Contribution of Meteorological Conditions to the Variation in Winter PM 2.5 Concentrations from 2013 to 2019 in Middle-Eastern China

: Severe air pollution events accompanied by high PM 2.5 concentrations have been repeatedly observed in Middle-Eastern China since 2013 and decreased in recent years. The reason for this caused widespread attention. The month of January was selected to represent the winter season annual changes in the winter PM 2.5 and meteorological conditions—including the upper-air meridional circulation index (MCI), winds at 700 and 850 hPa levels and surface meteorology—from 2013 to 2019. These conditions were analyzed to study the contribution of meteorology changing to the annual PM 2.5 changing on the regional scale. Results show that, based on values of upper-level MCI, the years 2014, 2015, 2017, and 2019 were deﬁned as meteorology-haze years and the years 2016 and 2018 were deﬁned as meteorology-clean years. A change in meteorological conditions may lead to a 26% change in PM 2.5 concentration between 2014 and 2013 (two meteorology-haze years) and 16–20% changes in PM 2.5 concentration between meteorology-haze years and meteorology-clean years. Changes in pollutant emissions may cause 21–47% changes in PM 2.5 concentration between each two meteorology-haze years. A comparison of two meteorology-clean years and pollutant emissions in 2018 may be reduced by 40% compared with 2016. Overall, changes in emissions had a greater inﬂuence on changes in PM 2.5 compared with meteorological conditions.


Introduction
Fine particulate matter with an aerodynamic diameter <2.5 µm (PM 2.5 ) is considered to be a major air pollutant that causes severe environmental problems and affects human health [1][2][3][4]. Air pollution has been an environmental issue in China for many decades. The visibility-retrieved PM 2.5 concentrations as well as haze events experienced an overall increasing trend throughout China from 1957 to 2014, with fluctuations in different time periods [5][6][7]. PM 2.5 concentrations and fog and haze events all have significant seasonal variations. PM 2.5 concentrations tend to be highest in winter and lowest in summer. Low visibility and days with haze and fog, accompanied by high PM 2. 5 concentrations, mostly occur in the winter [8].
The region of Mid-eastern China (MEC, defined as 34-41.5 • N, 112-120 • E), which includes the Hebei province, the cities of Tianjin and Beijing, and the northern part of Henan and Shandong heavily polluted areas, such as the Yangtze River Delta, are far from the study area. In addition, considering the weak winds and relatively stable atmospheric conditions in the winter, the effect of pollutants transport between different regions on variations in regional average PM2.5 concentration in MEC should be small. On the city scale, pollutants transport outside the city is considered to be very important for PM2.5 concentration in some cities [20]. This study is focused on the influence of meteorological conditions including the upper-air circulation pattern, wind speed, temperature, and ambient relative humidity in a regional PM2.5 concentration.
The unique topography of MEC has contributed to this region becoming one of the most polluted areas of China (Figure 1b). The western part of MEC is influenced by dry, warm winds from the eastern Taihang Mountains, which results in increased stability at the surface. The east coast of MEC often has lower PM2.5 concentrations under the influence of moderate to strong winds from the Bohai and Yellow seas [12]. Figure 1b shows the locations of the 264 PM2.5 monitoring sites selected for this study. The average PM2.5 concentrations in January were calculated from 2013 to 2019 for MEC and some large polluted cities (e.g., Beijing, Xingtai, and Tianjin) to analyze the overall trend in winter PM2.5 concentrations. A day is defined as an air pollution day based on the NAAQS if the 24-h average PM2.5 concentration is >75.0 μg m −3 ( Table 1). The number of days with different levels of air pollution was counted to show the trends.   The MEC contains almost all the heavily polluted cities in Central and Eastern China. Other heavily polluted areas, such as the Yangtze River Delta, are far from the study area. In addition, considering the weak winds and relatively stable atmospheric conditions in the winter, the effect of pollutants transport between different regions on variations in regional average PM 2.5 concentration in MEC should be small. On the city scale, pollutants transport outside the city is considered to be very important for PM 2.5 concentration in some cities [20]. This study is focused on the influence of meteorological conditions including the upper-air circulation pattern, wind speed, temperature, and ambient relative humidity in a regional PM 2.5 concentration.
The unique topography of MEC has contributed to this region becoming one of the most polluted areas of China (Figure 1b). The western part of MEC is influenced by dry, warm winds from the eastern Taihang Mountains, which results in increased stability at the surface. The east coast of MEC often has lower PM 2.5 concentrations under the influence of moderate to strong winds from the Bohai and Yellow seas [12]. Figure 1b shows the locations of the 264 PM 2.5 monitoring sites selected for this study. The average PM 2.5 concentrations in January were calculated from 2013 to 2019 for MEC and some large polluted cities (e.g., Beijing, Xingtai, and Tianjin) to analyze the overall trend in winter PM 2.5 concentrations. A day is defined as an air pollution day based on the NAAQS if the 24-h average PM 2.5 concentration is >75.0 µg m −3 (Table 1). The number of days with different levels of air pollution was counted to show the trends.
We also collected data from 491 ground meteorological observation sites and five sounding sites, including Beijing, Xingtai, Taiyuan, Zhengzhou, and Jinan ( Figure 1b). The wind speed, temperature (T, K), and dew-point temperature (T D , K) at 02:00, 05:00, 08:00, 11:00, 14:00, 17:00, 20:00, and 23:00 h local time in January 2013-2019 were used to analyze the influence of these meteorological factors on the PM 2.5 concentrations. Average wind speed of 850, 925, and 1000 hPa levels was calculated to represent the wind speed in the boundary layer.
Re-analysis meteorological data from the European Center for Medium-Range Weather Forecasts [36], including the geopotential height (dagpm), temperature (T, K), dew-point temperature (T D , K), and wind speed (m s −1 ) at heights of 500, 700, and 850 hPa. In addition, the surface and the sea-level pressure (hPa) at 02:00, 05:00, 08:00, 14:00, and 20:00 h local time from 2013 to 2019 were used to determine the climate characteristics and synoptic conditions from 2013 to 2019 and their effect on the PM 2.5 distribution.  Figure 2b shows the January average PM 2.5 concentrations at the eight typical stations from 2013 to 2019. The highest PM 2.5 concentrations in 2013 were observed at Xingtai and Baoding stations (330.8 and 266.9 µg m −3 , respectively), which are both located in the middle part of MEC. Beijing, which is the capital of China, also recorded serious air pollution with an average PM 2.5 concentration of 236.4 µg m −3 . The southern MEC, including Zhengzhou (219.9 µg m −3 ) and Jinan (230.7 µg m −3 ), was another highly polluted area. There were clear regional differences in the temporal changes in PM 2.5 concentrations. The annual change trend of PM 2.5 concentration at these middle stations (Beijing, Tianjin, Xingtai, Tangshan, and Baoding) was very similar with the trend of the whole MEC. Yet, the PM 2.5 trends at stations of Taiyuan and Zhengzhou were different from that of MEC (black box). The PM 2.5 concentrations at stations in the middle MEC all decreased from 2013 to 2016, but clearly increased in 2017, and then decreased to the lowest value in 2018. While PM 2.5 concentration in Zhengzhou was the lowest in 2014, it increased from 2014 to 2016. In general, the regional average PM2.5 concentration of MEC in 2019 was 83.4 μg m −3 lower than that of 2013 and PM2.5 concentrations in Beijing, Xingtai, and Baoding stations dropped by >140 μg m −3 , which suggests that air quality in MEC improved significantly during the study period. This phenomenon may be related to the control of anthropogenic emissions and more favorable meteorological conditions.

Meteorological Conditions in the Winter of 2013
The highest regional winter PM2.5 concentration was recorded in 2013 and, therefore, the meteorological conditions, including the upper air circulation pattern and surface meteorological conditions, were investigated further. Figure 3a,b show the average geopotential height and temperature fields at 500 and 850 hPa levels, respectively, in January 2013. The synoptic situation at 500 hPa height was dominated by the distinct zonal circulation. A weak high-pressure ridge was found at both 500 and 850 hPa height in the middle and high latitudes of Asia. The MEC region was located in the front of this high ridge, which indicates that this area was often influenced by the divergence and subsidence of winds, suggesting weak winds and very stable atmospheric conditions in the winter of 2013. Figure 3c shows the average sea-level pressure and temperature fields on the surface in January 2013. The isobaric lines were extremely sparse in MEC, which suggests low horizontal wind speeds. This led to weak dispersion abilities.
The distribution of wind speeds has a dramatic effect on the distribution of PM2.5 concentrations. Low wind speeds suggest that the atmosphere is relatively stable and the dispersion of local emissions is largely limited. High ambient relative humidity (RH) favors the hygroscopic growth of aerosols. Figure 3d shows the average RH on the surface at 8:00 h local time and the wind speed field on 29 pollution days in January 2013. The whole area was controlled by low wind speeds (<3 m s −1 ) and a high ambient RH (>84%). Meteorology conditions such as a turbulence condition and wind may have similar effects on water vapor and PM2.5, which is one of the reasons for high relative humidity and PM2.5 usually happening at the same time. High humidity is very favorable for increasing PM2.5 because of its contribution to the hygroscopic aerosol [8,12]. The highest PM2.5 concentrations in the winter of 2013 were found at Xingtai, Beijing, and Baoding stations in middle MEC. Similarly, a wind convergence zone (brown box in Figure 3c) from the northeast to the southwest was clearly visible in the middle MEC, which covered the most severely polluted stations (Beijing, Baoding, Shijiazhuang, Hengshui, Xingtai, and Zhengzhou). The wind convergence zone occurred in this region primarily as a result of the unique topography. The southeastern winds from the eastern plain and the northwestern winds weakened by the blocking effect of the Taihang Mountains converge to this wind convergence zone along the foot of the Taihang Mountains, which In general, the regional average PM 2.5 concentration of MEC in 2019 was 83.4 µg m −3 lower than that of 2013 and PM 2.5 concentrations in Beijing, Xingtai, and Baoding stations dropped by >140 µg m −3 , which suggests that air quality in MEC improved significantly during the study period. This phenomenon may be related to the control of anthropogenic emissions and more favorable meteorological conditions.

Meteorological Conditions in the Winter of 2013
The highest regional winter PM 2.5 concentration was recorded in 2013 and, therefore, the meteorological conditions, including the upper air circulation pattern and surface meteorological conditions, were investigated further. Figure 3a,b show the average geopotential height and temperature fields at 500 and 850 hPa levels, respectively, in January 2013. The synoptic situation at 500 hPa height was dominated by the distinct zonal circulation. A weak high-pressure ridge was found at both 500 and 850 hPa height in the middle and high latitudes of Asia. The MEC region was located in the front of this high ridge, which indicates that this area was often influenced by the divergence and subsidence of winds, suggesting weak winds and very stable atmospheric conditions in the winter of 2013. Figure 3c shows the average sea-level pressure and temperature fields on the surface in January 2013. The isobaric lines were extremely sparse in MEC, which suggests low horizontal wind speeds. This led to weak dispersion abilities.
The distribution of wind speeds has a dramatic effect on the distribution of PM 2.5 concentrations. Low wind speeds suggest that the atmosphere is relatively stable and the dispersion of local emissions is largely limited. High ambient relative humidity (RH) favors the hygroscopic growth of aerosols. Figure 3d shows the average RH on the surface at 8:00 h local time and the wind speed field on 29 pollution days in January 2013. The whole area was controlled by low wind speeds (<3 m·s −1 ) and a high ambient RH (>84%). Meteorology conditions such as a turbulence condition and wind may have similar effects on water vapor and PM 2.5 , which is one of the reasons for high relative humidity and PM 2.5 usually happening at the same time. High humidity is very favorable for increasing PM 2.5 because of its contribution to the hygroscopic aerosol [8,12]. The highest PM 2.5 concentrations in the winter of 2013 were found at Xingtai, Beijing, and Baoding stations in middle MEC. Similarly, a wind convergence zone (brown box in Figure 3c) from the northeast to the southwest was clearly visible in the middle MEC, which covered the most severely polluted stations (Beijing, Baoding, Shijiazhuang, Hengshui, Xingtai, and Zhengzhou). The wind convergence zone occurred in this region primarily as a result of the unique topography. The southeastern winds from the eastern plain and the northwestern winds weakened by the blocking effect of the Taihang Mountains converge to this wind convergence zone along the foot of the Taihang Mountains, which leads to low wind speeds and high aerosol loadings [12]. This explains why high PM 2.5 events are common in this region.
Atmosphere 2019, 10, 563 6 of 18 leads to low wind speeds and high aerosol loadings [12]. This explains why high PM2.5 events are common in this region. In general, the meteorological conditions in 2013 were very conducive for the accumulation of aerosol pollutants and the occurrence of high PM2.5 concentration events. The strong zonal circulation at 500 hPa height, weak wind speeds, and high ambient RH on the surface favored the highest PM2.5 concentrations in the winter of 2013.

Meteorological Conditions from 2014 to 2019 Compared with 2013
3.3.1. Spatial Distribution of Geopotential Height, Temperature, and Wind Fields at 500 and 850 hPa Levels The type and intensity of upper air circulation patterns have a strong influence on atmospheric conditions. When the synoptic situation is dominated by the zonal circulation, the circulation is relatively straight. Weak troughs and weak ridges usually appear in the westerly air flow. Affected by these, the MEC was often influenced by the divergence and subsidence of winds, which often results in stable atmospheric conditions and weak diffusion abilities. When the synoptic situation is dominated by the meridional circulation, deep troughs and strong ridges usually develop in the westerly air flow, which can cause strong cold air activity and high wind speeds. Figure 4a shows In general, the meteorological conditions in 2013 were very conducive for the accumulation of aerosol pollutants and the occurrence of high PM 2.5 concentration events. The strong zonal circulation at 500 hPa height, weak wind speeds, and high ambient RH on the surface favored the highest PM 2.5 concentrations in the winter of 2013.

Meteorological Conditions from 2014 to 2019 Compared with 2013
3.3.1. Spatial Distribution of Geopotential Height, Temperature, and Wind Fields at 500 and 850 hPa Levels The type and intensity of upper air circulation patterns have a strong influence on atmospheric conditions. When the synoptic situation is dominated by the zonal circulation, the circulation is relatively straight. Weak troughs and weak ridges usually appear in the westerly air flow. Affected by these, the MEC was often influenced by the divergence and subsidence of winds, which often results in stable atmospheric conditions and weak diffusion abilities. When the synoptic situation is dominated by the meridional circulation, deep troughs and strong ridges usually develop in the westerly air flow, which can cause strong cold air activity and high wind speeds. Figure 4a shows the average geopotential height and temperature fields in January from 2014 to 2019. Figure 4b shows the differences in these fields in 2014-2019 relative to 2013.   There were large changes in the synoptic situations at 500 hPa in 2016 and 2018 compared with those meteorology-haze years. The average synoptic situation in the mid-troposphere in 2016 and 2018 were dominated by the meridional circulation. As can be seen in Figure 4(a3), a strong center of high pressure occurred in the southern Ural Mountains in 2016. Strong north winds in front of the high ridge transported a large volume of cold air to Central and Eastern China and caused a considerable decrease in temperature. A deep low-pressure trough was present in the vicinity of Lake Baikal. As a result, MEC was often affected by the convergence and ascension of winds, which contributed to the vertical diffusion of pollutants. Similar geopotential height and temperature fields, although with a weaker ridge and trough, were present in 2018 (Figure 4(b5)), which suggests a relatively weaker upper air circulation in 2018 than in 2016. In general, MEC experienced frequent cold air events and strong winds in both 2016 and 2018, which contributed to the dispersion of pollutants. We, therefore, define 2016 and 2018 as meteorology-clean years.
The synoptic situations at 850 hPa height can also reflect the type of low-level circulation and the intensity of cold air affecting the MEC. Figure 5 shows the average geopotential height, temperature, and wind fields in January from 2014 to 2019.  There were large changes in the synoptic situations at 500 hPa in 2016 and 2018 compared with those meteorology-haze years. The average synoptic situation in the mid-troposphere in 2016 and 2018 were dominated by the meridional circulation. As can be seen in Figure 4(a3), a strong center of high pressure occurred in the southern Ural Mountains in 2016. Strong north winds in front of the high ridge transported a large volume of cold air to Central and Eastern China and caused a considerable decrease in temperature. A deep low-pressure trough was present in the vicinity of Lake Baikal. As a result, MEC was often affected by the convergence and ascension of winds, which contributed to the vertical diffusion of pollutants. Similar geopotential height and temperature fields, although with a weaker ridge and trough, were present in 2018 (Figure 4(b5)), which suggests a relatively weaker upper air circulation in 2018 than in 2016. In general, MEC experienced frequent cold air events and strong winds in both 2016 and 2018, which contributed to the dispersion of pollutants. We, therefore, define 2016 and 2018 as meteorology-clean years.
The synoptic situations at 850 hPa height can also reflect the type of low-level circulation and the intensity of cold air affecting the MEC. Figure 5 shows the average geopotential height, temperature, and wind fields in January from 2014 to 2019. The distributions of geopotential height and temperature fields at a height of 850 hPa from 2013 to 2019 were almost consistent with those at 500 hPa height. These fields clearly reflected the strength of low-level cold air and its affecting region. When the synoptic situation were dominated by the distinct zonal circulation in these meteorology-haze years (2013, 2014, 2015, 2017, and 2019), the MEC was mainly affected by the weak northwest air flow and the low-temperature region was located on the northeast side of the study area, which suggests that the northwest airflow accompanied by cold air mainly affected the high latitudes of Asia and had little effect on MEC. However, when the synoptic situation was dominated by the strong meridional circulation in these meteorology-clean years (2016 and 2018), the low-temperature area (red box) were very close to MEC with clearly lower temperature values and higher wind speed compared with those in meteorology-haze years. The MEC was under the control of strong northerly airflow accompanied by strong cold air from the low-temperature area. The lower temperature and higher wind speed in MEC in 2016 and 2018 suggest that the meteorological conditions in 2016 and 2018 were much more conducive to the diffusion of aerosol pollutants.
The synoptic situation at 500 hPa height was dominated by the strong meridional circulation in the winter of 2016 and strong centers of high and low pressure appeared at high latitudes ( Figure  4(a3)). We, therefore, selected two regions (50-65° N, 70-90° E and 50-65° N, 126-146° E, blue dashed box), which contains the high and low pressure centers, as the meridional circulation affected areas and the difference between the average geopotential heights of these two regions was defined as the Meridional Circulation Index (MCI). The same calculation method was used to obtain the MCI from 2013 to 2019. MCI will be very high if the circulation at 500 hPa height is dominated by the meridional circulation. The average temperature of the upper area (38-52° N, 95-125° E) of MEC (high-level air temperature) was calculated to represent the strength of the high-level cold air. Strong cold air corresponds to a low high-level air temperature. Figures 6a and 7a show the values of the MCI and the high-level air temperature at 500 and 400 hPa heights from 2013 to 2019. The MCI was highest in 2016 and corresponded to the strongest meridional circulation. Figure 6b compares the differences in the MCI from 2013 to 2019 with the highest value. Figure 7b shows the differences in the high-level air temperature from 2014 to 2019 The distributions of geopotential height and temperature fields at a height of 850 hPa from 2013 to 2019 were almost consistent with those at 500 hPa height. These fields clearly reflected the strength of low-level cold air and its affecting region. When the synoptic situation were dominated by the distinct zonal circulation in these meteorology-haze years (2013, 2014, 2015, 2017, and 2019), the MEC was mainly affected by the weak northwest air flow and the low-temperature region was located on the northeast side of the study area, which suggests that the northwest airflow accompanied by cold air mainly affected the high latitudes of Asia and had little effect on MEC. However, when the synoptic situation was dominated by the strong meridional circulation in these meteorology-clean years (2016 and 2018), the low-temperature area (red box) were very close to MEC with clearly lower temperature values and higher wind speed compared with those in meteorology-haze years. The MEC was under the control of strong northerly airflow accompanied by strong cold air from the low-temperature area. The lower temperature and higher wind speed in MEC in 2016 and 2018 suggest that the meteorological conditions in 2016 and 2018 were much more conducive to the diffusion of aerosol pollutants.
The synoptic situation at 500 hPa height was dominated by the strong meridional circulation in the winter of 2016 and strong centers of high and low pressure appeared at high latitudes (Figure 4(a3)). We, therefore, selected two regions (50-65 • N, 70-90 • E and 50-65 • N, 126-146 • E, blue dashed box), which contains the high and low pressure centers, as the meridional circulation affected areas and the difference between the average geopotential heights of these two regions was defined as the Meridional Circulation Index (MCI). The same calculation method was used to obtain the MCI from 2013 to 2019. MCI will be very high if the circulation at 500 hPa height is dominated by the meridional circulation. The average temperature of the upper area (38-52 • N, 95-125 • E) of MEC (high-level air temperature) was calculated to represent the strength of the high-level cold air. Strong cold air corresponds to a low high-level air temperature. Figures 6a and 7a show the values of the MCI and the high-level air temperature at 500 and 400 hPa heights from 2013 to 2019. The MCI was highest in 2016 and corresponded to the strongest meridional circulation. Figure 6b compares the differences in the MCI from 2013 to 2019 with the highest value. Figure 7b shows the differences in the high-level air temperature from 2014 to 2019 relative to 2013. The average synoptic situation in the mid-troposphere was dominated by the zonal circulation in the

Spatial Distribution of Wind Speed and RH Field
Both weak winds and a high ambient RH on the surface are regarded as important factors in high aerosol loadings. The wind speed has a great influence on its ability to disperse pollutants. The ambient RH influences the growth of aerosols by affecting their scattering and hygroscopic properties. The effects of these two factors on PM2.5 pollution events can be determined from Figure  8, which shows the average RH at 8:00 h of local time and wind speed field of air pollution days from 2014 to 2019, along with their differences from the conditions in 2013.

Spatial Distribution of Wind Speed and RH Field
Both weak winds and a high ambient RH on the surface are regarded as important factors in high aerosol loadings. The wind speed has a great influence on its ability to disperse pollutants. The ambient RH influences the growth of aerosols by affecting their scattering and hygroscopic properties. The effects of these two factors on PM2.5 pollution events can be determined from Figure  8, which shows the average RH at 8:00 h of local time and wind speed field of air pollution days from 2014 to 2019, along with their differences from the conditions in 2013.

Spatial Distribution of Wind Speed and RH Field
Both weak winds and a high ambient RH on the surface are regarded as important factors in high aerosol loadings. The wind speed has a great influence on its ability to disperse pollutants. The ambient RH influences the growth of aerosols by affecting their scattering and hygroscopic properties. The effects of these two factors on PM 2.5 pollution events can be determined from Figure 8 (Figure 8(b4)).
The wind speeds increased at most of the stations located in the middle and southern MEC (including Shijiazhuang, Xingtai, Hengshui, Zhengzhou, and Jinan) in 2014. This region of high winds extended to the northern MEC in 2015, which includes Baoding, Beijing, and Zhangjiakou. The wind speeds at the southern stations (including Xingtai, Jinan, and Zhengzhou) then decreased from 2015 to 2018. There was no significant change in the wind speeds at stations in the middle and eastern MEC (including Baoding, Beijing, Tianjin, and Tangshan) from 2015 to 2018, even though these were still higher than in 2013. Figure 9 shows the daily average PM2.5 concentrations, wind speed, temperature, and RH values at 8:00 h of local time of the overall MEC and the five stations from 2013 to 2019 in order to study the influence of surface factors on PM2.5 concentrations. Table 2 gives the values of regional average PM2.5 concentrations, wind speeds at 700 and 850 hPa levels, surface RH, and wind speeds of the whole MEC as well as the air pollution days, MCI, and high-level air temperature from 2013 to 2019. Winter (January) is a dry season in MEC in China. Considering precipitation in the study area was generally low in the winter and the difference in spatial distribution of precipitation was small (Figure ignored). Therefore, the influence in the removal effect of rainfall on regional PM2.5 concentration could be neglected.  (Figure 8(b4)).

Contribution of Meteorological Conditions and Emissions to Winter PM2.5 Concentrations
The wind speeds increased at most of the stations located in the middle and southern MEC (including Shijiazhuang, Xingtai, Hengshui, Zhengzhou, and Jinan) in 2014. This region of high winds extended to the northern MEC in 2015, which includes Baoding, Beijing, and Zhangjiakou. The wind speeds at the southern stations (including Xingtai, Jinan, and Zhengzhou) then decreased from 2015 to 2018. There was no significant change in the wind speeds at stations in the middle and eastern MEC (including Baoding, Beijing, Tianjin, and Tangshan) from 2015 to 2018, even though these were still higher than in 2013. Figure 9 shows the daily average PM 2.5 concentrations, wind speed, temperature, and RH values at 8:00 h of local time of the overall MEC and the five stations from 2013 to 2019 in order to study the influence of surface factors on PM 2.5 concentrations. Table 2 gives the values of regional average PM 2.5 concentrations, wind speeds at 700 and 850 hPa levels, surface RH, and wind speeds of the whole MEC as well as the air pollution days, MCI, and high-level air temperature from 2013 to 2019. Winter (January) is a dry season in MEC in China. Considering precipitation in the study area was generally low in the winter and the difference in spatial distribution of precipitation was small ( Figure  ignored). Therefore, the influence in the removal effect of rainfall on regional PM 2.5 concentration could be neglected.    The time series of daily PM 2.5 values shows that the air quality in January 2013 (Figure 9a) was only good during the first two days. Thereafter, most of the daily values reached the level of heavy pollution. The heaviest air pollution episode with a peak value of 296.3 µg m −3 was observed from January 6, 2013 to January 16, 2013. Corresponding to these high PM 2.5 values, high RH values and low wind speeds were also observed in 2013. The rapid increase in PM 2.5 concentrations in early January corresponded with the rapid increase in RH and decrease in wind speed. The correlation coefficients between PM 2.5 concentration and RH and wind speed were 0.44 and −0.49, passed a 95% and a 99% significant test, respectively, which suggests that short-term variations in PM 2.5 concentrations have a positive correlation with ambient RH and a negative correlation with the wind speed. Beijing ( Figure 9b) and Xingtai (Figure 9c), which are located in the middle part of MEC, as well as Jinan ( Figure 9d) and Zhengzhou (Figure 9f) stations, which are located in the southern MEC, all experienced the most severe air pollution in 2013. An extremely serious air pollution event was observed in Beijing during the period January 9-23, with a peak on January 12 of 838.5 µg m −3 . Xingtai was confirmed as the most polluted city in 2013, with an average PM 2.5 concentration of 329.3 µg m −3 (Figure 2b). The daily PM 2.5 value for 21 days in Xingtai exceeded 250.0 µg m −3 and reached the level of hazardous pollution. These frequent air pollution events in the three middle stations may be related to the high ambient RH with average values of the two stations close to 80% and weak winds with average values <2 m·s −1 .

Contribution of Meteorological Conditions and Emissions to Winter PM 2.5 Concentrations
The regional daily PM 2.5 concentrations in 2014 fluctuated around 150 µg m −3 and most of the PM 2.5 values were relatively lower than in 2013 (Figure 9a). Corresponding to the lower PM 2.5 values, most RH values were also lower than in 2013. The wind speeds in the first half of the month in 2014 showed no significant change, but increased thereafter. A higher MCI and lower high-level air temperature than in 2013 suggest that the upper air circulation pattern in 2014 was more favorable for the diffusion of aerosol pollutants. The average PM 2.5 at Beijing and Xingtai stations decreased by 118.6 and 74.6 µg m −3 , respectively (Figure 2b). Correspondingly, the average RH at the two stations dropped to <60% and the average surface wind speeds increased. On the regional scale, the regional winter PM 2.5 decreased by 26% in 2014 compared with that in 2013 ( Table 2). The meteorological factors in 2014 were also more conducive to the diffusion of pollutants. Considering the 87% higher MCI, and both 5% higher wind speed of 700 hPa height and the ground than in 2013, the 26% decrease in PM 2.5 in 2014 may be due to better meteorological conditions. The regional winter PM 2.5 value in 2015 was clearly lower than that in 2013 and 2014 (Figure 2a). The PM 2.5 values decreased further at these stations, except at station Zhengzhou, but there was no clear change in the wind speed, RH, and temperature in MEC. The largest decrease in the average PM 2.5 concentration at the six stations was at Xingtai (Figure 9c), where it was accompanied by a large increase in the average wind speed (3.0 m·s −1 ) and a significant decrease in RH (48.3%). However, the reduction in PM 2.5 concentrations at Beijing, Taiyuan, and Jinan stations did not correspond to better meteorological conditions on the surface. The winds at Jinan station, both on the surface and in the boundary layer, became weaker than in 2014. The regional average PM 2.5 value of MEC in 2015 (92.8 µg m −3 ) decreased by 29% compared with 2014 (130.5 µg m −3 ), while the wind speeds at 700 and 850 hPa height decreased by 16% and 6%, respectively. In addition, there were only small changes in surface RH and wind speed ( Table 2). It has been shown that the large-scale circulation in 2015 resulted in a poorer diffusion of pollutants, with a 52% lower MCI than in 2014. We suggest that the 29% reduction in PM 2.5 in 2015 compared with 2014 was the result of emissions reduction. Regional PM 2.5 concentration in 2015 was 47% lower than 2013, while MCI, wind speed at 700 hPa and 850 hPa, height in 2015 were 11%, 12%, and 6% lower than those in 2013. Therefore, the 47% reduction in PM 2.5 compared with 2013 was likely due to emissions reduction in MEC.
MEC was under the control of the strongest meridional circulation in 2016, which was one of the meteorology-clean years. However, some severe PM 2.5 air pollution events still occurred at the five stations in early January. The daily regional PM 2.5 concentration decreased sharply from 80.0 to 18.3 µgm −3 in the period of January 21 to 23, accompanied by increasingly strong winds and lower temperatures (Figure 9a), which suggests that this region was strongly affected by cold air. A clear decrease in temperature corresponding to strong winds has been observed at all these stations. The effect of this strong cold air led to only a small decrease in the average PM 2.5 concentrations at Beijing, Xingtai, and Taiyuan stations in 2016 compared with 2015. By contrast, the PM 2.5 concentrations at Jinan and Zhengzhou increased. In general, the MCI (36.8) 2016 was much higher than in 2015 (8.3). Wind speed at 700 and 850 hPa heights and the surface increased by 26%, 22%, and 13%, respectively. All these values were the highest in the time period studied here, while the PM 2.5 concentration in the winter of 2016 showed a small increase from those in 2015. The change of PM 2.5 concentration in 2016 cannot be explained simply by the changes of meteorological conditions, which suggests that the emission of pollutants was far higher in 2016 than in 2015. This means that, even if the meridional circulation and cold air at 500 hPa height were the strongest and the wind speeds were the highest in 2016, the high emission of pollutants led to a slight increase in PM 2.5 concentration.
The synoptic situation was influenced by the zonal circulation at 500 hPa height in the winter of 2017 and the atmospheric conditions became stable again. Correspondingly, some heavy air pollution episodes occurred from January 1 to 12 (Figure 9a). The average PM 2.5 concentrations in 2017 at Beijing and Xingtai stations also increased and exceeded those in 2015, but were still much lower than in 2014. Despite the poorer surface conditions at Jinan and Zhengzhou stations, with lower winds and lower RH than in 2016, the average PM 2.5 concentration decreased slightly, which suggests lower anthropogenic emissions in 2017. The average PM 2.5 concentration at Taiyuan in January 2017 was the highest in the study period. The daily average PM 2.5 concentration even exceeded 400 µg m −3 on 1 January, 2017. However, the average RH was the lowest at Taiyuan station in 2017 and the wind speeds were similar to previous years. Compared with other stations, the large increase in PM 2.5 concentration in Taiyuan may be related to differences in anthropogenic emissions and its unique topography. In addition, it is believed that the Taiyuan station was affected by severe air pollution in 2017 when related to high anthropogenic emissions. According to Table 2, the MCI wind speeds were at 700 and 850 hPa levels. The surface decreased by 65%, 7%, 17%, and 13%, respectively, in 2017 when compared to 2016, whereas the regional winter PM 2.5 only increased by 16%. This suggests that the 16% increase in PM 2.5 in 2017 should be entirely due to the poorer meteorological conditions rather than changes in anthropogenic emissions. Higher MCI and wind speed at 700 hPa and 850 hPa levels in 2017 compared with those in 2015 suggests that meteorological conditions in 2017 were more conducive to the diffusion of pollutants. However, PM 2.5 concentration in 2017 was 21% higher than in 2015. Thus, the 21% higher PM 2.5 in 2017 was likely due to higher pollutant emissions in MEC rather than changes in meteorology conditions. PM 2.5 concentration was lowest in 2015 and low from 2013 to 2017. The pollutant emissions were also likely the lowest from 2013 to 2017.
The regional PM 2.5 concentration decreased to its lowest value (76.6 µg m −3 ) in 2018 (Table 2), which was another meteorology-clean year. Similar to the period 21-23 January, 2016, the PM 2.5 concentration clearly decreased from 85.0 to 21.8 µg m −3 in the period of January 7-9, accompanied by strong winds and low temperatures. The average winter PM 2.5 concentration at Beijing station in 2018 dropped to 39.1 µg m −3 , and there were only five air pollution days in Beijing in 2018 when compared with 27 in 2013. It is believed that the air quality was greatly improved in the two cities. As another meteorology-clean year, the MCI in 2018 (27.3) was 20% lower than in 2016 (36.8) and the other meteorological factors in 2016 and 2018 were similar, although the air quality was better in 2018, with the lowest regional PM 2.5 concentration (76.6 µg m −3 , 20% lower than in 2016) and the minimum number of air pollution days (14 days) occurred from 2013 to 2019 (Table 2). This suggests that anthropogenic emissions in the winter of 2018 were likely 20% to 40% lower when compared to 2016.
The winter PM 2.5 concentration increased again in 2019. The PM 2.5 concentrations at the middle stations (including Beijing, Tianjin, Xingtai, and Tangshan) increased more significantly than in the southern stations (including Jinan and Zhengzhou) (Figure 2b). Beijing, Xingtai, and Jinan were observed to have heavy PM 2.5 air pollution events, with a peak value >300 µg m −3 from 7-15 January, 2019. This may be related to the more stable atmospheric conditions and much lower wind speeds compared with those of 2018 ( Figure 9a). Table 2 shows that the MCI (17.1) decreased by 20% compared with that in 2018 (36.8). Wind speeds of 700 (8.8 m·s −1 ) and 850 hPa (4.1 m·s −1 ) levels as well as the surface (2.0 m·s −1 ) decreased by 24%, 20%, and 20%, respectively, and all were the lowest from 2013 to 2019, whereas the winter PM 2.5 concentration only increased by 20% in 2019. This suggests that the 20% increase in PM 2.5 concentration in 2019 may be entirely due to poorer atmospheric conditions. Compared with the meteorological conditions in 2014, wind speeds at 700 and 850 hPa levels in 2019 were 22% and 16% lower than in 2014, respectively. The MCI and surface wind speed in 2019 were very similar to those in 2014 (17.4 and 2.1 m·s −1 , respectively), whereas the PM 2.5 concentration was 30% lower when compared to 2014. This suggests that the 30% decrease in PM 2.5 concentration is entirely due to a reduction in the emission of pollutants in 2019.
Overall, a change in meteorological conditions may cause a 26% change in PM 2.5 concentration between two meteorology-haze years and 16% to 20% changes in PM 2.5 concentration between the meteorology-haze year and the meteorology-clean year, respectively. Changes in pollutant emissions may cause 21% to 47% changes in PM 2.5 concentration between two meteorology-haze years. In a comparison of two meteorology-clean years, pollutant emissions in 2018 were reduced by 40% when compared with 2016. Changes in emissions had a greater influence on changing in PM 2.5 than changes in meteorological conditions.

Discussion
This study is mainly focused on the contribution of the meteorological conditions to regional average PM 2.5 concentration. The methods have some limitations and uncertainties. The quantitative contribution of meteorological conditions and emissions to regional PM 2.5 is basically reasonable and credible, but it is not completely accurate. Other meteorological factors within this study region, which includes some thermal circulations caused by pollutants transportation [37][38][39], atmospheric chemistry [40,41], or thermal circulations [32] are also important when we consider the difference of PM 2.5 concentrations on the city scale. However, these have not been studied in depth because of their small influence on regional PM 2.5 concentration. Our follow-up work may involve these.
The ambient RH might be the most complex factor affecting PM 2.5 concentrations [42][43][44]. The PM 2.5 concentration and the RH were both highest in 2013, which suggests a possible contribution of high RH to a PM 2.5 concentration. High ambient RH is very helpful to the hygroscopic growth of aerosols [12]. However, PM 2.5 concentration and RH are affected by some meteorological factors at the same time. If the local wind speed is relatively low, the diffusion of pollutants and water vapor will be limited to a certain extent, which easily results in both high PM 2.5 concentration and high RH. The influence mechanism of RH on different components of PM 2.5 is also different [43][44][45]. It is indicated that the influence of the RH on PM 2.5 concentration requires further study.

Conclusions
The annual changes in the trends of the winter PM 2.5 concentration in Middle-Eastern China from 2013 to 2019 were analyzed by The Meridional Circulation Index (MCI), which represents the type and strength of the upper air circulation pattern (a higher MCI value indicates more favorable meteorological conditions for high PM 2.5 concentrations), wind speed at 700 and 850hPa heights, and surface wind speed. RH and temperature were investigated to study the contribution of meteorology changing yearly and PM 2.5 changing on the regional and city scale. On the regional scale, the winter PM 2.5 concentration in 2013 (175.2 µg m −3 ) was the highest from 2013 to 2019. Corresponding to this, the meteorological conditions, including the MCI and surface factors, were also the most suitable for the accumulation of aerosols and haze pollution in the winter of 2013, referred to as a meteorology-haze year. The January average synoptic situation at 500 hPa height was dominated by the distinct zonal circulation with a low MCI value (9.3), which results in weak cold air and stable atmosphere conditions. The lowest wind speed (2.0 m s −1 ) and the highest ambient RH The average synoptic situation at 500 hPa height in the winter of 2016 was dominated by the distinct meridional circulation with the highest MCI value, which resulted in the strongest cold air and highest wind speed (2.4 m·s −1 ). Compared with 2016, the MCI was 77% lower in 2015, 75% lower in 2013, 65% lower in 2017, 54% lower in 2019, 53% lower in 2014, and 26% lower in 2018. It can be seen that the annual change in the PM 2.5 concentration is not completely consistent with the meteorological conditions due to changes in the emission of pollutants.
The regional winter PM 2.5 concentration showed a significantly decreasing trend from 2013 to 2015. The PM 2.5 concentration decreased by 26% in 2014 relative to 2013. The meteorological factors in 2014 were also more conducive to the diffusion of pollutants. Considering the 87% higher MCI and both 5% higher wind speeds of 700 hPa height and the surface relative to 2014, the 26% decrease in PM 2.5 concentration in 2014 may be due to better meteorological conditions. The PM 2.5 concentration decreased by 29% in 2015, whereas the MCI value in 2015 was 52% lower than in 2014, which suggests that the 29% reduction in the PM 2.5 concentration relative to 2014 was due to a reduction in emissions. It is worth noting that both MCI and wind speeds in 2016 were much higher than in 2015 and were the highest from 2013 to 2019, even though the PM 2.5 concentration in the winter of 2016 was slightly higher than that in 2015. This suggests that the emission of pollutants in 2016 was much higher than in 2015. This means that, even if the meridional circulation and cold air at 500 hPa height was the strongest and the wind speeds were the largest in 2016, the higher emission of pollutants led to a slight increase in PM 2.5 concentrations when compared to 2015. The MCI, wind speeds at 700, 850 hPa levels, and the surface decreased by 65%, 26%, 22%, and 13%, respectively, in 2017 when compared to 2016, whereas the winter PM 2.5 concentration only increased by 16% in 2017. This suggests that the 16% increase in the PM 2.5 concentration in 2017 may be due to worse meteorological conditions. Higher MCI and wind speed at 700 hPa and 850 hPa levels in 2017 compared with those in 2015 suggests