What caused ozone pollution during the 2022 Shanghai lockdown? Insights from ground and satellite observations

. Shanghai, one of China’s most important economic centres, imposed a citywide lockdown in April and May 2022 to contain a resurgence in cases of coronavirus disease 2019. Compared with the 2020 lockdown, the 2022 lockdown occurred in a warm season and lasted much longer, thereby serving as a relevant real-world test of the response of ambient ozone (O 3 ) concentrations to emission reductions in a 10 high-O 3 season. In this study, we analysed surface observations of O 3 and nitrogen dioxide (NO 2 ) concentrations and satellite-retrieved tropospheric NO 2 and formaldehyde (HCHO) column concentrations in the first 5 months of 2022 with comparisons to the year 2021. During the 2-month 2022 lockdown, the maximum daily 8-h average (MDA8) O 3 concentrations at one or more of the city’s 19 sites exceeded China’s air quality standard of 160 µg/m 3 21 times, with the highest value being 200 µg/m 3 . The city-average MDA8 15 O 3 concentration increased by 13% in April–May 2022 year-on-year, despite sharp declines in NO 2 surface and column concentrations (both by 49%) and a 19% decrease in the HCHO column concentration. These results show that the reductions in O 3 precursors and other pollutants during the 2022 lockdown did not prevent ground-level O 3 pollution. An analysis of meteorological data indicates that there were only small changes in the meteorological conditions and there was little transport of O 3 from the high-O 3 inland regions 20 during the 2022 lockdown, neither of which can account for the increased and high concentrations of O 3 that were observed during this period. The mean HCHO/NO 2 ratio in April–May increased from 1.11 in 2021 to 1.68 in 2022, and the correlation between surface O 3 and NO 2 concentrations changed from negative in 2021 to positive in 2022. These results indicate that the high O 3 concentrations


Introduction
Shanghai is a megacity with over 25 million residents and a land area of 6,341 km 2 . It serves as a financial, transport and logistics centre of mainland China. From 27 March to 1 June 2022, the city imposed a strict 2-35 month lockdown (LCD) to curb increases in the number of cases of coronavirus disease 2019 .
Official statistics indicate that both economic activities and human livelihood were severely disrupted by the 2022 Shanghai LCD (hereinafter '2022 LCD'). For example, in April, year-on-year road passenger traffic turnover decreased by 85.5% and road cargo turnover decreased by 73.1% (Shanghai Bureau of Statistics, 2022); total industrial output decreased by 61.6% and power generation decreased by 41.7% (National Bureau 40 of Statistics, 2022); and port cargo throughput at the Port of Shanghai, the world's largest seaport, decreased by 36.5% (Ministry of Transport of the People's Republic of China, 2022). As Shanghai is the most important transportation and logistics hub of the manufacturing-intensive Yangtze River Delta (YRD) region, the ripple effects of the LCD in Shanghai have disrupted supply chains of its surrounding provinces, other regions of China, and the world (Cao et al., 2022;Hale et al., 2022;He, 2022). 45 Such a reduction in human activity can be expected to drastically decrease emissions of air pollutants, as has been borne out by numerous studies of the 2020 LCD (Wang et al., 2022;Huang et al., 2021;Doumbia et al., 2021). Compared with the 2020 LCD in China, which was imposed in late January-early February, the 2022 3 LCD in Shanghai was imposed in the high-O3 season and lasted for longer, during which time the O3 concentrations exceeding the air quality standard were frequently observed. In the present study, we aimed 50 to understand how meteorology and non-linear chemistry influenced ground-level O3 concentrations during the 2022 LCD and whether the huge decrease in NOx emission could push a typical VOC-limited megacity to a NOx-limited regime. We first analysed surface observations of O3 and NO2 concentrations and satellitemeasured NO2 and HCHO column concentrations to assess changes in the concentrations of these species in Shanghai and its surrounding areas during the 2022 LCD. We then examined the roles of meteorological and 55 chemical conditions in determining O3 concentrations during the 2022 LCD by analysing small-and largescale meteorological data and chemical indicators of O3-formation regimes. We discuss the implications of our findings for future strategies aimed at reducing O3 concentrations, which is currently the dominant air pollutant in China in warm seasons.
2 Data and methodology 60

Surface measurement data
We obtained hourly concentrations of ground-level O3, NO2 and particulate matter of 2.5 µm and smaller (PM2.5) recorded at ~1,700 stations in China during 2019-2022 from the China National Environmental Monitoring Centre (http: //106.37.208.233:20035/). The ambient concentrations of O3, NO2 and PM2.5 are measured by an automated monitoring system at each site and reported to the China National Environmental 65 Monitoring Centre and published online after validation (Wang et al., 2014). O3, NO2, and PM2.5 are measured with UV photometry, chemiluminescence, and micro-oscillating balance or β absorption, respectively, with a detection limit of 2 ppb, 2 ppb and 2 µg/m 3 , respectively (https://www.mee.gov.cn). The national network has 10 stations in Shanghai in 2019-2020, most of which are situated within the city centre. Since 2021, the number of stations has increased to 20 with most newly added sites locating outside the city centre (refer to 70 Fig. 1 and Fig. S1 for their locations). (Note there were no data from one site (Dianshanhu) in 2021-2022, thus the data from 19 stations are available for years of 2021 and 2022). The maximum daily 8-h average 4 (MDA8) O3 concentration was calculated for each site as the highest value of the 24 8-h moving average O3 concentrations for a given day. The daily average concentrations of the other pollutants (NO2 and PM2.5) were also calculated from the hourly data. The 14-day moving averages of the surface pollutant concentrations  75 were calculated, as these had fewer fluctuations than daily concentrations and thus better revealed trends.
Finally, the daily average mixing ratios of Ox (= O3 + NO2) were calculated to account for the titration effect of nitric oxide (NO) on changes in O3 concentration.

Satellite and meteorological data
Satellite data were used to investigate the spatiotemporal variations in tropospheric formaldehyde (HCHO) 80 and NO2 column concentrations during the 2022 LCD period and the same period in 2021. We obtained Sentinel-5P Level-3 Offline products (HCHO and NO2) using the TROPOspheric Monitoring Instrument (TROPOMI) from the Google Earth Engine (GEE; https://earthengine.google.com/) cloud-based platform, which is an open-source processing system based on JavaScript (Ghasempour et al., 2021). The Sentinel-5P Level-3 Offline products were converted from the original Sentinel-5P Level-2 Offline data (at a resolution 85 https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-pressure-levels). We calculated monthly averages of the surface-observed meteorological data in 2021 and 2022 and the changes in gridded meteorological data between 2022 and 2021 to determine the spatiotemporal variations in meteorological conditions in Shanghai and its surrounding regions.

Backward trajectories analysis 105
The 24-h backward trajectories for Shanghai were calculated at 1-h intervals during April-May 2022 using MeteoInfoMap software (Wang, 2014(Wang, , 2019 and meteorological data from the Global Data Assimilation System (ftp://arlftp.arlhq.noaa.gov/pub/archives/gdas1/). The endpoints of the trajectories were 500 m above ground level of Shanghai (31.23°N, 121.47°E). Cluster classification was then conducted to divide the trajectories of O3 exceedance days into five groups based on the origins of air mass. 110

Site classification and regression analysis
We classified the aforementioned 19 environmental monitoring sites in Shanghai into four types based on land use and mean HCHO/NO2 ratios ( Fig. 1 and Fig. 8b). The type A sites (12 sites) are in the city centre (and almost all are within the Shanghai Outer Ring Expressway) and had the lowest HCHO/NO2 ratios; type B are sites in the city perimeter (4 sites) that had moderate HCHO/NO2 ratios; type C are semi-rural sites (2 115 sites) that had the largest HCHO/NO2 ratios; and type D site (one site) is located near the wetland park in the East Chongming Tidal Flat and is least affected by urban emissions but may be influenced by ship emissions.
Based on the above classification, we conducted regression analyses between the surface MDA8 O3 concentrations and daily mean NO2 concentrations for three types of sites (A, B and C) to examine the O3 formation regime during the 2022 LCD in Shanghai's city centre, city perimeter and semi-rural sites. ). Within Shanghai, the year-on-year increase in MDA8 O3 concentrations decreased from urban to semirural areas, with the largest increase recorded in the Putuo district in the city centre (32.9 µg/m 3 ) (Fig. 3f). both the NO2 and HCHO column concentrations were related to decreases in anthropogenic activity during the 2022 LCD. Due to a larger decrease in NO2 concentrations than in HCHO concentrations, the average HCHO/NO2 ratio increased from 2.57 in 2021 to 2.96 in 2022 in the YRD region (Fig. 4c). In Shanghai, the HCHO and NO2 column concentrations decreased by 19.4% and 49.2%, respectively, during the 2022 LCD (see Section 3.3.2 for a detailed discussion). In comparison, the TROPOMI-based HCHO and NO2 column 160 concentrations in February 2020 showed 17% and 38% decreases, respectively, in the YRD region compared with the same month in 2019 (Stavrakou et al., 2021). The much larger reduction in NO2 concentrations than in HCHO concentrations during the 2022 LCD is attributable to the fact that NOx is mainly emitted by transportation activities (which exhibited the greatest decrease during the 2022 LCD) and by power generation, whereas VOCs are derived from more diverse sources. In Shanghai, the main VOC sources are 165 vehicular exhaust, evaporation of fuels, paints and solvents, petrochemical industries, liquefied petroleum gas and biogenic sources (Lin et al., 2020;Han et al., 2022).

Effects of meteorological conditions on O3 pollution in Shanghai
A large body of literature has indicated that meteorological conditions can significantly affect O3 170 concentrations by altering emissions, chemical reaction rates, and distribution and removal processes (e.g., Lu et al., 2019;Liu and Wang, 2020;Liu et al., 2021;Jacob and Winner, 2009;Lin et al., 2008;He et al., 2017). To gain insights into the weather conditions during the 2022 LCD period, we compared several meteorological parameters recorded at surfaces in Shanghai (temperature, RH, and wind speed and direction) ( Fig. 5) and the ERA5 reanalysis data for large regions (Fig. 6). The results indicate that during the 2022 175 LCD, the average surface air temperature and RH decreased compared with the same period in 2021. That is, the mean temperature was 18.4 °C in 2022 versus 19.9 °C in 2021, and the mean RH was 71.4% in 2022 versus 74.2% in 2021 ( Fig. 5a-b). The ERA5 reanalysis data revealed there was a decrease in the surface (2m) temperature in Shanghai, whereas there was an obvious increase in the 2-m temperature in the northern part of the YRD region and a large decrease in the 2-m temperature the southern part of the YRD region (Fig.  180   6). The decrease in mean temperature in Shanghai during the 2022 LCD may have slowed O3 production by reducing chemical reaction rates and biogenic emissions. The ERA5 reanalysis data also revealed that during the 2022 LCD there were insignificant changes in cloud cover, downward UV radiation, boundary layer heights and total precipitation in Shanghai but considerable changes in these parameters in the surrounding areas (Fig. 6). 185 We also assessed whether there was a significant change in surface air flow in Shanghai during the 2022 LCD compared with the preceding year. An examination of surface winds in Shanghai showed that during the 2022 LCD, predominant surface winds were from the north-east-southeast sectors (Fig. 5c-d), which is consistent with the 24-h calculated back trajectories (Fig. S4). This indicates that for the majority of the 2022 LCD, Shanghai was upwind of other cities in the YRD region. Compared with the same period in 2021, 190 during the 2022 LCD there was an increase in the occurrence of northerly winds and a decrease in the 9 occurrence of westerly winds. To check for transport of O3 from high-O3 areas in the northwest direction, we examined surface wind flows during the 21 O3-exceedance days. Both surface wind flows and back trajectories indicated that during these days air mainly came from the northeast-east-southeast directions ( Fig. 7), indicating that air from other YRD cities contributed little to the high-O3 days in Shanghai during 195 the 2022 LCD.
The above results suggest that the increase in O3 concentrations in Shanghai during the 2022 LCD were not due to changes in meteorological conditions but were a result of local chemical production.

Effect of the changes in O3 formation regimes in Shanghai
The photochemical production of O3 is controlled by the non-linear chemistry of NOx (NO and NO2) and 200 VOCs (NRC, 1992). It is well known that in many cities, O3 concentrations decrease as VOC emissions decrease but increase as NOx emissions decrease (e.g., Wang et al., 2022). The literature has shown that Shanghai largely operates within a VOC-limited regime (e.g., Lin et al., 2020). We found observational evidence -the HCHO/NO2 ratios and the relationship between O3 and NO2 concentrations -that shows that the city remained in a VOC-limited regime before the 2022 LCD (both in the earlier months of 2022 and in 205 April-May of 2021) but may have transitioned near a NOx-VOC co-limited regime during the 2022 LCD.
The HCHO/NO2 ratio has been used as a proxy for O3 formation regimes, with low ratios indicating VOClimited conditions, high ratios indicating NOx-limited conditions, and intermediate values indicating a colimited regime. Figure 8 shows the temporal variations in city-average HCHO/NO2 ratios in January-May of 2021 and 2022 (Fig. 8a) and the spatial variation of the ratios in Shanghai (Fig. 8b). It indicates that the ratios 210 were comparable (0.56 versus 0.58) in the pre-2022-LCD months of 2021 and 2022 but significantly increased during the 2022 LCD (from 1.11 to 1.68) (Fig. 8a and Fig. S5). Moreover, the ratios were higher in the southern part than the northern part Shanghai, which houses the major urban districts (Fig. 8b).
Previous analysis of the historical OMI-HCHO/NO2 ratios in Shanghai indicated a general rising trend in the ratios with values ranging from 0.5-1.5 in 2005-2019 in warm seasons (April-September) (Itahashi et al., 215 2022;Li et al., 2021;Lee et al., 2022). For April and May, the months when the 2022 Shanghai LCD took place, the HCHO/NO2 ratios ranged from 0.73 to 1.36 were in 9 out of 10 years during 2010-2019 with a higher ratio (1.61) for year 2014 (Li et al., 2021). This shows that the HCHO/NO2 ratio during LCD 2022 (1.68) was high compared with that of the same months in the past decade, resulting from the sharply reduced NO2 column concentrations during the LCD. Several other megacities in East Asia have also seen increasing 220 HCHO/NO2 ratios during 2015-2019 (Itahashi et al., 2022;Lee et al., 2022).
To determine the regime transition threshold, we adopted an observation-based method similar to that which has been used by previous researchers (Jin et al., 2020;Wang et al., 2021;Schroeder et al., 2022): we plotted the city-average MDA8 O3 concentrations against the HCHO/NO2 ratio for the first 5 months of 2021 and 2022 (Fig. 9). The peak O3 concentrations increased as the HCHO/NO2 ratio increased and plateaued at 225 approximately 2, indicating that an HCHO/NO2 ratio of 2 was the threshold for transition from a VOC-limited to a co-limited regime. This value is similar to that determined (2.3) for other major Chinese cities (Wang et al., 2021); however, it is less than that determined (~3) for several US cities (Jin et al., 2020) and greater than that which has been derived (1) from model simulations (Duncan et al., 2010;Li et al., 2021). The cityaveraged HCHO/NO2 ratio was greater than 2 on 15 days during the 2022 LCD but on just 2 days in 2021 230 (see Fig. 9). Moreover, a spatial analysis shows that during the 2022 LCD, the southern part of Shanghai was in a VOC-NOx co-limited regime (with an HCHO/NO2 ratio > 2), whereas the city centre in the northern part remained in the VOC-limited regime (Fig. 8b).
The relationship between surface O3 and NO2 concentrations supports the inference made based on satellitederived HCHO/NO2 ratio data. It is known that O3 concentrations are negatively correlated with NOx, NOy 235 or NOz concentrations (NOy = NOx + oxidation products of NOx [NOz]) under a NOx-titrated condition; this relationship is slightly positive (with small ΔO3/ΔNOx ratios) in a VOC-limited regime and very positive (with large ΔO3/ΔNOx ratios) in a NOx-limited regime. For example, previous studies have suggested that an afternoon ΔO3/ΔNOz ratio (ppb/ppb) less than 4 corresponds to a VOC-limited regime, an afternoon ΔO3/ΔNOz ratio greater than 7 corresponds to a NOx-limited regime and an afternoon ΔO3/ΔNOz of 4-7 240 11 corresponds to a transition regime (Wang et al., 2017). Only NO2 or NOx (not other forms reactive nitrogen) are measured by most regular air-monitoring networks, including the China Environmental Monitoring Network used in this study. Figure 10 shows the scatter plot of the MDA8 O3 and daily average NO2 concentrations in three types of sites in Shanghai: the city centre (12 sites), the city perimeter (4 sites) and the semi-rural areas (2 sites). The VOC-limited regime in the city centre sites (with an average HCHO/NO2 245 ratio of 1.43) had the smallest O3/NO2 slope (3.4) and that in the transition regime sites (with an HCHO/NO2 ratio of 2.27) had the largest O3/NO2 slope (5). In comparison, in 2021, the peak O3 concentration had either The above results suggest that the increased O3 concentrations observed during the 2022 LCD were mainly due to increased O3 production, which resulted from a larger reduction in NOx emissions than in VOC emissions under a VOC-limited condition. Additionally, the decrease in particulate emissions, which reduces the uptake of radicals and O3 and increases radiation, could have increased O3 concentrations. Previous model simulations of the 2020 LCD in central China, which saw similar decreases in emissions (i.e., NOx: ~50%, 255 PM: ~30%), showed that the decrease in NOx emissions made a much larger contribution than the decrease in PM emissions to the increase in O3 concentrations (Liu et al., 2021). We believe that this also occurred during the 2022 LCD, i.e., the O3 concentration increase was mainly due to enhanced O3 production, which occurred as a result of a large reduction in NOx emissions. The O3 formation regime during the 2022 LCD remained VOC-limited in urban areas but entered co-limiting conditions outside the city centre. 260

Summary and implications
This study analysed the causes of frequent ground-level O3 pollution during the 2022 LCD (April-May 2022) in Shanghai and assessed the increases in O3 concentrations compared with the same periods in previous years using ground and satellite-based observations. We found that despite large reductions in the activities of transportation sectors and industries during the 2022 LCD, frequent exceedances of the O3 air quality 265 standard (~30% days) were observed at ground level. Moreover, the O3 concentrations were increased during the 2022 LCD compared with the same period in the preceding year. This increase resulted from a large reduction in NOx emissions (~50% from surface and satellite-based measurements) and a small reduction in VOCs (19% in HCHO column concentrations) within Shanghai. In contrast, meteorology and outside influences had insignificant effects on O3 concentrations during the 2022 LCD. Moreover, O3 formation 270 during the 2022 LCD remained in the VOC-limited regime at most urban and suburban sites but transitioned to a VOC-NOx co-limited regime in semi-rural areas.
Our findings on the O3 response to the 2022 LCD have implications for mitigating summer O3 pollution, which has become the predominant air-pollution problem during warm seasons in China. O3 pollution is also a persistent environmental hazard in the US and Europe even after a few decades of research and control (e.g., 275 Tao et al., 2022;Derwent and Parrish, 2022). Similar to Shanghai, many of the world's cities, such as Los Angeles and New York, are still in VOC-limited regimes (Jin et al., 2020;Tao et al., 2022). Our results show that drastically decreasing emissions from conventional fossil fuel-powered transportation sectors during the 2022 LCD can lead to increased and elevated O3 concentrations in VOC-limited urban areas. China is aiming to have its carbon emissions peak by 2030 and to achieve carbon neutrality by 2060, and other countries are 280 also committing to drastically reduce their carbon emissions. This will necessitate the rapid uptake of electric vehicles in many cities. However, although large-scale adoption of electric vehicles will greatly improve overall air quality, VOC emissions from other sectors will need to be decreased at the same time to prevent increases in O3 concentrations from the reductions in NOx (and particulate) emissions in the early stages of phase-out of conventional vehicles. Over time, the wide application and acceptance of renewable energy in 285 transportation and energy production will help cities reach the NOx-limited formation regime, alleviate ground-level O3 pollution and achieve carbon-reduction targets. The aggravated O3 pollution during the 2022 Shanghai LCD after large reductions in transportation emissions suggests that it may take considerably long time for some cities reach a NOx-limited regime. 13 Code and data availability. The code or data used in this study are available upon request from Tao Wang 290 (tao.wang@polyu.edu.hk).
Supplement. The supplement related to this article is available online at: Author contributions. TW initiated the research and designed the framework of data analysis. YT processed the data and made the figures. TW and YT analysed the results and wrote the paper.
Competing interests. One author (Tao Wang) is a member of the editorial board of Atmospheric Chemistry 295 and Physics. The peer-review process was guided by an independent editor, and the authors have no other competing interests to declare.   Forecast Reanalysis v5 data for the Yangtze River Delta and its neighbouring regions. 2-m temperature (t2m) (a), downward ultraviolet (UV) radiation at the surface (uvb) (b), total precipitation (tp) (c), relative humidity at 1,000 hPa (r) (d), total cloud cover (tcc) (e) and boundary layer height (blh) (f).   for 3 types of sites.