Combined Effects of Photochemical Processes, Pollutant Sources and Urban Configuration on Photochemical Pollutant Concentrations

Abstract

Rapid urbanization, dense urban configuration and increasing traffic emissions have caused severe air pollution, resulting in severe threats to public health. Particularly, photochemical pollution is associated with chemical transformation introducing more complexity. The understanding of the combined effects of pollutant sources, urban configuration and chemical transformation is still insufficient because most previous studies focused on non-reactive pollutant dispersions. In this study, we adopt a simplified street network model including complex photochemical reactions, i.e., the Model of Urban Network of Intersecting Canyons and Highways (MUNICH), with the real traffic and street data of a region in Guangzhou to investigate the combined effects of the three factors above on photochemical pollution. Our simulations show that the overall reduction in traffic emissions decreases NO_x pollution while increasing O₃ concentration. Controlling VOC emission can effectively mitigate O₃ pollution. Moreover, irregular building heights and arrangements can lead to certain hot spots of air pollution. High-rise buildings will obstruct ventilation and exacerbate pollution. If higher buildings have lower vehicle use, the deep canyon can offset the effect of lower emissions. In conclusion, urban planners and policy makers should avoid deep canyons and irregular street networks to achieve better pollutant dispersion and pay attention to controlling VOC emissions.

1. Introduction

With the rapid social and economic development in the past several decades, drastic urbanization with various urban landscapes is happening all around the world [1,2]. The tremendous urbanization has led to a continuously growing population and social wealth accumulation but also to the severe problem of urban air pollution [3,4,5]. Due to the deteriorated air quality, a major public health crisis has emerged in both developed and developing countries [6,7]. The deterioration of urban air quality is the result of the combination of pollutant sources, dynamical processes and chemical transformation [8]. Common urban air pollution includes passive (such as CO) and photochemical pollutants (NO, NO₂ and O₃). According to previous research, CO has strong toxicity that poses threats to human life, and CO₂ as a greenhouse gas will contribute to climate change [9]. The photochemical pollutant NO_x, directly emitted by cars, has high concentrations in urban areas and contributes to human lung cancer and heart diseases [9,10]. O₃ in the troposphere is harmful to the human respiratory system and can also cause eye diseases [9]. Therefore, it is of great significance to explore ways to mitigate urban air pollution, which requires further investigation of urban air pollution formation and its interaction with the urban landscape.

Traffic emission, as a major source of urban air pollution, has soared with rapid urbanization in recent years [11]. Particularly at the street level, urban air pollution from vehicle emissions plays a much more important role than background pollutant concentration due to its proximity and the complex building configuration surrounding it [12,13]. Many previous studies have investigated the dispersion of traffic-related air pollution in 3D urban areas or 2D street canyons [14,15,16,17]. However, most studies only consider constant meteorological conditions and pollutant emission rates, which is not realistic since these conditions are dynamically changing in real urban areas [18,19]. Therefore, it is worth exploring the variation in urban air pollution under real-time meteorological conditions and pollutant emission rates.

In addition, with rapid urbanization, urban geometry has become a key factor in urban air-pollutant concentration [20,21]. Complex building shapes and height variations increase atmospheric friction near the ground surface and hence reduce the overall wind speed in urban areas [22]. High concentrations of air pollutants could usually occur in less-ventilated regions. Most previous studies focused on the variations in passive air-pollutant concentration caused by different building configurations such as different aspect ratios and urban building densities [23,24,25], step-up and step-down canyons [26,27] and the lateral entrainment of buildings [28]. However, there is limited investigation on the variation in photochemical pollutants with complex photochemical reactions in the irregular arrangement of 3D urban areas. In addition, some studies claim that compact urban configuration often leads to reduced traffic emissions in prosperous areas as it attracts people to get around by public transit which causes less vehicle use [19,29,30,31]. However, there exists another opinion that the compact urban configuration would lead to high concentrations of air pollutants due to the worsened atmospheric dispersion [32]. Thus, it is of great interest to study the combined effects of pollutant sources and urban configuration on air-pollutant concentration.

For passive pollutants (CO and SF₆), the only two factors to consider in urban air pollution variation are pollutant sources and dynamical processes for their long reaction time to ignore the effects of chemical reactions [28,33,34,35]. However, for photochemical pollutants (NO, NO₂ and O₃), the chemical transformation process also plays an important role in the pollutant variation as chemical reactions can happen at a fast speed, changing the pollutant concentrations constantly [36,37,38]. Few studies have explored photochemical pollutant dispersion along with other factors in urban areas. Moreover, despite significant efforts in recent years to reduce traffic emissions and thus lower urban air pollution, which have resulted in notable decreases in concentration for most air pollutants, O₃ concentration has increased in some urban areas. For instance, most Chinese cities have experienced an annual growth rate of 1–9% in O₃ concentration [39,40]. Therefore, we need to pay particular attention to the variation in photochemical pollutants and the factors driving their formation and concentrations in urban areas.

Most previous studies adopted computational fluid dynamics (CFD) to investigate the dispersion of passive and photochemical pollutants in 2D street canyons and 3D urban areas via street-scale numerical simulations. Zhang et al. [23] investigated the effect of various building heights and densities on the dispersion of photochemical pollutants with simple photochemical reactions in 3D urban areas. Kwak et al. [41] found that the sensitivity of O₃ concentration to pollutant emission rates is higher than ambient wind speed by using a CFD model coupled with the carbon bond mechanism IV (CBM–IV). Though CFD has achieved promising results in urban air pollution studies, it requires a huge amount of computational time and computational resources to investigate the complex photochemical reactions in the whole urban area. On the contrary, street network models such as the Model of Urban Network of Intersecting Canyons and Highways (MUNICH) [42,43,44] have been developed for street-scale simulation, saving computational resources with a less-detailed simulation but slightly reduced accuracy. MUNICH has been adopted to investigate the concentrations of pollutants in a Paris suburb including NO_x and volatile organic compound (VOC) emissions [42,43,44].

This study adopts MUNICH to investigate the influence of urban structure on the concentrations of photochemical pollutants and to explore the sensitivity of pollutants from different sources in a relatively large-scale urban area with real pollution, traffic emission and urban structure data. The objective of this study is to investigate the combined effects of traffic emissions, dynamical processes and chemical transformation on photochemical pollutants (NO, NO₂ and O₃) in 3D urban areas by using the MUNICH model. The pollution sources are divided into two parts: traffic pollutant sources on the street and background pollutant sources including natural emissions and regional transportation. The structure of this paper is as follows: Section 1 is an introduction that includes the research background and objectives for the study. Section 2 introduces MUNICH, simulation settings and data sources. Section 3 discusses the simulated results in detail. Section 4 draws conclusions from our analysis and simulated results.

2. Methodology

In this section, we will introduce the common methods for urban air pollution simulation first and then the details of MUNICH. Then, we will further introduce our detailed settings for MUNICH including meteorological data, traffic emissions, background concentrations and building structures. Lastly, we will conduct a validation study for MUNICH to validate its accuracy against real data.

2.1. Street-Level Air-Pollutant Dispersion Simulation

According to the characteristic length of atmospheric movement, air quality models can be commonly categorized into four length scales: regional scale (~100 km), city scale (~10 km), neighborhood scale (~1 km) and street scale (~100 m) [45,46]. The dynamical processes of each street within urban areas are often within the urban canopy scheme and are usually presented as parameterizations in city-scale and regional-scale air quality models (such as Weather Research and Forecasting model (WRF)). However, the parameterization in larger-scale models usually ignores the complicated street-level pollutant dispersion from traffic emissions, which contributes more to urban air quality than background air pollution from other sources [47]. For instance, a study has demonstrated that NO_x concentration at a monitoring station near the Boulevard Alsace-Lorraine exceeded the background concentration by many times [12]. Therefore, it is worth paying attention to the influence of dynamical processes and traffic emissions within the streets on air quality variation at the neighborhood and street scale [46]. Computational fluid dynamics (CFD) is the most commonly used method with a good ability to simulate pollutant diffusion within urban street blocks [48,49,50]. However, most CFD simulations are computationally expensive. When considering complex chemical reactions and dispersion models on a larger scale, CFD models face limitations such as difficulty in scaling, slow simulation speed and high computational costs. Therefore, simplified street network models, such as the Model of Urban Network of Intersecting Canyons and Highways (MUNICH), have been developed for neighborhood-scale and city-scale simulations with much faster simulation speed, lower computational costs and sufficient modeling precision [42,43,44]. Meanwhile, street network models take into account the complex physical and chemical processes of pollutant dispersion, which may allow for a more accurate and comprehensive assessment of street-level air pollution.

2.2. MUNICH

MUNICH is a simplified street network model introduced by Kim et al. [42] and Lugon et al. [43]. The complex street micrometeorology is simplified by considering only vertical transfer between streets and upper roughness sublayers and horizontal transport between streets. Moreover, air-pollutant concentrations are assumed to be homogeneous in each street segment. The pollutant concentration in a street at steady state is calculated from the equilibrium of mass flux Q (μg·s⁻¹) [42], which is expressed as Equation (1).

$$Q_s+Q_{inflow}+Q_{chem}=Q_{vert}+Q_{outflow}+Q_{dep}$$

(1)

where $Q_s$ is the rate of emission source; $Q_{in/outflow}$ is the horizontal flux of pollutants between streets; $Q_{chem}$ is the chemical conversion rate of air pollutants (positive for generation and negative for destruction); $Q_{vert}$ is the vertical flux of pollutants by turbulent diffusion at the roof level of the entire street; and $Q_{dep}$ is the pollutant loss rate caused by atmospheric sedimentation.

2.2.1. Horizontal and Vertical Airflow in the Street Network

To model the pollutant dispersion at neighborhood and street levels, the airflow in MUNICH is divided into two segments, the horizontal segment and the vertical segment [44]. In each street section, the wind speed is calculated based on the wind speed and direction above the street and the characteristics such as aspect ratio ($a_r=\frac{H}{W}$) of the street. The horizontal wind directions are assumed to be along the street directions, and wind speeds vary with different streets. The pollutant concentration in each street is driven by two factors: the horizontal transfer processes at street level and the background concentrations above the roof level through vertical transfer processes.

Along the downwind direction, the horizontal flux $Q_{inflow}$ is the inflow rate of the air pollutants entering the street, and $Q_{outflow}$ is the outflow rate of the air pollutants leaving the street. The calculation of the flux $Q_{in/outflow}$ is shown in Equation (2) [42,44].

$$Q_{in/outflow}=U_{street}HW{C}_{street}$$

(2)

where $U_{street}$ is the horizontal transfer velocity (m·s⁻¹) calculated as the average wind speed along the street direction; $HW$ as exchange section (m²) is the product of the height and width; and $C_{street}$ is the average pollutant concentration in the street (μg·m⁻³). The $U_{street}$ in the simulation is derived with the exponential attenuation profile as expressed in Equation (3) [42].

$$U_{street}=U_{H,\phi }\times \frac{1}{A{a}_r}\left(1-exp\left[A{a}_r\left(\frac{z_{0_s}}{H}-1\right)\right]\right)$$

(3)

where $U_{H,\phi }=U_H\left|cos\phi \right|$, $U_H$ is the horizontal wind speed at the level of building height H (m s⁻¹), and $\left|cos\phi \right|$ is used to select street component [51]; $A{a}_r$ is the dimensionless attenuation coefficient; and $z_{0_s}$ is the soil roughness (m).

The vertical flux $Q_{vert}$ is mainly determined by the external flow condition based on turbulent transfer coefficient ($q_{vert}$) as Equation (4) [49]:

$$Q_{vert}=q_{vert}WL\frac{C_{street}-C_{background}}{H}=\frac{{\sigma }_{w}WL}{\sqrt{2}\pi \left(1+a_r\right)}\left(C_{street}-C_{background}\right)$$

(4)

where W and L are the street width and length (m), respectively; $C_{background}$ is the average concentration of background air pollution in the street (μg·m⁻³); $q_{vert}={\sigma }_w{l}_m=\frac{{\sigma }_{w}H}{\sqrt{2}\pi \left(1+a_r\right)}$ is the vertical transfer coefficient (m²·s⁻¹); and ${\sigma }_w$ is the standard deviation of the vertical wind velocity at roof level (m·s⁻¹). ${\sigma }_w$ is related to atmosphere stability and for a neutral atmosphere is calculated by Equation (5):

$${\sigma }_w=1.3u_{*}\times \left(1-0.8\frac{H}{PBLH}\right)$$

(5)

where $u_{*}$ is the friction velocity, and PBLH is the planetary boundary layer height (m) [52]. The configuration of ${\sigma }_w$ for stable and unstable atmospheric conditions can be found in reference [53].

2.2.2. Chemical Reactions and Deposition

In CFD studies, there are usually two approaches to simulate the variation in photochemical pollution in urban areas: the simple photochemical NO_x-O₃ mechanism [54,55] and the complex VOCs-NO_x-O₃ photochemical mechanism [15,56]. Both of them require high computational cost. In MUNICH, the CB05 chemical mechanism is used to calculate chemical fluxes $Q_{chem}$ involving 53 species and 155 chemical reactions [57]. The deposition flux $Q_{dep}$ is divided into dry and wet deposition. The dry deposition flux is the product of the pollutant concentration and deposition velocity [44,58], which depends on atmospheric conditions and surface properties related to various surface types including building roofs, building walls and pavements (streets and sidewalks). For the pollutant concentration, the background concentrations above the urban canopy are used in the estimation of dry deposition at building roofs while the concentrations within the street network are used for the building walls and ground. Wet deposition is the deposition to building roofs and pavement through precipitation [44]. Wet deposition on building roofs is calculated by precipitation intensity and background concentration above the urban canopy. Wet deposition to the pavement includes both background concentrations above the urban canopy and street network concentrations within the urban canopy.

2.3. Model Setup and Data

In this study, 31 main street segments in Tianhe District, Guangzhou, China, are selected as the simulated urban area (113.302° E–113.328° E, 23.125° N–23.151° N), which are quite representative of Chinese cities (Figure 1a,b). In the simulated urban area, the length data of each street were directly obtained by the distance between the starting point and the ending point of the street, which was sourced from the OpenStreetMap dataset (https://www.openstreetmap.org, last access: 2 September 2019) and intercepted by ARCGIS. The width data of streets were determined based on the grade of roads (generally 4–8 lanes) with each lane of 3.5 m. The height data (h) of each street were obtained from the World Urban Database and Access Portal Tools (WUDAPT) dataset [59]. In this study, regular arrangement (Figure 1c) is modeled from a real urban street network (Figure 1a) with individual buildings of uniform length (616.87 m), width (21 m) and height (19.82 m).

The traffic emission data in the simulated urban area were calculated by a real-time on-road emission model (ROE) with a bottom-up method [60]. The ROE model can calculate the emissions of carbon monoxide (CO), nitrogen oxide (NO_x: NO₂ and NO), volatile organic compounds (VOCs) and other pollutants in each street by using real-time traffic data provided by Gaode Map and vehicle emission factors. Note that the non-holiday traffic emission data in this study were the average of traffic emissions on 27–28 April and 2–3 May 2018. In addition to traffic emissions, the background NO₂ and O₃ concentrations were obtained from the observed data at a monitoring station closest to the simulated area using the average concentration from 27 April to 3 May 2018. The background NO concentration was determined by assuming the ratio of NO:NO₂ (1:4). Meteorological data including wind profile and boundary height were provided by WRF with a grid resolution of 1 km. The physical schemes adopted in WRF simulation were from Wu et al. [60]. Figure 1d shows the wind speed at the grid point where the simulation area is located from 28 April to 1 May. The range of boundary height is 73–1245 m during this period. The chemical mechanism of CB05 was used in MUNICH to simulate the complex photochemical reaction in the street network [57].

2.4. Validation of Photochemical Pollution in MUNICH against the Observed Data

In order to ensure the reliability of our numerical simulations, the performances of MUNICH model were evaluated using the observed concentration data prior to the simulation analysis. MUNICH was validated by comparing the simulated data against the monitoring station data. Figure 2 shows the comparison between the simulated and observed NO₂ and O₃ concentrations from 28 April to 2 May 2018. The simulation results align well with the observed NO₂ and O₃ concentrations during the simulation period. The simulation of NO₂ concentrations was in close agreement with the actual observed data and performed better than that of O₃. The concentration of O₃ was overestimated at nighttime and underestimated during the daytime. It should be noted that the simulated results were the average concentrations in the entire study area while the observed data were only obtained from a closet monitoring station. Some discrepancy between the simulated and observed concentrations is expected, but MUNICH can well capture the variation in NO₂ and O₃ concentrations, and the difference between simulated and observed data is not significant.

Regarding the validation of airflow field, we also believe that the accuracy is satisfactory. This is because the meteorological data used in MUNICH was provided by WRF, and its accuracy has been validated in numerous previous studies [61,62]. Additionally, the wind speed for each street segment is calculated using the widely accepted exponential attenuation profile [35,63]. Given the correlation between the concentration and the airflow field and the satisfactory validation of the concentration field, the airflow field in the street network can be considered acceptable.

In order to further validate the capability of MUNICH to simulate the pollutant concentrations at street level, we calculated several statistical metrics including normalized mean bias (NMB), normalized mean error (NME) and root-mean-squared error (RMSE) and compared them against those of the MEP Technical Guide for Air Quality Model Selection [60] and other air modeling studies [64,65,66] summarized in

Table 1. Statics values of the simulation in MUNICH compared to the observed data.

	O3			NO2
	This Study	Technical Guide	Other Modeling	This Study	Technical Guide	Other Modeling
OBS	30.8	-	-	60.9	-	-
SIM	35.8	-	-	59.3	-	-
MB	5.1	-	-	−1.6	-	-
ME	20.8	-	-	14.9	-	-
NMB	16.4%	−15~15%	−21.2~20.0%	−2.6%	−40~50%	−27.5~−6%
NME	67.7%	<35%	38.2~98%	24.4%	<80%	29.2~53.0%
RMSE	25.1	-	9.4~40.1	18.9	-	16~37.3

. Results show that the NMB, NME and RMSE of O₃ and NO₂ fall within the reference range established by the Technical Guide and other studies. In general, MUNICH shows good performance, and the numerical errors in the simulation are not significant. Therefore, the model of MUNICH can be used to study air quality variation in urban areas.

3. Results and Discussion

To investigate the combined effects of photochemical processes, pollutant sources and urban configuration on photochemical pollutant concentrations in urban areas, a comprehensive set of simulations were conducted using the MUNICH model in this study. This section will detail the simulated results followed by our discussions. In Section 3.1, we first analyze the diurnal variation in photochemical pollution (NO, NO₂ and O₃). In Section 3.2, the effect of varied traffic-related pollution (VOCs and NO_x) on photochemical pollutant concentrations is investigated. Section 3.3 discusses the dynamical processes of pollutants within the urban areas with the real irregular and ideal regular building arrangement. In Section 3.4, we analyze the effects of high-rise buildings on street-scale dynamical processes and also the simultaneous effects of high-rise building and traffic emission on pollutant concentrations. In Section 3.5, we explore the chemical transformation process of photochemical pollutants by varying the concentrations of background pollutants (NO_x and O₃).

3.1. Diurnal Variation in Pollutant Concentrations in Urban Areas

Different from the inert pollutants in urban areas, photochemical pollutants NO_x, VOCs and O₃ can have chemical reactions with each other and hence quickly impact their concentrations within streets with high diurnal variations. Therefore, it is of great significance to investigate the diurnal variation in photochemical pollutant concentrations in urban areas taking chemical reactions into consideration. The chemical equations of the reactions are in Equations (6)–(8):

$${\mathrm{NO}}_2+hv\left(285~375 \mathrm{nm}\right)\to \mathrm{NO}+\mathrm{O}\left({}_{ }{}^3\mathrm{P}\right)$$

(6)

$$\mathrm{O}\left({}_{ }{}^3\mathrm{P}\right)+{\mathrm{O}}_2+M\to {\mathrm{O}}_3+M$$

(7)

$${\mathrm{O}}_3+\mathrm{NO}\to {\mathrm{NO}}_2+{\mathrm{O}}_2$$

(8)

where M is mainly composed of oxygen and nitrogen molecules that exist as catalysts in the reactions without loss. The above reactions will quickly reach equilibrium and stop and are also adopted as the chemical reaction processes in the simple photochemical mechanism in CFD models. However, in real urban areas, the photochemical reactions are much more complex due to the significant effects of VOCs on photochemical pollution. The simplified equations of the VOCs-NO_x-O₃ reaction are in Equations (9) and (10):

$$\mathrm{VOCs}+\mathrm{OH}\to {\mathrm{RO}}_2+{\mathrm{H}}_2\mathrm{O}$$

(9)

$${\mathrm{RO}}_2+\mathrm{NO}\to \mathrm{RO}+{\mathrm{NO}}_2$$

(10)

R groups in ${\mathrm{RO}}_2$ and $\mathrm{RO}$ generally refer to alkyl groups that do not affect the oxidation processes. RO₂ generated by VOCs and OH will react with NO to generate NO₂ and reduce NO, which leads to an increase in O₃.

Figure 3a shows the diurnal traffic emission (NO_x and VOCs), and Figure 3b illustrates the diurnal variation in photochemical pollutant concentrations in urban areas. In general, the lowest concentrations of all pollutants in a day happen in the early morning. The average concentration of O₃ in the study region starts to increase after 9:00 and reaches maximum around 14:00 due to the elevated photochemical reaction rate and accumulated precursors such as NO_x and VOCs in the urban areas. The average concentration of NO_x in the study region increases during morning and evening emission peaks due to the intense human activities (traffic emission) and the reaction with O₃ (chemical transformation). Specifically, the average concentration of NO increases at 6:00 and 17:00 due to the traffic emission and decreases at 12:00 and 22:00 due to the drastic reaction with O₃. The lowest level of NO₂ region mean concentration in a day happens in the early morning at around 5:00 and continuously accumulates and reaches the maximum at 18:00.

3.2. Effects of Traffic Emissions on Pollutant Concentrations

Traffic emission is one of the important sources of urban pollution. Residential and business buildings close to streets are at a high risk of exposure due to traffic emissions. One of the major sources of VOCs and NO_x in urban areas is traffic emissions. Figure 4 shows the variation in the region mean concentrations of photochemical pollutants (NO, NO₂ and O₃) varying with the equally proportional reduction in traffic emission (VOCs and NO_x) in the urban area. As shown in Figure 4a,b, when the traffic emission decreases to 80%, 60%, 40%, 20% and 0, the mean spatiotemporal concentration of NO decreases by 6.88%, 6.67%, 6.42%, 6.08% and 5.62% and the mean spatiotemporal concentration of NO₂ decreases by 3.51%, 3.68%, 3.88%, 4.15% and 4.51%. However, as shown in Figure 4c, the mean spatiotemporal concentration of O₃ increases by 3.02%, 3.38%, 3.82%, 4.39% and 5.15% when the traffic emission decreases in the same way. The reason is the reduction in NO, which constitutes the largest proportion of traffic emissions and leads to the decrease in titration reaction (Equation (8)), resulting in an increase in O₃ concentration. The overall reduction in traffic emissions, while effective in controlling NO_x pollution, may result in worsening O₃ pollution.

To further understand the impact of traffic emissions on O₃ concentration, the variation in O₃ mean spatiotemporal concentration with various levels of VOCs and NO_x emission is investigated (Figure 4d). When the NO_x emission decreases from 100% to 0 with 100% VOC emission, the mean spatiotemporal concentration of O₃ increases from 35.07 μg/m³ to 44.71 μg/m³. When the VOC emission decreases from 100% to 0 with 100% NO_x emission, the mean spatiotemporal concentration of O₃ decreases from 37.33 μg/m³ to 35.07 μg/m³. In other words, reducing NO_x emission will increase the O₃ concentration, while reducing VOC emission can reduce O₃ concentration because RO₂ will decrease when VOC emissions are reduced (Equation (9)), which will lead to an increase in NO and a decrease in NO₂ (Equation (10)). The effect of VOC emission on O₃ concentration is smaller than that of NO_x, due to the larger proportion of NO_x in traffic emissions. In conclusion, a reduction in traffic emissions (both NO_x and VOCs) in equal proportion can effectively reduce NO_x pollution, but it will impose negative effects on O₃ pollution. In order to effectively reduce O₃ pollution, we need to pay more attention to controlling VOC emissions than NO_x.

3.3. Effects of Building Arrangement on Pollutant Concentrations

Various 3D urban configurations can impact the dynamical processes of pollutants in the urban area, which will lead to the variation in urban air pollution at street level. In this section, the effects of regular and irregular building arrangements on photochemical pollutant concentrations (NO, NO₂ and O₃) are investigated. The real irregular and ideal regular urban street networks are shown in Figure 1.

The mean spatiotemporal concentrations of NO and NO₂ in regular arrangement decrease by 10.88% and 5.11% compared with those in irregular arrangement with a similar total building volume (Figure 5a,b). In contrast, the mean spatiotemporal concentrations of O₃ in a regular arrangement increase by 4.42% compared to an irregular arrangement with an equivalent total building volume (Figure 5c). We argue that in urban areas with a regular building arrangement, traffic-related air pollution such as NO_x can be dispersed more efficiently from the street canyon. Meanwhile, the reduction in NO_x in regular streets may result in an increase in O₃ concentrations, as NO_x acts as a precursor for photochemical reactions. However, the changes in mean spatiotemporal pollutant concentrations due to street arrangement are not significant. We further investigate the impacts of uniform building heights on air-pollutant concentration in actual street arrangement and find that there exists little difference between uniform and varying building heights (Figure 5). The reason for the little difference may be the low building height and limited building height variation in this urban area.

To further investigate the effects of street arrangement on spatial variations in the daily mean concentrations of NO_x and O₃, we conducted simulations summarized in Figure 6. We found that in real irregular street networks, the daily mean concentrations of NO_x (2119–3373 μg/m³) and O₃ (685–945 μg/m³) at the street level both vary a lot spatially, while in ideal regular urban areas, there are no significant spatial variations in daily mean concentrations for NO_x (2302–2329 μg/m³) and O₃ (883–891 μg/m³). Comparing the distribution of NO_x in regular (Figure 6a) and irregular (Figure 6b) street networks, we can see that the streets facing the incoming wind in the regular street network have uniformly lower concentrations compared to those in the back. However, this pattern is not observed in the irregular street network where the distribution of concentration is more random. This is because the varying heights of buildings create conditions for flow turbulence and thus hot spots for NO_x pollution. Due to the titration reaction of NO (Equation (8)), the distribution of O₃ daily mean concentrations (Figure 6c,d) is almost in opposition to that of NO_x daily mean concentrations. In real irregular street networks, the variations in building heights and arrangement can lead to certain hot spots of air pollution in urban areas, which can cause particularly high exposure to harmful pollutants for the nearby residents.

3.4. Effects of Increasing Building Height on Pollutant Concentrations

Urbanization has led to an increasing population and building height in urban areas. Dense high-rise buildings can obstruct the dispersion of pollutants [31]. The effect of the increasing building heights in the real urban area on the concentrations of photochemical pollutants (NO, NO₂ and O₃) is discussed (Figure 7).

In the real urban area, when the building height is increased to 3, 5, 7 and 10 times its original height, the mean spatiotemporal concentrations of NO increase by 22.49%, 57.02%, 92.53% and 110.11% (Figure 7a, solid symbols), and the mean spatiotemporal concentrations of NO₂ increase by 9.41%, 20.64%, 31.17% and 35.57% (Figure 7b, solid symbols), respectively. The variation is greater for NO due to the large proportion of NO in traffic emissions. During the morning and evening peak hours, urban areas with taller buildings experience significantly higher NO_x concentrations, suggesting that taller buildings can more easily weaken the dynamical processes of traffic-related air pollution in urban areas. Due to the titration reaction of NO (Equation (8)), the mean spatiotemporal concentrations of O₃ decrease by 6.73%, 12.97%, 17.28% and 19.53% (Figure 7c, solid symbols) when the building height is increased to 3, 5, 7 and 10 times its original height, respectively. We can see that even with greatly increasing NO in high-rise street canyons, the relief of O₃ concentration is less significant compared to NO_x. The weakened dynamic processes caused by high-rise buildings reduce the dispersion of all air pollutants including O₃ and thus increase the exposure of nearby residents.

It is argued that in compact urban areas with high-rise buildings, people will live closer and travel less, which can reduce traffic emissions. To quantify the effects of reduced emissions and increased building heights on pollutant concentrations, an empirical model based on the form of STIRPAT (stochastic impact by regression on population, affluence and technology) was used to simulate the traffic emissions with varying three-dimensional urban configurations [67]. Note that the effects of urban height, volume, average surface area and floor area ratio on emissions were neglected. The formula is adapted as below:

$$ln\left(Emission\right)=-8.81-0.0072\ln \left(DEI\right)-0.382\ln \left(BCR\right)+0.407ln\left(SCD\right)$$

(11)

$$DEI=\sqrt{\left[{{\displaystyle \sum }}_{i=1}^n{\left(V_i-V_{avg}\right)}^2\right]/A_{urban}}$$

(12)

$$BCR={{\displaystyle \sum }}_{i=1}^n{A}_i/A_{urban}\times 100\%$$

(13)

$$SCD=\frac{V}{AH\times A_{urban}}\times 100\%$$

(14)

DEI is the distribution evenness index; BCR is the building coverage ratio; SCD is the spatial congestion degree; $V_i$ is the volume of urban buildings, and $V_{avg}$ is the average volume (m³); $A_i$ is the area of urban buildings, and $A_{urban}$ is the area of urban land use (m²). According to the formula, when the building height is increased to 3, 5, 7 and 10 times its original height, the traffic emission is 92%, 89%, 87% and 85% of its original emissions, respectively. According to Figure 7, it can be observed that increasing building height along with reduced emissions can alleviate NO_x concentration and slightly increase O₃ concentration due to the titration reaction of NO (Equation (8)). However, the concentration change is much smaller than that caused by increasing building heights. For example, in cases with H = 5h and 0.89 times the original emissions, the NO_x concentration is still higher than that in cases with H = 3h and original emissions. Therefore, compact urban areas with high-rise buildings can cause accumulated air pollution that harms residential health even if there is less traffic emission in compact areas.

3.5. Effects of Background Pollution on Pollutant Concentrations

In urban areas, in addition to traffic emissions, natural emissions and regional transport are also important pollutant sources. We view the air pollution from these other sources as background concentrations in comparison to traffic-related air pollution. In this subsection, we evaluate the impact of varying the background concentration of O₃ and NO_x, as well as the associated chemical transformation processes, on urban pollutant concentration.

The increasing levels of O₃ are set according to the O₃ pollution level in the Pearl River Delta region (PRD), Guangdong Province. It should be noted that we adopted the air quality index in China to set the level of O₃ in the simulation, which is typically in the range of 1–4 in the PRD region (

Table 2. The air quality index and O₃ concentration of different levels in China.

Air Quality Index (AQI)	Air Quality Index Levels	O3 (1 h Average) (μg/m3)
0–50	Level 1 (Good)	0–160
51–100	Level 2 (Moderate)	160–200
101–150	Level 3 (Lightly Polluted)	200–300
151–200	Level 4 (Moderately Polluted)	300–400
201–300	Level 5 (Heavily Polluted)	400–800
300+	Level 6 (Severely Polluted)	>800

). Figure 8 illustrates the diurnal variation in photochemical pollutant concentrations under different levels of background O₃. We can see that the general trends of diurnal O₃ and NO_x concentrations do not change with varying levels of background O₃ concentrations. However, as shown in Figure 8b, the difference between the simulated urban O₃ region mean concentration and background O₃ concentration under various levels of O₃ pollution is considerable. Compared with the level 1 background concentration (Figure 8b), the mean spatiotemporal concentration of O₃ simulated in the urban area increases by 1.17 μg/m³. In this case, the photochemical reactions in urban areas tend to produce more O₃. When the urban O₃ concentration increases to levels 2, 3 and 4 (Figure 8b), the total O₃ in the urban area is constantly consumed, leading to a decrease in the mean spatiotemporal concentration of O₃ by 9.91 μg/m³, 17.00 μg/m³ and 22.00 μg/m³, respectively. With the aggravation of O₃ pollution, O₃ in the urban area increases NO consumption. With high levels of O₃ concentration in the afternoon, we observed low levels of NO concurrently. When the O₃ pollution is aggravated to levels 2, 3 and 4, the mean spatiotemporal concentration of NO in the urban area decreases by 20.75%, 34.03% and 43.40% (Figure 8c), and that of NO₂ increases by 14.57%, 23.72% and 30.03%, respectively (Figure 8d). The increasing O₃ background concentration invokes more NO reaction and less NO₂ reaction, which leads to a decreased NO concentration but an increased NO₂ concentration.

Reversely, in our simulation, we adjusted the background O₃ precursors (NO and NO₂) to 100%, 75% and 50% (Figure 9) of their original level to explore the influence of background precursor variations on photochemical pollutant concentration. In scenarios of 75% and 50% of the background NO_x concentration, the total NO₂ concentration in the urban area reduced to 79% and 58%, while the total NO concentration in the urban area reduced to 84% and 68% (Figure 9a). It can be observed that the total NO_x concentration did not decrease as much as the background concentration reduction in urban areas. This is because the NO_x from traffic emission sources slightly negated the reduction in total NO_x concentration. The higher proportion of NO in traffic emission sources is the cause for the discrepancy in total NO and NO₂ concentrations. After the background concentration of NO_x was reduced to 75% and 50%, the daily average concentration of O₃ decreased by 4.04% and 8.64% (Figure 9b). As background NO_x concentration reduces, a larger decrease in total NO₂ concentration is observed compared to NO, leading to a decreased production of O₃, hence reduced O₃ concentration. In the afternoon, the decrease in background precursors has little effect on O₃ concentration (Figure 9b). This is due to the intense photochemical reaction and sufficient accumulation of O₃ precursors from traffic emissions rather than background concentration in urban areas at that time.

3.6. Limitations and Future Research

This study has a few limitations. First, the limited simulation region is specific to a certain climate and urban structure where the simulation results can differ from other regions. Second, the simulated time we selected was mainly in spring. Chemical transformation processes are influenced by ambient temperature. Thus, the photochemical reaction rate under the simulated time and region may be different from that under other conditions. Third, MUNICH does not take into account the influence of solar radiation and heat on the urban dynamical process, which is of importance for the dynamic diffusion process of pollutants, especially in deep street canyons.

In the future, we will further explore the impact of solar radiation and heat on photochemical pollutant dispersion in urban districts. Moreover, previous research has verified that urban geometrical parameters also significantly influence urban ventilation [68,69,70], the temporal/spatial characteristics of the urban thermal environment/solar shading and outdoor thermal comfort [71,72], as well as the related urban energy consumption for summer cooling. Future research on sustainable urban planning should consider both the issue of urban air quality and that of urban ventilation/thermal comfort and urban energy use. In addition, measuring the health effects of exposure during commuting is worth investigation to calculate the precise trade-off of environmental health consequences between less traffic and more compact urban design [73].

4. Conclusions

In this study, the combined effect of pollutant sources, dynamical processes and chemical transformation on the variation in photochemical pollutant concentrations in urban areas was investigated by using the MUNICH model and real-world street geometry, traffic and background air-pollutant concentration data in Guangzhou, China. Specifically, we conducted extensive simulations to investigate the concentrations of urban photochemical pollutants (O₃ and NO_x) under various background and traffic-related pollution (VOCs and NO_x) levels and different building and street configurations. Based on our results, we find that among all the factors considered in simulations, building height has the most significant impact on urban photochemical pollutant concentration. Deeper street canyons would tremendously increase traffic-related air pollution levels due to the weak dynamical processes within them and hence the reduced dispersion of pollutants. Specifically, the increase in building height enhances the resistance to the dynamical process, resulting in elevated concentrations of traffic pollutants but reduced O₃ concentrations in urban areas. Even if traffic emissions are reduced with the more compact streets, this reduction cannot offset the negative effects of weakened dispersion in deep street canyons with tall buildings.

Another important factor that influences photochemical pollution is varied background concentration. Due to the chemical transformation process, the increasing background O₃ pollution leads to more NO reaction and less NO₂ reaction, which will relieve NO pollution but exacerbate NO₂ pollution. With the increased background NO_x pollution, the larger reduction in NO₂ than NO will lead to less production of O₃ and hence relieved O₃ pollution. Reducing background NO_x has a smaller effect on the concentration of O₃ due to the large portion of NOx from traffic emissions during the day. Moreover, traffic emissions play an important role in urban photochemical pollution variations. Reducing the overall traffic emissions can effectively reduce NO_x pollution, though it has slight adverse effects on O₃ pollution. If VOC and NO_x emissions are reduced in equal proportion, the titration reaction of NO will reduce more due to the large proportion of NO in traffic emissions. Reducing VOC emissions can effectively reduce O₃ concentration, but its effect is less than that of NO_x emission due to the small proportion of VOCs in traffic emission. The street arrangement can also impact pollutant dispersion in urban areas, with irregular building heights and arrangements leading to hot spots of air pollution, causing high exposure to harmful pollutants for nearby residents. However, the effect of regular and irregular street arrangement and building heights on mean spatiotemporal pollutant concentrations is not significant.

This study has provided valuable insights for urban planners and policy makers in urban photochemical pollution mitigation. First, deep and irregular urban configurations should be avoided to achieve better pollutant dispersion. Moreover, policy makers should pay attention to controlling VOC emissions in urban areas, which can significantly alleviate the O₃ concentration. Last but not least, background pollution from natural emissions and long-distance transport needs to be monitored and tackled.