Changes in Air Quality and Drivers for the Heavy PM2.5 Pollution on the North China Plain Pre- to Post-COVID-19

Under the clean air action plans and the lockdown to constrain the coronavirus disease 2019 (COVID-19), the air quality improved significantly. However, fine particulate matter (PM2.5) pollution still occurred on the North China Plain (NCP). This study analyzed the variations of PM2.5, nitrogen dioxide (NO2), sulfur dioxide (SO2), carbon monoxide (CO), and ozone (O3) during 2017–2021 on the northern (Beijing) and southern (Henan) edges of the NCP. Furthermore, the drivers for the PM2.5 pollution episodes pre- to post-COVID-19 in Beijing and Henan were explored by combining air pollutant and meteorological datasets and the weighted potential source contribution function. Results showed air quality generally improved during 2017–2021, except for a slight rebound (3.6%) in NO2 concentration in 2021 in Beijing. Notably, the O3 concentration began to decrease significantly in 2020. The COVID-19 lockdown resulted in a sharp drop in the concentrations of PM2.5, NO2, SO2, and CO in February of 2020, but PM2.5 and CO in Beijing exhibited a delayed decrease in March. For Beijing, the PM2.5 pollution was driven by the initial regional transport and later secondary formation under adverse meteorology. For Henan, the PM2.5 pollution was driven by the primary emissions under the persistent high humidity and stable atmospheric conditions, superimposing small-scale regional transport. Low wind speed, shallow boundary layer, and high humidity are major drivers of heavy PM2.5 pollution. These results provide an important reference for setting mitigation measures not only for the NCP but for the entire world.


Introduction
Air pollution over the past decade has been an increasing concern for the public and central government, especially concerning persistent and severe fine particulate matter (PM 2.5 ) pollution [1][2][3]. Extensive epidemiological studies have shown that long-term exposure to air pollution can increase the incidence of respiratory and cardiovascular diseases [4][5][6]; therefore, prevention and control measures to mitigate air pollution are urgently required. As a response, the State Council promulgated the "Atmospheric Pollution Prevention and Control Action Plan" in September 2013, which targeted reducing PM 2.5 pollution during 2013-2017. Subsequently, the State Council issued the "Three-year Action Plan on Defending the Blue Sky", which further focused on greater achievements in alleviating air pollution during 2018-2020. Specifically, the North China Plain (NCP) is the most polluted region in China, and tremendous pollution controls have been implemented to achieve the targets set by the State Council on the NCP. Major control measures included shutting down small and polluting factories, strengthening industrial and vehicle emission standards, upgrading industrial boilers, and promoting clean energy [7,8]. As air pollution is a regional issue, a regional joint mitigation measure called "2 + 26" strategies, which comprised two megacities (Beijing and Tianjin) and 26 neighboring cities on the NCP [9], was promulgated in September 2017. After the two stages of clean air action plans, air pollution has improved significantly [10][11][12][13][14].
Unexpectedly, Wuhan City, the capital of Hubei Province in China, reported the appearance of coronavirus disease 2019  in late December 2019 [15,16].  and has also been suffering from heavy PM 2.5 pollution. Moreover, seven cities within Henan are located in the Beijing-Tianjin-Hebei (BTH) transmission corridor. However, only a few studies investigated the variations and formation mechanisms of PM 2.5 pollution in Henan. Liu et al. [58] revealed that severe haze events were tightly related to adverse meteorological, atmospheric transformation processes, and regional transport. Song et al. [73] indicated that PM 2.5 deterioration during PM 2.5 pollution episodes was associated with local emissions and secondary formation on the south edge of the NCP, which was different from the northern edge. Moreover, studies on heavy PM 2.5 pollution episodes during active-to post-COVID-19 remain lacking.
To fill the abovementioned gaps, this study aims to comprehensively assess the changes in concentrations of PM 2.5 and gaseous pollutants from 2017 to 2021 based on the observation data over the NCP, focusing on Beijing on the northern edge of the NCP and Henan on the southern edge of the NCP. Further, the drivers for the heavy PM 2.5 pollution episodes were explored in Beijing and Henan pre-to post-COVID-19. Specifically, the three main objectives of the study were to: (1) analyze the variation trends of PM 2.5 and gaseous pollutants during 2017-2021 in Beijing and Henan; (2) determine the PM 2.5 pollution status when anthropogenic emissions significantly decreased, especially between 2019 and 2021; (3) explore the formation mechanism of heavy PM 2.5 pollution on the northern and southern edges of the NCP between 2019-2021. The findings gained from the present study would provide an important theoretical reference for future air pollution prevention strategies over the NCP.

Study Region
Beijing, as the capital of China, is the political and cultural center and is located on the northern edge of the NCP. With the dramatic economic development, Beijing has become one of the well-developed megacities. Along with economic development, the ensuing problem of air pollution was not ignored. Notably, the high emissions plants were relocated out of Beijing, and cleaner production techniques were employed to alleviate industrial emissions [52]. Luo et al. [74] indicated that vehicular and secondary sources have become much more dominant recently. In contrast, Henan is located on the southern edge of the NCP and is a developing province. Henan is one of the largest energy-consuming provinces in China, with coal as its primary energy source [54]; hence, Henan faces greater challenges in balancing development and the ecological environment. Industrial, transportation, and energy consumption contribute to substantial PM 2.5 emissions [75]. Moreover, Henan is the second largest agricultural province in China, accounting for 25% of China's wheat output every year [76], and planting and agriculture are important emission sources of air pollution.

Ambient Air Pollutants and Meteorological Dataset
The ground-level monitoring network of ambient air pollutants established by the Ministry of Ecology and Environment of China began operating in 2013. In the present study, ground-level hourly concentrations of five ambient air pollutants in Beijing and Henan were employed, including PM 2.5 , NO 2 , SO 2 , CO, and O 3 . PM 2.5 was measured via the micro-oscillating balance method. For gaseous pollutants, the NO 2 , SO 2 , CO, and O 3 were monitored using chemiluminescence, ultraviolet fluorescence, non-dispersive infrared absorption, and ultraviolet spectrometry, respectively. The ambient air pollutant data was released by the China National Environmental Monitoring Centre, and was collected from the online database (https://quotsoft.net/air/#archive, accessed on 27 July 2022). The daily-average concentrations of PM 2.5 , NO 2 , SO 2 , and CO were computed as the 24-h mean values for each site. As for O 3 , the maximum daily 8-h average (MDA8) concentration was calculated. Moreover, the monthly and yearly concentrations were acquired based on the daily levels.
The present study aims to highlight the variation trends of PM 2.5 and gaseous pollutants and the drivers for the persistent PM 2.5 pollution episodes during pre-to post-COVID The meteorological data include boundary layer height (BLH), 10 m wind components u and v, temperature and dewpoint temperature at 2 m during 2019-2021, and were obtained from the European Center for Medium-Range Weather Forecasts (ECMWF) ERA5 hourly reanalysis dataset (http://cds.climate.copernicus.eu/cdsapp#/home, accessed on 27 July 2022), with a resolution of 0.25° × 0.25°. Further, relative humidity (RH) was calculated employing temperature and dewpoint temperature at 2 m. Additionally, meteorological inputs to the backward trajectories were 1° × 1° Global Data Assimilation System (GDAS) data from the National Centers for Environmental Prediction (NCEP) reanalysis (available at ftp://arlftp.arlhq.noaa.gov/pub/archives/gdas1, accessed on 27 July 2022).

Potential Sources Analysis
To investigate the origin of airborne particles during heavy PM2.5 pollution episodes, 24-h backward trajectories were calculated four times each day (00:00, 06:00, 12:00, and 18:00 UTC) at the height of 500 m above the ground in Beijing (39.87° N, 116.43° E) and Henan (34.75° N, 113.63° E), based on the Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model developed by the National Oceanic and Atmospheric Administration (NOAA) Air Resources Laboratory (ARL) [77]. Based on the backward tra- The meteorological data include boundary layer height (BLH), 10 m wind components u and v, temperature and dewpoint temperature at 2 m during 2019-2021, and were obtained from the European Center for Medium-Range Weather Forecasts (ECMWF) ERA5 hourly reanalysis dataset (http://cds.climate.copernicus.eu/cdsapp#/home, accessed on 27 July 2022), with a resolution of 0.25 • × 0.25 • . Further, relative humidity (RH) was calculated employing temperature and dewpoint temperature at 2 m. Additionally, meteorological inputs to the backward trajectories were 1 • × 1 • Global Data Assimilation System (GDAS) data from the National Centers for Environmental Prediction (NCEP) reanalysis (available at ftp://arlftp.arlhq.noaa.gov/pub/archives/gdas1, accessed on 27 July 2022).

Potential Sources Analysis
To investigate the origin of airborne particles during heavy PM 2.5 pollution episodes, 24-h backward trajectories were calculated four times each day (00:00, 06:00, 12:00, and 18:00 UTC) at the height of 500 m above the ground in Beijing (39. [77]. Based on the backward trajectories, the potential source contribution function (PSCF) method was applied to evaluate the potential geographic origins for PM 2.5 pollution episodes in Beijing and Henan [78]. Prior to conducting the PSCF analysis, the geographic region covered by the backward trajectories was divided into an array of 0.5 • × 0.5 • grid cells. The PSCF is a conditional probability function [66]. Specifically, the PSCF value in the grid (i, j) is calculated as M ij /N ij , where i and j are the latitude and longitude, N ij is the number of endpoints that fall in the grid (i, j), and M ij is the number of polluted trajectory endpoints in the grid (i, j). The daily PM 2.5 concentration exceeding 75 µg m −3 refers to a polluted day according to the new China National Ambient Air Quality Standards issued in 2012. In the present study, the polluted trajectory was thus determined to exceed a threshold concentration of 75 µg m −3 . High PSCF values represent higher probabilities of grid cells being potential geographic origins, making a greater contribution to air pollution within a receptor region. When calculating the PSCF values, Zeng and Hopke [79] found some grid cells only had one endpoint (N ij = 1). If this endpoint corresponds to a polluted trajectory, the PSCF values for these grid cells will be 1. However, the confidence in these PSCF values is very low. Therefore, to reduce the PSCF uncertainties caused by the small values of N ij , an arbitrary weight factor (W ij ) was multiplied by the PSCF value [80]. The weighted PSCF (WPSCF) was calculated as WPSCF ij = PSCF ij × W ij . The W ij is defined as follows [81,82], (1) .5%, and 7.9%, respectively, compared to the previous year. Similarly, the PM 2.5 concentrations in Henan decreased by 6.0%, 4.8%, 13.3%, 9.6%, respectively. For Beijing, the decline rate (17.6%) from 2018 to 2019 was the highest. However, the highest decline rate (−13.3%) in Henan was observed from 2019 to 2020, with a one-year delay compared to Beijing. Furthermore, it is noted that the average PM 2.5 concentrations in Henan were significantly higher than that in Beijing.  As for NO2, the annual-average concentrations in Beijing showed a decreasing trend during 2017-2020, from 44 μg m −3 in 2017 down to 28 μg m −3 in 2020. Particularly, the reduction in 2020 was the largest, with a decrease rate of 22.2%. However, the annual average concentrations of NO2 rebounded slightly in 2021, increasing by 3.6% relative to 2020. Unlike Beijing, the annual concentrations of NO2 in Henan exhibited a continuous decreasing trend during the study period, from 41 μg m −3 in 2017 to 27 μg m −3 in 2021. The most pronounced reduction occurred in 2020 (−14.3%), followed by 2021 (−10.0%). In con- As for NO 2 , the annual-average concentrations in Beijing showed a decreasing trend during 2017-2020, from 44 µg m −3 in 2017 down to 28 µg m −3 in 2020. Particularly, the reduction in 2020 was the largest, with a decrease rate of 22.2%. However, the annual average concentrations of NO 2 rebounded slightly in 2021, increasing by 3.6% relative to 2020. Unlike Beijing, the annual concentrations of NO 2 in Henan exhibited a continuous decreasing trend during the study period, from 41 µg m −3 in 2017 to 27 µg m −3 in 2021. The most pronounced reduction occurred in 2020 (−14.3%), followed by 2021 (−10.0%). In contrast to PM 2.5 , NO 2 pollution levels were higher in Beijing than in Henan, except in 2020.

Results and Discussions
For SO 2

Monthly Variations
The monthly variations of PM 2.5 and gaseous pollutants (NO 2 , SO 2 , CO, and O 3 ) in Beijing and Henan were further analyzed ( Figure 3). Notably, the monthly average concentrations of air pollutants during 2017-2019 were calculated as the historical levels before COVID-19, and the present study highlighted the monthly variations of air pollutants during and post-COVID-19 (2020-2021). The monthly variations of PM 2.5 concentrations exhibited a U-shaped pattern in Henan, which was consistent with the results of Yao et al. [54]. Unlike Henan, the monthly average concentrations of PM 2.5 in Beijing fluctuated, and differences between winter and summer were relatively small, without showing a pronounced U-shaped pattern. This may be attributed to a series of strict air pollution control measures in Beijing during recent years, which has dramatically improved PM 2.5 pollution, especially in winter with higher PM 2.5 concentrations.
In 2020, the average PM 2.5 concentration in Beijing increased slightly by 6.91% to 62.47 µg m −3 in February relative to January. Subsequently, the PM 2.5 concentration reduced sharply (−44.35%) in March compared to February and remained relatively stable before increasing in October. Gao et al. [83] also revealed a similar phenomenon; that is, PM 2.5 concentration began to decline in March of 2020. Moreover, the monthly PM 2.5 concentrations showed a decreasing trend in November and December. Notably, the monthly average concentration in December was relatively low, at 28.51 µg m −3 . The monthly concentrations in 2021 were generally at a lower level compared to 2017-2020, except for March. It is worth noting that PM 2.5 concentration in March of 2020 was significantly lower than in historical years (2017-2019) and 2021. However, the PM 2.5 concentration in March of 2021 reached a peak (86.59 µg m −3 ). In Henan, the PM 2.5 concentration in February dropped significantly by 47.54% relative to January 2020. Compared with the corresponding monthly averages in 2020, the monthly averages in 2021 in Henan generally showed minor fluctuations. [54]. Unlike Henan, the monthly average concentrations of PM2.5 in Beijing fluctuated, and differences between winter and summer were relatively small, without showing a pronounced U-shaped pattern. This may be attributed to a series of strict air pollution control measures in Beijing during recent years, which has dramatically improved PM2.5 pollution, especially in winter with higher PM2.5 concentrations. In 2020, the average PM2.5 concentration in Beijing increased slightly by 6.91% to 62.47 μg m −3 in February relative to January. Subsequently, the PM2.5 concentration reduced sharply (−44.35%) in March compared to February and remained relatively stable before increasing in October. Gao et al. [83] also revealed a similar phenomenon; that is, PM2.5 concentration began to decline in March of 2020. Moreover, the monthly PM2.5 concentrations showed a decreasing trend in November and December. Notably, the monthly average concentration in December was relatively low, at 28.51 μg m −3 . The monthly concentrations in 2021 were generally at a lower level compared to 2017-2020, except for March. It is worth noting that PM2.5 concentration in March of 2020 was significantly lower than in historical years (2017-2019) and 2021. However, the PM2.5 concentration in March of 2021 reached a peak (86.59 μg m −3 ). In Henan, the PM2.5 concentration in February dropped significantly by 47.54% relative to January 2020. Compared with the corresponding As for gaseous pollutants, the monthly-average concentrations of NO 2 , SO 2 , and CO were generally higher from October to March than those from April to September, while O 3 showed an opposite trend due to its formation mechanism. The increase in temperature and solar radiation favored the photochemical formation of O 3 from VOCs and NO x between April and September [84,85]. Meanwhile, the reduction in NO x levels contributed to higher O 3 levels by hindering the NO x titration and/or the effect of radical terminating reactions [19]. It is worth noting that the NO 2 concentrations in February decreased generally compared with that in January. Especially in 2020, the monthly-average concentration of NO 2 in February dropped sharply by 33.58% in Beijing and subsequently remained stable, while that in August started to increase significantly. In contrast with 2020, the monthly NO 2 concentration in March during the period of historical years (2017-2019) and 2021 rebounded rapidly. Notably, the monthly NO 2 concentration between February and May in 2020 was the lowest compared to 2017-2019 and 2021. Similar to Beijing, NO 2 concentration in Henan decreased sharply by 54.49% in February 2020 relative to January. However, NO 2 concentration rebounded rapidly in March, increasing by 67.06%. The monthly averages of NO 2 in 2021 generally decreased relative to 2017-2020, while the averages in January and February of 2021 were higher than in 2020, especially in February.
Generally, the NO 2 concentration is lower in February relative to January, as many household heating and industrial activities decline around this time of year due to the Spring Festival [70]. However, the sharper drop in NO 2 concentration during February 2020 in Beijing and Henan was mainly attributed to lockdown measures to constrain COVID-19 [70]. Furthermore, a study has revealed that air pollution in Beijing and its neighboring provinces recovered more slowly as a consequence of the extension of the lockdown [86]. This also validates our finding that NO 2 concentrations in Beijing rebounded more slowly.
With regard to SO 2 , the variation trend of monthly average concentration in 2020 was generally similar to that of NO 2 in Beijing and Henan. Overall, the monthly average concentration of SO 2 in 2021 showed a downward trend compared to 2020 in Beijing. For Henan, the monthly average concentration deteriorated in January and February of 2021 relative to 2020; the monthly concentration between March and December in 2021 presented a decreasing trend relative to the corresponding months in 2020. As for CO, the monthly concentration in  (Table 1). In Beijing, the days with severe pollution accounted for 1.12% in 2017. It is worth noting that there have been no severe pollution days in Beijing since 2018. Moreover, the percentage of days with excellent levels increased significantly, with 50.3% (2019), 57.8% (2020), and 69.4% (2021) in the last three years, respectively. However, heavy pollution episodes still occurred during 2019-2021, although their proportion was significantly lower than in 2017 (4.5%) and 2018 (4.2%). In Henan, severe pollution episodes have been eliminated since 2019. The proportion of days with excellent levels increased significantly, while the proportion of good levels showed a decreasing trend. The days with good levels gradually changed to excellent days. Like Beijing, Henan persistently suffered from heavy PM 2.5 pollution from 2017 to 2021, with a lower proportion in 2020 (2.2%) and 2021 (1.9%). In comparison to gaseous precursors, the composition of PM 2.5 is more complicated. As mentioned above, PM 2.5 pollution has been mitigated under a set of policies and control measures in recent years, while haze still occurred on the NCP. Notably, PM 2.5 pollution was observed in January and February of 2020, when anthropogenic emissions decreased significantly due to the outbreak of COVID-19 and the Spring Festival [66,71]. Therefore, this study next focused on the period of pre-to post-COVID-19 (2019-2021) and analyzed the drivers for the PM 2.5 pollution episodes on the northern and southern edges of the NCP in the past three years.  (Table 2). Notably, HPE6 lasted for eight days, followed by HPE2 (seven days), and the remaining pollution incidents lasted for five days. The evolution of PM 2.5 concentration, wind direction (WD), wind speed (WS), BLH, and RH during six pollution episodes is shown in Figure 4. Generally, the HPEs were initially caused by southerly or southeasterly winds, which transported air pollutants from southern regions to Beijing [87][88][89]. Hence, PM 2.5 began accumulating in Beijing, and PM 2.5 concentration gradually increased. Subsequently, the BLH dropped significantly compared to the clean period, and the RH increased simultaneously, above 60%. Notably, the highest RH exceeded 80%. The WS was low, and the average WS was less than 2 m s −1 . Therefore, PM 2.5 pollution further deteriorated due to the stagnant meteorological conditions [66,70,90]. Moreover, Zhang et al. [91] found that the interaction between the continuous accumulation of PM 2.5 and further deterioration of meteorological conditions resulted in the explosive rise of PM 2.5 in the middle period of HPEs. It is worth noting that the hourly mean of PM 2.5 concentrations during HPE6 peaked among six pollution episodes, increasing sharply from 110 µg m −3 on the evening of March 14 to exceeding 600 µg m −3 on the morning of March 15. The explosive increase in PM 2.5 could be explained by the regional transport under favorable meteorology with high BLH and wind and low humidity. This differs from the previous studies that the twoway feedback between unfavorable meteorology and accumulated PM 2.5 is the dominant mechanism for the occurrence of explosive increase [91].
Furthermore, the variations in gaseous pollutants were investigated during the six HPEs ( Figure 5). The gaseous pollutants (excluding O 3 ) were generally characterized by remarkably elevated concentrations. It was observed that the SO 2 and CO concentrations during the six HPEs increased by 6.44-56.55% and 32.89-114.32%, respectively, compared with the corresponding monthly average concentrations. For NO 2 , except for HPE3, the concentrations increased by 18.59-81.08% during the pollution episodes. In contrast, the NO 2 concentration decreased by 16.38% during the HPE3, mainly due to the drastic reduction in traffic caused by the COVID-19 lockdown. Gao et al. [83] revealed that diffusive sources of NO 2 were mainly related to vehicular traffic. As for O 3 , the concentrations decreased by 3.08-51.98% during the period of HPE1, HPE4, HPE 5, and HPE6, whereas the concentrations increased by 6.61% and 46.19%, respectively, in the HPE2 and HPE3. A significant increase in O 3 concentration in the HPE3 was associated with the reduced NO 2 that hindered the reaction between NO and O 3 , resulting in the increased atmospheric oxidizing capacity [92][93][94]. A similar phenomenon was also observed in an air pollution episode in Shijiazhuang during the COVID-19 outbreak [66]. Several studies also revealed that the drop in NO x could not avoid PM 2.5 pollution during the COVID-19 outbreak [18,95]. The variations in gaseous precursors indicated that the secondary formation was a critical process during pollution events.  It is worth noting that the hourly mean of PM2.5 concentrations during HPE6 peaked among six pollution episodes, increasing sharply from 110 μg m −3 on the evening of March 14 to exceeding 600 μg m −3 on the morning of March 15. The explosive increase in PM2.5 could be explained by the regional transport under favorable meteorology with high BLH and wind and low humidity. This differs from the previous studies that the two-way feedback between unfavorable meteorology and accumulated PM2.5 is the dominant mechanism for the occurrence of explosive increase [91].
Furthermore, the variations in gaseous pollutants were investigated during the six significant increase in O3 concentration in the HPE3 was associated with the reduced NO2 that hindered the reaction between NO and O3, resulting in the increased atmospheric oxidizing capacity [92][93][94]. A similar phenomenon was also observed in an air pollution episode in Shijiazhuang during the COVID-19 outbreak [66]. Several studies also revealed that the drop in NOx could not avoid PM2.5 pollution during the COVID-19 outbreak [18,95]. The variations in gaseous precursors indicated that the secondary formation was a critical process during pollution events.  The ratio of PM 2.5 /CO is an indicator of secondary pollutants to primary emissions [65]. Specifically, the ratio of PM 2.5 /CO during six HPEs increased by 46.96%, 68.09%, 88.37%, 53.22%, 35.13%, and 42.46%, respectively, compared with the corresponding monthly average PM 2.5 /CO ratio. These pronounced increases in PM 2.5 /CO ratio further revealed the great contribution of secondary formation to heavy PM 2.5 pollution in Beijing. Consistently, studies have revealed that the persistent severe haze pollution in BTH and eastern China was associated with stronger secondary formation under a stagnant atmosphere despite the great reductions in primary emissions during the COVID-19 lockdown [65,96,97].
As illustrated in Figure 6, the potential source regions of PM 2.5 during six PM 2.5 pollution episodes were investigated by applying the WPSCF model. The relatively small areas with high WPSCF values were observed in HPE1, HPE3, and HPE5, indicating the crucial influence from surrounding areas, including western and southern Hebei, as well as Tianjin. In comparison, the wide banded areas with high WPSCF values were observed during the period of HPE2, HPE4, and HPE6, especially during HPE6, and the potential source regions were mainly in the southern Hebei, which indicated the industrial emissions contributed to this HPE because Hebei is the largest industrial cluster in China [98]. The potential source regions of PM 2.5 for the HPE2, HPE4, and HPE6 revealed that long-distance regional transport played a significant role in causing PM 2.5 pollution. Hence, more strengthened regional collaborative air pollution control in Beijing and the surrounding provinces is suggested. Overall, regional transport, secondary formation, along with adverse meteorology work jointly to the formation and evolution of heavy PM 2.5 pollution in Beijing, which is a complex process.
88.37%, 53.22%, 35.13%, and 42.46%, respectively, compared with the corresponding monthly average PM2.5/CO ratio. These pronounced increases in PM2.5/CO ratio further revealed the great contribution of secondary formation to heavy PM2.5 pollution in Beijing. Consistently, studies have revealed that the persistent severe haze pollution in BTH and eastern China was associated with stronger secondary formation under a stagnant atmosphere despite the great reductions in primary emissions during the COVID-19 lockdown [65,96,97].
As illustrated in Figure 6, the potential source regions of PM2.5 during six PM2.5 pollution episodes were investigated by applying the WPSCF model. The relatively small areas with high WPSCF values were observed in HPE1, HPE3, and HPE5, indicating the crucial influence from surrounding areas, including western and southern Hebei, as well as Tianjin. In comparison, the wide banded areas with high WPSCF values were observed during the period of HPE2, HPE4, and HPE6, especially during HPE6, and the potential source regions were mainly in the southern Hebei, which indicated the industrial emissions contributed to this HPE because Hebei is the largest industrial cluster in China [98]. The potential source regions of PM2.5 for the HPE2, HPE4, and HPE6 revealed that longdistance regional transport played a significant role in causing PM2.5 pollution. Hence, more strengthened regional collaborative air pollution control in Beijing and the surrounding provinces is suggested. Overall, regional transport, secondary formation, along with adverse meteorology work jointly to the formation and evolution of heavy PM2.5 pollution in Beijing, which is a complex process.

Analysis of PM2.5 Episodes on the Southern Edge of the NCP
For Henan, the formation mechanism of PM2.5 pollution between 2019 and 2021 was also explored with a detailed analysis of six HPEs (Table 3 and Figure 7). In contrast with Beijing, the duration of PM2.5 pollution in Henan was longer. Specifically, HPE1 lasted twenty-four days, including six consecutive days with PM2.5 levels reaching heavy pollution (over 150 μg m −3 ). In 2020, from January to early February, there were three persistent

Analysis of PM 2.5 Episodes on the Southern Edge of the NCP
For Henan, the formation mechanism of PM 2.5 pollution between 2019 and 2021 was also explored with a detailed analysis of six HPEs (Table 3 and Figure 7). In contrast with Beijing, the duration of PM 2.5 pollution in Henan was longer. Specifically, HPE1 lasted twenty-four days, including six consecutive days with PM 2.5 levels reaching heavy pollution (over 150 µg m −3 ). In 2020, from January to early February, there were three persistent PM 2.5 pollution events for 29 days. Moreover, from November to December of 2020, a seventeen-day PM 2.5 pollution incident occurred. Except for HPE4, the average WS during pollution episodes was below 2 m s −1 with predominantly northerly winds, which led to the accumulation of air pollutants in Henan from the northern areas with high emission loads along the weak northerly winds. Furthermore, the average values of RH during six HPEs were 75%, and the BLH also decreased remarkably; thereby, PM 2.5 accumulated in Henan under constant high humidity and stable atmospheric conditions, resulting in persistent PM 2.5 pollution. In addition to unfavorable meteorological conditions, the socio-economic factors, such as economic development level, industrial structure, energy consumption, scientific level, and land-use structure contributed to the longer duration of PM 2.5 pollution in Henan compared to Beijing [99,100], which was also indicated in Section 2.1. Jiang et al. [101] suggested that the central region, including Henan, should endeavor to finish industrial restructuring and remove the backward production capacity. PM2.5 pollution events for 29 days. Moreover, from November to December of 2020, a seventeen-day PM2.5 pollution incident occurred. Except for HPE4, the average WS during pollution episodes was below 2 m s −1 with predominantly northerly winds, which led to the accumulation of air pollutants in Henan from the northern areas with high emission loads along the weak northerly winds. Furthermore, the average values of RH during six HPEs were 75%, and the BLH also decreased remarkably; thereby, PM2.5 accumulated in Henan under constant high humidity and stable atmospheric conditions, resulting in persistent PM2.5 pollution. In addition to unfavorable meteorological conditions, the socioeconomic factors, such as economic development level, industrial structure, energy consumption, scientific level, and land-use structure contributed to the longer duration of PM2.5 pollution in Henan compared to Beijing [99,100], which was also indicated in Section 2.1. Jiang et al. [101] suggested that the central region, including Henan, should endeavor to finish industrial restructuring and remove the backward production capacity.  The variations in gaseous precursors are present in Figure 8. The CO concentrations during the six HPEs increased by 4. 26   Further, the ratio of PM 2.5 /CO during the six HPEs increased by 17.21%, 4.55%, 17.27%, 10.71%, 25.86%, and 35.44%, respectively, compared with the corresponding monthly ratio. Unlike Beijing, the impact of secondary formation on PM 2.5 pollution in Henan was relatively small. This is consistent with the findings by Huang et al. [65], revealing the relatively higher secondary PM 2.5 production in the BTH region. Some studies have revealed the dominant proportions of primary emission to the PM 2.5 pollution in southern parts of the NCP compared to the secondary formation [73], which was consistent with our results.
The WPSCF map for the six PM 2.5 pollution episodes in Henan showed that the potential source regions, except for Henan itself, were mainly in the areas adjacent to Henan, including southern Hebei, southern Shanxi, and western Shandong (Figure 9). This indicated that PM 2.5 pollution sources were featured by local dispersion and small-scale regional transport rather than long-distance regional transport, and similar conclusions were also found by Song et al. [73]. Overall, PM 2.5 pollution was dominated by local accumulation with a higher proportion of primary emissions under the persistent high humidity and stable atmospheric conditions, while superimposed regional transport on a small scale.

Conclusions
This study evaluated the change trends of PM2.5 and gaseous pollutants from 2017 to 2021 in Beijing and Henan and further explored the drivers for the PM2.

Conclusions
This study evaluated the change trends of PM 2.5 and gaseous pollutants from 2017 to 2021 in Beijing and Henan and further explored the drivers for the PM 2.5 pollution episodes pre-to post-COVID-19 (2019-2021) based on the air pollutant observation dataset. The main findings are as follows: (1) The annual concentrations of PM 2.5 , NO 2 , SO 2 , and CO decreased year by year during 2017-2021 under a series of clean air action plans, whereas the exception was NO 2 in Beijing in 2021, which increased slightly by 3.6% relative to 2020. During 2017-2021, the concentrations of PM 2.5 , SO 2 , and CO in Henan were higher than in Beijing. In contrast, NO 2 concentration was the opposite, except in 2020. The differences in pollutant levels between Beijing and Henan are related to pollution emissions, development levels, and other socio-economic indicators. Unlike other gaseous pollutants, the 90th percentiles of MDA8 O 3 values began to decrease significantly in 2020 in Beijing and Henan.
(2) The lockdown measures to constrain COVID-19 significantly improved air quality, and the concentrations of PM 2.5 , NO 2 , SO 2 , and CO decreased sharply in February 2020. The exceptions were PM 2.5 and CO in Beijing, which exhibited a delayed decrease in March caused by adverse meteorological conditions and transported pollutants emitted by non-stop industries and fireworks and reached the lowest values relative to March of 2017-2019 and 2021. (3) Overall, the PM 2.5 pollution conditions have improved significantly. However, Beijing and Henan still suffered from heavy PM 2.5 pollution between 2019 and 2021. For Beijing, the formation and evolution of PM 2.5 pollution were caused by initial regional transport and following secondary formation under adverse meteorology. Unlike Beijing, PM 2.5 elevation in Henan was caused by local accumulation with a dominated proportion of primary emissions under adverse atmospheric conditions, superimposing regional transport on a small scale. Hence, the heavy PM 2.5 pollution on the NCP was highly heterogeneous, and stagnant weather, such as low wind speed, shallow boundary layer, and high humidity, is one of the major drivers of heavy PM 2.5 pollution on the NCP. (4) The formation and evolution of elevated PM 2.5 pollution are affected by multiple factors. A balanced and coordinated strategy in regulating various air pollutants, the critical role of meteorology, and strengthened regional collaborative air pollution control should be considered when setting mitigation measures on the NCP.
Understanding the contributions of different sources to PM 2.5 pollution and identifying the dominant emission sources are essential while formulating an effective air pollution strategy. However, the present study has some limitations. The contributions of sources to PM 2.5 pollution were not examined in detail in the present study, especially for the period of heavy PM 2.5 pollution. Therefore, subsequent studies shall perform source apportionments of PM 2.5 to reveal the emission sources with more detailed information.

Conflicts of Interest:
The authors declare no conflict of interest.