on atmospheric ozone formation regimes, most of the urban areas in China are in VOCs-limited zones, with anthropogenic VOCs (especially reactive aromatics and alkenes) playing a dominant role. However, some variations were found in the chemistry regime of atmospheric ozone formation in different regions.
