Spatiotemporal Distribution of PM 2 . 5 and O 3 and Their Interaction During the Summer and Winter Seasons in Beijing , China

This study analyzed the spatiotemporal variations in PM2.5 and O3, and explored their interaction in the summer and winter seasons in Beijing. To this aim, hourly PM2.5 and O3 data for 35 air quality monitoring sites were analyzed during the summer and winter of 2016. Results suggested that the highest PM2.5 concentration and the lowest O3 concentration were observed at traffic monitoring sites during the two seasons. A statistically significant (p < 0.05) different diurnal variation of PM2.5 was observed between the summer and winter seasons, with higher concentrations during daytime summer and nighttime winter. Diurnal variations of O3 concentrations during the two seasons showed a single peak, occurring at 16:00 and 15:00 in summer and winter, respectively. PM2.5 presented a spatial pattern with higher concentrations in southern Beijing than in northern areas, particularly evident during wintertime. On the contrary, O3 concentrations presented a decreasing spatial trend from the north to the south, particularly evident during summer. In addition, we found that PM2.5 concentrations were positively correlated (p < 0.01, r = 0.57) with O3 concentrations in summer, but negatively correlated (p < 0.01, r = −0.72) with O3 concentrations in winter.


Introduction
In recent years, serious air pollution has become one of the most important environmental problems along with the development of society and economy, especially in economically developed cities, such as Beijing, Shanghai, Guangzhou, and other big cities [1].The major pollutants include particulate matter (e.g., PM 2.5 and PM 10 ), ground-level ozone (O 3 ), nitrogen oxides (NOx), sulfur dioxide (SO 2 ), carbon monoxide (CO), and Volatile Organic Compounds (VOCs) [2,3].Many studies have suggested that high levels of air pollutants can cause some adverse effects [4][5][6][7][8][9][10][11].The Chinese government has been dedicated to air pollution monitoring and control over last couple of decades, and ambient air quality has been widely measured in big cities since 1980s.In February 1988, the air pollution index (API), based on the concentrations of three atmospheric pollutants, namely total suspended particles (TSP), sulfur dioxide (SO 2 ), and nitrogen dioxide (NO 2 ), was introduced to assess the status of ambient air quality in China.On 5 June 2000, the China National Environmental Monitoring Center began to report API, defined according to the National Ambient Air Quality Standard (NAAQS-China, GB3095-1996) [12], by considering particulate matter (PM 10 ), sulfur dioxide (SO 2 ), and nitrogen dioxide (NO 2 ).From 18 January 2013, the Beijing municipal government has promulgated the new National Ambient Air Quality Standards (NAAQS-China, GB3095-2012) [13], and publicized real time monitoring data of Air Quality Index (AQI) and six major atmospheric pollutants (PM 2.5 , O 3 , PM 10 , CO, NO 2 , and SO 2 ).
In particular, regional complex atmospheric pollution characterized by high levels of O 3 and PM 2.5 has become a major challenge in China [14,15].O 3 is regarded as a secondary pollutant, primarily produced through a series of complex chemical reactions between Volatile Organic Compounds (VOCs) and nitrogen oxides (NO X ) under solar radiation in the wavelength range of 200 to 300 nm [16,17].The formation of O 3 is a very complicated process which is affected by many factors, including precursor emissions (e.g., NO X, CO, and VOCs), local climate conditions (e.g., temperature, relative humidity, solar radiation, and wind direction and speed), and atmospheric chemical processes [18][19][20][21].Fine particulate matter with an aerodynamic diameter of less than 2.5 µm (PM 2.5 ) can scatter and/or absorb solar radiation and influence the surface radiation budget [22], and these scattering and/or absorbing properties are inherently related to its chemical composition.The carbonaceous components (e.g., OC and EC) and water-soluble ions (e.g., SO 4 2− , NO 3 − , and NH 4 + ) are important chemical constituents of particulate matter [23,24].In general, PM 2.5 in urban areas originates mainly from emission sources such as biomass burning, traffic-related emissions, road dust, and industrial and agricultural production as well as regional transported aerosols [25,26], while the remaining portions originate from natural sources [27].To reduce emissions of air pollutants and improve air quality in China, the "Air Pollution Prevention and Control Action Plan" was formulated and released by the Chinese government on September 10, 2013.A recent study reported that its implementation has favored the reduction of air pollutants (PM 2.5 , PM 10 , SO 2 , and NO X ) [28], but O 3 concentration showed an increasing trend from 2013 to 2017 [29][30][31].Meanwhile, other studies also suggested that the implementation of air pollution control measures during some international events such as the APEC Meeting, the G20 summit, and some political events can result in the reduction in concentration of PM 2.5 , while O 3 concentration increased [32,33].As a result, it has been suggested that PM 2.5 and O 3 pollution and their interaction should be further investigated [34][35][36].In fact, although many studies have reported on PM 2.5 and O 3 pollution, most of them focused on spatiotemporal distribution and their relationships with weather conditions; research on the interaction of PM 2.5 and O 3 in the atmosphere is still rare.Feng et al. [37] pointed out that high aerosol concentrations can significantly decrease photolysis frequencies and reduce O 3 concentration by more than 50 µg/m 3 in the summertime.In addition, previous studies also reported a significant positive correlation between PM 2.5 and O 3 concentrations during the summertime [38,39].The discrepancy in these studies suggests that the interaction between O 3 and PM 2.5 in the atmosphere may vary over time across space, which should be further investigated.
In this paper, we analyzed hourly PM 2.5 and O 3 data collected at 35 air monitoring sites in Beijing during the summer (June-August 2016) and winter (December 2016-February 2017) seasons.Our primary objectives are thus to (i) analyze the spatiotemporal distribution of PM 2.5 and O 3 in Beijing and (ii) explore the interaction of O 3 and PM 2.5 in the atmosphere during the summer and winter seasons.

Study Area
Beijing, situated in the north of China, is the capital city of China.It covers a total area of about 16,800 square kilometers with a population of more than 20 million.The city is constituted of 16 administrative subdivisions including a total of 14 urban and suburban districts and two rural counties.The vegetation coverage is mainly distributed in Northern Beijing, including Changping, Yanqing, Huairou, Miyun, and Pinggu Districts.Southern Beijing presents lower vegetation coverage due to the influence of urbanization.Beijing has experienced continuous economic growth and rapid urbanization over the past several decades.However, the rapid development of the economy has generated a general decline in environmental quality; Beijing is currently considered as one of the cities with the most severe air pollution in China, and even globally [40].

Data Sources
In this study, PM 2.5 and O 3 hourly concentration data for Beijing were collected from a web platform (http://zx.bjmemc.com.cn/) made available from by Beijing Municipal Environmental Protection together with other air pollutants (PM 10 , SO 2 , NO 2 , CO) for 35 air monitoring sites (Figure 1) since 2013 [15].We divided the 35 air monitoring sites into four categories according to their functions (Table 1) [41].In particular, hourly PM 2.5 and O 3 concentrations were obtained for summer 2016 (June-August 2016) and winter 2016/2017 (December 2016-February 2017).In addition, the data of daily total radiation were obtained by the China Meteorological Data Sharing Service System Administration (available online: http://data.cma.cn/site/index.html)from December 2016 to February 2017.

Data Analysis
Daily (24-h) mean concentrations of PM 2.5 and O 3 concentrations were calculated in order to study the seasonal variations.We calculated the diurnal variations of PM 2.5 and O 3 concentrations during two seasons by averaging the concentrations at same time points of all days per season.The limit value for daily PM 2.5 concentration was defined as the new National Ambient Air Quality Standards (NAAQS-China, GB3095-2012) (75 µg/m 3 ).For O 3 , we used the daily maximum 8-h O 3 concentration (160 µg/m 3 ) and the daily maximum 1-h O 3 concentration (200 µg/m 3 ) (Available online: http://www.mee.gov.cn/).
For spatial comparison, we calculated the average concentrations of PM 2.5 and O 3 in each region of Beijing according to the location of each site [42].The regions of Beijing were defined based on geographical administrative divisions, as shown in Table 1.
We used Inverse Distance Weighted (IDW) to interpolate the air pollution data from the 35 monitoring stations to general spatial maps for spatial comparison.This method for spatial interpolation was applied to characterize spatial distributions of pollutant concentrations in some studies [43][44][45][46].The IDW-based interpolation maps of PM 2.5 and O 3 were produced using a random 70% of points [15] and validated with the remaining 30% of data.In addition, the root mean square error of PM 2.5 and O 3 concentration maps were calculated as an indication of the accuracy of the interpolation maps.Spatial distribution maps of PM 2.5 and O 3 concentrations were produced using ArcGIS ArcMap 10.0 software (Available online: http://www.esri.com/).
In order to further study secondary formation of PM 2.5 under different O 3,max in summer, CO is usually considered as a tracer of primary combustion source, and the PM 2.5 /CO ratio is an indicator to qualitatively infer the contribution of secondary aerosols [47].
In order to analyze the correlation between PM 2.5 and O 3 , the statistical analysis was performed using two-tailed test, which showed that these data did not obey normal distribution.Then, Pearson correlation coefficients were calculated in order to quantify the association between PM 2.5 and O 3 .All p-values were two-tailed and p < 0.05 was considered statistically significant.All analyses were carried out using the Origin software (OriginLab 9.0, Origin Lab Corporation, Northampton, MA, USA).

PM 2.5 and O 3 Concentrations
As shown in Figure 2, the average PM 2.5 concentrations during the summer and winter were 58.02 µg/m 3 and 108.13 µg/m 3 respectively, with 29 days in summer and 46 days in winter exceeding the current NAAQS limit for PM 2.5 .For O 3 , the average concentration in summer was 91.26 µg/m 3 with 44 days exceeding the current NAAQS limit, and the average concentration in winter was 30.33 µg/m 3 with no days over the limit.
error of PM2.5 and O3 concentration maps were calculated as an indication of the accuracy of the interpolation maps.Spatial distribution maps of PM2.5 and O3 concentrations were produced using ArcGIS ArcMap 10.0 software (Available online: http://www.esri.com/).
In order to further study secondary formation of PM2.5 under different O3, max in summer, CO is usually considered as a tracer of primary combustion source, and the PM2.5/CO ratio is an indicator to qualitatively infer the contribution of secondary aerosols [47].
In order to analyze the correlation between PM2.5 and O3, the statistical analysis was performed using two-tailed test, which showed that these data did not obey normal distribution.Then, Pearson correlation coefficients were calculated in order to quantify the association between PM2.5 and O3.All p-values were two-tailed and p < 0.05 was considered statistically significant.All analyses were carried out using the Origin software (OriginLab 9.0, Origin Lab Corporation, Northampton, MA, USA).

PM2.5 and O3 Concentrations
As shown in Figure 2, the average PM2.5 concentrations during the summer and winter were 58.02 μg/m 3 and 108.13 μg/m 3 respectively, with 29 days in summer and 46 days in winter exceeding the current NAAQS limit for PM2.5.For O3, the average concentration in summer was 91.26 μg/m 3 with 44 days exceeding the current NAAQS limit, and the average concentration in winter was 30.33 μg/m 3 with no days over the limit.PM2.5 presented the highest concentrations at traffic sites both in summer and in winter, followed by urban, suburban, and background sites (60.51, 54.28, and 54.94 μg/m 3 in summer and 105.51, 102.16, and 109.62 μg/m 3 in winter, respectively), as shown in Figure 3. On the contrary, O3 concentrations were lowest at traffic monitoring sites and higher at background/suburban sites.Specifically, O3 concentrations were 73.52 (24.04) μg/m 3 at traffic sites in the summer (winter), and reached 92.13 (29.76), 96.82 (34.02), and 94.31 (34.42) μg/m 3 at urban, suburban, and background sites, respectively.

Diurnal Variations in PM2.5 and O3 Concentrations
PM2.5 and O3 concentrations presented a distinct diurnal pattern over the seasons across sites (see Figure 4).During summer, PM2.5 concentrations were higher in the daytime than in the nighttime, with the highest concentrations always recorded at traffic sites.In general, during summer, PM2.5 concentrations began to rise early in the morning at around 4:00-5:00, reaching a peak around 8:00-9:00, successively decreasing until reaching the minimum at around 16:00-17:00, followed by a new rise until 21:00-23:00.Except for background sites, other sites showed similar characteristics, with two valley values at 4:00-5:00 and 16:00-17:00, and two peak values at 8:00-9:00 and 20:00-23:00, respectively; this is consistent with previous observations [48].The highest PM2.5 concentrations in urban, suburban, background, and traffic sites were 66.92 μg/m 3 , 58.66 μg/m 3 , 62.03 μg/m 3 , and 70.68 μg/m 3 , respectively.The lowest PM2.5 concentration in suburban and background sites was ~50 μg/m 3 , while it was approximately 55-60 μg/m 3 at the other two sites types.On the contrary, PM2.5 concentrations presented the opposite pattern during winter, with lower concentrations in the daytime than in the nighttime.For suburban and background sites, PM2.5 concentrations gradually reduced from 0:00-8:00, and then increased to a small peak at around 9:00-10:00; the lowest PM2.5 concentrations appeared at 15:00.For urban and traffic sites, PM2.5 concentrations began to decrease at 0:00-1:00 and reached the lowest level (85-95 μg/m 3 ) at 11:00, followed by a sharp increase to the maximum level (125-145 μg/m 3 ) around 23:00-1:00.   4).During summer, PM 2.5 concentrations were higher in the daytime than in the nighttime, with the highest concentrations always recorded at traffic sites.In general, during summer, PM 2.5 concentrations began to rise early in the morning at around 4:00-5:00, reaching a peak around 8:00-9:00, successively decreasing until reaching the minimum at around 16:00-17:00, followed by a new rise until 21:00-23:00.Except for background sites, other sites showed similar characteristics, with two valley values at 4:00-5:00 and 16:00-17:00, and two peak values at 8:00-9:00 and 20:00-23:00, respectively; this is consistent with previous observations [48].The highest PM 2.5 concentrations in urban, suburban, background, and traffic sites were 66.92 µg/m 3 , 58.66 µg/m 3 , 62.03 µg/m 3 , and 70.68 µg/m 3 , respectively.The lowest PM 2.5 concentration in suburban and background sites was ~50 µg/m 3 , while it was approximately 55-60 µg/m 3 at the other two sites types.On the contrary, PM 2.5 concentrations presented the opposite pattern during winter, with lower concentrations in the daytime than in the nighttime.For suburban and background sites, PM 2.5 concentrations gradually reduced from 0:00-8:00, and then increased to a small peak at around 9:00-10:00; the lowest PM 2.5 concentrations appeared at 15:00.For urban and traffic sites, PM 2.5 concentrations began to decrease at 0:00-1:00 and reached the lowest level (85-95 µg/m 3 ) at 11:00, followed by a sharp increase to the maximum level (125-145 µg/m 3 ) around 23:00-1:00.Although the O3 concentrations showed a statistically significant difference (p < 0.05) between the two seasons, traffic monitoring sites always present lower O3 concentrations than the other three site types.Interestingly, the four monitoring sites presented similar diurnal variation in the summer: the lowest O3 concentrations were observed at 7:00 in the morning.The lowest value was less than 20 μg/m 3 in traffic monitoring sites, while values were between 30 and 40 μg/m 3 at the other three site types.O3 concentrations for all four sites increased from 7:00 until around 15:00-16:00, when they reached their peak.The highest O3 concentration in urban, suburban, and background sites was around 160-170 μg/m 3 , while it was about 130 μg/m 3 at traffic sites.During the nighttime, O3 concentrations exhibited a decrease.After midnight, they presented a relatively stable phase, in agreement with previous observations [49].The pattern of the diurnal variations of O3 in winter was similar to that in summer, although with reduced fluctuation.The minimum O3 concentrations appeared at 08:00 (except for background sites) followed by an increase to the highest value at 15:00, and then a gradual decrease until the next morning.

Spatial Variations in PM2.5 and O3 Concentrations
Figure 5 illustrates average concentrations of PM2.5 and O3 in each region.PM2.5 concentration was significantly higher (p < 0.01) in winter than in summer, while O3 concentration was significantly lower (p < 0.01) in winter than in summer.The highest PM2.5 concentrations in the two seasons were observed in Southeast Beijing.PM2.5 concentrations were much lower in northwest and northeast of Beijing than in the rest of the regions in the two seasons.Northwest and northeast presented the lowest PM2.5 concentrations in summer and the highest O3 concentrations in winter.Although the O 3 concentrations showed a statistically significant difference (p < 0.05) between the two seasons, traffic monitoring sites always present lower O 3 concentrations than the other three site types.Interestingly, the four monitoring sites presented similar diurnal variation in the summer: the lowest O 3 concentrations were observed at 7:00 in the morning.The lowest value was less than 20 µg/m 3 in traffic monitoring sites, while values were between 30 and 40 µg/m 3 at the other three site types.O 3 concentrations for all four sites increased from 7:00 until around 15:00-16:00, when they reached their peak.The highest O 3 concentration in urban, suburban, and background sites was around 160-170 µg/m 3 , while it was about 130 µg/m 3 at traffic sites.During the nighttime, O 3 concentrations exhibited a decrease.After midnight, they presented a relatively stable phase, in agreement with previous observations [49].The pattern of the diurnal variations of O 3 in winter was similar to that in summer, although with reduced fluctuation.The minimum O 3 concentrations appeared at 08:00 (except for background sites) followed by an increase to the highest value at 15:00, and then a gradual decrease until the next morning.

Spatial Variations in PM 2.5 and O 3 Concentrations
Figure 5 illustrates average concentrations of PM 2.5 and O 3 in each region.PM 2.5 concentration was significantly higher (p < 0.01) in winter than in summer, while O 3 concentration was significantly lower (p < 0.01) in winter than in summer.The highest PM 2.5 concentrations in the two seasons were observed in Southeast Beijing.PM 2.5 concentrations were much lower in northwest and northeast of Beijing than in the rest of the regions in the two seasons.Northwest and northeast presented the lowest PM 2.5 concentrations in summer and the highest O 3 concentrations in winter.Overall, PM2.5 pollution was significantly higher in the southern part of Beijing than in the northern area (p < 0.01), and this pattern was especially obvious in winter.On the contrary, O3 concentrations presented decreasing concentrations from north to south, especially during summer.Overall, PM2.5 pollution was significantly higher in the southern part of Beijing than in the northern area (p < 0.01), and this pattern was especially obvious in winter.On the contrary, O3 concentrations presented decreasing concentrations from north to south, especially during summer.Overall, PM 2.5 pollution was significantly higher in the southern part of Beijing than in the northern area (p < 0.01), and this pattern was especially obvious in winter.On the contrary, O 3 concentrations presented decreasing concentrations from north to south, especially during summer.

Interaction between PM 2.5 and O 3 during the Summer and Winter Seasons
To better understand the interaction of PM 2.5 and O 3 in the atmosphere, we examined the correlations of daily mean PM 2.5 and O 3 concentrations (see Figure 7).The results indicated that PM 2.5 concentrations were positively correlated (p < 0.01, r = 0.57) with O 3 concentrations in summer, but negatively correlated (p < 0.01, r = −0.72)with O 3 concentrations in winter.
To study the effect of O 3 concentrations on PM 2.5 concentrations in summer, the diurnal variation of PM 2.5 was displayed under four O 3,max (daily maximum O 3 concentration) levels (see Figure 8) [38]: exceeding 200 µg/m 3 , from 160 to 200 µg/m 3 , from 100 to 160 µg/m 3 , and below 100 µg/m 3 .When O 3,max was greater than 200 µg/m 3 , PM 2.5 concentration was the highest and the diurnal variation was significant, with the maximum PM 2.5 concentration observed at 10:00.When O 3,max was between 160 and 200 µg/m 3 , PM 2.5 concentration was relatively high and the maximum concentration was observed at 08:00.When O 3,max was between 100 to 160 µg/m 3 and less than 100 µg/m 3 , PM 2.5 concentrations were low with the maximum hourly concentrations occurring at 23:00 and 00:00, respectively.On the whole, PM 2.5 levels were higher when O 3 levels were higher in summer.In addition, we also averaged the PM 2.5 /CO ratio under different O 3,max in summer, as shown in Figure 9.When O 3,max exceeded 200 µg/m 3 , 160-200 µg/m 3 , 100-160 µg/m 3 , and below 100 µg/m 3 , respectively, the average PM 2.5 /CO ratios were 0.083, 0.075, 0.055, and 0.048, respectively.

Interaction between PM2.5 and O3 during the Summer and Winter Seasons
To better understand the interaction of PM2.5 and O3 in the atmosphere, we examined the correlations of daily mean PM2.5 and O3 concentrations (see Figure 7).The results indicated that PM2.5 concentrations were positively correlated (p < 0.01, r = 0.57) with O3 concentrations in summer, but negatively correlated (p < 0.01, r = −0.72)with O3 concentrations in winter.

Interaction between PM2.5 and O3 during the Summer and Winter Seasons
To better understand the interaction of PM2.5 and O3 in the atmosphere, we examined the correlations of daily mean PM2.5 and O3 concentrations (see Figure 7).The results indicated that PM2.5 concentrations were positively correlated (p < 0.01, r = 0.57) with O3 concentrations in summer, but negatively correlated (p < 0.01, r = −0.72)with O3 concentrations in winter.

Discussion
Undoubtedly, in recent years, atmospheric compound pollution characterized by PM2.5 and O3 has caused negative effects on atmospheric visibility, human health, and climate change [50][51][52].Particulate matter in the atmosphere can not only affect air quality and human health, but also affects the earth's radiation budget [53].In addition, Feng et al. [54] showed that current O3 concentrations across China were sufficiently high to affect the yield and quality of agricultural crops.In addition, O3 can also make an impact on human health.Numerous epidemiological studies found that O3 had an association with premature mortality [55].A recent study reported that O3 has replaced PM2.5 as the chief pollutant in economically developed areas of China (e.g., Beijing-Tianjin-Hebei, Yangtze To study the effect of PM 2.5 concentration on O 3 concentration in winter, we analyzed the diurnal variation of O 3 under four PM 2.5 levels, selected on the basis of previous researches by Jia et al. [38]: exceeding 115 µg/m 3 , from 75 to 115 µg/m 3 , from 35 to 75 µg/m 3 , and below 35 µg/m 3 .When PM 2.5 level was greater than 115 µg/m 3 , O 3 concentration was low with small hourly variation.When PM 2.5 levels were between 75 to 115 µg/m 3 and between 35 to 75 µg/m 3 , O 3 concentrations further increased, with the previously described hourly variations.When PM 2.5 level was less than 35 µg/m 3 , O 3 concentration was the highest with large hourly variation.In general, O 3 levels were lower when PM 2.5 levels were higher in winter.We also averaged the daily total radiation under different PM 2.5 levels in winter, as shown in Figure 10.When To study the effect of PM2.5 concentration on O3 concentration in winter, we analyzed the diurnal variation of O3 under four PM2.5 levels, selected on the basis of previous researches by Jia et al. [38]: exceeding 115 μg/m 3 , from 75 to 115 μg/m 3 , from 35 to 75 μg/m 3 , and below 35 μg/m 3 .When PM2.5 level was greater than 115 μg/m 3 , O3 concentration was low with small hourly variation.When PM2.5 levels were between 75 to 115 μg/m 3 and between 35 to 75 μg/m 3 , O3 concentrations further increased, with the previously described hourly variations.When PM2.5 level was less than 35 μg/m 3 , O3 concentration was the highest with large hourly variation.In general, O3 levels were lower when PM2.5 levels were higher in winter.We also averaged the daily total radiation under different PM2.

Discussion
Undoubtedly, in recent years, atmospheric compound pollution characterized by PM2.5 and O3 has caused negative effects on atmospheric visibility, human health, and climate change [50][51][52].Particulate matter in the atmosphere can not only affect air quality and human health, but also affects the earth's radiation budget [53].In addition, Feng et al. [54] showed that current O3 concentrations across China were sufficiently high to affect the yield and quality of agricultural crops.In addition, O3 can also make an impact on human health.Numerous epidemiological studies found that O3 had an association with premature mortality [55].A recent study reported that O3 has replaced PM2.5 as the chief pollutant in economically developed areas of China (e.g., Beijing-Tianjin-Hebei, Yangtze

Discussion
Undoubtedly, in recent years, atmospheric compound pollution characterized by PM 2.5 and O 3 has caused negative effects on atmospheric visibility, human health, and climate change [50][51][52].Particulate matter in the atmosphere can not only affect air quality and human health, but also affects the earth's radiation budget [53].In addition, Feng et al. [54] showed that current O 3 concentrations across China were sufficiently high to affect the yield and quality of agricultural crops.In addition, O 3 can also make an impact on human health.Numerous epidemiological studies found that O 3 had an association with premature mortality [55].A recent study reported that O 3 has replaced PM 2.5 as the chief pollutant in economically developed areas of China (e.g., Beijing-Tianjin-Hebei, Yangtze River Delta, and Pearl River Delta) [56].In China, the annual average concentrations of O 3 and PM 2.5 were 49.00 µg/m 3 and 63.70 µg/m 3 in 2014, whereas in 2015, concentrations of O 3 and PM 2.5 increased and decreased by 7.40% and 12.00%, respectively [57].Therefore, studies on the evolution of PM 2.5 and O 3 and their relationships over time across space are essential for a better understanding and planning of policies for air pollution control.
In this study, we found that PM 2.5 concentrations in Beijing were lower in summer than in winter, while O 3 concentrations were higher in summer than in winter.Higher PM 2.5 concentrations were observed in winter, which is attributed to a large number of anthropogenic emissions from biomass burning and combustion of fossil fuels and unfavorable weather conditions (e.g., low temperature, weak solar radiation, and low boundary layer height) for the dilution and diffusion of PM 2.5 [8,34,58].Meteorological factors and O 3 precursors are linked to O 3 formation [59][60][61].Generally, higher temperatures and enhanced solar radiation can result in an elevated rate of photochemical reaction that promotes the production of ozone [7,62], which explains why O 3 concentration is higher in summer than in winter, as observed in this study.In addition, relative humidity is also an important factor determining O 3 levels.Enhanced relative humidity is favorable for aerosol formation [63], and high aerosol concentration can decrease photochemical activities and inhibit the formation of O 3 [64,65].The higher PM 2.5 concentrations observed at traffic monitoring sites during both the winter and summer seasons are linked to their vicinity to traffic emission sources [48], which increases PM 2.5 pollution.However, the results also indicated that traffic monitoring sites had much lower O 3 concentrations than other sites.This is due to the high concentration of primary NO present at traffic sites, which favor O 3 titration, while at rural areas reduced NO concentrations together with higher VOCs emissions (biogenic) favor O 3 formation [32,66].In particular, some studies showed that local O 3 production was controlled by the VOC/NO X ratio [67].In this case, O 3 concentration increases with increasing VOC concentration, but decreases with increasing NOx concentration.Usually, urban sites show relatively low VOC/NOx ratios due to high NO X emission [68].In addition, O 3 is also generated during the transport of its precursors from urban areas to suburban areas [69,70].Overall, the higher VOC/NOx ratios and/or transport of O 3 and precursors from urban areas can explain the observation of higher O 3 levels at suburban sites than at urban sites [71].
The higher daytime PM 2.5 concentrations in summer can be explained by the stronger atmospheric oxidation favoring secondary aerosol production [72][73][74].The forenoon peak of PM 2.5 concentrations is probably due to enhanced anthropogenic activity during rush hour [34], and the decreasing PM 2.5 concentrations after sunrise is mainly caused by the higher boundary layer height, which favors the dilution or diffusion of PM 2.5 .The boundary layer height usually remains high for longer time in summer afternoons, leading to a plateau in PM 2.5 concentrations [75].The early nighttime peak may result from a combination of the reduced boundary height and enhanced anthropogenic activity during rush hours [42].Overall, the lower nighttime PM 2.5 concentrations in summer are likely associated with the reductions in source activities and removal of particles by dry deposition [76], a mechanism enhanced for hygroscopic particles by the increase in relative humidity.In winter, PM 2.5 concentrations were higher during nighttime than during daytime.The boundary layer height usually decreases early in the afternoon due to decreased solar radiation, resulting in low PM 2.5 concentrations during noontime [76].Decreased boundary layer height and wind speed and increased source activity during the afternoon rush hour lead to higher PM 2.5 concentrations during evening hours [34,76].However, the diurnal variations of O 3 concentrations during the two seasons were almost the same, with a single-peak occurring at 16:00 and 15:00 in summer and winter, respectively.During both seasons, O 3 concentration peaks result from higher solar radiation and temperatures promoting photochemical formation of O 3 [72].
Spatially, PM 2.5 concentrations in Southern Beijing were higher than in Northern Beijing during the two seasons, and this pattern was especially evident in winter.This can result from regional transport of PM 2.5 from neighboring heavily polluted cities of Southern Beijing (e.g., Tianjin, Langfang, Baoding, and Zhangjiakou) [49,77,78], while Northern Beijing is surrounded by mountains.In addition, the lower population density with reduced human activity in Northern Beijing could contribute to lower PM 2.5 concentration [8].Conversely, O 3 concentrations showed a decreasing trend from the north to the south during the two seasons; this pattern was more evident in summer.A previous study showed that AOD in Southern Beijing was higher than in Northern Beijing [79], suggesting that surface solar radiation was stronger in Northern Beijing than in Southern Beijing, which favored photochemical O 3 production.Further, higher biogenic VOCs emissions due to the presence of vegetation in Northern Beijing can favor O 3 formation [80].On the other hand, the higher NOx concentrations observed in Southern Beijing favor O 3 titration [81,82].
A significant positive correlation in summer and a significant negative correlation in winter were observed between PM 2.5 and O 3 .The formation mechanisms and sources of PM 2.5 are very complex.Recent studies have suggested that secondary aerosols might dominate the PM 2.5 mass [83].We compared the PM 2.5 /CO ratio under different O 3,max in summer, and found that the PM 2.5 /CO ratio increased with rising in O 3,max , reflecting the fact that the contribution of secondary aerosols increased with rising in O 3,max .Thus, the significant positive correlation between PM 2.5 and O 3 is likely due to the fact that high O 3 concentrations with strong atmospheric oxidation in summer can promote the enhanced photochemical formation of secondary fine particulates [35,38].Additionally, we also compared the daily total radiation under different PM 2.5 levels in winter, and found that the daily total radiation decreased with rising in PM 2.5 , suggesting that high PM 2.5 concentrations in winter could potentially weaken the intensity of solar radiation reaching the Earth's surface and inhibit the formation of O 3 , leading to a negative relation between PM 2.5 and O 3 .

Conclusions
In summary, we analyzed the spatiotemporal distribution characteristics in PM 2.5 and O 3 in Beijing and investigated their interaction mechanisms in summer and winter based on hourly PM 2.5 and O 3 data for all 35 sites.The results showed that PM 2.5 concentrations were significantly lower (p < 0.05) in summer than in winter, while O 3 concentrations were significantly higher (p < 0.05) in summer than in winter.Among sites, traffic monitoring sites showed the highest PM 2.5 concentration and the lowest O 3 concentration during the two seasons.A remarkably different diurnal variation of PM 2.5 concentration was observed between summer and winter (p < 0.05), with the higher concentration during daytime summer and nighttime winter.Diurnal variations in O 3 concentration during the two seasons were observed as a single peak at 16:00 and 15:00 in summer and winter, respectively.Spatially, PM 2.5 concentrations decreased in gradient from the south to the north.On the contrary, O 3 concentrations in Northern Beijing were higher than those in Southern Beijing.
Our findings also confirmed that high O 3 concentrations with strong atmospheric oxidation in summer could promote the enhanced photochemical formation of secondary fine particulates, and the enhanced PM 2.5 levels in winter could potentially suppressed the intensity of surface solar radiation and inhibit the formation of O 3 .
According to the discussion and analysis of this study, we propose the following suggestions for future direction of air pollution control.First, although Beijing currently has 35 air quality monitoring stations, their spatial distribution is not uniform.As shown in Figure 1, most of the stations are sited in the urban areas of Beijing and thus more monitoring stations should be established in suburban areas.Second, previous studies have shown that air pollution in Beijing is strongly influenced by regional atmospheric pollutant transportation from the surrounding areas.Consequently, regional cooperation and joint defense and control are effective ways to control air pollution.Third, Beijing's government has implemented a series of air pollution control policies and regulations to improve air quality in recent years.These measures have favored the reduction of PM 2.5 , but O 3 concentration showed an increasing trend.Thus, it is recommended that more studies be conducted on the pollution characteristics, sources, and formation processes of O 3 .Finally, this study reveals the interaction between PM 2.5 and O 3 , and indicates that the coordinated control of PM 2.5 and O 3 is the future research direction in the field of atmospheric environment in Beijing.
In conclusion, the formation process and interaction of PM 2.5 and O 3 are very complex, as they are not only affected by primary emissions but also affected by other influencing factors including meteorological conditions, transformation of precursors, and atmospheric oxidant capacity, suggesting further investigation is needed.

Figure 1 .
Figure 1.Locations of 35 ambient air quality monitoring sites in Beijing.

Figure 2 .
Figure 2. The daily average concentrations of PM2.5 and O3 in the two seasons in Beijing (the green dots indicate O3 1h max and the red dots indicate O3 8h max).

Figure 2 .
Figure 2. The daily average concentrations of PM 2.5 and O 3 in the two seasons in Beijing (the green dots indicate O 3 1h max and the red dots indicate O 3 8h max).

PM 2 .
5 presented the highest concentrations at traffic sites both in summer and in winter, followed by urban, suburban, and background sites (60.51, 54.28, and 54.94 µg/m 3 in summer and 105.51, 102.16, and 109.62 µg/m 3 in winter, respectively), as shown in Figure 3. On the contrary, O 3 concentrations were lowest at traffic monitoring sites and higher at background/suburban sites.Specifically, O 3 concentrations were 73.52 (24.04) µg/m 3 at traffic sites in the summer (winter), and reached 92.13 (29.76), 96.82 (34.02), and 94.31 (34.42) µg/m 3 at urban, suburban, and background sites, respectively.

Figure 3 .
Figure 3. PM2.5 and O3 concentrations in different kinds of air monitoring sites ((a) PM2.5 concentration during summer, (b) PM2.5 concentration during winter, (c) O3 concentration during summer, and (d) O3 concentration during winter).The whiskers above and below the boxes indicate the 100th and 0th percentiles, the upper and lower boundaries of the boxes indicate the 75th and 25th percentiles, the squares indicate the mean values, and the lines in the boxes indicate the median values.

Figure 3 .
Figure 3. PM 2.5 and O 3 concentrations in different kinds of air monitoring sites ((a) PM 2.5 concentration during summer, (b) PM 2.5 concentration during winter, (c) O 3 concentration during summer, and (d) O 3 concentration during winter).The whiskers above and below the boxes indicate the 100th and 0th percentiles, the upper and lower boundaries of the boxes indicate the 75th and 25th percentiles, the squares indicate the mean values, and the lines in the boxes indicate the median values.

3. 2 .
Diurnal Variations in PM 2.5 and O 3 Concentrations PM 2.5 and O 3 concentrations presented a distinct diurnal pattern over the seasons across sites (see Figure

Figure 4 .
Figure 4. Average diurnal variations in PM 2.5 and O 3 concentrations: (a) PM 2.5 concentration during summer; (b) PM 2.5 concentration during winter; (c) O 3 concentration during summer; and (d) O 3 concentration during winter.

Figure 5 .
Figure 5. Average concentrations of PM 2.5 and O 3 in each region of Beijing.(a) Concentration of PM 2.5 .(b) Concentration of O 3 .

Figure 6 .
Figure 6.Spatial distribution in PM 2.5 and O 3 concentrations in Beijing: (a) PM 2.5 concentration during summer; (b) PM 2.5 concentration during winter; (c) O 3 concentration during summer; and (d) O 3 concentration during winter.

Figure 7 .
Figure 7. Scatter plots of PM2.5 vs. O3 during the summer and winter seasons (a) in summer and (b) in winter.

Figure 7 .
Figure 7. Scatter plots of PM 2.5 vs. O 3 during the summer and winter seasons (a) in summer and (b) in winter.

Figure 7 .
Figure 7. Scatter plots of PM2.5 vs. O3 during the summer and winter seasons (a) in summer and (b) in winter.

Figure 10 .
Figure 10.Average daily total radiation under different PM2.5 levels in winter.

Figure 9 .
Figure 9. Average PM 2.5 /CO ratio under different O 3,max in summer.

Figure 10 .
Figure 10.Average daily total radiation under different PM2.5 levels in winter.

Figure 10 .
Figure 10.Average daily total radiation under different PM 2.5 levels in winter.

Table 1 .
Category of each monitoring site in Beijing city.