The Different Impacts of Emissions and Meteorology on PM 2.5 Changes in Various Regions in China: A Case Study

: Emissions and meteorology are signiﬁcant factors affecting aerosol pollution, but it is not sufﬁcient to understand their relative contributions to aerosol pollution changes. In this study, the observational data and the chemical model (GRAPES_CUACE) are combined to estimate the drivers of PM 2.5 changes in various regions (the Beijing–Tianjin–Hebei (BTH), the Central China (CC), the Yangtze River Delta (YRD), and the Pearl River Delta (PRD)) between the ﬁrst month after COVID-19 (FMC_2020) (i.e., from 23 January to 23 February 2020) and the corresponding period in 2019 (FMC_2019). The results show that PM 2.5 mass concentration increased by 26% (from 61 to 77 µ g m − 3 ) in the BTH, while it decreased by 26% (from 94 to 70 µ g m − 3 ) in the CC, 29% (from 52 to 37 µ g m − 3 ) in the YRD, and 32% (from 34 to 23 µ g m − 3 ) in the PRD in FMC_2020 comparing with FMC_2019, respectively. In the BTH, although emissions reductions partly improved PM 2.5 pollution ( − 5%, i.e., PM 2.5 mass concentration decreased by 5% due to emissions) in FMC_2020 compared with that of FMC_2019, the total increase in PM 2.5 mass concentration was dominated by more unfavorable meteorological conditions (+31%, i.e., PM 2.5 mass concentration increased by 31% due to meteorology). In the CC and the YRD, emissions reductions ( − 33 and − 36%) played a dominating role in the total decrease in PM 2.5 in FMC_2020, while the changed meteorological conditions partly worsened PM 2.5 pollution (+7 and +7%). In the PRD, emissions reductions ( − 23%) and more favorable meteorological conditions ( − 9%) led to a total decrease in PM 2.5 mass concentration. This study reminds us that the uncertainties of relative contributions of meteorological conditions and emissions on PM 2.5 changes in various regions are large, which is conducive to policymaking scientiﬁcally in China.

In addition to emissions, meteorological conditions also have important impacts on PM 2.5 concentration by changing the ventilation rate, dry/wet deposition, chemical conversion loss rate, etc. [21][22][23]. Zonal westerly airflow and high-pressure ridge are two major background circulations that cause a reduction in the height of the planetary boundary layer (PBLH), affecting the formation of aerosol pollution in the Beijing-Tianjin-Hebei (BTH) [24]. Under normal circumstances, the appearance of weaker wind speed (WS), higher relative humidity (RH), and temperature inversion can be conducive to the accumulation of aerosol pollution in Beijing, Nanjing, Guangzhou, Sichuan Basin, etc. [25][26][27][28]. Some studies show that increasing wind speed will enhance PM 2.5 mass concentration by a transmission and convergence process [29,30]. It should also be noted that when the PM 2.5 accumulates to a certain level, it will further worsen the meteorological conditions [31,32]. A significant "two-way feedback" effect between PM 2.5 and unfavorable meteorological conditions for aerosol pollution diffusion is determined by studying the formation of HPEs in multiple cities in China [33][34][35].
Emissions and meteorological conditions will both affect the changes in PM 2.5 mass concentration, but the question of which parameter is more important is still the focus of research [21,[36][37][38][39]. In the past few years, the Chinese government has implemented a series of air pollution control measures, such as the "Action Plan on Prevention and Control of Air Pollution", the "Three-year Action Plan for Blue Skies", etc. [40][41][42][43]. Under the influence of control measures, the PM 2.5 pollution has been greatly improved. Most studies show that emissions reductions play a dominant role (about 70-80%) in the improvement of air quality from 2013 to 2017 [21,38,39]. However, some studies also pointed out that the contributions of changed meteorological conditions to the improvement of PM 2.5 reach about 50% in winter [44,45]. Moreover, even in the background of continuous emissions reductions, HPEs still occur in many cities in China due to unfavorable meteorological conditions for aerosol pollution diffusion and secondary formation [46][47][48]. All of these fully illustrate the complicated non-linear relationship between PM 2.5 concentration, meteorological conditions, and pollutants emissions. During the COVID-19 outbreak period, various anthropogenic emissions showed a significant decrease [49][50][51][52], e.g., large reductions in NO 2 were caused by the significantly reduced traffic emissions [52,53]. This special lockdown situation provided a better opportunity to study the relationship between PM 2.5 and emissions and meteorological conditions in China.
Although many studies have carried out some work about the reasons for changes in emissions, major air pollutants concentration, and aerosol composition during the COVID-19 outbreak in China [53][54][55][56][57], they are only for a certain element or region. The comprehensive study of the causes of PM 2.5 changes during the COVID-19 outbreak is still poor, especially the comparisons in various regions. In this study, air pollutants data, meteorological data, and the GRAPES_CUACE model are used to investigate reasons for PM 2.5 mass concentration changes during the first month (from 23 January to 23 February 2020) after COVID-19 (FMC_2020) compared with the corresponding historical period in 2019 (FMC_2019) in China. The discussions help further understand the causes of aerosol pollution changes in various regions and provide scientific technique support for regional aerosol pollution control in China.

Air Pollutants Data
Hourly major air pollutants (PM 2.5 , CO, NO 2 , and SO 2 ) mass concentration in FMC_2019 and FMC_2020 are collected by an automated ambient air quality monitoring system (an air quality monitoring sub-station, a quality assurance laboratory, etc.) from the Ministry of Ecology and Environment of China. This system can automatically monitor, collect, process, and store PM 2.5 and gaseous pollutants data, then transmit data to the central computer, and finally collate data by quality assurance process (invalid data should have raw records). The mass concentration units are µg m −3 for PM 2.5 , NO 2 , SO 2 , and mg m −3 for CO. For the missing data, if the sample size is sufficient, we choose to directly remove the missing data. If the sample size is insufficient, we use adjacent data to supplement [58].

Meteorological Data
Meteorological data come from National Centers for Environmental Prediction (NCEP) Final (FNL) analysis data. This product is conducted by the Global Data Assimilation System (GDAS), which continuously collects observational data from the Global Telecommunications System (GTS), and other sources. The FNL analysis data are made with the same model which NCEP uses in the Global Forecast System (GFS), but the FNL analysis data are delayed so that more observational data (e.g., satellite data and radar data) can be used. We use the FNL analysis data in FMC_2019 and FMC_2020 with the resolution of 0.25 • × 0.25 • in this paper. These data mainly include geopotential height, sea level pressure, PBLH, temperature (T), RH, and WS.

Study Regions
This study focuses on the comparative analysis of the four major megacity clusters (the BTH, the Yangtze River Delta (YRD), the Pearl River Delta (PRD), and the Central China (CC)) ( Figure 1). The aerosol pollution in the BTH ( itor, collect, process, and store PM2.5 and gaseous pollutants data, then transmit data to the central computer, and finally collate data by quality assurance process (invalid data should have raw records). The mass concentration units are µg m −3 for PM2.5, NO2, SO2, and mg m −3 for CO. For the missing data, if the sample size is sufficient, we choose to directly remove the missing data. If the sample size is insufficient, we use adjacent data to supplement [58].

Meteorological Data
Meteorological data come from National Centers for Environmental Prediction (NCEP) Final (FNL) analysis data. This product is conducted by the Global Data Assimilation System (GDAS), which continuously collects observational data from the Global Telecommunications System (GTS), and other sources. The FNL analysis data are made with the same model which NCEP uses in the Global Forecast System (GFS), but the FNL analysis data are delayed so that more observational data (e.g., satellite data and radar data) can be used. We use the FNL analysis data in FMC_2019 and FMC_2020 with the resolution of 0.25° × 0.25° in this paper. These data mainly include geopotential height, sea level pressure, PBLH, temperature (T), RH, and WS.

Study Regions
This study focuses on the comparative analysis of the four major megacity clusters (the BTH, the Yangtze River Delta (YRD), the Pearl River Delta (PRD), and the Central China (CC)) (

Statistical Analysis
In a certain period, primary PM 2.5 and gas pollutants generally have the same anthropogenic emissions, and the contribution ratio of emissions to them is relatively stable, so gas pollutants can be used to estimate impacts of emissions changes on PM 2.5 [59][60][61]. For the primary source of PM 2.5 , the influence of combustion sources is mainly studied. Among the main gas pollutants, CO is selected as an indicator of the primary com- bustion sources [62,63], because CO has a long lifetime and is less affected by chemical reactions [63][64][65][66]. In addition, we use a linear relationship to fit CO and PM 2.5 mass concentration with a determination coefficient (r 2 ). The r 2 reflects how much the changes of CO mass concentration can account for PM 2.5 changes, as shown in Equation (1): where i is the ith sample, . y i and y i represent the PM 2.5 mass concentration calculated by the linear fitting equation and the actual PM 2.5 mass concentration, respectively, and y is the mean PM 2.5 mass concentration.
Here we assume that the changes of air pollutants are only affected by emissions and meteorological conditions, which is a general method in many studies. Therefore, the changes in CO and PM 2.5 mass concentration under similar meteorological conditions are used to represent the contributions of emissions.
The assimilation system of GRAPES_CUACE model is formed by the Three-Dimensional Variational Assimilation (3DVAR) and the Four-Dimensional Variational Assimilation (4DVAR), which is optional. In this study, we use 3DVAR to perform data assimilation [71]. The physical and chemical schemes selected in this model are the same as those in Zhang et al. [76]. These schemes are listed in Table 1. There are 7 species of aerosol that can be modeled, including sulfates, soil dust, black carbon, organic carbon, sea salts, nitrates, and ammonium salts. The initial and boundary conditions are provided by FNL reanalysis data (0.25 • × 0.25 • ), and the emissions used in the model are Multi-resolution Emission Inventory for China (MEIC) of Tsinghua University in 2017 [7,59]. The anthropogenic emissions used in the model are Multi-resolution Emission Inventory for China (MEIC) of Tsinghua University in 2017. MEIC covers more than 700 anthropogenic emissions in the whole of China, which has 10 major atmospheric pollutants and carbon dioxide (SO 2 , NO x , CO, NMVOC, NH 3 , PM 2.5 , PM 10 , BC and OC, CO 2 ). The natural emissions do not enter into the current model. Usually, the contributions of anthropogenic emissions to aerosol pollution are much greater than those of natural emissions. In the future, we will consider natural emissions in the model to simulate air pollution accurately. The simulated area includes the entire Chinese region and the two simulated periods are FMC_2019 and FMC_2020. The model has a horizontal resolution of 0.15 • × 0.15 • and a vertical resolution of 31 levels (from 1000 hPa to 1 hPa).
It is worth noting that we cannot evaluate the simulations in this study because the emissions and the simulated period are not in the same years. However, the model simulations by using the emissions and the meteorology in the same year have been widely validated in the paper of Zhang et al. [76], which indicates that the GRAPES_CUACE model can be used to simulate changes in meteorology and PM 2.5 and the results are credible. The descriptions of sensitive experiments are shown in Table 2 and the difference between the two experiments (EXP1 and EXP2) can determine the impact of meteorological conditions on PM 2.5 changes.

Impacts of Anthropogenic Emissions on PM2.5 Mass Concentration
Changes from FMC_2019 to FMC_2020 Figure 3a,b is the spatial distribution of CO in China. Compared with FMC_2019, CO mass concentration has a small decrease in FMC_2020. The overall CO mass concentration in the CC, the YRD, and the PRD has decreased by 14, 12, and 21%, but the CO mass concentration in the BTH has increased by 10% ( Table 3). The changes in CO mass concentration in these four regions are mainly affected by primary sources and meteorological con-

Impacts of Anthropogenic Emissions on PM 2.5 Mass Concentration
Changes from FMC_2019 to FMC_2020 Figure 3a,b is the spatial distribution of CO in China. Compared with FMC_2019, CO mass concentration has a small decrease in FMC_2020. The overall CO mass concentration in the CC, the YRD, and the PRD has decreased by 14, 12, and 21%, but the CO mass concentration in the BTH has increased by 10% ( Table 3). The changes in CO mass concentration in these four regions are mainly affected by primary sources and meteorological conditions. To eliminate the influence of meteorological conditions and the possibility of transportation [85], we select periods for analysis that are associated with stable and similar meteorological conditions between FMC_2019 and FMC_2020 in four regions as much as possible (Table 4). For example, the meteorological conditions affecting BTH during 28-29 January and 1-2 February 2019 are similar to those during 9-11 February and 19-20 February 2020. The difference of mean atmospheric compositions between the periods in FMC_2019 and FMC_2020 can indicate the net effectiveness of large-scale lockdown measures after the COVID-19 outbreak. The same methods are also used in the CC, the YRD, and the PRD. The long-range transport of CO can be excluded in Figures S3-S6. The comparison results show that CO mass concentration from primary sources decreases by about 4, 18, 15, and 10% in the BTH, the CC, the YRD, and the PRD in FMC_2020 compared with FMC_2019 (Table 4), which means that the combustion sources reductions are relatively smaller in the BTH and larger in the CC than other regions.    In addition, Figure 4 shows the linear correlation between CO and PM 2.5 . CO and PM 2.5 in the BTH, the CC, and the YRD have a strong positive correlation (r 2 > 0.6), while the correlation in the PRD is weak (r 2 < 0.3). All of these passed the 0.05 significance test. These show that PM 2.5 and CO in the BTH, the CC, and the YRD have similar sources, which means combustion sources have large contributions to PM 2.5 . PM 2.5 in the PRD may be mainly affected by non-combustion sources, secondary aerosol, etc. Through calculations by the linear regression equations with different r 2 (Figure 4), compared with FMC_2019, the primary PM 2.5 from combustion sources increases by about 15% in the BTH, 18% in the CC, and decreases by 38% in the YRD, 83% in the PRD, which is consistent with previous studies about primary PM 2.5 [51,56]. Furthermore, the correlations analysis also finds that the increase in primary PM 2.5 from combustion sources in the BTH is closely related to the increase in the contribution of coal combustion (r 2 between SO 2 and PM 2.5 increases from 0.22 in FMC_2019 to 0.46 in FMC_2020). In the CC, the increase in the primary PM 2.5 from combustion sources is mainly related to NOx emissions (r 2 between NO 2 and PM 2.5 increases from 0.14 in FMC_2019 to 0.42 in FMC_2020), but it has been confirmed that the total primary emissions of PM 2.5 decrease (Table 4), indicating a more significant decrease in PM 2.5 from non-combustion sources. The lower contributions from combustion sources in the YRD and the PRD are mainly related to the decreased contributions of SO 2 related emissions (r 2 between SO 2 and PM 2.5 decreases from 0.44 (0.64) in FMC_2019 to 0.17 (0.18) in FMC_2020 in YRD (PRD)).  In China, nitrate and sulfate are the most important components of secondary particulate matter in PM2.5, which is partly affected by NOx and SO2 in the air [86,87]. Figure  3c-f shows the spatial distribution of NO2 mass concentration and SO2 mass concentration in China in FMC_2019 and FMC_2020. It can be seen that compared with FMC_2019, NO2 mass concentration and SO2 mass concentration in China in FMC_2020 decrease significantly. The overall average NO2 mass concentration in the BTH, the CC, the YRD, and the PRD regions decrease by 23, 47, 47, and 42%, and the SO2 mass concentration decrease by 19, 31, 22, and 19%, respectively (Table 3). Correspondingly, the primary emissions of NOx and SO2 in the four regions in FMC_2020 also show a downward trend, which is also confirmed by other studies [54]. The ratio of SO2/NO2 is one indicator of air pollution sources from mobile sources and stationary sources [88,89]. Figure 5 is the values of SO2/NO2. The highest value of SO2/NO2 is in the BTH, followed by the CC, the PRD, and the YRD, which shows that industrial emissions and coal combustion have a more significant impact on air pollution in the BTH in winter. At the same time, compared with two years, the value of SO2/NO2 in the BTH has changed a little, indicating that the impacts of mobile sources and stationary sources on air pollution are similar between FMC_2019 and FMC_2020. In the other three regions, the values of SO2/NO2 all increase in FMC_2020, indicating that air pollution is mainly influenced by stationary sources. The value of PM2.5/CO is also often used to measure the impacts of aerosol secondary formation processes on PM2.5 [58,66]. Therefore, values of PM2.5/CO are calculated in the BTH, the CC, the YRD, and the PRD in FMC_2019 (0.049, 0.074, 0.053, and 0.037) and FMC_2020 (0.054, 0.062, 0.046, and 0.031). The degree of correlation between CO and PM2.5 is not very high (r 2 < 0.3) in the PRD, so aerosol secondary formations are not well determined by the ratio and are, therefore, not very meaningful. The secondary aerosol formations in the BTH and the CC show higher contributions to PM2.5 than those in the PRD and the YRD. Compared with FMC_2019, the weakening secondary aerosol formation processes in the air in FMC_2020 are mainly related to the substantial decrease in NOx and SO2. It is worth noting that although primary emissions of NOx and SO2 in the In China, nitrate and sulfate are the most important components of secondary particulate matter in PM 2.5 , which is partly affected by NOx and SO 2 in the air [86,87]. Figure 3c-f shows the spatial distribution of NO 2 mass concentration and SO 2 mass concentration in China in FMC_2019 and FMC_2020. It can be seen that compared with FMC_2019, NO 2 mass concentration and SO 2 mass concentration in China in FMC_2020 decrease significantly. The overall average NO 2 mass concentration in the BTH, the CC, the YRD, and the PRD regions decrease by 23, 47, 47, and 42%, and the SO 2 mass concentration decrease by 19, 31, 22, and 19%, respectively (Table 3). Correspondingly, the primary emissions of NOx and SO 2 in the four regions in FMC_2020 also show a downward trend, which is also confirmed by other studies [54]. The ratio of SO 2 /NO 2 is one indicator of air pollution sources from mobile sources and stationary sources [88,89]. Figure 5 is the values of SO 2 /NO 2 . The highest value of SO 2 /NO 2 is in the BTH, followed by the CC, the PRD, and the YRD, which shows that industrial emissions and coal combustion have a more significant impact on air pollution in the BTH in winter. At the same time, compared with two years, the value of SO 2 /NO 2 in the BTH has changed a little, indicating that the impacts of mobile sources and stationary sources on air pollution are similar between FMC_2019 and FMC_2020. In the other three regions, the values of SO 2 /NO 2 all increase in FMC_2020, indicating that air pollution is mainly influenced by stationary sources. The value of PM 2.5 /CO is also often used to measure the impacts of aerosol secondary formation processes on PM 2.5 [58,66]. Therefore, values of PM 2.5 /CO are calculated in the BTH, the CC, the YRD, and the PRD in FMC_2019 (0.049, 0.074, 0.053, and 0.037) and FMC_2020 (0.054, 0.062, 0.046, and 0.031). The degree of correlation between CO and PM 2.5 is not very high (r 2 < 0.3) in the PRD, so aerosol secondary formations are not well determined by the ratio and are, therefore, not very meaningful. The secondary aerosol formations in the BTH and the CC show higher contributions to PM 2.5 than those in the PRD and the YRD. Compared with FMC_2019, the weakening secondary aerosol formation processes in the air in FMC_2020 are mainly related to the substantial decrease in NOx and SO 2 . It is worth noting that although primary emissions of NOx and SO 2 in the BTH have also been reduced, the secondary contributions to PM 2.5 are strengthened, which is consistent with previous studies [19,90]. For example, although the NOx emissions have decreased, the secondary aerosol formation process is still strengthened under unfavorable meteorological conditions for aerosol pollution diffusion, mainly due to the increased particle acidity (pH), which offsets the emissions reductions [19,90]. From the perspective of the contributions of all emissions changes to PM2.5, we calculate the PM2.5 mass concentration under similar meteorological conditions (Table 4). Compared with FMC_2019, PM2.5 mass concentration decreases by 6, 25, 32, and 21% in the BTH, the CC, the YRD, and the PRD in FMC_2020. To a certain extent, there are still many uncertainties in the calculations. On the one hand, the impacts of meteorological conditions cannot be completely eliminated. On the other hand, the secondary formation processes of aerosol are greatly affected by meteorological conditions. For example, there are more secondary aerosols generated under unfavorable meteorological conditions for aerosol pollution diffusion [90]. All of these will lead to errors. However, there is a certain reference value when comparing the contributions of the two-year emissions changes to PM2.5. Further research is needed.

Effects of Changes in Meteorological Conditions on PM2.5 Mass Concentration from FMC_2019 to FMC_2020
Meteorological conditions also have a significant impact on affecting PM2.5 pollution [21,46]. Figure 6 shows the background circulation patterns in FMC_2019 and FMC_2020. In FMC_2019, there is a strong Siberian-Mongolian high pressure with a central value of more than 1040 hPa. Eastern China is affected by this high-pressure system, causing clean and cold air from the north. At the same time, the 500 hPa geopotential height lines are relatively dense and the development of a high-pressure ridge is weak, which leads to poor atmospheric stability. In this case, it is beneficial to the horizontal and vertical diffusion of air pollutants. In FMC_2020, the Siberian-Mongolian high pressure becomes weaker and the central value is about 1035 hPa. The sea-level pressure in the four regions in China decreases significantly and becomes uniform. However, in the PRD, the gradient of sea-level pressure becomes larger, which is beneficial to the diffusion of air pollutants. At 500 hPa, the geopotential height lines are less dense than those in FMC_2019, and the development of the high-pressure ridge is strengthened, which leads to relatively strong atmospheric stability. These unfavorable meteorological conditions are conducive to the accumulation of aerosol pollution.
The meridional circulation index (MCI) is used to represent the intensity of atmospheric circulation [91,92]. The higher value of MCI means that it is easier for cold air from the north to flow to the south, and the exchange of airflow between north and south is enhanced, which is conducive to the diffusion of aerosol pollution. In this paper, we  (Table 4). Compared with FMC_2019, PM 2.5 mass concentration decreases by 6, 25, 32, and 21% in the BTH, the CC, the YRD, and the PRD in FMC_2020. To a certain extent, there are still many uncertainties in the calculations. On the one hand, the impacts of meteorological conditions cannot be completely eliminated. On the other hand, the secondary formation processes of aerosol are greatly affected by meteorological conditions. For example, there are more secondary aerosols generated under unfavorable meteorological conditions for aerosol pollution diffusion [90]. All of these will lead to errors. However, there is a certain reference value when comparing the contributions of the two-year emissions changes to PM 2.5 . Further research is needed.

Effects of Changes in Meteorological Conditions on PM 2.5 Mass Concentration from FMC_2019 to FMC_2020
Meteorological conditions also have a significant impact on affecting PM 2.5 pollution [21,46]. Figure 6 shows the background circulation patterns in FMC_2019 and FMC_2020. In FMC_2019, there is a strong Siberian-Mongolian high pressure with a central value of more than 1040 hPa. Eastern China is affected by this high-pressure system, causing clean and cold air from the north. At the same time, the 500 hPa geopotential height lines are relatively dense and the development of a high-pressure ridge is weak, which leads to poor atmospheric stability. In this case, it is beneficial to the horizontal and vertical diffusion of air pollutants. In FMC_2020, the Siberian-Mongolian high pressure becomes weaker and the central value is about 1035 hPa. The sea-level pressure in the four regions in China decreases significantly and becomes uniform. However, in the PRD, the gradient of sea-level pressure becomes larger, which is beneficial to the diffusion of air pollutants. At 500 hPa, the geopotential height lines are less dense than those in FMC_2019, and the development of the high-pressure ridge is strengthened, which leads to relatively strong atmospheric stability. These unfavorable meteorological conditions are conducive to the accumulation of aerosol pollution. by 38% (from 21 to 13) and 29% (from 24 to 17), respectively. This shows that the meridional circulation is significantly weaker in FMC_2020 and impacts of clean and cold air from the north on PM2.5 are weakened, which is not conducive to the removal of aerosol pollution. The influence of the meridional circulation on the four regions of China is weakened one by one from the BTH to the PRD.  According to existing research, PBLH, temperature inversion, RH, and WS are the meteorological factors closely related to PM2.5 [28,93], among which the difference of temperature between 900 and 1000 hPa (T900−T1000) refers to the temperature inversion and atmospheric stability [55,93]. Figure 8 and Table 5 show the fractional changes of the PBLH, T900−T1000, RH at 1000 hPa (RH1000), and WS at 1000 hPa (WS1000) between FMC_2020 and FMC_2019. It can be seen that in the BTH in FMC_2020, the PBLH has decreased by 24% and the lower PBLH allows PM2.5 to mix in a smaller range; the The meridional circulation index (MCI) is used to represent the intensity of atmospheric circulation [91,92]. The higher value of MCI means that it is easier for cold air from the north to flow to the south, and the exchange of airflow between north and south is enhanced, which is conducive to the diffusion of aerosol pollution. In this paper, we choose two regions for calculating MCI (50-65 • N, 70-90 • E and 50-65 • N, 126-146 • E, white boxes in Figure 6), which is consistent with the previous research [93]. Figure 7 shows the values of MCI at 400 and 500 hPa in FMC_2019 and FMC_2020. It can be seen that compared with FMC_2019, the MCI at 400 and 500 hPa in FMC_2020 has decreased by 38% (from 21 to 13) and 29% (from 24 to 17), respectively. This shows that the meridional circulation is significantly weaker in FMC_2020 and impacts of clean and cold air from the north on PM 2.5 are weakened, which is not conducive to the removal of aerosol pollution. The influence of the meridional circulation on the four regions of China is weakened one by one from the BTH to the PRD. by 38% (from 21 to 13) and 29% (from 24 to 17), respectively. This shows that the meridional circulation is significantly weaker in FMC_2020 and impacts of clean and cold air from the north on PM2.5 are weakened, which is not conducive to the removal of aerosol pollution. The influence of the meridional circulation on the four regions of China is weakened one by one from the BTH to the PRD.  According to existing research, PBLH, temperature inversion, RH, and WS are the meteorological factors closely related to PM2.5 [28,93], among which the difference of temperature between 900 and 1000 hPa (T900−T1000) refers to the temperature inversion and atmospheric stability [55,93]. Figure 8 and Table 5 show the fractional changes of the PBLH, T900−T1000, RH at 1000 hPa (RH1000), and WS at 1000 hPa (WS1000) between FMC_2020 and FMC_2019. It can be seen that in the BTH in FMC_2020, the PBLH has decreased by 24% and the lower PBLH allows PM2.5 to mix in a smaller range; the According to existing research, PBLH, temperature inversion, RH, and WS are the meteorological factors closely related to PM 2.5 [28,93], among which the difference of temperature between 900 and 1000 hPa (T 900 −T 1000 ) refers to the temperature inversion and atmospheric stability [55,93]. Figure 8 and Table 5 show the fractional changes of the PBLH, T 900 −T 1000 , RH at 1000 hPa (RH 1000 ), and WS at 1000 hPa (WS 1000 ) between FMC_2020 and FMC_2019. It can be seen that in the BTH in FMC_2020, the PBLH has decreased by 24% and the lower PBLH allows PM 2.5 to mix in a smaller range; the T 900 −T 1000 has increased by 13% and the stronger atmospheric stability is not conducive to the vertical diffusion of PM 2.5 ; the RH increases by 53% and the hygroscopic growth of PM 2.5 is strengthened; the WS decrease by 4% and horizontal diffusion of PM 2.5 is weakened. All of these meteorological factors are beneficial to the accumulation of air pollutants in the BTH. However, the four meteorological factors have different impacts on PM 2.5 changes in the CC, the YRD, and the PRD. The ±10% difference is used to take as a threshold of significance [55], which means that when the absolute values of fractional changes in meteorological factors are more than 10%, the meteorological factors may have a significant impact on PM 2.5 mass concentration. In the CC, the absolute values of fractional changes of PBLH, RH, and WS are less than 10%, and only T 900 −T 1000 decreases by 22%. In the YRD, the absolute values of fractional changes of PBLH and RH are less than 10%, and T 900 −T 1000 and WS decrease 17 and 26%, respectively. In general, the comprehensive effects of these changed meteorological elements on PM 2.5 changes are small in the CC and the YRD. In the PRD, PBLH increases by 8%, T 900 −T 1000 decreases by 20%, WS increases by 18%, and RH increases a little (3%), all of which are beneficial to the diffusion of PM 2.5 .
Atmosphere 2022, 13,222 12 of 19 T900−T1000 has increased by 13% and the stronger atmospheric stability is not conducive to the vertical diffusion of PM2.5; the RH increases by 53% and the hygroscopic growth of PM2.5 is strengthened; the WS decrease by 4% and horizontal diffusion of PM2.5 is weakened. All of these meteorological factors are beneficial to the accumulation of air pollutants in the BTH. However, the four meteorological factors have different impacts on PM2.5 changes in the CC, the YRD, and the PRD. The ±10% difference is used to take as a threshold of significance [55], which means that when the absolute values of fractional changes in meteorological factors are more than 10%, the meteorological factors may have a significant impact on PM2.5 mass concentration. In the CC, the absolute values of fractional changes of PBLH, RH, and WS are less than 10%, and only T900−T1000 decreases by 22%. In the YRD, the absolute values of fractional changes of PBLH and RH are less than 10%, and T900−T1000 and WS decrease 17 and 26%, respectively. In general, the comprehensive effects of these changed meteorological elements on PM2.5 changes are small in the CC and the YRD. In the PRD, PBLH increases by 8%, T900−T1000 decreases by 20%, WS increases by 18%, and RH increases a little (3%), all of which are beneficial to the diffusion of PM2.5.
The contributions of changed meteorological conditions to PM 2.5 mass concentration can be derived from the difference between the two sensitive experiments. From Figure 9a-c, compared with FMC_2019, the overall meteorological conditions of FMC_2020 cause PM 2.5 mass concentration to increase by 35% (from 31 to 42 µg m −3 ) in the BTH, 5% (from 28 to 29.5 µg m −3 ) in the CC, and 6% (from 34 to 36 µg m −3 ) in the YRD, which indicates that the meteorological conditions in FMC_2020 are more conducive to the accumulation of aerosol pollution. Besides, PM 2.5 mass concentration decreases by 8% (from 12 to 11 µg m −3 ) in the PRD. This indicates that the meteorological conditions are more conducive to the diffusion of aerosol pollution.
The contributions of changed meteorological conditions to PM2.5 mass concentration can be derived from the difference between the two sensitive experiments. From Figure  9a-c, compared with FMC_2019, the overall meteorological conditions of FMC_2020 cause PM2.5 mass concentration to increase by 35% (from 31 to 42 µg m −3 ) in the BTH, 5% (from 28 to 29.5 µg m −3 ) in the CC, and 6% (from 34 to 36 µg m −3 ) in the YRD, which indicates that the meteorological conditions in FMC_2020 are more conducive to the accumulation of aerosol pollution. Besides, PM2.5 mass concentration decreases by 8% (from 12 to 11 µg m −3 ) in the PRD. This indicates that the meteorological conditions are more conducive to the diffusion of aerosol pollution.

Relative Contributions of Emissions and Meteorological Conditions to PM2.5 Changes
Assuming that the relationship between meteorology, emissions and PM2.5 is linear is a widely used method to calculate the relative contributions of meteorology and emis-

Relative Contributions of Emissions and Meteorological Conditions to PM 2.5 Changes
Assuming that the relationship between meteorology, emissions and PM 2.5 is linear is a widely used method to calculate the relative contributions of meteorology and emissions to PM 2.5 changes [94,95]. In Sections 3.2 and 3.3, the changes of PM 2.5 mass concentration caused by changed emissions and meteorological conditions are estimated, so the Equations (2) and (3) can be used to calculate the relative contributions of them to actual PM 2.5 mass concentration changes.
ReCON(emi) = AC(PM 2.5 ) × CON(emis) CON(met) + CON(emi) where ReCON(met) and ReCON(emi) represent the relative contributions of meteorological conditions and emissions to PM 2.5 mass concentration. AC(PM 2.5 ) represents the actual changes of PM 2.5 mass concentration. CON(met) and CON(emi) represent the contributions of changed meteorological conditions and emissions to PM 2.5 mass concentration changes. The calculated results are shown in Figure 9d. The changed meteorological conditions and emissions cause PM 2.5 mass concentration to increase by 31% and decrease by 5% in the BTH, respectively, which indicates that meteorological conditions dominate the increase in PM 2.5 mass concentration. In the CC and the YRD, the changed emission causes the PM 2.5 mass concentration to decrease by 33 and 36%, respectively, which is much greater than the 7 and 7% increase from meteorology, reflecting the primary effect of the emissions. In the PRD, emissions and meteorology have caused PM 2.5 mass concentration to decrease by 23 and 9%.

Conclusions
The relative contributions of meteorological conditions and emissions to PM 2.5 changes are hot issues of air quality research in China. This paper uses air pollutants data, meteorological data, and the GRAPES_CUACE model to analyze and compare the relative contributions of emissions and meteorology on PM 2.5 changes in four different regions (the BTH, the CC, the YRD, and the PRD).
The results show that compared with FMC_2019, the PM 2.5 mass concentration in FMC_2020 in four regions has changed significantly. Among them, PM 2.5 mass concentration increases by 26% in the BTH while it decreases by 26, 29, and 32% in the CC, the YRD, and the PRD, respectively. These changes are mainly caused by the combined effects of changed emissions and meteorological conditions.
In terms of emissions, compared with FMC_2019, CO emissions have the smallest decrease in the BTH (4%), followed by the PRD (10%), the YRD (15%), and the CC (18%) in FMC_2020 by calculating the difference of CO mass concentration during periods associated with similar stable weather conditions. Besides, through using the linear regression equations with different r 2 , the contributions of combustion sources to primary PM 2.5 in FMC_2020 increase by about 15% in the BTH and 18% in the CC, while decrease by 38% in the YRD and 83% in the PRD. The increased contributions in the BTH and the CC are caused by higher contributions of coal burning and emissions related to NO2, respectively. The decreased contributions of combustion sources in the YRD and the PRD are mainly related to the lower contributions of emissions related to SO 2 .
For meteorology, compared with FMC_2019, the Siberian-Mongolian high-pressure is weakened, the high-pressure ridge at 500 hPa is strengthened, and the geopotential height lines are less dense in FMC_2020, which is not conducive to the diffusion of PM 2.5 in the BTH, the CC, and the YRD. However, the gradient of sea level pressure becomes larger in the PRD region, which is conducive to the diffusion of air pollutants. The MCI changes show that the air ventilation between the north and the south of China is weakened, which is not conducive to the diffusion of aerosol pollution in four regions. Moreover, PBLH decreases by 24%, the T 900 −T 1000 increases by 13%, the RH increases by 53%, and the WS decreases by 4% in the BTH, which are all beneficial to the accumulation of aerosol pollution. However, meteorological factors have no significant changes in the CC and the YRD. In the PRD, meteorological factors are beneficial to the diffusion of aerosol pollution.
In general, in the BTH, the observed PM 2.5 mass concentration increase is mainly attributed to unfavorable meteorological conditions (+31%) which eliminate the −5% changes of PM 2.5 mass concentration caused by emissions reductions. In the CC and the YRD, the decrease in PM 2.5 mass concentration is dominated by emissions reductions (−33 and −36%), which are greater than the positive contributions of meteorological conditions (+7 and +7%). In the PRD, the decrease in PM 2.5 mass concentration is caused by emissions reductions (−23%) and favorable meteorological conditions (−9%). The changes in air pollutants in further pre-COVID years (Figures S1 and S2 from the Supplementary Materials) will continue to be discussed in future studies.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/atmos13020222/s1, Figure S1: Distributions of monthly average PM 2.5 mass concentration (µg m −3 ) in China; Figure S2: Distributions of monthly average CO mass concentration (mg m −3 ) in China; Figure S3: The values of CO/NO x during FMC_2019 and FMC_2020 in the BTH; Figure S4: As in Figure S3, but in the CC; Figure S5: As in Figure S3, but in the YRD; Figure S6: As in Figure S3, but in the PRD.