Spatial and Temporal Characteristics of Environmental Air Quality and Its Relationship with Seasonal Climatic Conditions in Eastern China during 2015–2018

Exploring the relationship between environmental air quality (EAQ) and climatic conditions on a large scale can help better understand the main distribution characteristics and the mechanisms of EAQ in China, which is significant for the implementation of policies of joint prevention and control of regional air pollution. In this study, we used the concentrations of six conventional air pollutants, i.e., carbon monoxide (CO), sulfur dioxide (SO2), nitrogen dioxide (NO2), fine particulate matter (PM2.5), coarse particulate matter (PM10), and ozone (O3), derived from about 1300 monitoring sites in eastern China (EC) from January 2015 to December 2018. Exploiting the grading concentration limit (GB3095-2012) of various pollutants in China, we also calculated the monthly average air quality index (AQI) in EC. The results show that, generally, the EAQ has improved in all seasons in EC from 2015 to 2018. In particular, the concentrations of conventional air pollutants, such as CO, SO2, and NO2, have been decreasing year by year. However, the concentrations of particulate matter, such as PM2.5 and PM10, have changed little, and the O3 concentration increased from 2015 to 2018. Empirical mode decomposition (EOF) was used to analyze the major patterns of AQI in EC. The first mode (EOF1) was characterized by a uniform structure in AQI over EC. These phenomena are due to the precipitation variability associated with the East Asian summer monsoon (EASM), referred to as the “summer–winter” pattern. The second EOF mode (EOF2) showed that the AQI over EC is a north–south dipole pattern, which is bound by the Qinling Mountains and Huaihe River (about 35° N). The EOF2 is mainly caused by seasonal variations of the mixed concentration of PM2.5 and O3. Associated with EOF2, the Mongolia–Siberian High influences the AQI variation over northern EC by dominating the low-level winds (10 m and 850 hPa) in autumn and winter, and precipitation affects the AQI variation over southern EC in spring and summer.


Introduction
The World Health Organization (WHO) has stated that over 90% of the world's population are affected by high levels of air pollution, and nearly 7 million premature deaths related to air pollution occur each year [1]. China is one of the most severely air polluted countries in the world [2,3], which may significantly impact the health of its citizens [4][5][6][7] and economic development [8,9]. Therefore, atmospheric contamination has attracted the attention of the Chinese public and government, and become one of the most important environmental issues in current societies [10].
Environmental air quality (EAQ) is mainly affected by the combined effects of human activities and climatic conditions [11]. On a regional scale, the emission of air pollutants is a significant contributor to ambient air pollution [12]. In the past decade, due to the continuous expansion of Chinese intensive industry and the increase in production capacity and energy consumption, EAQ has deteriorated severely [13]. The levels of primary and the EAQ in the northern plains of China in the autumn and winter of 2013 [39]. Compared with 2014, the climatic/meteorological conditions in 2015 were adverse for the spreading of air pollutants in some regions in China [40]. Furthermore, the frequency and duration of severe haze weather events in northern China were also related to specific climatic conditions in winter [41]. The rapid decline in the extent of Arctic sea-ice, which is caused by global warming, and reduced precipitation and surface winds could have intensified the haze pollution in eastern China [42], which can explain the 45%-67% of interannual to decadal haze pollution variabilities after 2000 [43]. Hence, climatic conditions may play a non-negligible role in influencing the EAQ on a large scale.
The above studies explored the spatial distribution of EAQ in major Chinese cities in recent years and its relationship with climatic/meteorological conditions, providing important theoretical references and practical recommendations for this study. However, due to the monitoring data of EAQ being sparse in time and space, the major spatial and temporal distribution of EAQ in China and its response to climatic conditions remain less understood. Therefore, the aim of this study was to understand the major spatial and temporal variations of AQI, using 4-year (January 2015-December 2018) EAQ data released by the MEP for nearly 1300 monitoring stations in Eastern China (EC), and their associations with climatic conditions, hoping to provide a scientific reference for Chinese EAQ governance.

Data Sources
The hourly air pollution data of SO 2 , NO 2 , PM 10  We acquired the climate data from the ERA-interim (ERAI, https://www.ecmwf. int/en/forecasts/datasets/reanalysis-datasets/era-interim, accessed on 1 May 2019) reanalysis data of the European Center for Medium-Range Weather Forecast (ECMWF), which is available monthly. The ERAI reanalysis has improved significantly the global atmospheric records of mass, moisture, energy, and angular momentum compared with ERA-40 [44]. Compared with other reanalysis data, ERAI reanalysis products are closer to the real observation data of China, in climatic elements such as temperature, atmospheric circulations, and cloud water distribution [45][46][47]. These efforts ensure the trustworthiness of the conclusions made in this study. Given the main objectives of this study, the spatial resolution of climatic elements derived from the ERAI was 0.5 • × 0.5 • , and the period was from January 2015 to December 2018, the same as the air pollution data (more information in Table 1).

Methods
According with the aims of the study, the original air pollution monitoring data derived from the stations was screened and processed. First, we computed daily averages by averaging hourly CO, SO 2 , NO 2 , PM 2.5 , and PM 10 , and maximum average 8 h O 3 values in a day. Missing values in the data were replaced by linear interpolation. For a certain type of monitoring data, if the number was less than eight points in a day, it was considered as the missing value for that day. Second, we calculated the total missing days for the six types of air pollutant during 2015 to 2018. For each monitoring station, if the number of missing days for a type of air pollutant was less than 5% of the year, it would be classified as an "effective monitoring station". Otherwise, the station was classified as an "invalid monitoring station". The locations of the effective (red dot) and invalid (blue dot) monitoring stations during 2015 to 2018 are shown in Figure 1. The number of effective monitoring stations for each air pollutant for every year was greater than 1300 (the total number of monitoring stations is 1615), accounting for more than 80% (

Methods
According with the aims of the study, the original air pollution monitoring data derived from the stations was screened and processed. First, we computed daily averages by averaging hourly CO, SO2, NO2, PM2.5, and PM10, and maximum average 8 h O3 values in a day. Missing values in the data were replaced by linear interpolation. For a certain type of monitoring data, if the number was less than eight points in a day, it was considered as the missing value for that day. Second, we calculated the total missing days for the six types of air pollutant during 2015 to 2018. For each monitoring station, if the number of missing days for a type of air pollutant was less than 5% of the year, it would be classified as an "effective monitoring station". Otherwise, the station was classified as an "invalid monitoring station". The locations of the effective (red dot) and invalid (blue dot) monitoring stations during 2015 to 2018 are shown in Figure 1. The number of effective monitoring stations for each air pollutant for every year was greater than 1300 (the total number of monitoring stations is 1615), accounting for more than 80% (Table 2), and covering 369 Chinese cites by statistical calculations. To ensure the reliability of the study results, air pollutant data derived from invalid monitoring stations each year were removed in this study.  Due to the most effective monitoring stations being located in EC (Figure 1), this study mainly focused on characteristics of EAQ in EC (105-135° E, 20-55° N). To better explore the major spatial and temporal characteristics of monthly AQI, as well as its relationship with climate conditions in EC during 2015-2018, the data were processed as follows: (1) Interpolation of air pollutants data (daily mean derived from the effective monitoring stations during 2015-2018) to a regular 0.5° × 0.5° grid using the cubic spline method, which was the same as the spatial resolution of the climate data covering the EC; (2) calculation of daily average AQI using interpolated air pollutant data for each grid point of EC, referring to Chinese ambient air quality standards (GB3095-2012, https://www.mee.gov.cn/ywgz/fgbz/bz/bzwb/dqhjbh/dqhjzlbz/201203/W0201204103302 32398521.pdf, accessed on 5 February 2021) issued by CMEP. The calculation formula of AQI is:  Due to the most effective monitoring stations being located in EC (Figure 1), this study mainly focused on characteristics of EAQ in EC (105-135 • E, 20-55 • N). To better explore the major spatial and temporal characteristics of monthly AQI, as well as its relationship with climate conditions in EC during 2015-2018, the data were processed as follows: (1) Interpolation of air pollutants data (daily mean derived from the effective monitoring stations during 2015-2018) to a regular 0.5 • × 0.5 • grid using the cubic spline method, which was the same as the spatial resolution of the climate data covering the EC; (2) calculation of daily average AQI using interpolated air pollutant data for each grid point of EC, referring to Chinese ambient air quality standards (GB3095-2012, https://www.mee.gov.cn/ywgz/ fgbz/bz/bzwb/dqhjbh/dqhjzlbz/201203/W020120410330232398521.pdf, accessed on 5 February 2021) issued by CMEP. The calculation formula of AQI is: AQI = max IAQI SO 2 , IAQI NO 2 , IAQI CO , IAQI PM 2.5 , IAQI PM 10 , IAQI O 3 where p represents the pollutant p (SO 2 , NO 2 , CO, PM 2.5 , PM 10 , and O 3 ); IAQI p represents the individual AQI (IAQI); C P is the measured concentration of p; BP Hi represents the high threshold of C P interval; BP LO is the low threshold of C P interval; IAQI Hi is the IAQI values of BP Hi ; and IAQI Lo is the IAQI values of BP LO . The maximum IAQI p of all pollutants is chosen as the overall AQI. According to the stage of development of China's socio-economic conditions, the IAQI threshold interval was set as the medium-term target recommended by the WHO (Table 3). (3) The monthly AQI was simply calculated by averaging the non-missing daily AQI for the month. Table 3. Individual air quality index (IAQI) and corresponding concentration limit of pollution items (HJ633-2012).  0  50  50  40  2  50  35  100  100  150  80  4  150  75  160  150  475  180  14  250  115  215  200  800  280  24  350  150  265  300  1600  565  36  420  250  800  400  2100  750  48  500  350  1000  500  2620  940  60  600  500  1200 To obtain the major spatial and temporal characteristics of AQI distribution in EC during 2015-2018, one technique used for such studies is to decompose the spatial AQI of EC into patterns called empirical orthogonal functions (EOF), and these patterns are ranked according to how much of the total variance they explain. Each EOF is multiplied by a time-dependent coefficient, which is called the principal component (PC) of that EOF. Often a few of these EOFs with their PCs explain most of the variability, and in some cases a particular PC has a temporal fingerprint that matches the climatic signal. EOF has been widely used in meteorology to analyze various meteorological elements. Its fundamental function is decomposing the variable matrix X i×j , which consists of the AQI of j times (time scale of samples) at i spatial points, into the linear combination of a spatial eigenvectors matrix V and their associated time coefficients matrix T:

IAQI
For nth AQI value x mn at mth point, EOF expansion is to decompose x mn into the sum of products of spatial functions and temporal functions: The eigenvector characterizes the spatial pattern of a reginal AQI field. The associated time coefficients delineate the temporal variation character of the spatial pattern characterized by the eigenvector. A positive associated time coefficient indicates that the pattern is the major trend of a variable at that moment; conversely, a negative associated time coefficient indicates that the variable displays an opposite trend of the pattern; and the larger the absolute value of the associated time coefficient, the more significant the associated spatial pattern. For more information about EOF method, we refer to the study by North et al. [48].
If not specified otherwise, here, spring means March-April-May, summer means June-July-August, autumn means September-October-November, and winter means December-January-February.

Spatial and Temporal Characteristics of AQI
The spatial patterns of the seasonal average AQI in EC in 2015 are shown in Figure 2a. Generally, the EAQ was significantly better than other seasons in summer, and worst in winter in EC in 2015. The regional average concentrations of the six air pollutants all reached their lowest in summer and the highest in winter (with the exception of O 3 ). The Asian monsoon system, which brings more precipitation to EC in summer, offers increased air purification capacity. Therefore, the AQI was relatively low in summer ( Figure 2a). Starting in the autumn, using coal for heating begins in northern EC, and the local EAQ was gradually poor. Furthermore, the climate in EC is mainly controlled by the East Asian winter monsoon in winter, which results in a strong vertical temperature inversion layer, making air pollutants difficult to spread. Therefore, the AQI was relatively high in winter. The above results are consistent with previous research [49]. Moreover, as the largest fraction of energy consumption is in the Beijing-Tianjin-Hebei region [50,51], the highest AQI of each season was mainly distributed in the North China Plain. In contrast, because of being large and sparsely populated, and its animal husbandry production mode, the AQI was relatively low in the central and eastern Inner Mongolia border regions (Figure 2a). The distributions of AQI were relatively low in southeast coastal regions as well, which have abundant rainfall and high air humidity all year round. There was a strong anticorrelation between air humidity and AQI values [29]. Compared with 2015, the AQI in spring and summer did not change significantly in 2016 (Figure 2b). In autumn, the AQI distributed in Hebei, Shanxi, and Shaanxi increased significantly, while it decreased in Northeast China. In winter, the distributions of AQI showed a significant increase in Hebei, Shanxi, and parts of southern EC. In 2017, the distributions of AQI increased in the summer in Shanxi and Northern EC relative to 2015 (Figure 2c). In winter, the AQI distribution in Hebei decreased significantly, while it increased significantly in southern EC. In 2018, air pollution generally increased in northern EC from the spring with reference to 2015 (Figure 2d), especially in the central and northern regions of Inner Mongolia. The distributions of AQI also increased significantly in Hebei, Shanxi, and central-northern Inner Mongolia in summer. In autumn, air quality improved significantly in the northeast and the North China Plain, which was the most polluted area in 2015. In winter, the EAQ in EC improved overall, and the distributions of AQI increased only in the central part of Shanxi Province. larger the absolute value of the associated time coefficient, the more significant the associated spatial pattern. For more information about EOF method, we refer to the study by North et al. [48].
If not specified otherwise, here, spring means March-April-May, summer means June-July-August, autumn means September-October-November, and winter means December-January-February.

Spatial and Temporal Characteristics of AQI
The spatial patterns of the seasonal average AQI in EC in 2015 are shown in Figure 2a. Generally, the EAQ was significantly better than other seasons in summer, and worst in winter in EC in 2015. The regional average concentrations of the six air pollutants all reached their lowest in summer and the highest in winter (with the exception of O3). The Asian monsoon system, which brings more precipitation to EC in summer, offers increased air purification capacity. Therefore, the AQI was relatively low in summer ( Figure  2a). Starting in the autumn, using coal for heating begins in northern EC, and the local EAQ was gradually poor. Furthermore, the climate in EC is mainly controlled by the East Asian winter monsoon in winter, which results in a strong vertical temperature inversion layer, making air pollutants difficult to spread. Therefore, the AQI was relatively high in winter. The above results are consistent with previous research [49]. Moreover, as the largest fraction of energy consumption is in the Beijing-Tianjin-Hebei region [50,51], the highest AQI of each season was mainly distributed in the North China Plain. In contrast, because of being large and sparsely populated, and its animal husbandry production mode, the AQI was relatively low in the central and eastern Inner Mongolia border regions (Figure 2a). The distributions of AQI were relatively low in southeast coastal regions as well, which have abundant rainfall and high air humidity all year round. There was a strong anti-correlation between air humidity and AQI values [29]. Compared with 2015, the AQI in spring and summer did not change significantly in 2016 (Figure 2b). In autumn, the AQI distributed in Hebei, Shanxi, and Shaanxi increased significantly, while it decreased in Northeast China. In winter, the distributions of AQI showed a significant increase in Hebei, Shanxi, and parts of southern EC. In 2017, the distributions of AQI increased in the summer in Shanxi and Northern EC relative to 2015 (Figure 2c). In winter, the AQI distribution in Hebei decreased significantly, while it increased significantly in southern EC. In 2018, air pollution generally increased in northern EC from the spring with reference to 2015 (Figure 2d), especially in the central and northern regions of Inner Mongolia. The distributions of AQI also increased significantly in Hebei, Shanxi, and central-northern Inner Mongolia in summer. In autumn, air quality improved significantly in the northeast and the North China Plain, which was the most polluted area in 2015. In winter, the EAQ in EC improved overall, and the distributions of AQI increased only in the central part of Shanxi Province. In recent years, with the strengthening of air pollution governance efforts, the air quality in most areas of EC has been improving [52,53]. The concentration of SO 2 , CO and NO 2 , decreased year by year over the period of 2015-2018 (figures not shown), especially the SO 2 concentration decreased significantly in EC. This indicates that the strict control of SO 2 industrial emissions has been effective [49]. Although the concentration of particulate matter (PM 2.5 and PM 10

The Major Modes of the AQI Distribution
It can be seen from the previous section, that the seasonal mean states of AQI show some differences year by year. Would this change in the mean state affect the monthly variations? To address this question, EOF analysis was applied to the monthly AQI in EC over the period of 2015-2018. The first two EOF modes were statistically distinguished from the rest of the EOF modes based on the significant test proposed by North et al. [54]. Figure 3 shows the spatial pattern of the leading EOF (EOF1) of the AQI in EC (Figure 3a) and the corresponding PC time series (PC1) for 2015-2018 (Figure 3c). The fractional variance explained by the EOF1 was 52.9%, accounting for more than half of the total variance, which represents a robust mode of spatial and temporal variabilities of the AQI in EC. The spatial pattern of EOF1 was characterized by a uniform structure throughout EC (Figure 3a), which exhibited a relatively large variation in the northern plains of China. The PC1 shown in Figure 3c indicates an obvious annual cycle variation from 2015 to 2018, which means that EAQ was better in summer and worse in winter in EC, consistent with the previous research results [55,56]. Therefore, the EOF1 of the AQI distribution in EC suggests an anti-correlation between winter and summer, which refers to this case as the "winter-summer" mode. Furthermore, the spatial pattern of the mean AQI in the summer half-year (May-September) minus that in the winter half-year (November-next March) was consistent with the spatial pattern of EOF1 during 2015-2018 (spatial correlation coefficient (CC) was 0.94, p < 0.01). Likewise, the same procedure of EOF analysis was applied to the six conventional air pollutants (figures not shown). The EOF1 of each air pollutant was similar to the EOF1 of AQI over EC (except for the leading mode of O 3 , which had a significant anti-correlation with the EOF1 of AQI). This means that the "winter-summer" mode is a common feature of the variations in each air pollutant in EC. viron. Res. Public Health 2021, 18, x 9 of 17 The spatial structure of the second mode of AQI (EOF2) in EC is a north-south dipole pattern, characterized by the EAQ "north good (poor) and south poor (good)" (Figure 4a). The explained variance of EOF2 was 14.4%. The corresponding PC time series of EOF2 (PC2) was not significantly characterized by cycles and trends (Figure 4c). However, since the spring of 2017, the amplitude of PC2 intensity has significantly increased. By calculating the regional average IAQI in northern and southern EC (with 35° N as the boundary), respectively, the IAQIs of the six air pollutants all showed significant anti-correlations between northern and southern EC (CCs < −0.99, p < 0.01). Relative to EC-average IAQIs of CO, PM10, SO2, and NO2, the IAQIs in each month showed negative values over the southern EC and positive values over the northern EC ( Figure 5). This indicates that these air pollutants over northern EC are always greater than those over southern EC. However, the IAQI of PM2.5 and O3 averaged over southern and northern EC in each month showed alternating positive and negative changes relative to these averages over EC ( Figure 5). Therefore, the EOF2 of AQI was probably caused by the combined effect of PM2.5 and O3. By comparing the difference between the PC2 and the IAQIs (the average values over the northern EC minus the average values over the southern EC) of O3 and PM2.5, respectively (Figure 6), the changes of O3 during 2015-2018 follow that of the PC2 closely (CC = −0.58, p < 0.01). By combining the changes of PM2.5, the CC with PC2 was significantly increased (CC = −0.83, p < 0.01). Furthermore, the mean and standard deviation of O3 concentration over EC has shown an increased intensity since 2017, especially in summer. This indicates that O3 has become more serious and widespread in EC during 2015-2018, which is also The spatial structure of the second mode of AQI (EOF2) in EC is a north-south dipole pattern, characterized by the EAQ "north good (poor) and south poor (good)" (Figure 4a). The explained variance of EOF2 was 14.4%. The corresponding PC time series of EOF2 (PC2) was not significantly characterized by cycles and trends (Figure 4c). However, since the spring of 2017, the amplitude of PC2 intensity has significantly increased. By calculating the regional average IAQI in northern and southern EC (with 35 • N as the boundary), respectively, the IAQIs of the six air pollutants all showed significant anti-correlations between northern and southern EC (CCs < −0.99, p < 0.01). Relative to EC-average IAQIs of CO, PM 10 , SO 2 , and NO 2 , the IAQIs in each month showed negative values over the southern EC and positive values over the northern EC ( Figure 5). This indicates that these air pollutants over northern EC are always greater than those over southern EC. However, the IAQI of PM 2.5 and O 3 averaged over southern and northern EC in each month showed alternating positive and negative changes relative to these averages over EC ( Figure 5). Therefore, the EOF2 of AQI was probably caused by the combined effect of PM 2.5 and O 3 .     In summary, the combined effect of O3 and PM2.5 is the reason for the AQI northsouth dipole pattern during 2015-2018. In recent years, although the governance of PM10 has had remarkable achievements, O3 and PM2.5 have gradually become important air pollutants that affect EAQ in EC. The PM2.5 and O3 of atmospheric composite pollution, and the north-south dipole pattern of AQI may become more and more serious in EC in the future.  In summary, the combined effect of O3 and PM2.5 is the reason for the AQI northsouth dipole pattern during 2015-2018. In recent years, although the governance of PM10 has had remarkable achievements, O3 and PM2.5 have gradually become important air pollutants that affect EAQ in EC. The PM2.5 and O3 of atmospheric composite pollution, and the north-south dipole pattern of AQI may become more and more serious in EC in the future. In summary, the combined effect of O 3 and PM 2.5 is the reason for the AQI north-south dipole pattern during 2015-2018. In recent years, although the governance of PM 10 has had remarkable achievements, O 3 and PM 2.5 have gradually become important air pollutants that affect EAQ in EC. The PM 2.5 and O 3 of atmospheric composite pollution, and the north-south dipole pattern of AQI may become more and more serious in EC in the future.

The Relationship between Climatic Conditions and the Major Modes of AQI
To understand the relationship between climatic conditions and the major modes of AQI in EC, the climatic element fields were regressed onto the PCs of AQI during 2015-2018 for discussion. The residual terms of the regression were needed.
The EOF1 of AQI in EC refers to as the "summer-winter" mode ( Figure 3a). Previous studies have shown that autumn and winter were heating seasons in northern China, and the emission of air pollutants in the north increased significantly in these seasons [49]. This means that human activities have made important contributions to this mode. However, the impact of climatic conditions in EC cannot be ignored. Figure 7a shows the PC1 of AQI in EC associated (regressed) with precipitation from 2015 to 2018. The spatial structure is characterized by a nearly uniform increase (or decrease) of precipitation across the EC (the black cross means an interval at a 99% confidence level, which is used hereafter). The interannual cycle of PC1 was more consistent with cycles of temperature and precipitation in EC, indicating that the monsoonal climate also has some impact on the EAQ. That is, in summer precipitation increases significantly, which has a strong capability for air purification. The same pattern is also reflected in vertical convection (Figure 7b). The "summer-winter" mode of AQI and the strength of convection change conversely, and significantly. This means that when the AQI decreases (increases) in summer, the atmospheric vertical motion in latitude increases (decreases). On the contrary, the thermal inversion layer is generally strengthened (weakened) in winter in EC, and the atmospheric vertical motion is reduced (enhanced). Furthermore, the AQI change in the mid-latitude region (30-40 • Figure 3a) is more sensitive to the strengthening of vertical convection, which is in line with the previous studies [20,56]. Atmospheric circulation is also one of the main climatic factors affecting the AQI. In summer, the climate in southern EC is affected by warm and humid air currents in the Indian Ocean, and the climate in northern EC is affected by water vapor transport from the nearby Pacific Ocean (Figure 7c,d). The enhanced land-sea thermal contrast may bring more water vapor from the Pacific to EC, especially in coastal areas, and the EAQ is relatively good. Conversely, with the dry and cold air in northern China passing through the EC in winter, the thermal inversion layer becomes thicker, and the vertical convection is weak, then the precipitation is reduced, and results in the EAQ being relatively poor in winter. Similarly, the low-level wind field also shows the typical characteristics of a sea-land breeze (Figure 7d). In summary, the surface and the low-level wind fields can bring warm and humid air from nearby oceans into EC, and the increased water vapor can deposit and absorb air pollutants, which leads to lower AQI in EC in summer. In winter, northern EC begins the heating season with a significant increase in pollution, and the East Asian winter monsoon climatic condition causes strong northerly winds, which bring air pollutants from northern to southern EC. In addition, due to the continuous consumption of energy in winter in northern EC, the overall AQI values over the entirety of EC are relatively higher in winter. The spatial structure of AQI EOF2 in EC is characterized by a north-south dipole pattern (Figure 4a). The associations of PC2 and various climatic element fields indicate that the north-south dipole pattern of AQI is closely related to the intensity of the Mongolia-Siberian high (Figure 8a). The dipole pattern, i.e., the AQI in the south is high and that in the north is low, usually occurs in autumn and winter (Figure 4c). It is usually associated with the gradual increase of the Mongolia-Siberian high (Figure 8a), which causes stronger cold airflow from north to south, bringing air pollutants from northern EC to southern EC (Figure 8c,d). Meanwhile, the climate in southern EC is not in the rainy season and the air purification capacity is relatively weak (Figure 8b). This would induce the pattern that the regional AQI is increased in southern EC, while being decreased in northern EC. On the contrary, the dipole pattern, i.e., the AQI in the north is high and that in the south is low, mostly happens in spring and summer (Figure 4c). In these seasons, the temperature begins to increase in EC and precipitation generally begins and continues in southern EC. The land-sea thermal gradient progressively increases between the temperature over the land in EC and the northern Indian Ocean. The low-level and surface winds are predominantly south-western (Figure 8c,d). Summer Asian monsoon precipitation lands in southern EC from May (Figure 8b), and the air purification capacity over southern EC is strengthened. At this time, the monsoon precipitation in northern EC has not yet arrived (generally concentrated in July and August) and the Mongolia-Siberian high is gradually weakening and disappearing. The seasonal northerly wind has decreased. Therefore, the AQI in the southern EC has decreased and that in the northern EC has increased. Therefore, AQI EOF2 is manifested as the influence of climate on the migration and purification of air pollutants. The AQI in southern EC is closely related to precipitation changes in spring and summer, and the AQI in northern EC is significantly related to the intensity of the Mongolia-Siberian high in autumn and winter. The spatial structure of AQI EOF2 in EC is characterized by a north-south dipole pattern (Figure 4a). The associations of PC2 and various climatic element fields indicate that the north-south dipole pattern of AQI is closely related to the intensity of the Mongolia-Siberian high (Figure 8a). The dipole pattern, i.e., the AQI in the south is high and that in the north is low, usually occurs in autumn and winter (Figure 4c). It is usually associated with the gradual increase of the Mongolia-Siberian high (Figure 8a), which causes stronger cold airflow from north to south, bringing air pollutants from northern EC to southern EC (Figure 8c,d). Meanwhile, the climate in southern EC is not in the rainy season and the air purification capacity is relatively weak (Figure 8b). This would induce the pattern that the regional AQI is increased in southern EC, while being decreased in northern EC. On the contrary, the dipole pattern, i.e., the AQI in the north is high and that in the south is low, mostly happens in spring and summer (Figure 4c). In these seasons, the temperature begins to increase in EC and precipitation generally begins and continues in southern EC. The land-sea thermal gradient progressively increases between the temperature over the land in EC and the northern Indian Ocean. The low-level and surface winds are predominantly south-western (Figure 8c,d). Summer Asian monsoon precipitation lands in southern EC from May (Figure 8b), and the air purification capacity over southern EC is strengthened. At this time, the monsoon precipitation in northern EC has not yet arrived (generally concentrated in July and August) and the Mongolia-Siberian high is gradually weakening and disappearing. The seasonal northerly wind has decreased. Therefore, the AQI in the southern EC has decreased and that in the northern EC has increased. Therefore, AQI EOF2 is manifested as the influence of climate on the migration and purification of air pollutants. The AQI in southern EC is closely related to precipitation changes in spring and summer, and the AQI in northern EC is significantly related to the intensity of the Mongolia-Siberian high in autumn and winter.

Conclusions
EAQ is one of the most concerning environmental issues in China, and understanding the characteristics of large-scale EAQ is essential for the implementation of the joint prevention and control of regional air pollution. In this study, we used the concentrations of six major air pollutants derived from about 1300 environmental monitoring stations in EC (from January 2015 to December 2018) to analyze the monthly average AQI according to national standards (GB3095-2012). The main seasonal spatial and temporal characteristics of the EAQ and its relationship with climatic conditions are as follows: 1. Generally, the AQI decreased in each season in EC from 2015 to 2018, but in spring and summer, the AQI distributions in north-central Inner Mongolia and Shanxi Province increased from 2015-2018. 2. From 2015 to 2018, the leading AQI mode in EC was characterized by the "summer and winter" mode, which was uniform across the EC. The AQI in EC generally increased during the winter half-year, and decreased during the summer half-year. In addition to human activities, climatic conditions, such as the East Asian monsoon, may play an important role in changes in EAQ patterns over the EC. 3. From 2015 to 2018, the second EOF mode of AQI in EC shows the characteristics of a north-south dipole pattern. That is, as the AQI in southern EC increased (decreases), it decreased (increases) in northern EC. PM2.5 and O3 are the main air pollutants associated with the AQI dipole pattern. In terms of a mechanism of climate conditions, the AQI in southern EC is closely related to precipitation changes in spring and summer, and the AQI in northern EC is significantly related to the intensity of the Mongolia-Siberian high in autumn and winter.
Air pollution is commonly affected by human activities and climatic factors. Human activities are the intrinsic factors, and climatic/meteorological conditions are the extrinsic factors. This was the point of departure for the present study. Based on the analysis of observational data, this study also shows that seasonal climatic conditions are one of the

Conclusions
EAQ is one of the most concerning environmental issues in China, and understanding the characteristics of large-scale EAQ is essential for the implementation of the joint prevention and control of regional air pollution. In this study, we used the concentrations of six major air pollutants derived from about 1300 environmental monitoring stations in EC (from January 2015 to December 2018) to analyze the monthly average AQI according to national standards (GB3095-2012). The main seasonal spatial and temporal characteristics of the EAQ and its relationship with climatic conditions are as follows: 1.
Generally, the AQI decreased in each season in EC from 2015 to 2018, but in spring and summer, the AQI distributions in north-central Inner Mongolia and Shanxi Province increased from 2015-2018.

2.
From 2015 to 2018, the leading AQI mode in EC was characterized by the "summer and winter" mode, which was uniform across the EC. The AQI in EC generally increased during the winter half-year, and decreased during the summer half-year. In addition to human activities, climatic conditions, such as the East Asian monsoon, may play an important role in changes in EAQ patterns over the EC.

3.
From 2015 to 2018, the second EOF mode of AQI in EC shows the characteristics of a north-south dipole pattern. That is, as the AQI in southern EC increased (decreases), it decreased (increases) in northern EC. PM 2.5 and O 3 are the main air pollutants associated with the AQI dipole pattern. In terms of a mechanism of climate conditions, the AQI in southern EC is closely related to precipitation changes in spring and summer, and the AQI in northern EC is significantly related to the intensity of the Mongolia-Siberian high in autumn and winter.
Air pollution is commonly affected by human activities and climatic factors. Human activities are the intrinsic factors, and climatic/meteorological conditions are the extrinsic factors. This was the point of departure for the present study. Based on the analysis of observational data, this study also shows that seasonal climatic conditions are one of the main factors affecting EAQ in EC, especially in the spreading and migration of air pollutants. However, due to the short period of EAQ monitoring in China, the data are not extensive enough to perform reasonable analyses for large-scale AQI changes on annual to decadal timescales. Nonetheless, this study can provide a reference for understanding the relationship between climatic conditions and EAQ.

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