ENSO modulation of summertime tropospheric ozone over China

Ozone (O3) is one of the most critical pollutants affecting air quality in China in recent years. In this study, different impacts of the El Niño–Southern Oscillation (ENSO) warm/cold phases on summertime tropospheric O3 over China are examined based on model simulations, ground measurements, and reanalysis data. Summertime surface O3 concentrations in China show a positive correlation with ENSO index during years 1990–2019, with the largest increases by 20% over southern China in El Niño (warm phase) relative to La Niña (cold phase) years. The ENSO modulation extends to the middle and even upper troposphere. Our analysis indicates that O3 flux convergence associated with weakened southerlies is the primary reason for the increase in tropospheric O3 over southern China. In addition, the O3 increase during El Niño years is mainly from domestic emissions in China. This study highlights the potential significance of ENSO in modulating tropospheric O3 concentrations in China, with great implications for O3 pollution mitigation.


Introduction
Tropospheric ozone (O 3 ) is now the most critical air pollutant in China apart from particulate matter in recent years (Fu et al 2019. In the recent decade, daily maximum 8 h average (MDA8) O 3 concentration in China in the warm season increased at a rate of 5% per year, which is faster than any other regions of the world (Lu et al 2018(Lu et al , 2020. Tropospheric O 3 is produced through the oxidation of carbon monoxide (CO) and volatile organic compounds (VOCs) with the sunlight in the existence of nitrogen oxides (NO x ). High O 3 concentration harms human health (Jerrett et al 2009, Malley et al 2017 and reduces crop yields (Yue et al 2017, Mills et al 2018. Therefore, it is of great significance to understand factors that modulate the variation of O 3 concentrations in China. In addition to anthropogenic emissions of precursor gases (Qu et al 2020) and atmospheric oxidation capacity (Lu et al 2019a), meteorological conditions also exert various impacts on O 3 concentrations (Yin et al 2019). High temperature along with intense solar radiation enhances both photochemical reactions and natural precursor emissions of O 3 (Jacob and Winner 2009). Thus, O 3 pollution is generally severe in summer with strong sunlight. Relative humidity (RH) is negatively correlated with O 3 concentration due to the strong uptake of O 3 by trees under high RH conditions (Kavassalis and Murphy 2017) and water vapor from marine air also serves as a chemical sink of O 3 at low NO x conditions (Lu et al 2018). The existence of clouds can reduce O 3 concentrations via decreasing the solar radiation and hence photochemical reactions (Jeong and Park 2013). Through changing the air stagnant condition and atmospheric transport, winds influence O 3 concentrations in both local and downwind areas (Doherty et al 2013, Zhang et al 2017. Gong et al (2019) revealed that O 3 pollution events in northern China occurred with high temperature, low RH, an anomalous high in the mid-troposphere, and downdraft in the lower troposphere. Changes in meteorological conditions are strongly associated with large-scale atmospheric circulation anomalies. Yang et al (2014) found that the year-by-year changes in the East Asian summer monsoon strength caused an interannual variation in near-surface O 3 concentrations over central-eastern China by 2%-5%, attributed to the changing transboundary transport of O 3 , with high O 3 levels appearing in strong monsoon years. A stronger western Pacific subtropical high can bring in more humid air, resulting in more clouds and lower temperature in southern China, which decreases O 3 concentrations there Wang 2017, Yin et al 2019).
El Niño-Southern Oscillation (ENSO) is the dominant mode of variability in Earth's climate system at interannual time scales (Ropelewski and Halpert 1987). It features anomalous high sea surface temperatures (SSTs) over the tropical eastern Pacific Ocean during the warm phase (i.e. El Niño) and anomalous low SST during the cold phase (i.e. La Niña). ENSO can influence atmospheric circulations and weather conditions across the globe and further change global O 3 distributions. Numerous studies have examined the ENSO influences on tropical O 3 (e.g. Oman et al 2011, Xie et al 2014, Olsen et al 2016 and the transpacific transport of O 3 McElroy 2010, Xue et al 2021).
Although ENSO signal weakens in boreal summer season, when an ENSO event is developing, until its mature phase in boreal winter of the year, ENSO has been reported to have remarkable impacts on summertime O 3 in the mid-latitudes of the Northern Hemisphere (Zhang et al 2015, Shen and Mickley 2017, Xu et al 2017, Wie et al 2021. Based on surface ozone observations and meteorological variables from reanalysis data, Xu et al (2017) found that surface O 3 concentrations decreased during El Niño and increased during La Niña years (including summer) over the continental United States over 1993-2013, which was attributed to ENSO-induced changes in chemical processes and dynamic transport. Shen and Mickley (2017) also reported an El Niño-O 3 connection in eastern United States and found that El Niño reduced moisture transport into the Atlantic coast states but enhanced moisture flux into the southcentral states, leading to surface O 3 increases and decreases, respectively, over these two regions in summer. Zhang et al (2015) investigated the influence of ENSO on the total O 3 column (TOC) in China using satellite data and found that summertime TOC increased over northeastern and northwestern China during El Niño events, while it decreased over southern and southwestern China as well as in the middle reach of the Yellow River during La Niña events. Based on satellite measurements and chemistry-climate model simulations, Wie et al (2021) examined the variability of tropospheric column O 3 over East Asia and found that that O 3 increased in the troposphere 4-5 months after the La Niña peak, associated with anomalous O 3 transport from higher latitudes. However, few studies have quantified the roles of different physical and chemical processes in determining the impact of ENSO on summertime O 3 variability in China. In addition, ENSO-driven O 3 changes near the surface in China, which is more directly linked to air quality and human health than the total column burden, were rarely studied in previous works.
In this study, the impact of ENSO on summertime tropospheric O 3 concentrations in China and its mechanisms are investigated using global chemical transport model simulations over 1990-2019 driven by meteorological fields from reanalysis data. Details of the model and experimental design are described in section 2. Section 3 quantifies the influence of ENSO on near-surface O 3 concentrations in China. The mechanisms behind the impacts are analyzed in section 4. Section 5 summarizes the main results and discusses potential uncertainties in this study and implications for future studies.

Model description
O 3 concentrations are simulated in this study by using GEOS-Chem model v12.9.3 with a horizontal resolution of 2 • latitude × 2.5 • longitude and 47 vertical layers from the surface to 0.01 hPa. GEOS-Chem is a global model of atmospheric chemistry with fully coupled O 3 -NO x -hydrocarbon-aerosol chemical mechanisms (Pye et al 2009, Mao et al 2013, Sherwen et al 2016. The boundary layer mixing is represented by a nonlocal scheme (Lin and McElroy 2010) and the stratospheric O 3 chemistry uses the LINOZ scheme (McLinden et al 2000). Meteorological fields driving the model in this study are from NASA Modern-Era Retrospective analysis for Research and Applications, Version 2 (MERRA-2). Additionally, the geopotential height data from the National Center for Environmental Prediction and National Center for Atmospheric Research reanalysis are used in our analysis due to the lack of geopotential heights in the MERRA-2 reanalysis products. The tendencies of different chemical and physical processes of tracers are written out from GEOS-Chem, which can be used to quantify the relative contribution of individual processes to variations in O 3 concentrations.

Emissions
Global baseline anthropogenic emissions utilized in this study are obtained from the community

Experiments
In this study, O 3 concentrations are estimated for years 1990-2019 in the baseline simulation (BASE) using the GEOS-Chem model driven by MERRA-2 meteorological fields. In order to quantify the impact of ENSO on O 3 in China, the effect from emission changes is removed by keeping anthropogenic, biomass burning and natural emissions at their year 2017 level in the BASE simulation. But we note that the year-to-year emissions are different and choosing 2017 as the base year may cause a small bias in the ENSO influence in magnitude. The quantitative impact of ENSO on the interannual variation of summertime O 3 in China can be obtained from the BASE experiment. Unless specified otherwise, results in this study are derived from the BASE experiment. Another sensitivity simulation, which is same as BASE but with anthropogenic emissions of O 3 precursors (CO, NO x and VOCs) from China turned off (NO_CHN), is also conducted to understand the role of China domestic emissions in the ENSO-induced O 3 variation by differing the two simulations (BASE-NO_CHN).

ENSO index
The Niño 3.4 index is employed to characterize the phase and intensity of ENSO events. It is calculated as the area averaged SST anomaly within the region of 5 • S-5 • N and 170-120 • W based on MERRA-2 reanalysis in this study. Niño 3.4 index averaged over June-July-August (JJA) is used to explore the relationship between ENSO and O 3 in China during the boreal summer.

Impact of ENSO on the interannual variation in O 3 over China
To investigate the connection between ENSO and summertime O 3 in China, the empirical orthogonal function (EOF) analysis of GEOS-Chem JJA near-surface O 3 concentrations is conducted. As shown in figure S1(a) (available online at stacks. iop.org/ERL/17/034020/mmedia), the leading EOF can explain 23% of interannual variability of nearsurface O 3 concentrations in Asia and positive pattern of EOF appears in the whole China, especially over southern China. The principal component (PC) of the leading EOF of near-surface O 3 concentrations is positively correlated with the JJA mean Niño 3.4 index (figure S1(b)). The correlation coefficient between the leading PC and Niño 3.4 index is +0.5 and statistically significant at a 99% confidence level, indicating that ENSO may play a vital role in the interannual variations in JJA near-surface O 3 concentrations in southern China, although note that the leading-mode EOF only explains 23% of the O 3 variance over China.
The spatial distribution of the correlation coefficients between the simulated JJA mean near- Note that, there are two types of ENSO, eastern Pacific (EP) type and central Pacific (CP) type (Kao and Yu 2009). We also calculated the correlation coefficient (R) between EP/CP Index based on Kao and Yu (2009) and O 3 concentrations in southern China (figure S2). CP Index is positively correlated with O 3 (R = 0.30), while EP Index does not have discernable correlation (0.03). Considering the higher frequency of CP El Niño events in recent decades relative to past centuries (Freund et al 2019), ENSO could be more influential on O 3 in China. We also tested the Niño 1 + 2 (0.21), Niño 3 (0.50) and Niño 4 (0.52) Indices and found that all the correlation coefficients between these indices and O 3 are lower than Niño 3.4 (0.55). ENSO may have a cross-correlation with O 3 (Shen et al 2017). The correlations between January-February-March/March-April-May mean Niño 3.4  (a)). The dotted areas in (a) indicate statistical significance with 95% confidence from a two-tailed Student's t test. Red and green bars in (b) represent O3 concentrations in the strongest three El Niño (1997, 2002 and 2015) and La Niña (1998, 1999 and 2010), respectively. The correlation coefficient (R) between the Niño 3.4 index and near-surface O3 concentrations in southern China is shown in the top-right corner of panel (b). and JJA O 3 concentrations over southern China are also estimated and they only present weak negative relationships, with a correlation coefficient of −0.26 and −0.10, respectively.
To quantitatively examine the impact of climate variabilities on air pollutants, composite differences between two opposite phases of the climate variability are commonly used in previous studies (e.g. Yang et al 2014). In this study, the top three strongest El Niño years (1997, 2002 and 2015) and La Niña years (1998, 1999 and 2010) during 1990-2019 are firstly selected ( figure 1(b)). However, the simulated JJA near-surface O 3 concentration slightly increased over southern China in the 1999 La Niña year, which is inconsistent with the positive correlation between Niño 3.4 index and O 3 in China. It is likely due to the abnormal winds in JJA 1999 compared to the other two La Niña years (1998 and 2010). In general, the meridional component of winds over southern China shows anomalous northerlies during El Niño and southerlies during La Niña relative to the climatological mean (figure S3), with a correlation coefficient of −0.43 between JJA Niño 3.4 index and 850 hPa meridional wind speed (figure S4). In JJA 1999, possibly dominated by other internal variabilities rather than the La Niña signal, southern China was controlled by anomalous northerlies relative to the climatological mean, leading to a different O 3 transport and thus different O 3 concentration changes in this particular year compared to the other two La Niña years. Therefore, we only use the other two La Niña years (1998 and 2010) in the following composite analysis. Figure 2 shows the composite differences in JJA near-surface O 3 concentrations between the selected three El Niño and two La Niña years and figure  S5 gives the composite differences of the El Niño and La Niña years compared to the climatological mean. O 3 levels in the El Niño years are higher across the whole China, with the largest increases exceeding 9 ppbv (or 20% of the climatological mean) over southern China, compared to the concentrations during the La Niña years. The spatial pattern of the composite differences is identical to those of EOF (figure S1(a)) and correlation coefficients ( figure 1(a)). The ENSO modulation of O 3 in China is not only within the surface layer, but also extends to the middle and even upper troposphere (figures 2(b) and (d)). Averaged over the latitude range of 20-32 • N, O 3 concentrations increase from the surface to 300 hPa during El Niño compared to La Niña years. The maximum increase by 9 ppbv (5%-10%) arises between 105 • E and 110 • E around 700 hPa over southern China.

Mechanisms of ENSO impacts on O 3 in China
ENSO influences the chemical and physical processes of O 3 , including net chemical production, horizontal advection, vertical convection, diffusion and dry deposition, through changing meteorological parameters. The role of each process can be quantified by the integrated process rate analysis to identify the dominant factors that influence the variations of O 3 . The process source/sink rates are summarized in table S1. In general, JJA O 3 in the troposphere over southern China is formed locally by chemical productions and dissipated due to horizontal advection, diffusion and dry deposition indicated in table S1. Figure 3 shows the composite differences in various meteorological parameters in JJA between the selected El Niño and La Niña years. Relative to La Niña years, an anomalous low at 500 hPa is located in southern China during El Niño years ( figure 3(b)), which leads to a decrease in air temperature (figures 3(d) and (g)) (Wu et al 2010) and is unfavorable for the O 3 chemical production (Gong et al 2019). Although the decrease in cloud cover (figure 3(c)) allows more solar radiation reaching the surface (figure 3(f)), partly compensating the less O 3 chemical production due to lower temperature, the net chemical production of O 3 in the troposphere is still lower by −24.8 Gg d −1 over southern China during El Niño compared to La Niña years (table S1). Therefore, the anomalous increase in O 3 during El Niño years is due to the physical processes, especially the horizontal advection, which represents the only net gain (28.1 Gg d −1 ) inducing the increase in O 3 mass in the troposphere over southern China.
In summer, meteorological fields in China are dominated by the East Asian summer monsoon, with prevailing southwesterlies over southern China. Compared to the La Niña years, southwesterlies are weakened during the El Niño years (figure 3(a)) (Zhang et al 2015), leading to an anomalous O 3 flux convergence in southern China. Table 1 summarizes the horizontal and vertical O 3 mass fluxes, estimated as the product of O 3 concentration and wind speed, from the surface to 500 hPa, where O 3 concentrations have the largest change ( figure 2(b)). Due to the weakened southerlies from the surface to the mid-troposphere during the El Niño years, import of O 3 from the south to southern China is reduced by 4.1 Tg, but the export of O 3 out to the north is reduced by 11.2 Tg, compared to La Niña years. The net change in meridional transport results in a O 3 accumulation over southern China. The bigger O 3 flux change in the north of southern China than in the south is due to the higher O 3 concentrations over land than over the coastal areas. The less import of O 3 from the west of southern China (1.4 Tg) and less export to the east (1.0 Tg) cause nearly no change in the zonal transport. Net upward O 3 transport from the lower troposphere (1.1 Tg) due to the horizontal convergence slightly restrains the surface concentration increase. Therefore, O 3 flux convergence is identified as the primary reason for the increase in JJA O 3 levels over southern China during El Niño years, relative to La Niña years.

Conclusions and discussions
Additionally, natural emissions are fixed during the simulations, which can also be perturbed during the ENSO events and should be considered in future studies. Recent works have reported that, besides anthropogenic emissions, the biogenic emissions, including those from urban green spaces, exert a significant role in O 3 formation (Gao et al 2022, Ma et al 2022. The potential influences of fixed biogenic emissions on O 3 deserve further investigation. The impact of ENSO on O 3 is investigated in this study using the MERRA-2 driven GEOS-Chem simulations together with statistical methods. The ENSO modulation could be disturbed by other climate signals. The pure ENSO influences can be more cleanly isolated using general circulation models driven by prescribed SSTs with and without ENSO signal (Zeng et al 2021). The composite differences in meteorological parameters between El Niño and La Niña years from MERRA-2 reanalysis are validated here by comparing with ERA5 reanalysis (figure S7). The changes in meteorological parameters resemble each other, suggesting that MERRA-2 reanalysis is credible in representing the changes in meteorological variables during ENSO events.
The data that support the findings of this study are openly available at the following URL/DOI: https://doi.org/10.5281/zenodo.5338946.