Characteristics, Secondary Formation and Regional Contributions of PM 2.5 Pollution in Jinan during Winter

: Air pollution is an increasing threat to human health in China. In this study, daily PM 10 and PM 2.5 samples were collected simultaneously at the Jinan Environmental Monitoring Station (EMS)in Jinan, China from 15 November 2016 to 15 March 2017. The aim of this work was to improve the understanding of the characteristics and sources of air particles and determine di ﬀ erent levels of PM 2.5 pollution and its constituent elements, water-soluble ions and carbonaceous species. Nitrate (NO − ), organic materials (OM) and sulfate (SO 42 − ) were identiﬁed as the three main components of PM 2.5 pollution. With increasing pollution level, the contributions of SO 42 − , NO 3 − and NH 4 + increased at greater rates, unlike that of OM. The proportion of SO 42 − exceeded that of NO 3 − and became predominant in severe PM 2.5 pollution (SP; 250 µ g m − 3 ≤ PM 2.5 ≤ 500 µ g m − 3 ). This work demonstrates that SO 42 − has a dominant role in SP level and, consequently, requires greater research attention. It is demonstrated that relative humidity (RH) enhances the rate of sulfate formation more than that of nitrate. Therefore, under the current Chinese emergency response measures, it is necessary to further reduce emissions of SO 2 and NO 2 . Four clusters of backward trajectories identiﬁed dominant pollution vectors originating from highly industrialized areas that exacerbate the poor air quality in Jinan. It is, therefore, necessary to undertake regional control measures to reduce pollutant emissions.


Introduction
In recent years there have been frequent heavy pollution days in the central and eastern regions of China. Consequently, Chinese authorities and the public are increasingly demanding more stringent prevention and control measures for atmospheric pollution [1]. Serious air pollution not only lowers air quality but also negatively affects human health, both physically and mentally. Therefore, China's continuously severe air pollution (SP) has caused wide public concern [2]. There have been many studies on air pollution in major areas of China [3][4][5][6][7][8][9][10][11], of which a large number have proven that secondary inorganic species play an important role in haze formation [3,4,[9][10][11].
Jinan is located in central Shandong Province and is its capital city. The city extends south to Mount Tai (which is the province's highest peak at 1000 m) and north to the Yellow River. Jinan City
The pollution conditions in Jinan were divided into five categories as follows. Using the Technical Regulation on Ambient Air Quality Index, we categorized PM2. 5  Atmospheric particulate matter mass reconstruction can estimate the effects of aerosols from different sources on ambient air quality based on the proportions of compounds in different components [22]. In this study, the chemical mass reconstruction method was used to classify organic matter (OM), EC, mineral dust (MD), trace elements (TE), SO4 2− , NO3 − , NH4 + and Cl − in PM2.5. The mass of MD was estimated on the basis of oxides of Al, Si, Ca, Fe, Ti, K and Mg as follows [11]: The OM content included undetected H, S, N and O and was estimated by multiplying the OC content by a conversion factor (CF) corresponding to the organic molecular carbon weight per carbon weight. The CF ranged from 1.6 to 2.1. In previous work [22][23][24], CFs of 1.6 ± 0.2 appear to be more accurate for urban aerosols [23]. In Jinan, CFs of 1.4 and 1.8 have been used for urban research before [16,25]. The OM was estimated to be 1.4 times the OC according to our previous study in Jinan [16].
In MD, except for the above elements, the sum of all other element concentrations was defined as the TE concentration. "Other" were considered unidentified mass, measurement or experimental errors et al [26].
Atmospheric particulate matter mass reconstruction can estimate the effects of aerosols from different sources on ambient air quality based on the proportions of compounds in different components [22]. In this study, the chemical mass reconstruction method was used to classify organic matter (OM), EC, mineral dust (MD), trace elements (TE), SO 4 2− , NO 3 − , NH 4 + and Cl − in PM 2.5 .
The mass of MD was estimated on the basis of oxides of Al, Si, Ca, Fe, Ti, K and Mg as follows [11]: The OM content included undetected H, S, N and O and was estimated by multiplying the OC content by a conversion factor (CF) corresponding to the organic molecular carbon weight per carbon weight. The CF ranged from 1.6 to 2.1. In previous work [22][23][24], CFs of 1.6 ± 0.2 appear to be more accurate for urban aerosols [23]. In Jinan, CFs of 1.4 and 1.8 have been used for urban research before [16,25]. The OM was estimated to be 1.4 times the OC according to our previous study in Jinan [16]. In MD, except for the above elements, the sum of all other element concentrations was defined as the TE concentration. "Other" were considered unidentified mass, measurement or experimental errors et al [26].  It is obvious that there are two extreme PM 2.5 peaks in Figure 2 coinciding with high concentrations of SO 4 2− . The general features of the main PM 2.5 components during the sampling periods are listed in m −3 , and the average mass concentrations of PM2.5 during the sampling period were 107.1 ± 75.4 μg m -3 . It is obvious that there are two extreme PM2.5 peaks in Figure 2 coinciding with high concentrations of SO4 2− . The general features of the main PM2.5 components during the sampling periods are listed in Table 1. The main component in PM2.5 was NO3 − , with an average concentration ± standard deviation of 21.9 ± 15.6 μg m −3 . The second-highest proportions were of SO4 2− and OM, at concentrations of 19 ± 21.2 and 19 ± 11.4 μg m −3 . Therefore, NO3 − , SO4 2− and OM are the three major species of PM2.5. On average, PM2.5 accounted for 62% of PM10, which indicates that atmospheric particulate matter pollution was mainly dominated by PM2.5 in Jinan during winter. The high percentage of PM2.5 infers that the primary source is of combustion/secondary aerosol origin, as there was a low fraction of crustal material. Some 110 valid samples were collected during the sampling period, and there were 45, 30, 16, 11 and 8 days of C, S, M, HP and SP PM2.5 pollution, respectively. Eight heavy pollution days were identified as18-20 November 2016 and 2, 4-6, and 18 January 2017. Figure 3 displays a PM2.5 composition spectrum characteristic diagram. NO3 − (21.9 g m −3 ), SO4 2− (19.0 μg m −3 ), OC (13.6 μg m −3 ) and NH4 + (11.6 μg m −3 ) were the four dominating components and comprised nearly 60% of the total mass. In addition, a correlation analysis between PM2.5, gaseous pollutants and meteorological parameters is shown in Table S1. PM2.5 exhibited good positive relationships with CO, SO2, NO2 and RH, but not O3, which is consistent with previous results [8,27].

General Characteristics of PM 2.5 Pollution in Winter
The mass closures of PM2.5 are presented in Figure 4. According to the results of PM2.5 reconstruction, NO3 − , OM and SO4 2− were the three main components in the constructed PM2.5 and, on average, accounted for 19.9%, 19.1% and 15.7% of the total PM2.5, respectively. There were almost the same proportions of MD and NH4 + , at 10.2% and 9.9%, respectively. The small fraction of MD is consistent with the inferred result that there was a high percentage of PM2.5. The proportion of EC (5.0%) in PM2.5 was slightly greater than that of Cl − (4.1%). Other unanalyzed substances, such as water, or experimental analysis errors [26], accounted for 14.9% of the PM2.5.    and comprised nearly 60% of the total mass. In addition, a correlation analysis between PM 2.5 , gaseous pollutants and meteorological parameters is shown in Table S1. PM 2.5 exhibited good positive relationships with CO, SO 2 , NO 2 and RH, but not O 3 , which is consistent with previous results [8,27].
Atmosphere 2020, 11, x FOR PEER REVIEW 5 of 13    The mass closures of PM 2.5 are presented in Figure 4. According to the results of PM 2.5 reconstruction, NO 3 − , OM and SO 4 2− were the three main components in the constructed PM 2.5 and, on average, accounted for 19.9%, 19.1% and 15.7% of the total PM 2.5 , respectively. There were almost the same proportions of MD and NH 4 + , at 10.2% and 9.9%, respectively. The small fraction of MD is consistent with the inferred result that there was a high percentage of PM 2.5 . The proportion of EC (5.0%) in PM 2.5 was slightly greater than that of Cl − (4.1%). Other unanalyzed substances, such as water, or experimental analysis errors [26], accounted for 14.8% of the PM 2.5 .

Chemical Compositions of Different Pollution Levels
Atmosphere 2020, 11, x FOR PEER REVIEW 5 of 13

Chemical Compositions of Different Pollution Levels
The mass concentrations of OM, EC, MD, TE, SO 4 2− , NO 3 − , NH 4 + and Cl − increased with increasing concentrations of PM 2.5 ; however, their proportions varied. Figure 5 lists the average percentages of the main constituents in PM 2.5 at different pollution levels. Differences from the C-level to SP-level are noticeable, especially for OM, SO 4 2− and NO 3 − . As shown in Figure 5, the average proportion of OM declined consistently as the pollution level increased. This decrease in OM from the C to HP level in Jinan in winter is the same at that observed in Beijing [28,29], indicating that there is a decreasing contribution of carbonaceous matter with increasingly severe pollution levels.
The average percentages of NO 3 − in the five pollution categories (from lowest to highest) were 16.9%, 21.8%, 22.9%, 24.8% and 15.9%. The peak value coincides with HP level pollution, which suggests an inverted V-shaped trend in the dataset. Figure S1 emphasizes the changes in the proportions of SO 4 2− and NO 3 − over the whole sampling period. Figure S1b also shows an inverted V-shaped curve. The potential origin of this unusual relationship is that higher PM Other proportion had small increases up to the HP level and more obvious increases from HP to SP. These may be attributed to measurement errors, improper multiplier(s), missing source(s) and particle-bound water [26]. PM 2.5 pollution becomes more hygroscopic, with abundant inorganic water-soluble ions, as RH increases [30]. Particle-bound water in PM 2.5 maybe the "other" species that caused a significant increase in its concentration at high PM 2.5 concentrations. All these data indicate that the formation of heavy haze is mainly promoted by secondary inorganic species, especially SO 4 2− .

Meteorological Conditions
Wind speed (WS) and relative humidity (RH) are the most important meteorological factors influencing the mass concentrations of PM2.5, as demonstrated in previous studies [10,11]. In this study, RH and WS had great impacts on PM2.5, which is consistent with the study of Tian [11]. PM2.5 concentrations (Figure 6a) showed a decreasing dependence on WS but increasing dependence on RH ( Figure 6b). As shown in Figure 6, WS was an influential factor and had a negative relationship with PM2.5 concentration, with a Spearman correlation coefficient (ρ) of 0.2. Compared with WS, RH had a greater impact on PM2.5 concentration and was positively correlated with it (ρ = 0.6). Although eight serious PM 2.5 pollution days were observed, all the data show that sulfate contributed most to severe PM 2.5 pollution. The molar ratio of ammonium to sulfate was used to infer their existent forms in the particles. As shown in Table S2, the concentration of NH 4 + was strongly

Meteorological Conditions
Wind speed (WS) and relative humidity (RH) are the most important meteorological factors influencing the mass concentrations of PM 2.5 , as demonstrated in previous studies [10,11]. In this study, RH and WS had great impacts on PM 2.5 , which is consistent with the study of Tian [11]. PM 2.5 concentrations (Figure 6a) showed a decreasing dependence on WS but increasing dependence on RH ( Figure 6b). As shown in Figure 6, WS was an influential factor and had a negative relationship with PM 2.5 concentration, with a Spearman correlation coefficient (ρ) of 0.2. Compared with WS, RH had a greater impact on PM 2.5 concentration and was positively correlated with it (ρ = 0.6).

Meteorological Conditions
Wind speed (WS) and relative humidity (RH) are the most important meteorological factors influencing the mass concentrations of PM2.5, as demonstrated in previous studies [10,11]. In this study, RH and WS had great impacts on PM2.5, which is consistent with the study of Tian [11]. PM2.5 concentrations (Figure 6a) showed a decreasing dependence on WS but increasing dependence on RH ( Figure 6b). As shown in Figure 6, WS was an influential factor and had a negative relationship with PM2.5 concentration, with a Spearman correlation coefficient (ρ) of 0.2. Compared with WS, RH had a greater impact on PM2.5 concentration and was positively correlated with it (ρ = 0.6).   The ratio of SO 4 2− obviously increased with increasing RH in PM 2.5 , with ρ = 0.7. As shown in Figure 7b, the low coefficient values seem to indicate that RH is slightly positively related to nitrate. Aqueous reactions may make a more significant contribution to the formation of sulfate than of nitrate with increasing PM 2.5 concentration. Hence, high RH was more favorable for the formation of sulfate. In this study, Figure 7 demonstrates that hygroscopic secondary inorganic ions significantly increased with increasing RH during winter, which is consistent with previous studies [10,32].  Figure 7 shows the relationships between RH and the proportions of SO4 2− and NO3 − in PM2.5. The ratio of SO4 2− obviously increased with increasing RH in PM2.5, with ρ = 0.7. As shown in Figure  7b, the low coefficient values seem to indicate that RH is slightly positively related to nitrate. Aqueous reactions may make a more significant contribution to the formation of sulfate than of nitrate with increasing PM2.5 concentration. Hence, high RH was more favorable for the formation of sulfate. In this study, Figure 7 demonstrates that hygroscopic secondary inorganic ions significantly increased with increasing RH during winter, which is consistent with previous studies [10,32].

Secondary Formation
Sulfate and nitrate mainly originate from the conversion of SO2 and NO2 gaseous precursors [33]. The sulfate oxidation rate (SOR) and nitrate oxidation rate (NOR) were used to evaluate the degrees of SO2 and NO2 conversion in the atmosphere. The larger the SOR and NOR values, the more SO2 and NO2 are converted to sulfate and nitrate. A value of SOR > 0.1 is often used to indicate the presence of secondary conversion [34,35]. SOR and NOR were calculated based on the following formulas: SOR = n-SO4 2− /(n-SO4 2− + n-SO2) and NOR = n-NO3 − /(n-NO3 − + n-NO2), where n-SO4 2− , n-SO2, n-NO3 − and n-NO2 represent the molecular concentrations of sulfate, sulfur dioxide, nitrate and nitrogen dioxide, respectively. Figure 8 presents the relationships between the daily average values of SOR, NOR, RH and O3 in PM2.5. SOR and NOR demonstrate opposite trends with RH and O3. The concentration of O3 decreased with increasing RH, indicating weak photochemical reactivity. At RH < 40%, there was a lesser influence of SOR, but SOR increased rapidly from <0.2 up to nearly 0.5 at RH > 60% in PM2.5. At the same time, we quantified the molecular ratio of SO4 2− to SO2, which reflects sulfur partitioning between the particle and gas phases in PM2.5. This had the same trend as that of SOR to RH, as shown in Figure S2. Similar SO4 2− evolution has been observed in Xi'an and Beijing [36]. Figure 8(b) shows that O3 contributed more to NOR than RH in PM2.5, as the trend in O3 was consistent with that of NOR at RH < 60%.
Meteorological conditions were recorded by the EMS during the sampling period. The average ozone (O3) concentrations at the EMS were 40.4, 45.5, 29.6, 29.0 and 7.7 μg m −3 for pollution levels of C to SP, while RH levels were 40.1%, 47.4%, 59.3%, 65.1% and 81.5%, respectively. The average SOR values from the C to SP pollution levels were 0.09, 0.13, 0.20, 0.24 and 0.33, while those of NOR were 0.11, 0.21, 0.24, 0.30 and 0.27, respectively. We can see that the formation rate of NO3 − was faster than that of SO4 2− from the C-level to HP-level; however, differences occurred at the SP-level. With decreases in O3 concentration accompanied by increases in RH, the NO3 − formation rate was slower than that of SO4 2− at the SP-level. SO4 2− can be formed through oxidation of SO2 by hydroxyl radicals in a gas phase reaction or by oxidants (e.g. H2O2, O3) in an aqueous phase reaction [37]. Nitrate is

Secondary Formation
Sulfate and nitrate mainly originate from the conversion of SO 2 and NO 2 gaseous precursors [33]. The sulfate oxidation rate (SOR) and nitrate oxidation rate (NOR) were used to evaluate the degrees of SO 2 and NO 2 conversion in the atmosphere. The larger the SOR and NOR values, the more SO 2 and NO 2 are converted to sulfate and nitrate. A value of SOR > 0.1 is often used to indicate the presence of secondary conversion [34,35]. SOR and NOR were calculated based on the following formulas: SOR = n-SO 4 2− /(n-SO 4 2− + n-SO 2 ) and NOR = n-NO 3 − /(n-NO 3 − + n-NO 2 ), where n-SO 4 2− , n-SO 2 , n-NO 3 − and n-NO 2 represent the molecular concentrations of sulfate, sulfur dioxide, nitrate and nitrogen dioxide, respectively. Figure 8 presents the relationships between the daily average values of SOR, NOR, RH and O 3 in PM 2.5 . SOR and NOR demonstrate opposite trends with RH and O 3 . The concentration of O 3 decreased with increasing RH, indicating weak photochemical reactivity. At RH < 40%, there was a lesser influence of SOR, but SOR increased rapidly from <0.2 up to nearly 0.5 at RH > 60% in PM 2.5 . At the same time, we quantified the molecular ratio of SO 4 2− to SO 2 , which reflects sulfur partitioning between the particle and gas phases in PM 2.5 . This had the same trend as that of SOR to RH, as shown in Figure S2. Similar SO 4 2− evolution has been observed in Xi'an and Beijing [36]. Figure 8b shows that O 3 contributed more to NOR than RH in PM 2.5 , as the trend in O 3 was consistent with that of NOR at RH < 60%.
Atmosphere 2020, 11, x FOR PEER REVIEW 9 of 13 predominantly formed by the gas-phase reaction of NO2 and OH radicals and by heterogeneous reactions of nitrate radicals (NO3) [38]. With increases in pollution level, photochemical reactions in the gas phase are suppressed [10]. RH plays an important role in SO2 conversion, which is consistent with the relationship between SO4 2− and RH (Figures 7a and S2). It is further suggested that this part of SO4 2− increases due to aqueous phase secondary formation at the SP-level. However, the oxidation of NO2 was weakened. Concentrations of NO2 and SO2 increases by 30% and 50% respectively from the HP to SP level. Figure 5 shows that nitrate and sulfate concentrations increased by 13% and 131% from the HP to SP level, respectively. NO2 is not only a precursor of nitrate but is also an important oxidant in sulfate formation during severe pollution days [39]. This can also explain the trends in nitrate and sulfate concentrations from the HP to SP level. Thus, the dual functions of RH and NO2 accelerate the formation of sulfate during SP-level pollution.

Regional Transport
In order to study the impact of regional transport on Jinan (36°39'47" N, 117°3'18" E) at the different pollution levels, 24 h backward trajectories from 15 November 2016 to 15 March 2017 were Meteorological conditions were recorded by the EMS during the sampling period. The average ozone (O 3 ) concentrations at the EMS were 40.4, 45.5, 29.6, 29.0 and 7.7 µg m −3 for pollution levels of C to SP, while RH levels were 40.1%, 47.4%, 59.3%, 65.1% and 81.5%, respectively. The average SOR values from the C to SP pollution levels were 0.09, 0.13, 0.20, 0.24 and 0.33, while those of NOR were 0.11, 0.21, 0.24, 0.30 and 0.27, respectively. We can see that the formation rate of NO 3 − was faster than that of SO 4 2− from the C-level to HP-level; however, differences occurred at the SP-level. With decreases in O 3 concentration accompanied by increases in RH, the NO 3 − formation rate was slower than that of SO 4 2− at the SP-level. SO 4 2− can be formed through oxidation of SO 2 by hydroxyl radicals in a gas phase reaction or by oxidants (e.g. H 2 O 2 , O 3 ) in an aqueous phase reaction [37]. Nitrate is predominantly formed by the gas-phase reaction of NO 2 and OH radicals and by heterogeneous reactions of nitrate radicals (NO 3 ) [38]. With increases in pollution level, photochemical reactions in the gas phase are suppressed [10]. RH plays an important role in SO 2 conversion, which is consistent with the relationship between SO 4 2− and RH (Figure 7a and Figure S2). It is further suggested that this part of SO 4 2− increases due to aqueous phase secondary formation at the SP-level. However, the oxidation of NO 2 was weakened. Concentrations of NO 2 and SO 2 increases by 30% and 50% respectively from the HP to SP level. Figure 5 shows that nitrate and sulfate concentrations increased by 13% and 131% from the HP to SP level, respectively. NO 2 is not only a precursor of nitrate but is also an important oxidant in sulfate formation during severe pollution days [39]. This can also explain the trends in nitrate and sulfate concentrations from the HP to SP level. Thus, the dual functions of RH and NO 2 accelerate the formation of sulfate during SP-level pollution.

Regional Transport
In order to study the impact of regional transport on Jinan (36 • 39 47" N, 117 • 3 18" E) at the different pollution levels, 24 h backward trajectories from 15 November 2016 to 15 March 2017 were calculated. An altitude of 100 m AGL was set as the average flow field of the atmospheric boundary layer of the study area and start times of 00:00, 06:00, 12:00 and 18:00 UTC each day were used. The backward trajectory clusters were calculated by the TrajStat model, which is a plugin in MeteoInfo [40], and 472 effective trajectories of the simulation were clustered. There were 286 "polluted trajectories" during days where the PM 2.5 concentration was >75 µg m −3 . Four main transmission paths were obtained. The hourly PM 2.5 concentration data were imported into the TrajStat model obtained from the Municipal Environmental Monitoring Centre.
As shown in Figure 9 and Table S3,  and its surrounding cities can influence each other. Previous studies have shown that Shandong Province is the most important contributor to particulate matter pollution in Tianjin [41]. In April 2017, China implemented the National Research Program for Key Issues in Air Pollution Control, China. The main causes of heavy pollution in the BTH region and surrounding areas have been determined to be local accumulation, regional transport and secondary formation [42]. This heavy pollution episode in Jinan is a comprehensive result of multiple causes rather than a single one. Hence, to effectively reduce pollution, regional joint prevention is necessary and inevitable.

Conclusions
The chemical characteristics and formation of five different PM2.5 pollution levels were investigated in winter, from 15 November 2016 to 15 March 2017, in Jinan, along with local meteorological parameters. Sulfate was observed to have a dominant role in severe PM2.5 pollution.
Daily PM2.5 concentrations varied from 16.5 to 413.6 μg m −3 with a mean ± SD of 107.1 ± 75.4 μg m −3 . The three main components in PM2.5 pollution were NO3 − , OM and SO4 2− which, on average, accounted for 19.9%, 19.1% and 15.7% of the total PM2.5, respectively. The fraction of OM decreased (from 21.2% to 14.6%) and the proportion of SO4 2− increased (from 13.6% to 24.6%) as the pollution level increased from the C-level to SP-level, while that of NO3 − had an inverted V-shaped relationship with pollution level and peaked with HP-level pollution (NO3 − proportions = 16.9%, 21.8%, 22.9%, 24.8% and 15.9% with increasing pollution levels, respectively). The most obvious PM2.5 characteristic was that the concentration and mass percentage of SO4 2− increased significantly from the HP to SP level and far exceeded those of NO3 − , thereby becoming the largest constituent of SP-level PM2.5 pollution.
The RH increased according to PM2.5 pollution level, which is favorable for the formation of inorganic species. As the RH increased, so did the proportion of sulfate in PM2.5. The proportion of NO3 − appeared to have a minor increasement as RH increases. SOR and NOR demonstrated opposite variations with RH and O3. The linear relationship between SOR and RH was similar to that between Winter air pollution in Jinan is likely to mostly come from these surrounding areas, given their high-emission intensities. Jinan is located in the central part of Shandong Province between 36 • 01 and 37 • 32 N latitude and 116 • 11 to 117 • 44 E longitude, which is about 420 km from Beijing. Jinan and its surrounding cities can influence each other. Previous studies have shown that Shandong Province is the most important contributor to particulate matter pollution in Tianjin [41]. In April 2017, China implemented the National Research Program for Key Issues in Air Pollution Control, China. The main causes of heavy pollution in the BTH region and surrounding areas have been determined to be local accumulation, regional transport and secondary formation [42]. This heavy pollution episode in Jinan is a comprehensive result of multiple causes rather than a single one. Hence, to effectively reduce pollution, regional joint prevention is necessary and inevitable.

Conclusions
The chemical characteristics and formation of five different PM 2.5 pollution levels were investigated in winter, from 15  The RH increased according to PM 2.5 pollution level, which is favorable for the formation of inorganic species. As the RH increased, so did the proportion of sulfate in PM 2.5 . The proportion of NO 3 − appeared to have a minor increasement as RH increases. SOR and NOR demonstrated opposite variations with RH and O 3 . The linear relationship between SOR and RH was similar to that between SO 4 2− and RH in PM 2.5 . The variation in NOR was the same as that of nitrate. At the HP to SP level, RH was more favorable to sulfate conversion than to nitrate. RH increased with decreases in O 3 , which is more favorable for the aqueous phase formation of sulfate. At the same time, NO 2 , an important oxidant, was favorable to sulfate formation, leading to faster sulfate production and worsening air pollution in a repeating cycle. Therefore, according to current emergency response measures, it is necessary to further reduce the emissions of SO 2 and NO 2 . Finally, 24 h air mass backward trajectories were determined to study the impact of regional transport on Jinan at different pollution levels. Four clusters were obtained, indicating that 26.3% of the air mass came from the north and was relatively clean. The other three clusters, which were the main transmission paths affecting air quality in Jinan, came either from a coastal area or highly industrialized cities. They accounted for 73.7% of the total air trajectories and had slow air mass movement.
In general, RH corresponds with higher PM 2.5 concentrations and more gaseous pollutants, leading to faster sulfate production and more severe haze pollution in Jinan in winter. Meanwhile, pollution in Jinan is also affected by regional transport, especially from other cities in Shandong Province. The Chinese government has begun to undertake regional collaboration to effectively control pollution emissions in the BTH and surrounding cities. As an important city in the BTH pollution transmission channel, Jinan should start reducing its emissions.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4433/11/3/273/s1. Table S1. Correlation analysis between PM 2.5 , gaseous pollutants and meteorological parameters; Table S2. Matrix of correlation coefficients between different ions in PM 2.5 ; Table S3. The concentration and proportion of PM 2.5 in each cluster; Figure S1. Relationship between the proportion of SO 4 2− , NO 3 − and PM 2.5 concentrations; Figure S2.