Factors Influencing O3 Concentration in Traffic and Urban Environments: A Case Study of Guangzhou City

Ozone (O3) pollution is a serious issue in China, posing a significant threat to people’s health. Traffic emissions are the main pollutant source in urban areas. NOX and volatile organic compounds (VOCs) from traffic emissions are the main precursors of O3. Thus, it is crucial to investigate the relationship between traffic conditions and O3 pollution. This study focused on the potential relationship between O3 concentration and traffic conditions at a roadside and urban background in Guangzhou, one of the largest cities in China. The results demonstrated that no significant difference in the O3 concentration was observed between roadside and urban background environments. However, the O3 concentration was 2 to 3 times higher on sunny days (above 90 μg/m3) than on cloudy days due to meteorological conditions. The results confirmed that limiting traffic emissions may increase O3 concentrations in Guangzhou. Therefore, the focus should be on industrial, energy, and transportation emission mitigation and the influence of meteorological conditions to minimize O3 pollution. The results in this study provide some theoretical basis for mitigation emission policies in China.


Introduction
Air pollution has become a crucial issue in China due to rapid economic development [1]. The Chinese government has exerted a significant effort to reduce air pollution in recent years. As a result, fine particulate matter (PM 2.5 ) has significantly decreased due to strict emission mitigation policies [2]. Ozone (O 3 ) has become the most prevalent pollutant in China. The O 3 concentration has increased by 10.6% from 2015 to 2021 in 339 [3,4]. Excessive exposure to O 3 can be extremely harmful to human health, causing substantial damage and irritation to the eyes, respiratory tract, and lungs [5][6][7].
Many studies have focused on O 3 pollution in China, investigating the spatiotemporal variations [8][9][10][11][12], secondary formation mechanism [13][14][15], emission sources [16][17][18][19], and other factors. The Pearl River Delta (PRD) is one of the most developed regions in China and has experienced significant O 3 pollution. The O 3 concentration has increased in the PRD since 2015 [20]. The O 3 pollution is the highest in autumn in the PRD due to high temperatures, strong solar radiation, and low relative humidity (RH) [21][22][23][24][25]. In addition, several studies confirmed the "weekend effect" [26,27] in China, i.e., the O 3 concentration is higher on weekends than during working days in Beijing [28], Shanghai [29,30], and Guangzhou [31]. There are two reasons. First, the nitric oxide (NO) concentration is lower during the weekend due to fewer traffic emissions. Therefore, the inhibitory effect of NO on O 3 is weaker, and more O 3 is generated. Second, fewer aerosol particles are emitted during the weekend, resulting in less scattering and absorption of solar radiation. As a result, more O 3 is formed due to the stronger solar radiation during weekends [32].
There are three major sources of near-ground O 3 precursors: traffic emissions [33], industry emissions, and emissions by power plants [34]. Mitigating O 3 pollution has become a crucial issue in the PRD region in recent years [35]. However, it is challenging to control O 3 pollution due to the complex O 3 generation mechanism [36]. After absorbing ultraviolet light, tropospheric O 2 decomposes into two O atoms. The O atoms are combined with O 2 to form O 3 (Equations (1) and (2)). In urban areas, NO 2 in traffic emissions is the main precursor of O 3 (Equation (3)). O 3 rapidly oxidize NO to form NO 2 , known as the titration effect (Equation (4)).
In these processes, a dynamic equilibrium exists during the formation and consumption of O 3 by NO X . However, alkoxy radicals (RO) and hydroperoxyl radicals (HO 2 ) generated by the reaction of volatile organic compounds (VOCs) and hydroxyl (OH) radicals in the atmosphere also react with NO (Equations (5)-(8)), destroying the dynamic balance between NO X and O 3 and increasing the O 3 concentration.
If large amounts of NO X are emitted, HO and RO 2 react predominantly with NO 2 (Equations (9) and (10)); if small amounts of NO X are emitted, the free radical reaction dominates (Equations (11) and (12)). According to the formation mechanism of O 3 , the O 3 concentration is closely related to the NO X and VOCs concentrations because of the highly nonlinear relationship between O 3 and its precursors. Therefore, it is more difficult to mitigate O 3 than other pollutants.
The O 3 concentration depends on the O 3 formation process and diffusion [37][38][39]. Accordingly, the photochemical reaction rate [40], human activities, and meteorological conditions are the three dominant factors affecting the local O 3 concentration [41,42]. Many studies have demonstrated that low cloudiness [43,44], intense solar radiation [45], high temperature [46,47], and low RH [48] can accelerate the O 3 production rate [49,50]. High road network density [51], frequent motor vehicle braking, rapid acceleration, and high traffic flow [52] lead to high NO X emissions [53]. Wind speed and direction can affect the horizontal distribution of O 3 in local areas, and a low wind speed facilitates O 3 accumulation [54,55].
Traffic emissions are the main pollutant source in urban areas. NO X and VOC from traffic emissions are the main precursors of O 3 . Therefore, it is necessary to investigate the relationship between traffic conditions and O 3 pollution. However, there are very few studies focusing on the influence of traffic situations on O 3 . We investigate the potential relationship between the O 3 concentration and traffic conditions at roadside and urban background stations in Guangzhou, one of the largest cities in the PRD and China. The results provide a scientific reference for policymakers to establish emission mitigation policies.

Study Area and Measurement Data
Guangzhou is one of the largest cities in China, with a developed economy, dense population, and advanced manufacturing industries. The atmospheric pollutant concentrations were obtained from three national monitoring stations: two roadside stations (Yangji station (YJ station) and Huangsha station (HS station)) and one urban background station (Luhu station (LH station)) ( Figure 1). The YJ station is located at an intersection of the main road (Zhongshan road) in the city center, about 5 m higher above ground. The HS station is located on a three-layer viaduct. The measurement instruments were installed between the second and third layers, about 20 m above the ground. The LH station is situated in Luhu Park, allowing us to compare air pollution in traffic and an urban park. The national measurement data were obtained from Guangzhou Ecological Environment Bureau (http://sthjj.gz.gov.cn/, accessed on 1 July 2021). The temporal resolution of the measurement data is one hour.
Meteorological data were obtained from Guangzhou Weather website (http://www. tqyb.com.cn/gz/weatherLive/autoStation/, accessed on 1 July 2021), including ambient temperature, wind speed, wind direction, solar radiation, and RH. The dynamic traffic data were obtained from the Guangzhou Municipal Bureau of Transportation (http://jtj.gz.gov. cn/jtcx/lkcx/, accessed on 1 July 2021). All the data were quality-controlled and covered the period from January to June 2021.

Analysis Approaches
A stepwise regression model was used to investigate the relationship between the potential impact factors and O 3 concentration. Stepwise regression analysis automatically selects the most important variables to establish a predictive or explanatory model. The influencing factor are incorporated into the model one by one, and the statistical significance was evaluated. The insignificant factors were removed from the model.  Meteorological data were obtained from Guangzhou Weather website (http://www.tqyb.com.cn/gz/weatherLive/autoStation/, accessed on 1 July 2021), including ambient temperature, wind speed, wind direction, solar radiation, and RH. The dynamic traffic data were obtained from the Guangzhou Municipal Bureau of Transportation (http://jtj.gz.gov.cn/jtcx/lkcx/, accessed on 1 July 2021). All the data were quality-controlled and covered the period from January to June 2021.

Analysis Approaches
A stepwise regression model was used to investigate the relationship between the potential impact factors and O3 concentration. Stepwise regression analysis automatically selects the most important variables to establish a predictive or explanatory model. The

Temporal Variations of NO 2 and O 3 3.1.1. Daily Variations
Generally, pollutant concentrations are affected by several factors, such as emission sources, meteorological conditions, and pollutant formation mechanisms. The median diurnal variation of O 3 and NO 2 during the cold (from January to March) and warm (from April to June) seasons is shown in Figure 2. Similar diurnal patterns of O 3 are observed at the three stations. The O 3 concentration is low from 22:00 to the early morning on the following day. Then, it rapidly increases from around 8:00 in the morning and reaches the maximum around 14:00-16:00. As the solar radiation increases during the daytime, the O 3 concentration increases [56,57] (Equations (1) and (2)). However, the O 3 concentration remains low during the night. There are two reasons. First, less O 3 is generated in the absence of sunlight. Second, NO can react with O 3 to form NO 2 and O 2 during the night (Equation (4)), which is referred to as the titration effect of NO.
The diurnal variation of NO 2 differs from that of O 3 . As shown in Figure 2d-f, the NO 2 concentration is lower at 3:00-4:00 and 12:00-16:00 and higher at 6:00-8:00 and 20:00-22:00. The highest NO 2 concentration occurs at 20:00-22:00. The NO 2 concentration shows an increasing trend from 04:00-8:00 at the two roadside stations (HS and JY) because of traffic emissions. This increasing trend is not observed at the urban background station (LH). The solar radiation increases after 08:00. NO 2 reacts with VOCs to produce O 3 , resulting in a decreasing trend at all three stations. The NO 2 concentration increases after 16:00 due to lower solar radiation and a decrease in the photochemical reaction [58][59][60]. During the night, the NO 2 concentration increases again due to the titration effect [61].
The seasonal difference in the pollutant concentration is larger for NO 2 than for O 3 , as shown in Figure 2d-f. The NO 2 concentration is higher in the cold season (from January to March) than in the warm season (from April to June). The decisive factor influencing the seasonal variation of the NO 2 concentration is solar radiation. The average solar radiance in Guangzhou is 1352 kJ/ m 2 in the cold season and 1806 kJ/ m 2 in the warm season. Lower solar radiation leads to less O 3 generation and less NO 2 consumption. Another possible factor may be the lower RH in winter. In Guangzhou, the average RH is 59.04% and 86.2% in the cold and warm seasons, respectively [62,63]. A higher RH results in a stronger photochemical reaction and a lower NO 2 concentration in the warm season. Another possible explanation is the seasonal change in the planetary boundary layer height. It is 717 m in winter and 1239 m in summer in Guangzhou [64,65]. A lower planetary boundary layer accumulates NO 2 , resulting in a higher NO 2 concentration [66]. However, the seasonal difference in the O 3 concentration is smaller than that of the NO 2 concentration. The reason is that O 3 is a secondary pollutant whose concentration is controlled by highly complex and nonlinear secondary formation mechanisms.

Temporal Variations of NO2 and O3
3.1.1. Daily Variations Generally, pollutant concentrations are affected by several factors, such as emission sources, meteorological conditions, and pollutant formation mechanisms. The median diurnal variation of O3 and NO2 during the cold (from January to March) and warm (from April to June) seasons is shown in Figure 2. Similar diurnal patterns of O3 are observed at the three stations. The O3 concentration is low from 22:00 to the early morning on the following day. Then, it rapidly increases from around 8:00 in the morning and reaches the maximum around 14:00-16:00. As the solar radiation increases during the daytime, the O3 concentration increases [56,57] (Equations (1) and (2)). However, the O3 concentration remains low during the night. There are two reasons. First, less O3 is generated in the absence of sunlight. Second, NO can react with O3 to form NO2 and O2 during the night (Equation (4)), which is referred to as the titration effect of NO. The diurnal variation of NO2 differs from that of O3. As shown in Figure 2d-f, the NO2 concentration is lower at 3:00-4:00 and 12:00-16:00 and higher at 6:00-8:00 and 20:00-22:00. The highest NO2 concentration occurs at 20:00-22:00. The NO2 concentration shows an increasing trend from 04:00-8:00 at the two roadside stations (HS and JY) because of traffic emissions. This increasing trend is not observed at the urban background station (LH). The solar radiation increases after 08:00. NO2 reacts with VOCs to produce O3, resulting in a decreasing trend at all three stations. The NO2 concentration increases after 16:00 due to lower solar radiation and a decrease in the photochemical reaction [58][59][60]. During the night, the NO2 concentration increases again due to the titration effect [61].
The seasonal difference in the pollutant concentration is larger for NO2 than for O3, as shown in Figure 2d-f. The NO2 concentration is higher in the cold season (from January to March) than in the warm season (from April to June). The decisive factor influencing the seasonal variation of the NO2 concentration is solar radiation. The average solar radiance in Guangzhou is 1352 kJ/ m 2 in the cold season and 1806 kJ/ m 2 in the warm season. Lower solar radiation leads to less O3 generation and less NO2 consumption. Another

Weekly Variations
The weekly variations in the O 3 and NO 2 concentrations at the three stations are illustrated in Figure 3. In general, the weekly trends of the O 3 and NO 2 concentrations are similar at three stations, but the average concentrations are different. As shown in Figure 3a, the O 3 concentration is significantly higher on weekends (Saturday and Sunday) than on weekdays (from Monday to Friday), indicating the weekend effect of O 3 . It is believed to be related to a change in the proportion of O 3 precursor emissions and other pollutant emissions from human activities [67]. Fewer human activities on weekends lead to lower PM 2.5 and a lower aerosol optical thickness and radiation extinction. Therefore, the O 3 concentrations are higher on the weekend than on weekdays due to stronger photochemical reactions [68,69]. Moreover, high traffic flow during the morning rush hour results in a rapid increase in the NO concentration, inhibiting O 3 formation on weekdays [70,71].
Differences in the O 3 concentration are observed at the three stations. The highest O 3 concentration was measured at the LH station, followed by the two roadside stations YJ and HS. The reason is the surrounding environment. The LH station is located in Luhu Park. VOCs generated by biological sources compete with NO, reducing the inhibition of NO on O 3 and leading to a higher O 3 concentration [72,73]. The YJ station is surrounded mostly by business and entertainment areas with frequent human activities. Large amounts of NO X are emitted from traffic inhibited O 3 formation. In addition, the titration effect of NO is stronger at the YJ station, leading to a slightly lower O 3 concentration at the YJ station than at the LH station. The HS station is a roadside station located near a park. It has higher vegetation cover than the YJ station.
The weekly variation in the NO 2 concentration shows a significantly different pattern than that of the O 3 concentration. The NO 2 concentration is slightly higher on weekdays than on the weekend due to higher anthropogenic emissions, especially traffic emissions in urban areas [74][75][76][77]. The NO 2 concentration is the highest at the HS station, followed by the YJ and LH stations, which is consistent with the traffic emissions and the local environment of the three stations.
at three stations, but the average concentrations are different. As shown in Figure 3a, the O3 concentration is significantly higher on weekends (Saturday and Sunday) than on weekdays (from Monday to Friday), indicating the weekend effect of O3. It is believed to be related to a change in the proportion of O3 precursor emissions and other pollutant emissions from human activities [67]. Fewer human activities on weekends lead to lower PM2.5 and a lower aerosol optical thickness and radiation extinction. Therefore, the O3 concentrations are higher on the weekend than on weekdays due to stronger photochemical reactions [68,69]. Moreover, high traffic flow during the morning rush hour results in a rapid increase in the NO concentration, inhibiting O3 formation on weekdays [70,71]. Differences in the O3 concentration are observed at the three stations. The highest O3 concentration was measured at the LH station, followed by the two roadside stations YJ and HS. The reason is the surrounding environment. The LH station is located in Luhu Park. VOCs generated by biological sources compete with NO, reducing the inhibition of NO on O3 and leading to a higher O3 concentration [72,73]. The YJ station is surrounded mostly by business and entertainment areas with frequent human activities. Large  Figure 4 shows the scatterplots of the O 3 and NO 2 concentrations during the daytime (07:00-19:00) and nighttime (20:00-06:00) at the three stations. The linear regression model has a negative slope for all three stations during the daytime and nighttime, indicating that the NO 2 concentration decreases as the O 3 concentration increases. However, differences are observed between daytime and nighttime. In the daytime, NO 2 is consumed, and O 3 is produced (Equations (2) and (3)). However, without a photochemical reaction during nighttime, O 3 is converted to NO 2 due to the titration effect (Equation (4)), leading to a lower O 3 concentration. Due to the highly nonlinear and complex O 3 formation mechanism, the R 2 value is low for all fitting results. The R 2 value is larger during nighttime at all three stations due to the absence of the photochemical reaction, the titration effect of NO, and weaker vertical diffusion [78,79]. The nighttime fitting degree is better at the LH station than at the roadside stations. The reason might be the surrounding environment of the LH station. The vegetation cover is higher; thus, vegetation respiration is stronger at night. Consequently, the NO 2 and O 3 concentrations are relatively stable, leading to a better fitting degree.

Synergistic Variation of O 3 and NO 2
The fitted results of the three stations are similar. However, the dominant emission sources differ at the three stations. This result indicates no significant effect of traffic emissions on the O 3 concentration at the roadside stations. Due to the absence of VOCs, a dynamic equilibrium exists between O 3 and NO X in the atmosphere. Thus, O 3 is not accumulated and does not exceed the air pollution standard [80,81]. However, the reaction between VOCs and NO weakens the inhibitory effect of NO on O 3 , resulting in high O 3 pollution [82]. Controlling NO X emissions does not mitigate O 3 pollution. Moreover, Guangzhou is in the VOC-limitation area [83,84]. Limiting vehicle emissions to reduce the NO X concentration may even increase the O 3 concentration. Therefore, the focus should be on industrial, energy, and transportation emission mitigation and the influence of meteorological conditions to minimize O 3 pollution.  The fitted results of the three stations are similar. However, the dominant emission sources differ at the three stations. This result indicates no significant effect of traffic emissions on the O3 concentration at the roadside stations. Due to the absence of VOCs, a dynamic equilibrium exists between O3 and NOX in the atmosphere. Thus, O3 is not accumulated and does not exceed the air pollution standard [80,81]. However, the reaction between VOCs and NO weakens the inhibitory effect of NO on O3, resulting in high O3 pollution [82]. Controlling NOX emissions does not mitigate O3 pollution. Moreover, Guangzhou is in the VOC-limitation area [83,84]. Limiting vehicle emissions to reduce the NOX concentration may even increase the O3 concentration. Therefore, the focus should be on industrial, energy, and transportation emission mitigation and the influence of meteorological conditions to minimize O3 pollution.

Pearson Correlation and Stepwise Regression Analyses
Pearson correlation analysis and stepwise regression analysis were conducted to describe the relationship between the pollutant concentration and other factors, such as meteorological parameters and dynamic traffic parameters. Tables 1 and 2 show the results of Pearson's correlation analysis and stepwise regression analysis, respectively. Pearson's correlation shows the correlation between the O 3 concentration and potential factors, and the stepwise regression model determines the significant impact factors. The beta values are used to quantify the contribution of the variables. Briefly, the O 3 concentration is positively correlated with solar radiation, temperature, and travel-time ratio and negatively correlated with the NO 2 concentration, wind speed, and vehicle speed ( Table 1). The stepwise regression model shows that the significant factors affecting O 3 concentration are temperature, NO 2 concentration, and RH. As shown in Table 1, the O 3 concentration positively correlates with the travel-time ratio. The travel time ratio is the ratio of the actual travel time to the ideal travel time in smooth traffic flow. The larger the ratio, the higher the degree of traffic congestion. The NO X and VOC emissions are higher during frequent vehicle braking than during uniform driving. Thus, more O 3 precursors are emitted, leading to a significant positive correlation between O 3 concentration and travel-time ratio. The temperature is positively correlated with O 3 concentration as a result of O 3 formation. The negative correlation between the NO 2 and O 3 concentrations has already been discussed in Section 3.2.1. Moreover, a negative correlation is observed between vehicle speed and O 3 concentration. The fuel consumption is higher at higher speeds than at lower speeds, resulting in more precursor emissions and a higher O 3 concentration. Wind speed and O 3 concentration are negatively correlated because of the dilution effect. The O 3 concentration is lower at higher RH due to wet deposition. Moreover, an increase in RH significantly reduces the number of oxygen atoms, reducing the amount of O 3 generation.

Impact Factors Daytime Nighttime
Temperature (

Case Study
As discussed in the previous section, traffic emissions affect the O 3 concentration but are not the dominant factor. Many studies demonstrated that solar radiation was a significant factor influencing O 3 formation. A case study was conducted to quantify the influence of solar radiation on O 3 concentration in Guangzhou. Two weeks were selected: 1 February to 7 February 2021, with sunny weather, and 24 February to 2 March 2021, with cloudy weather.
The pollutant concentrations and related parameters are listed in Table 3. The O 3 concentration is substantially different on sunny and cloudy days at all three stations, indicating the predominant influence of solar radiation. The O 3 concentration is 2-3 times higher on sunny days than on cloudy days in the daytime and nighttime. However, there are no large differences in the NO 2 concentration. In the daytime, there are no differences in the NO 2 concentration between sunny and cloudy days. However, the nighttime NO 2 concentration is 1.5 to 2 times higher on sunny days than on cloudy days. More O 3 is formed during sunny days, leading to a stronger titration effect and a higher NO 2 concentration during the nighttime on sunny days. It should be noted that the NO 2 concentration is lower at the LH station than at the two roadside stations during the daytime. However, the O 3 concentration is similar at all three stations due to the lower inhibitory effect of NO, as discussed in Section 3.1.2. This finding confirms our results, i.e., traffic emissions contribute significantly to O 3 generation, but the contribution is not higher at roadside stations than at the urban background station. The scatterplots of the NO 2 and O 3 concentrations in the two periods at YJ and HS are shown in Figure 5. The colored dots indicate the dynamic traffic conditions. The linear regression results demonstrate that the negative correlation between the NO 2 and O 3 concentrations is stronger during the daytime than during the nighttime at both stations due to the stronger photochemical reaction strength. Furthermore, no significant relationship is observed between the O 3 concentration and dynamic traffic conditions.

Conclusions
This study evaluated the factors influencing the O3 concentration in traffic and urban background environments. The diurnal and weekly variation of the O3 and NO2 concentrations demonstrated a similar pattern at the three stations. These results were attributed to differences in the O3 generation mechanism, meteorological conditions, and emission sources. However, no significant differences in the O3 variation were observed between the three stations, implying that the O3 concentration was not significantly higher in the traffic environment than in the urban background environment. Since Guangzhou is located in a VOC-limited area, the lower O3 concentration in the urban background area is due to the lower inhibition of NO on O3.
Pearson correlation analysis and stepwise regression analysis were used to describe the relationship between the pollutant concentration and the influencing factors, such as

Conclusions
This study evaluated the factors influencing the O 3 concentration in traffic and urban background environments. The diurnal and weekly variation of the O 3 and NO 2 concentrations demonstrated a similar pattern at the three stations. These results were attributed to differences in the O 3 generation mechanism, meteorological conditions, and emission sources. However, no significant differences in the O 3 variation were observed between the three stations, implying that the O 3 concentration was not significantly higher in the traffic environment than in the urban background environment. Since Guangzhou is located in a VOC-limited area, the lower O 3 concentration in the urban background area is due to the lower inhibition of NO on O 3 .
Pearson correlation analysis and stepwise regression analysis were used to describe the relationship between the pollutant concentration and the influencing factors, such as meteorological and dynamic traffic parameters. Traffic and meteorological parameters (temperature, solar radiation, RH, and precipitation) were significantly correlated with the O 3 concentration at the two roadside stations. It was concluded that traffic emissions contributed to O 3 pollution in the urban area but were not the decisive factor, while the meteorological factors also influenced the O 3 concentration.
A case study was conducted for two weeks to quantify the influence of solar radiation on O 3 concentration in Guangzhou. On sunny days, the O 3 concentration exceeded 90 µg/m 3 at the three sites. It was 2 to 3 times higher than during cloudy days due to meteorological conditions. The dynamic traffic condition (travel-time ratio) had no significant relationship with the O 3 and NO 2 concentrations at the two roadside stations.
This study analyzed the temporal variation of O 3 and its precursor NO 2 at roadside and urban background environments in Guangzhou and its influencing factors. The results confirmed that limiting traffic emissions might increase O 3 concentrations in Guangzhou. Therefore, emission mitigation should be performed, i.e., industrial, energy, and transportation emission mitigation, and the influence of meteorological conditions should be considered to minimize O 3 pollution. However, some limitations exist in this study. Due to a lack of NO and VOCs data, the relationship between O 3 concentration and NO and VOCs was not analyzed. In future, a mobile measurement focusing on O 3 will be carried out in Guangzhou, and a more detailed analysis will be performed.

Data Availability Statement:
The data presented in this study are available on request from the corresponding website.