Ground-level ozone pollution in China: a synthesis of recent findings on influencing factors and impacts

Ozone (O3) in the troposphere is an air pollutant and a greenhouse gas. In mainland China, after the Air Pollution Prevention and Action Plan was implemented in 2013—and despite substantial decreases in the concentrations of other air pollutants—ambient O3 concentrations paradoxically increased in many urban areas. The worsening urban O3 pollution has fuelled numerous studies in recent years, which have enriched knowledge about O3-related processes and their impacts. In this article, we synthesise the key findings of over 500 articles on O3 over mainland China that were published in the past six years in English-language journals. We focus on recent changes in O3 concentrations, their meteorological and chemical drivers, complex O3 responses to the drastic decrease in human activities during coronavirus disease 2019 lockdowns, several emerging chemical processes, impacts on crops and trees, and the latest government interventions.


Introduction
While ozone (O 3 ) in the stratosphere is beneficial for life on Earth by filtering out Sun's harmful ultraviolet radiation, at ground level, O 3 adversely affects human health and vegetation. O 3 also regulates the oxidative capacity of the troposphere, and is a powerful greenhouse gas that contributes to climate warming. Elevated O 3 concentrations in the lower part of the atmosphere remain a persistent environmental problem in many of the world's urban and industrialised regions [1][2][3]. In mainland China (henceforth 'China'), the rigorous control of emissions following the implementation of the Air Pollution Prevention and Action Plan (2013-2017) has led to remarkable reductions in air pollution caused by sulphur dioxide, particulate matter and nitrogen oxides [4]. However, ground-level O 3 concentrations in major population centres have been increasing in recent years [5,6], which has led to a surge in the numbers of studies on O 3 in the past six years. From this body of work, new findings on the factors influencing O 3 production and distribution and on the O 3 impacts have emerged. On the policy side, mitigating O 3 pollution is now on the agenda of the national and local governments planning the next phase of air-pollution controls. Therefore, the time is ripe for a synthesis of recent findings concerning the key chemical and physical influences on ground-level O 3 concentrations, which will help provide scientific support for the development of evidence-based O 3 control strategies.
Several reviews on the O 3 situation in China have been conducted in the past five years. The first comprehensive review [3] covered the fundamental chemical and meteorological processes involved in O 3 production and transport, the history of O 3 monitoring and research in China, the O 3 concentrations in major city-clusters, the relationships between O 3 concentrations and their precursors, early findings on the roles of nitrous acid (HONO) and nitryl chloride (ClNO 2 ) in O 3 production, and the impacts of O 3 on crops, forests and human health. Subsequently, five additional topical reviews were published [7][8][9][10][11]. Lu et al [7] reviewed findings on the O 3 formation regimes and their temporal (inter-annual, seasonal and diurnal) changes for period up to 2016. Lu et al [8] focused on the impacts of meteorology and climate on natural emissions of O 3 precursors, chemistry and deposition, and transport patterns. Fu et al [9] reviewed impacts of emission, multiscale meteorology, land use change, aerosol effect on O 3 , and the influence of tropospheric O 3 change on radiative forcing and temperature. Liu et al [10] reviewed the methodological aspects of O 3 and precursor apportionment, including back trajectories, observationbased chemical models, receptor models for volatile organic compounds (VOCs), emission-based chemical transport model, and their respective results. Xu [11] provided a short review of the impacts of O 3 on crop yield and economic loss, tree risk exposure to O 3 , and cardio-vascular mortality with O 3 exposure dose. The O 3 spatial distribution and changes across China from 2013 up to 2019 were presented in three of the above reviews [9][10][11], and the decadal trends of O 3 at limited number of sites were reviewed in two of them [9,11].
The present review focuses on the multiple novel insights that were gained in the recent six years, but were not covered by the prior reviews. They include (1) the recent situation of O 3 -formation regimes amid decreasing NO x and increasing VOC emissions, (2) the recent results on the impact of HONO and ClNO 2 on O 3 production and the new findings on the impact of other reactive halogen (Cl 2 and BrCl) in polluted regions, (3) the winter photochemistry of O 3 (an understudied topic), (4) the complex responses of O 3 to reduced emissions under recent control measures and during China's coronavirus disease 2019 (COVID-19) lockdown, (5) the impact of O 3 on the quality of grain crops, tree productivity and water use efficiency. Additionally, we update the results on surface O 3 changes across China to the year of 2021, which reveals possible signs of O 3 levelling off or decrease in some urban areas, and (6) introduce recent government efforts to control O 3 in summer. We also give a graphic summary of multi-scale transport patterns.

Literature search and results
We adopted an evidence-based approach to systematically search three research databases (Web of Science, Scopus and Google Scholar) for English-language, peer-reviewed studies that were published (online) between 1 January 2000 and 30 June 2021, using three search methods. Here, we included the papers published before 2015 in the search with the aim to show the rapid development of ozone research in China. The first search method used keywords that combined 'ozone' or 'O 3 ' with the name of a specific region, namely 'China' , 'Chinese' , 'Pearl River Delta' (or 'PRD'), 'Yangtze River Delta' (or 'YRD'), 'North China Plain' (or 'NCP'), 'Beijing-Tianjin-Hebei' (or 'BTH'), 'Jing-Jin-Ji' (or 'JJJ'), 'Sichuan Basin' (or 'SCB'), Tibetan Plateau (or 'TP'), 'Beijing' , 'Shanghai' , 'Guangzhou' or 'Hong Kong' . This returned 656 articles that had been published since 2000. The second search method targeted publications from active O 3 -research groups across 21 universities (or research institutes) that were known to the authors (table S1 (available online at stacks.iop.org/ERL/17/ 063003/mmedia)). The search yielded 54 additional papers that had not been found in the previous search. At this stage, we also identified four additional papers that cited Wang et al [3]. The third search method used the keywords 'O 3 ' and 'COVID-19' , and produced 37 papers. A total of 751 relevant papers were identified (see appendix for the complete list), 571 of which were published between 1 January 2015 and 30 June 2021. Figure 1 shows the number of O 3 -related papers published from 2000 to the end of the June 2021, which highlights the rapid growth of O 3 research in China since 2015. Ozone studies in China have focused on the three most urban and industrial conglomerates: the PRD (including Hong Kong) in the south, the YRD in the east, and the Jing-Jin-Ji (Beijing-Tianjin-Hebei) region in the NCP. In recent years, an increasing proportion of publications have focused on the northern parts of China (figure 1, insert). This trend reflects the fact that earlier O 3 studies had mainly been conducted in southern parts of the country, whereas research (and mitigation) in the north had been focused on the more pressing problems of sulphur and particulate pollution [3]. Increasing recognition of the serious O 3 pollution in Beijing [12,13] has motivated more studies on O 3 in the NCP. There have also been more publications on O 3 processes and impacts across the whole of China since 2013, when data on ambient concentrations of O 3 and other regulated pollutants were made available by the China National Environmental Monitoring Center network.
The topics of investigation in O 3 studies in China have changed over the years. Due to limited research funding, earlier studies were largely exploratory in nature. Subsequent rapid increases in research facilities and the research community, and a large increase in research funding, has led to recent in-depth and rigorous studies on specific processes contributing to O 3 concentrations and the impacts of O 3 on vegetation and human health.  Illustration depicting the sources of ground-level ozone (O3) and radicals that initiate oxidation of VOC and CO in polluted regions. The 'conventional' radicals are labelled in red, and understudied halogen radicals and HONO (an OH source) are shown in blue. The O3 loss pathways, radical termination steps and their depositions are not shown.

Brief review of the factors affecting ground-level O 3 concentrations
The basic chemical and physical factors affecting the formation and distribution of ground-level O 3 have been well established, and reader is referred to previous review articles [3,[14][15][16][17][18][19][20][21][22]. Here, we only give a brief summary. Ground-level O 3 is produced via the chemical reactions of the precursors NO x (nitric oxide (NO) and nitrogen dioxide (NO 2 )), VOCs, and CO (carbon monoxide) under sunlight (figure 2). In urban and industrial regions, NO x is mainly emitted from the combustion of fossil fuels, such as in vehicles and industrial settings, whereas VOCs are released from more diverse sources such as vehicle exhaust, evaporating fuels and solvents, consumer products, and trees. In polluted regions, the photochemical production of O 3 begins with OH radical (or an analogue, such as Cl atom) reacting with VOC (and CO) to form organic peroxy radical (RO 2 ) and hydroperoxyl radical (HO 2 ), which subsequently oxidise NO to NO 2 . Then, photolysis of NO 2 produces O atoms, which react with O 2 to form O 3 (figure 2, also see Wang et al [3]). Therefore, the sources and sinks of OH radical profoundly affect the rate at which O 3 is formed. It has been well established that O 3 has a non-linear relationship with its precursors; that is, NO x can lead to either decreases or increases in the concentration of O 3 , depending on the relative ratio of NO x to VOC. In general, the production of O 3 in urban areas with high NO x /VOC ratios is VOC-limited, such that a reduction in NO x emissions will tend to increase O 3 concentrations due to the decreased titration of O 3 and radicals by NO x . In contrast, in rural areas where NO x /VOCs ratios are often low, decreasing NO x emissions will decrease O 3 concentrations. Aerosols can also influence O 3 concentrations by altering solar irradiance, and via the chemical reactions that occur on aerosol surfaces. In addition to in-situ photochemistry, O 3 is transported from higher altitudes of the atmosphere to ground level, and from one region to another. Meteorological conditions can also strongly affect O 3 production and distribution by changing patterns of air transport, the wet and dry depositions of gases and aerosols, and the rates of chemical reactions and natural emissions.

Decadal change
Long-term (>10 years) changes in O 3 concentrations over China have been summarized in recent reviews [9,11], and factors contributing to the O 3 changes were reviewed by Fu et al [9]. The reader is referred to these reviews for the detail results on the decadal change in O 3 . Here, we give a brief summary. In mainland China, O 3 measurements lasting longer than 15 years were limited, and were mainly conducted at several regional background sites [9,11]. Additionally, O 3 decadal changes have been reported in major cities such as Beijing and Shanghai and in the PRD region [9,11]. The results from these relatively long measurements indicate increasing surface O 3 in or near the developed regions of China. The increasing rate was typically lower at rural sites than at urban cores in a given region (e.g. JJJ), and the surface O 3 concentrations at several rural sites (Lin'an, Longfengshan, Gucheng and Hok Tsui) appeared to have stopped increasing or even decreased during the recent decade [23,24]. The decreases in O 3 concentrations in the rural areas may be a consequence of decreased NO x emissions since 2011. Figure 3 shows a marked decrease in NO x emissions after 2011, but continued increase in VOC emissions, which is in contrast to the decrease in both NO x and VOC emissions in the U.S. and Europe (figure 3).

Urban O 3 since 2013
After public release of nation-wide data from the China National Environmental Monitoring Center network since 2013, a much clearer picture has emerged on the O 3 changes in urban areas across China. Numerous recent studies have analysed the data to show rapidly increasing urban O 3 concentrations during years up to 2019 [9,11]. In the present review, we extend the prior analysis of the trend result to the year of 2021. Figure

O 3 formation regime
Previous findings on O 3 formation regimes up to the year of 2016 have been reviewed by Wang et al [3] and Lu et al [7], which indicated that in most urban areas, chemical production of O 3 was controlled by VOCs. Observation-based results from the past 5 years suggest that this situation has largely remained despite decrease in NO x emissions and increase in VOC emissions, e.g. in urban areas of Shanghai [28], Nanjing [29], and Wuhan [30]. The recent studies have expanded to less-developed cities NMVOCs are the total mass of non-methane hydrocarbons and oxygenated VOCs. The emission data of NOx and NMVOC for China were obtained from Li et al [25], Zheng et al [26] and the China statistical yearbook [27]. The anthropogenic emission data of NOx and NMVOC for the U.S. and Europe were obtained from USEPA (www.epa.gov/air-emissions-inventories/airpollutant-emissions-trends-data) and EEA (www.eea.europa.eu/data-and-maps/dashboards/air-pollutant-emissions-dataviewer-4), respectively. such as Dongying (in NCP) [31], Xuzhou, Yancheng, and Nantong (in YRD) [32], Shantou (in Guangdong province) [33] and Weinan (in FWP) [34]. For most of the newly-studied cites, urban areas were found in a VOC-limited regime, with two exceptions in Weinan and Nantong, both of which were in a mixed-limited regime [32,34]. However, transition from a VOClimited [35] to a mixed-limited O 3 formation regime [28] was indicated in some suburban areas of the YRD, mainly owing to decreases in NO x emissions in the recent years. As to altitude dependence, an analysis of airborne measurements in the NCP showed that O 3 formation became more sensitive to NO x emissions with increasing heights [36].
Recent analysis of in-situ measurements indicated that alkenes and aromatics were the major reactive VOC groups for O 3 production in urban areas. For example, alkenes were the dominant VOCs in the NCP [37], FWP [34], and SCB [38], while aromatics were more important in the YRD [28,39] and PRD [40]. The importance of aromatics to O 3 formation typically increased along with decreasing latitudes. These results are generally similar to those obtained from the early studies [3,7], indicating insignificant changes in VOC species that controlled O 3 formation in these regions. Biogenic VOCs play an important role in some occasions [39].
Recent emission-based model studies [41] and analysis of satellite data [42] have suggested some changes in the O 3 formation regime over eastern China, with the areas of mixed-limited regime becoming larger due to NO x emission reductions. Quantitatively, Wang et al [41] found that the areas in mixed-limited regime in JJJ, YRD, and PRD expanded by 17.1%, 20.8%, and 20.3%, respectively, from 2012 to 2016 according to the simulations with a Weather Research Forecast-Community Multiscale Air Quality (WRF-CMAQ). The expansion mainly occurred in suburban and rural areas, whereas urban areas in Beijing, Tianjin, Shanghai, Nanjing, and Guangzhou remained in a VOC-limited regime. The results from emission-based models are generally consistent with the findings from observation-based analysis, which indicate that the reduction in NO x emissions in the past decade has affected the O 3 formation regime, but is insufficient to change the typical urban areas from a VOC-limited to a NO x -limited regime.

Roles of multi-scale atmospheric transport
O 3 is a secondary pollutant with a lifetime that ranges between several days and several weeks [16], and thus can be transported from one place to another. In this review, we consider three scales at which the atmospheric transport of O 3 occurs: the intercontinental-, long-range-and regional scales. Intercontinental transport of O 3 occurs between, for instance, Europe and Asia. Long-range transport of O 3 occurs from other Asian countries to China and from the stratosphere to ground level. We separate regional transport of O 3 into super-regional (i.e. inter-regional) and intra-regional scales, which represent transport between different regions and different areas within the same region, respectively. Previous reviews have summarized the findings of the impact of multiscale meteorological conditions on O 3 accumulation and transport up to the year of 2018 [8,9] and the source  [10]. The present review synthesizes the main transport patterns, synoptic conditions, and dynamic features in a graphic form (figure 6) and update the results published in the past two years. We also review the findings on intercontinental transport, which was not covered in the prior reviews.

Intercontinental and long-range transport
Several studies have investigated the impact of intercontinental transport on O 3 in China [43][44][45]. Their results showed that the strength and location of the westerlies strongly influenced the transport pathway and seasonal variation of the amount of O 3 transported to China. Using Goddard Earth Observing System (GEOS)-Chem model and measurement data from 31 locations, Ni et al [45] quantified the contributions of other continents to O 3 in China in a spring season. Their simulations showed that Europe contributes 2.1-3.0 ppb to surface O 3 over China's northern border and that North America contributes 0.9-2.7 ppb of surface O 3 over most parts of China. These studies also indicated that intercontinental contributions are generally larger in the middle and upper troposphere than at surface.
It has been known that the transport of O 3 between China and other Asian countries can be influenced by the Asian monsoons [8,9]. Longrange transport of increased emissions from Southeast Asia, under the influence of the Asian Summer Monsoons, has been suggested to contribute to increased O 3 concentrations in remote or rural sites in western and southern China [23,46,47]. Anthropogenic emissions in Japan and Korea have also been shown in model studies to contribute to the O 3 concentrations over China's eastern coasts. For example, GEOS-chem simulations indicated a contribution of 0.6-2.1 ppb in a spring season [45]. A recent analysis of a week-long O 3 episode during autumn in eastern China showed that under the influence of eastward movement of the Mongolian high-pressure system, transport of O 3 produced from emissions in Japan and Korea could contribute up to 30 ppb of O 3 (or ∼45%) at the peak, according to WRF-CAMQ simulations [48].
Stratosphere-troposphere exchange has been known to affect tropospheric O 3 [8]. Recent studies have provided additional evidence to show that the stratosphere intrusion (SI) can be a source of O 3 in the lower troposphere. Analysis of ozonesonde data showed that the elevated concentrations of surface O 3 in Lhasa on the Tibetan Plateau were associated with the SI [49] and that the SI also affected the groundlevel O 3 in the PRD (Yangjiang and Hong Kong) during some periods of a year [47,50]. Model studies have further demonstrated the considerable role that the SI events played in O 3 pollution over eastern and southern China [49,[51][52][53].

Regional transport
The tropical cyclones (low-pressure systems) over the western pacific, continental anti-cyclones (highpressure system) and the West Pacific Subtropical High are the typical synoptic conditions leading to the formation and transport of O 3 pollution in China [9]. Recent studies have provided additional result on their impact on O 3 pollution in central (Wuhan), eastern (YRD) and southern (PRD) China [54][55][56][57][58][59][60][61][62]. Smaller-scale circulations, such as sea-land breezes (SLBs) and mountain-valley breezes (MVBs), can strongly affect O 3 concentrations in the coastal and mountainous areas under weak synoptic forcing [3]. Several recent studies presented more cases to demonstrate the impact of the SLBs on O 3 concentrations in coastal areas of the Bohai Gulf [63], the East China Sea [64] and the South China Sea [60,65]. MVBs have also been shown to strongly influence the diurnal variation in O 3 concentrations at Mt Waliguan, Mt Huang and Mt Tai Mo Shan in the western, central and southern China, respectively [66][67][68][69].
Recent model studies have quantified interregional transport [70][71][72][73]. The surface O 3 concentrations in the JJJ region were strongly influenced by transport from the province of Shandong, Henan, Jiangsu, and Anhui, with a collective contribution up to 36% of the O 3 concentrations in JJJ [74]. Similarly, emissions from non-YRD regions have been found to be the dominant contributor to O 3 pollution in the YRD, with the peak contribution of 63% [71,75,76]. In the PRD region, intercity transport played an important role, with Guangzhou and Foshan being the major sources of emissions [72,77]. In SCB, downward transport of upper stratospheric air from the Tibetan Plateau was found to be a key contributor to ground-level O 3 pollution over the city clusters of the basin [73]. These studies highlight the importance of cross-boundary transport in developing O 3 mitigation strategies for a specific city/region.

HONO and photolabile halogens
As discussed in section 1, any factors affecting the sources and sinks of OH or its analogues also affect the photochemical production of O 3 . In the past decade, the photolysis of HONO has gained increasing recognition as an important source of OH, not only in the morning but also throughout the day [3,15,78]. The results of limited studies that were conducted up to year of 2016 were given in a previous review [3].
In the past 5 years, the number of HONO studies in China has increased rapidly, allowing assessment of the HONO impact on OH and O 3 in more and diverse areas. The available studies thus far have shown very high daytime concentrations of HONO (on order of a few ppbs) in many urban and rural areas of China, in both summer and winter [79][80][81][82][83][84][85][86][87][88]. Calculations with observed HONO and O 3 concentrations show that the photolysis of HONO dominated the OH primary sources in the early morning and contributed 40%-50% at noon during photochemical episodes in southern China [89]. Similarly, the HONO photolysis dominated the OH primary sources at an agricultural site in the NCP, and was nearly five times of the production rates from O 3 photolysis after fertilization [87]. Chemistry-transport models (CTMs) incorporating known HONO sources have attempted to quantify the HONO contribution to O 3 concentrations [90][91][92][93]. Simulations with a Weather Research Forecast-Chemistry (WRF-Chem) model showed that HONO enhanced the O 3 peak concentrations by 4%-10% in the HK-PRD region of southern China during a summer O 3 episode [90]; another study using a WRF-CAMQ model showed 34% increase in regional O 3 concentrations in a heavy winter pollution episode in the same region [91]. In other regions of China, WRF-Chem simulations suggested monthly O 3 increase of 5%-44% in Beijing-Tianjin-Hebei region in August [92]. Another WRF-Chem study indicated 6%-13% increase in surface O 3 concentrations in the developed regions of eastern China in July [93]. Direct comparison of these model results is difficult due to the differences in the treatment of the HONO sources and other aspects of model configurations in these studies. Nonetheless, the results of these studies have demonstrated important impact of HONO as a source of OH on O 3 production in polluted regions.
The high HONO concentrations (>10 ppb) reported in China were mainly attributed to the heterogeneous reactions of NO 2 on ground and aerosol surfaces, and to the photolysis of nitrate (NO 3 − ) aerosols [91]. However, the kinetic parameters in the real atmospheric conditions remain uncertain (i.e. the NO 2 uptake coefficient, HONO production yields and NO 3 − photolysis rates). Additionally, recent field observations have shown that agricultural fertilisation can increase the emissions of HONO from soil [87,94,95]. Most state-of-the-art air quality models cannot properly simulate this HONO source due to a lack of information on the key factors determining HONO emissions from fertilised soils, such as fertiliser types, and biotic and abiotic factors. A very recent study developed a parametrisation linking soil HONO emissions to three commonly used fertilisers in China, and the temperature and water-holding capacity of soil samples, and incorporation of this parameterization in a WRF-CMAQ model improved simulations of HONO and increased regionally averaged O 3 concentrations by 8% in the agricultureintensive NCP [96].
Other radicals that can produce O 3 are chlorine (Cl) or bromine (Br) atoms. Ambient ClNO 2 is a potential source of Cl radical (and a reservoir for NO 2 ). The early findings on the impact of ClNO 2 observed in 2013-2014 in China using a box model were given in a previous review [3], which showed that Cl produced by the photolysis of ClNO 2 increased daytime O 3 production by up to 41% in top of the PBL of southern China during winter [97], and by up to 13% at a polluted rural site (Wangdu) in NCP [98]. Since then, ClNO 2 has been measured in more sites/seasons across China [99] and more model studies have been conducted. Recent measurements during winter at several sites in northern China showed that ClNO 2 concentrations can increase during daytime, leading to greater daytime O 3 production [99]. A WRF-Chem model incorporating laboratory-determined heterogeneous production of ClNO 2 was used to show that there was a 5%-6% increase in O 3 concentrations within the PBL of NCP and the YRD region [93] in the summer of 2014. The same model has been used to evaluate the effect of the ClNO 2 and HONO chemistry on the designation of the O 3 formation regimes, and the result showed that the 'new' nitrogen chemistry changed the O 3 sensitivity regime for nearly 40% of the simulated area with human influence in China, mainly from VOC-sensitive or NO x -sensitive regimes to mixed-sensitive regime [100]. However, current parametrisations of ClNO 2 production (including the uptake coefficient of nitrogen pentoxide (N 2 O 5 ) and the yields of ClNO 2 ) are subject to large uncertainties and need further improvement [101,102].
Molecular chlorine (Cl 2 ) is another potentially important photolytic source of Cl. Cl 2 mixing ratios of up to 400 ppt were observed at a polluted rural site in Wangdu in northern China during the summer of 2014, and corresponding box model calculations suggested that Cl 2 and ClNO 2 increased the O 3 production rate by 19% [103]. A recent study during the coal-burning period of winter documented surprisingly high concentrations of bromine chloride (BrCl) and Cl 2 at a site in Wangdu, and calculations from an observation-constrained model indicated that Cl and Br atoms collectively increased the total O x (O 3 + NO 2 ) mixing ratios at this site by 50% [104]. Another very recent study reported unprecedented levels of Cl 2 (up to 1 ppb) at a polluted coastal site of Hong Kong, and such high concentrations of Cl 2 increased O x daytime production by 16% [105]. However, existing CTMs cannot account for observed daytime Cl 2 or BrCl concentrations because they lack appropriate mechanistic and kinetic information. Nonetheless, several CTM studies have attempted to examine the impact of previously known chlorine sources on the atmospheric oxidation capacity and O 3 concentrations and suggested potential importance of halogen chemistry in the lower troposphere [106][107][108][109].

Active winter photochemistry in northern China
It is typically assumed that the photochemical processes that produce O 3 are slow during winter in places like the NCP, owing to the substantially lower temperatures and levels of solar radiation during winter than in the warm seasons. However, several field studies in Beijing and its surrounding areas have indicated that photochemical processes could be rapid during polluted days in winter, which has been attributed to the presence of very high concentrations of HO x precursors (i.e. HONO and oxygenated VOCs) and halogen atom precursors (BrCl and Cl 2 ), and to the rapid recycling of OH due to high NO concentrations [104,[110][111][112][113]. Chemical models have suggested that rapid photochemical process was likely to occur throughout the NCP region and its surrounding areas [113,114]. In these studies, the noontime rates of O 3 production were 15-100 ppb h −1 under high NO x conditions, which is comparable to those during the summer. However, the concentration of O 3 was supressed due to the simultaneous and rapid loss of O 3 by NO titration, the subsequent loss of NO 2 via gas-phase reactions with OH and the heterogeneous loss of N 2 O 5 , which led to the rapid production of nitric acid and secondary aerosols [112,114]. Thus, decreasing emissions VOCs would slow the winter photochemical processes that generate O 3 and thereby help to alleviate the severity of the winter haze covering Beijing and a large part of the NCP.

Emissions change in 2013-2017
In 2013, the Chinese government implemented the Air Pollution Prevention and Control Action Plan to mitigate the severe haze in many parts of the country, and consequently, anthropogenic emissions of sulphur dioxide (SO 2 ), NO x , CO and 2.5 µm or smaller particulate matter (PM 2.5 ) in China were reduced by 59%, 21%, 23%, and 33% from 2013 to 2017, respectively [4]. However, the emissions of VOCs and ammonia increased slightly in the same period [26]. Data from national environmental monitoring stations has shown corresponding decreases in the ambient concentrations of SO 2 , NO x and PM, but increases in ground-level O 3 concentrations in numerous urban areas [1,6,115,116] (see figures 4 and 5). A key policy-relevant conclusion that can be drawn from the above studies is that while the nationwide control measures from 2013 to 2017 have successfully reduced emissions of primary pollutants, they have also led to increased O 3 concentrations in urban areas of China, due to the non-linear dependence of O 3 on NO x and aerosol feedbacks. Therefore, the control of VOC emissions should be included in future O 3 control strategies. A positive outcome of these control measures for O 3 is that O 3 concentrations in rural areas have been reduced according to model simulations [116] and observations at several non-urban sites [24], which can be attributed to NO x -limited O 3 formation regime that exists in rural areas [23,116].

Effects of China's COVID-19 lockdown on ground-level O 3 concentrations
The outbreak of coronavirus disease 2019 (COVID-19) pandemic in 2019, and attempts to control it, have severely impacted human activities worldwide. China was the first country that imposed nationwide measures (from 23 January to 13 February 2020) that aimed to prevent the spread of severe acute respiratory syndrome coronavirus 2, which causes COVID-19. These various restrictions drastically reduced emissions of air pollutants [120,121], which offered an unprecedented opportunity to investigating the effects of large and nationwide decreases in emissions on O 3 concentrations. A number of studies have analysed data on ground-level O 3 concentrations based on data from China National Environmental Monitoring Center network, using different methodologies. Some directly compared ambient O 3 concentrations during the COVID-19 lockdown period to those during pre-lockdown periods (in the same year or in previous years), whereas others have combined statistical transport models and CTMs to separate the effects of emission reductions and meteorological factors on O 3 concentrations. The emissions of air pollutants during the COVID-19 lockdown period have been estimated based on activity data or estimates (that were available at the time) for the transportation and power-generation industries and the residential sector [120][121][122]. The reductions in emissions were found to vary between regions, and were estimated to be ∼45%-54% for NO x and ∼22%-43% for VOCs in northern, central and southern China, with larger decreases in VOCs in the south [122].
Most studies of changes in O 3 concentrations during the COVID-19 lockdown have focused on the increases in ground-level O 3 concentrations that occurred in many of the urban areas in northern and central China during their lockdown period amid large decreases in the concentrations of primary pollutants such as NO 2 and SO 2 in the same areas [123][124][125]. A few studies have also investigated the decrease in O 3 concentrations in southern provinces such as Guangdong and Guangxi [122,126]. Studies using models and data on the estimated emission reductions during the COVID-19 lockdown period have found that meteorological factors had strong (or stronger than emissions) effects on O 3 concentrations during this period [122,125]. These findings re-affirm the critical role of meteorological factors in driving short-term air quality, as proposed in many previous studies on air-quality changes subsequent to short-term reductions in emissions [122]. The model simulations have attributed increases or decreases in O 3 concentrations to the non-linear relationship between O 3 concentrations and those of its precursors, which led to an increase in O 3 concentrations in most cities due to O 3 titration by NO x under NO xsaturated or VOC-limited conditions, and a decrease in O 3 concentrations in the south due to the lower NO x to VOC ratios there. Model-calculated indicators of the O 3 -formation regime (i.e. the ratio of the production rate of H 2 O 2 to that of HNO 3 ) suggest that most of the populated and industrialised parts of the NCP and YRD, and the core of the PRD remained a typical VOC-limited O 3 -formation regime during the COVID-19 lockdown, but other large parts of these regions transitioned to a mixed regime. The reduction in VOC emissions in northern China during the COVID-19 lockdown were insufficient to counteract the increase in O 3 concentrations due to decreased NO x titration [122]. Moreover, the increased O 3 concentrations in many northern cities could have enhanced the oxidative capacity of the atmosphere and consequently the production of secondary aerosols in the winter haze in these cities, such as Beijing [121].

Impacts of O 3 on crops and trees
There was limited research on crop responses to O 3 in China in the earlier years, but a rapid increase in studies on rice and wheat and those involving experimental and model investigations after 2008. A recent review highlighted the magnitudes of crop yield loss and tree biomass reduction across China or some regions of China due to rising O 3 concentration [11]. Currently, there has been a general consensus that O 3 pollution in China has substantially decreased crop yields [11,[127][128][129][130]. A very recent assessment, which is based on the latest O 3 dose-yield response relationship obtained from the Asian field experiments and hourly O 3 concentrations in 2018-2020 at 1400 air quality stations in China, suggests that the ambient O 3 concentration in China has caused the national yield loss of wheat, hybrid rice, inbred rice and maize by 32.8%, 29.8%, 12.2% and 8.6%, respectively [131]. These new estimates of crop yield loss were a bit higher than the previous studies, in which O 3 dose response relationship was based on the manipulation experiment from one or two sites [127][128][129][130]. While numerous studies have focused on crop yield [11], little is known about how O 3 pollution affects grain quality in China. The few available studies on grain quality have largely focused on wheat or rice. O 3 pollution has been shown to affect the starch content and composition of grain [132,133], which influences grain processing, appearance and cooking quality. A study on Chinese hybrid rice cultivars found that decreased grain cooking quality and increased proportions of chalky and cracked kernels are associated with high O 3 concentrations [134]. In addition, elevated O 3 concentrations have been found to lead to decreases in the total grain content of nutrients per plant, but have also been found to increase grain concentrations of protein and mineral nutrients (e.g. potassium, calcium and manganese) [135,136]. This increase in grain nutrient concentrations is related to advanced grain filling or the 'concentration effects' of yield loss under elevated O 3 concentrations [136,137]. These findings suggest that although O 3 pollution can improve grain nutrient quality in some cases, it generally leads to a deterioration of grain quality.
Ground-level O 3 pollution has also been shown to affect the health of Chinese forests. It has been suggested that over 98% of forested areas in China are threatened by O 3 pollution [138]. However, only a few studies have estimated the impacts of O 3 on Chinese forest productivity. Northern temperate forests in China are exposed to relatively higher O 3 concentrations than sub-tropical forests [139]. However, as tropical and subtropical evergreen forests have longer growing seasons (and are thus exposed to O 3 for longer durations) but are more resistant to O 3 than temperate deciduous forests, O 3 pollution was found to reduce tree biomass by similar amounts in these two forest types (13% vs. 11%) [138]. In addition, broadleaved forests have generally been shown to be more vulnerable to elevated O 3 concentrations than needle-leaved forests [140], probably due to the lower O 3 uptake per unit leaf mass by thicker or denser leaves [141].
The ratio of photosynthetic carbon uptake to water vapour loss, or water-use efficiency (WUE), is a key forest function that is commonly used to describe the coupling of the carbon and water cycles in terrestrial ecosystems. Evidence from studies on individual plants suggests that elevated O 3 concentrations can influence photosynthetic CO 2 assimilation and stomatal conductance, thereby ultimately altering WUE [142,143].

Government regulations and efforts to control O 3
To improve regional air quality, the Chinese central government has enacted a series of long-term regulations including the Air Pollution Prevention and Control Action Plan (2013-2017) and the Blue Sky Protection Campaign (2018-2020) (www.mee.gov.cn, accessed 1 December 2021). These regulations have set specific targets for reducing emissions and/or ambient PM 2.5 concentrations, or the number of heavy pollution days. The regulations in 2018 mandated a 15% reduction in the emissions of NO x and SO 2 and a 10% reduction in the emissions of VOC in 2020, relative the emissions in 2015. The implementation of these regulations in various economic sectors has led to significant decreases in the concentrations of PMs (including PM 2.5 and PM 10 ) and other routinely monitored pollutants such as NO x , CO, and SO 2 [26,115,144], but have not been effective in reducing O 3 pollution. This is in part because previous control measures mainly aimed to alleviate the notorious haze or PM 2.5 pollution in China (especially in northern China during the winter), but assigned a lower priority to reductions in O 3 concentrations. Nonetheless, during some important large-scale events held in China, such as the 2016 Group of Twenty (G20) Summit in Hangzhou, specific measures to reduce NO x and VOCs emissions were implemented to reduce O 3 pollution, and the responses of ambient O 3 concentrations to these reductions varied-decreased in some areas but increased others [122,145].
In its 14th Five-Year Plan for 2021-2025, China has identified VOC emission management as a major target, in parallel with reductions in NO x emissions (www.mee.gov.cn, accessed 1 December 2021). Specifically, as part of the O 3 Pollution Prevention and Control Action Campaign, which was launched by the Ministry of Ecology and Environment (MEE) to mitigate increasing summertime O 3 pollution, VOC emission reduction measures were implemented in many cities between June and September in 2020 and 2021. Inspection teams led by officials from MEE, Provincial and City's Bureau of Ecology and Environment and assisted by scientists, visited major VOC-emitting industries (such as petrochemical, chemical, solvent, packaging and fuel storage industries), and helped them to find ways to reduce their VOC emissions. The results from ambient monitoring suggest that these measures were effective, and that O 3 concentrations decreased in the summer of 2020 and 2021 in major cities, compared to previous years ( figure 4). This was especially true for Jing-Jin-Ji and its surrounding regions. It is expected that such top-down enforcement will continue in the coming years. It is important that these somewhat ad hoc approaches are developed into a more systematic and clearly formulated long-term effort.

Summary and recommendations
In this work, we have synthesised the findings of studies on the O 3  • A number of studies have assessed the atmospheric transport of O 3 and its precursors in different regions of China. These studies have highlighted the important roles of various scales of transport, including inter-regional transport between eastern China and the PRD, between the YRD and the Jing-Jin-Ji region, and intraregional (or intercity) transport within these populated and industrialised regions. These results re-affirm the need for cooperation between regional governments to reduce O 3 pollution. Their findings provide strong scientific evidence that confirms the negative impacts of O 3 on the crop yields and grain quality. The available studies have also shown that deciduous species are at more risk to O 3 than evergreen species, and increased O 3 concentrations can affect water use of trees. • In terms of policy, although previous control regulations and measures focused on PM 2.5 , those in the past two years have paid increasing attention to O 3 . The enforced control of VOC emissions in the two most recent summers has had the desired effect on ambient O 3 concentrations, which appear to show signs of decreasing in many O 3 -affected cities.
We would like to provide the following recommendations for future research and policy development.
• Data on long-term ambient VOC compositions in different regions of China is needed. While China National Environmental Monitoring Center network has provided unprecedented data on the concentrations of O 3 , NO 2 and other regulated pollutants since 2013, there remains a lack of data on the concentrations and composition of ambient VOCs. Despite the reported establishment of dozens of VOC measurement sites across the country, their data have not been made available. This impedes efforts to assess the roles of VOCs in O 3 formation and to validate VOC emission inventories. • More research is needed on the role of newly discovered chemical processes in O 3 formation, given the rapid changes that will occur in chemical environments in future years (i.e. the expected large decrease in the emissions of NO x and VOCs, and the continuing reductions in the emissions of PM 2.5 ). Air quality models should incorporate these processes to enable more accurate predictions. • More research is needed to quantify the complex relationship between O 3 and PM 2.5 in different regions. This will help support the development of a co-control strategy for O 3 and PM 2.5 .
Research for future O 3 -control policies should consider the expected transformative changes in energy production and transportation industries in response to the national commitment to peak carbon emission in 2030 and carbon neutrality by 2060.

Data availability statement
The data that support the findings of this study are available upon reasonable request from the authors.

Author contributions
T W coordinated the review. Section 1: T W; section 2: J D, Y Z, and T W; section 3: T W; section 4: Y Z, Y T, L X, and T W; section 5: J D and T W; sections 6-8: T W; section 9: Z F; section 10: L X, Y Z, and T W; section 11: T W, Z F, and L X. All authors commented on the full manuscript.