How Sensitive Morphological Parameters Influence on the PM_2.5 Diffusion: An Empirical Study of Two Neighborhoods in Central Beijing

Abstract

Air quality is highly related to the health of a human being. Urban morphology has a significant influence on air quality; however, the specific relationship between urban morphology characteristics and air quality at the neighborhood scale has yet to be investigated, especially the vegetation effect on PM_2.5 concentration and diffusion. The relevant morphological parameters based on the affected pathways of urban morphology on air quality were selected, and the sensitivity degree and laws of the selected morphological parameters to PM_2.5 were quantified by numerical simulation, bivariate correlation analysis, and regression analysis. The results showed that Building Density (BD), Block Envelope Degree (BED), Average Building Volume (ABV), Average Building Floors (ABF), Standard Deviation of Building Height (SDH) and Greenbelt Coverage Rate (GCR) were Sensitive Morphological Parameters (SMPs). A positive and cosine curve trend of BD and BED with PM_2.5 was observed. GCR was significant to dust retention along with vertical canopy height. When ABV = 40,000 m³ and ABF = 20F, the lowest PM_2.5 concentration was examined, while increased SDH could promote airflow and enhance the capacity of PM_2.5 diffusion. Finally, morphology-optimization strategies were proposed at the neighborhood scale: (1) Decreasing the BED along the street; (2) considering the species of vegetation with the appropriate height and increasing the GCR; (3) increasing the ABF of neighborhoods appropriately while controlling the ABV and distinguishing the internal SDH of neighborhoods. The study could apply the scientific basis for the planning and design of healthy and livable cities.

1. Introduction

Rapid urbanization has resulted in air pollution issues that had a negative impact on many sectors of human lives. According to the Beijing State of the Environment Bulletin 2020 [1], the number of days that met air quality standards in 2020 was 276 days, accounting for 75.4%. Although the overall air quality has improved compared to the previous period, the distribution of pollution still showed north–south differences, and the concentration of pollutants such as PM_2.5 and PM10 in ecological zones in the north and northwest is significantly lower than that in the southern high-density built-up areas and high-density population areas, which showed a “Low-Northwest while High-Southeast” situation. As urban planning became a more essential component of the development of livable cities, how to enhance air quality by optimizing urban morphology evolved into a focus of investigation in relevant disciplines [2,3].

Based on the foregoing, studies on urban morphology and air quality were steadily established [4]. Studies mainly involve two scales: regional-urban [5] and neighborhoods [6]. At the regional-urban scale, most scholars had focused on exploring the intrinsic effects of different urban morphological features on air quality. Research elements included physical spatial characteristics such as city size and urban shape, [7] vegetation cover [8], and non-physical characteristics such as population density and employment density [9]. In addition, studies have also been conducted to analyze the correlation between urban morphology and air quality from a spatial and temporal perspective [10].

At the neighborhood scale, most studies had conducted comparative studies of neighborhoods with different morphological characteristics for the correlation between urban morphology and pollutants [11,12] with air pollutant monitoring data from urban observatories. However, the specific relationship between urban morphological parameters and PM_2.5 concentration as well as diffusion has not been investigated clearly. In addition, the generalization of urban morphological characteristics needed a systematic and comprehensive framework. Since the selection of vegetation and non-physical morphological indicators is relatively weak in existing studies, they are unable to provide universal laws.

In recent years, the development of simulation techniques such as Computational Fluid Dynamics (CFD) has provided the technical support to establish the correlation between urban morphology and pollutant dispersion at the neighborhood scale with the Fluent simulation software being the most widely utilized. Studies could be divided into two categories. First, ideal-neighborhood simulation based on traditional settlement patterns [13] was constructed (lineal type, point group type, etc.). Different building combinations [14] or vegetation layouts [15] were explored individually, and the effect of different morphological features on the dispersion of pollutants based on simulation results was qualitative or quantitative analyzed. Secondly, a simulation based on actual cases was constructed. Different urban design schemes for the same neighborhoods [16] or comparisons of different neighborhoods [17] have been studied to promote the air quality. In general, the correlation between urban morphology and air quality had been gradually established; however, the quantitative guidance was limited. The simulation studies of ideal neighborhoods were separated from the complicated morphology of the building arrangement in the actual environment, and the simulation data were based on empirical data. In addition, in terms of modeling, the impact of the integrated neighborhood environment of buildings and vegetation on pollution dispersion has not been considered in previous studies, while the simulation studies of actual cases were aimed at promoting the air quality of specific public spaces and neighborhoods, which lacked the general application.

The study, which focused on two typical residential neighborhoods with different features in central Beijing, explored the quantitative rules of affection between urban morphology and air quality. We devised an urban morphology and air quality mechanism for selecting morphological parameters. Through the neighborhood-scale CFD simulation, which includes the calibration of vegetation factors and multi-source data, the Sensitive Morphological Parameters (SMPs) impacting air quality (PM_2.5, for example) at the neighborhood scale were filtrated before statistical models. Therefore, the quantitative rules of the influence of SMPs on the “pollutant-wind environment” could be estimated.

2. Data and Methods

2.1. Study Area

Taking a traditional residential area in central Beijing as the study area, the study selected a low-rise residential neighborhood (Neighborhood A) and a high-rise residential neighborhood (Neighborhood B) as the core study area based on the street network, buildings’ layout, and its group form (Figure 1a). Neighborhood A, which was built in the 1990s, is dominated by enclosed low-medium-rise residential buildings with schools and underlying retail; Neighborhood B, which was built in the early twenty-first century, is dominated by row-slab high-rise residential buildings with a few underlying retails. The aforementioned two neighborhoods differ in terms of building periods, functional placement, and spatial arrangement, which might illustrate the main features of Beijing’s residential neighborhood morphology.

2.2. Wind Environment and Pollutant Monitoring

The objective of the field monitoring is to count the concentration levels of pollutants in the neighborhoods as well as the characteristics of the wind environment and to conduct a preliminary analysis of the differences in pollutant concentrations and wind environment between neighborhoods, which serves as a foundation for testing and validating simulation results. The XL68 intelligent environmental monitoring equipment is chosen to monitor PM_2.5 concentration (resolution: 1 μg/m³, range: 0~1000 μg/m³) and wind speed (resolution: 0.1 m/s, range: 0~60 m/s) of the neighborhoods at monitoring points P1 and P2 (z = 3 m, z = 2 m) (Figure 1b). Details of equipment are shown in Table S1.

2.3. “Pollutant-Wind Environment” Model Setting

Micro-scale CFD numerical simulations have been widely used in the simulation of outdoor wind environments and pollutant dispersion [17]. ANSYS FLUENT 21.0 based on the finite volume method was adopted for numerical simulations, and the governing equation was the Reynolds-averaged Navier–Stokes equation. The standard K-ε turbulence model was adopted to simulate the airflow [18]. Pollutants and air were considered as continuous phases, and the pollutant concentrations were solved with the component transport model [19,20].

2.3.1. Computational Domain and Grid Generation

The calculation domain was constructed according to the method specified by the European Cooperation in the Field of Scientific and Technical Research (COST) [21], keeping a minimum of 5H for vertical distances (H—the maximum building height) and 5H for horizontal and horizontal distances (Figure 2). At the same time, an unstructured meshing method based on a hexahedron was adopted to save computational costs. Three sets of coarse–medium–fine meshes were divided, and grid irrelevance was tested. The final grid was 2.4 × 10⁸ for Neighborhood A and 2.6 × 10⁸ for Neighborhood B.

2.3.2. Boundary Condition

The incoming wind speed is exponentially distributed with height [22]. For the simulation, the calculation domain entry was set as the velocity-inlet boundary condition and adopted a user-defined function:

$$\mathrm{U}=\frac{{\mathrm{U}}_{*}}{\mathsf{\kappa }}\ln (\frac{{\mathrm{z}+\mathrm{z}}_0}{{\mathrm{z}}_0})$$

(1)

• U—horizontal wind speed at height z(m), m/s
• U_*—ground friction speed, m/s
• κ—Von·Karman constant, κ = 0.42
• z₀—surface roughness, z₀ = 0.25

PM_2.5 was mainly emitted from traffic emissions and was relatively stable by default. Pollutants were emitted vertically upwards at 0.5 m/s, and the source intensity was from the nearest urban monitoring station on the simulation day. The zero static gauge pressure outlet was used for the downstream boundary condition, and the standard wall functions with roughness modification were used for the building surface and the bottom of the computational domain. The roughness height was 0.0025–0.003 m, and the roughness constant was 0.75. Symmetry boundary conditions were served to the side-face computational domain and the upper-face computational domain [23]. Detailed boundary conditions are shown in Figure 3.

The study treated the canopy section of the tree with a porous medium due to the influence of trees on the surrounding flow field in reducing wind speed and increasing flow disturbance. According to the relevant literature [24], the modeling of the influence of tree canopy on the flow field was accomplished by adding source terms to the momentum equation, the K equation and the ε equation, respectively. The porosity was 0.7, the inertial resistance was 0.18, and the viscous resistance was 1.67. Meanwhile, pollutant sorption and deposition by trees were adjusted to a constant value, and the rate of deposition was determined by wind speed and pollutant concentration [25]:

$${\mathrm{Y}}_{\mathrm{PM}2.5}=\mathrm{v}·\mathrm{d}·\mathrm{LAD}·\mathrm{t}$$

(2)

• Y_PM2.5—Pollutant adsorption capacity per unit area (μg/m²)
• v—Adsorption rate (m/s)
• d—Pollutant concentration (μg/m³)
• LAD—leaf area density; (m²/m³)
• t—Adsorption time (s)

2.3.3. Solution Settings

The finite volume method was used to discretize the control equation, solved by the SIMPLE algorithm, and the second-order upwind algorithm was adopted. In the initial condition setting, the ground observation data of the Beijing meteorological station on typical dates (

Table 1. Meteorological data of national control stations on simulation dates.

Date	Wind Direction	Wind Speed (m/s)	PM2.5 (μg/m3)	Temperature (°C)
7.10	E	2.0	44	24.6
7.13	SE	1.7	55	28.6
10.14	N	1.8	9	9.2
10.26	NW	1.6	30	8.8

) were used as the initial conditions for the simulation. PM_2.5 monitoring concentrations close to those of national control stations were used as the basis for selecting typical dates, and four typical dates with typical meteorological characteristics during the monitoring period were selected to establish the CFD numerical model.

2.4. Selection and Extraction of Urban Morphological Parameters

2.4.1. Selection of Urban Morphological Parameters

Based on the research framework of urban morphology influencing air pollutant transport (Figure 4) and the generalization of existing studies, a system of six categories of urban morphological characteristics, including size, density, function, structure, shape and immaterial morphological characteristics, were constructed.

The six morphological characteristics stated above have been proved to have a direct or indirect impact on air quality and the urban microclimate. In particular, urban size affects the urban microclimate, pollutant emissions, and dispersion transport, and pollutant concentrations increase significantly as urban size increases [26]. The association between land use and air pollutants is more obvious, and air pollution is severe in industrial sites and commercial districts with greater emission sources [27]. In contrast, green spaces and water bodies can improve the local microclimate [28] and reduce the concentration of PM_2.5 in the region [29]. Different density characteristics demonstrate a broad range of heterogeneity in the routes of effect on air quality, such as the impact of building density on the alteration of local wind fields, which influences pollution dispersion [30,31], and the impact of road density on traffic pollutants [9]. Urban layout structure is a major factor affecting the wind environment, and a large number of scholars have conducted detailed studies on the elements of layout structure with wind environment and pollutant levels [32,33]. Most of the studies on shape features have focused on investigating the effects of different building shapes and combinations of building morphologies on the wind environment and pollutant dispersion [34]. Furthermore, since physical urban morphology is the spatial projection of non-physical urban morphology on land use [35], the adoption of the immaterial morphological indicator is highly relevant to the overall morphology of the neighborhoods.

Thus, ten morphological parameters were selected for investigation based on the principles of neighborhood scale, potential impact on pollutant levels, ease of implementation at the control and design stages, and the interaction mechanisms between the preceding morphological features and air quality, as well as the research progress of relevant literature (

Table 2. Selection of characteristic morphological parameters of neighborhoods.

Morphological Characteristics	Morphological Parameters	Calculation Method
Size Characteristics	Total building area, TBA	$\mathrm{TBA}={{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{S}}_{\mathrm{i}}{*\mathrm{F}}_{\mathrm{i}}$ Si—Building single-story area Fi— buildings Floors
	Floor area ratio, FAR	$\mathrm{FAR}=\mathrm{TBA}/\mathrm{SA}$ SA—Neighborhoods area
Functional Characteristics	Greenbelt coverage rate, GCR	$\mathrm{GCR}=\mathrm{TGA}/\mathrm{SA}$ TGA—Area of horizontal vegetation projection
Density Characteristics	Building density, BD	$\mathrm{BD}=\mathrm{BBA}/\mathrm{SA}$ BBA—Building footprint
Structural Characteristics	Block envelope degree, BED	$\mathrm{BED}=\mathrm{TBP}/\mathrm{TSP}$ TBP—Building envelope perimeter TSP—Neighborhood perimeter
	Space openness, SO	$\mathrm{SO}=(1 - \mathrm{BD})/\mathrm{FAR}$
Shape Characteristics	Average building volume, ABV	$\mathrm{ABV}=\frac{{{\displaystyle \sum }}_{\mathrm{i} =1}^{\mathrm{n}}{\mathrm{V}}_{\mathrm{i}}}{\mathrm{n}}$ Vi—Building volume
	Average building floors, ABF	$\mathrm{ABF}=\mathrm{FAR}/\mathrm{BD}$
	The standard deviation of building height, SDH	$\mathrm{SDH}=\sqrt{\frac{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{{(\mathrm{h}}_{\mathrm{i}} - \overline{\mathrm{h}})}^2}{\mathrm{n}}}$ $\overline{\mathrm{h}}$—Average building height
Immaterial Characteristics	Population density, PD	$\mathrm{PD}=\mathrm{TP}/\mathrm{SA}$ TP—Neighborhoods population

).

2.4.2. Extraction of Urban Morphological Parameters

The subject area’s similarities and differences can be quantitatively investigated by the division of calculating units. The neighborhoods were divided into 20 units according to road boundaries, spatial structure divisions and the size of typical urban neighborhoods by a 200 × 200 m grid (Figure 5). The Beijing Institute of Surveying and Mapping was used to obtain the 3D environmental data of the neighborhoods, while the sociological data were obtained from the field survey. The selected morphological parameters were calculated separately in ArcGIS 10.5.

2.5. Selection of Indicators for the Evaluation of “Pollutants–Wind Environment”

The “wind speed ratio” [23] evaluation index refers to the ratio of the wind speed at the actual selected location to the incoming wind speed on that day, which is often used to reflect the degree of influence of different areas or buildings on the wind speed, as a way to evaluate the condition of the wind environment in the region. “Pollutant concentration ratio” means the ratio of the concentration of pollutants in different areas to the concentration of incoming pollutants, which is used to quantify the relative level of pollution in a local area.

$$V{R}_w=\frac{V_p}{V_{\infty }}$$

(3)

$$C{R}_p=\frac{C_p}{C_{\infty }}$$

(4)

In Equations (3) and (4), $V{R}_w$ is the wind speed ratio; $V_p$ is the wind speed value at a certain height in a region (m/s); $C{R}_p$ is the pollutant concentration ratio; $V_p$ is the average concentration of pollutants at a certain height in a region (μg/m³); $V_{\infty }$ is the average concentration of incoming pollutants in a region (μg/m³).

2.6. Statistic Analysis Model

The Pearson bivariate correlation analysis was used to filtrate the Sensitive Morphological Parameters (SMPs) with a high correlation to PM_2.5 and wind speed. To avoid the effect of excessive differences in the morphology of different neighborhoods, a correlation analysis was used between the GCR (A) in Neighborhood A and the GCR (B) in Neighborhood B separately.

The Curvilinear Regression was used to estimate the influence pattern of SMPs with PM_2.5 and wind speed. SMPs in 20 calculation units are used as independent variables, and the simulation results of the $C{R}_p$ and $V{R}_w$ at 1.5 m height are used as dependent variables.

3. Results

3.1. Monitoring Results

Field monitoring is shown in Figure 6. Overall, trends in pollutant concentrations within neighborhoods are influenced by overall urban background concentrations, and the two neighborhoods are relatively close but again show part of the local variability.

3.2. Simulation Results and Error Analysis

In the correlation test (Figure 7), the two data sets showed a high correlation (R2 = 0.82, 0.77). In the paired t-test (

Table 3. P1 and P2 simulated data were paired with measured data for the sample test.

Comparing Criteria	Paired Differences					t	sig. (2-Tailed)
	Means	Std. Deviation	Std. Error Mean	95% Confidence Interval of the Difference
				Lower	Upper
Velocity	0.00875	0.026424	0.009342	−0.013341	0.030841	0.937	0.380
PM2.5	8.27500	13.066615	4.619746	−2.648964	19.198964	1.79	0.116

), the data significance (P) was greater than 0.05 for both groups at 95% confidence, which means that there was no significant difference before and after the simulation.

As a result, the CFD numerical model developed can predict the neighborhoods’ PM_2.5 concentration and wind environment more accurately. It can be used to predict a neighborhoods’ PM_2.5 concentration and wind environment under different morphological parameters.

3.3. Sensitive Morphological Parameters Filtrating

Table 4. Correlation analysis of morphological parameters with PM_2.5 and wind speed.

Analysis of Variables	$\mathit{C}{\mathit{R}}_{\mathit{p}}$		$\mathit{V}{\mathit{R}}_{\mathit{w}}$
	R	P	R	P
TBA	−0.246	0.325	0.114	0.268
FAR	−0.246	0.325	0.114	0.268
BD	0.443	0.025	−0.709 **	0.000
SO	0.221	0.379	−0.217	0.388
BED	0.401	0.035	−0.636 **	0.001
ABV	−0.564 **	0.005	0.505 *	0.012
ABF	−0.628 **	0.002	0.684 **	0.000
SDH	−0.612 **	0.001	0.573 **	0.004
GCR (A)	−0.912 **	0.001	−0.818 **	0.007
GCR (B)	−0.726 **	0.001	−0.810 **	0.003
PD	0.162	0.520	−0.128	0.296

** The correlation is significant at the 0.01 level (two-tailed). * The correlation is significant at the 0.05 level (two-tailed).

shows the results of the correlation analysis between morphological parameters and the $C{R}_p$ and $V{R}_w$ of different neighborhoods. Among the 10 morphological parameters, BD, BED, ABV, ABF, SDH and GCR showed strong sensitivity (p < 0.05) to $C{R}_p$ and $V{R}_w$ with correlation coefficients between 0.4 and 0.8. It is suggested that there is a slight interaction between morphological parameters and pollutant concentrations and wind speeds, and alterations in these morphological parameters can result in more sensitive responses in PM_2.5 concentrations and wind conditions, which are Sensitive Morphological Parameters (SMPs). In contrast, the four parameters TBA, FAR, SO and PD were less sensitive (p > 0.05) to the PM_2.5 concentration ratio and wind speed ratio, with correlation coefficients below 0.3. It is suggested that alterations in these morphological features do not appear to produce more sensitive changes in PM_2.5 concentration and wind environment response.

Furthermore, the correlation coefficient R suggests that each SMP demonstrates an inverse trend of association between wind speed and PM_2.5. Among them, BD, BED and GCR show a positive correlation with PM_2.5 and a significant negative correlation with wind speed. ABV, ABF and SDH show part of negative correlation with PM_2.5 and a significant positive correlation with wind speed. Therefore, the SMPs influence the transport and dispersion of air pollutants primarily via altering the wind environment in the surrounding area: increased wind speed enhances the transport and dispersion of air pollutants, resulting in reduced air pollutant concentrations. In general, the wind environment is still the main factor influencing the dispersion of pollutants. In the case of identical neighborhoods’ land uses, the different morphological parameters affect the dispersion of pollutants mainly indirectly by influencing the wind environment.

4. Discussion

4.1. Trend Analysis of the Influence of SMPs

Figure 8 shows the interaction curves of the six SMPs with PM_2.5 and wind speed. It also demonstrates that the effect of each SMP on PM_2.5 concentration and wind speed has an inverse connection within a particular interval, but the outcomes of other intervals on PM_2.5 concentration and wind speed are different.

According to Figure 8, wind speed is the most important element influencing PM_2.5 dispersion, and high wind speeds promote pollution transport and dispersion. The spatial morphological characteristics of distinct neighborhoods define their internal wind environment under the same incoming wind speed conditions, which has an impact on the transport dispersion and concentration distribution of air pollutants. Furthermore, the different impact results are exhibited by the morphological parameters in different intervals. It is claimed that there are parameter intervals for each morphological parameter that are more positive (or less positive) to the transmission and dispersion of air contaminants.

4.1.1. BD and BED

The relationship between BD, BED and PM_2.5 and wind speed all show a trend of sine and cosine curves (Figure 8a,c). As BD or BED rises, PM_2.5 shows a trend of decreasing, while wind speeds show a decrease followed by an increase, then increasing and then decreasing. In particular, PM_2.5 is lowest when BD is around 10% and continues to rise above 10%, reaching a maximum of PM_2.5 at around 20%. PM_2.5 is lowest at a BED of around 3 and highest at 5. When BD = 20% or BED = 5, the corresponding wind speed is at its minimum. The effect of BD and BED on the wind environment can be further seen by comparing the local area wind speed vectors for Neighborhood A and Neighborhood B (Figure 9): BD and BED of Neighborhood A with the relatively smooth internal wind environment is higher than Neighborhood B. This is because when the BD and BED are within a specific range, the transport of pollutants from outside is considerably restricted. However, when the BD and BED rise to a certain level, the neighborhood’s wind environment tends to stabilize, which is not conducive to the migration and dispersion of atmospheric pollutants and creates a cumulative effect. Therefore, keeping the building density and block envelope degree of neighborhoods within a reasonable range can assist in enhancing air quality.

4.1.2. GCR

As the GCR increases, PM_2.5 concentration and wind speed all showed a decreasing trend (Figure 8b). This is because wind speed is reduced and dust is suppressed by vegetation in the vertical zone beneath the canopy, with a progressive decrease in the zone above the canopy. As an example, Figure 10 illustrates the local vegetation XZ plane wind speed and PM_2.5 distribution in Neighborhood A, which suggests that the influence of vegetation on pollutant concentrations and wind speed is related to the height of the vegetation and its dust retention effect is most noticeable in the vertical zone beneath the canopy, with a progressive decrease in the zone above the canopy.

4.1.3. ABF, ABV and SDH

ABF, ABV and SDH demonstrated varying degrees of positive and negative correlations with PM_2.5 and wind speed, respectively. When ABV = 40,000 m³ and ABF = 20F, the lowest PM_2.5 concentration was observed. The rise in ABF indirectly increases the building separation and enhances the air circulation within and outside the neighborhoods. When ABF exceeds 20, wind speed is further increased, but the static wind zone is formed on the leeward side of the building, which hinders the diffusion of pollutants (Figure 8d). As the ABV increases, PM_2.5 concentration shows a tendency to decrease and then increase (Figure 8e). This is because, within a certain volume range, increasing ABV helps to increase the open space and reduces the weakening effect on wind speed due to a large number of buildings. However, excessive ABV might increase the static wind area in the neighborhood and limit the effect of wind on pollution transmission.

The results reveal that the difference in height between buildings on the windward and leeward sides of the neighborhoods affects wind speed in the direction of incoming airflow differently (Figure 8f). Taking the example of four street valleys in the typical YZ plane of two neighborhoods (Figure 11), rising valley 1 and valley 4 are more effective in diffusing PM_2.5 in vertical space than falling valley 2 and valley 3. Therefore, reasonable SDH regulation promotes airflow rising, establishing a pleasant neighborhood wind environment, and boosting pollutant dispersion in vertical space. Moreover, angular flow zones are created with building heights up to a certain level, which enhances the dispersion of pollutants to the leeward side.

4.2. Morphology-Optimization Strategies for Pollutant Dispersion at Neighborhood-Scale

Based on the findings of the preceding investigation, strategies for optimizing neighborhood morphology based on air quality improvements were proposed. First, the relationship between BD, BED and PM_2.5 show a trend of sine and cosine curves. Therefore, BD should be reasonably controlled in the neighborhoods, and the BED of residential buildings along the street should be reduced while increasing the commercial buildings along the street (Figure 12a). Second, in terms of functional morphological characteristics, GCR showed a reduced influence on wind speed and PM_2.5 below the canopy, and the reducing effect decreases as the vertical height above the canopy increases. Therefore, it is critical to plant a diverse variety of vegetation species of varying heights (Figure 12b). Furthermore, it is important to increase the GCR while ensuring the functional integrity of the neighborhoods. Third, in terms of the shape morphological characteristics, ABF and ABV showed a trend of increasing and then decreasing with PM_2.5 values. Therefore, when the intensity of development is identified, specific buildings’ heights should be increased, and the individual building’s masses and number of total buildings in the neighborhood should be limited (Figure 12c). Additionally, the difference of building heights should be reasonably delineated (Figure 12d), according to the SDH curve fitting, to promote the climbing of incoming winds.

5. Conclusions

This study aims to investigate the mechanisms underlying the correlation between urban form and atmospheric pollutants (PM_2.5, for example), and two types of typical residential neighborhoods in Beijing were selected as the study area. Morphological parameters were selected according to the research pathway of urban morphology affecting air quality, and a sensitivity analysis of morphological parameters with PM_2.5 and wind speed was carried out through field monitoring and CFD numerical simulation.

In the sensitivity filtrating, six morphological parameters, such as BD, BED, etc., showed high sensitivity to PM_2.5 concentrations and wind speed within the neighborhood, which are called the Sensitive Morphological Parameters (SMPs). The different correlations of SMPs between PM_2.5 and wind speed were observed. This demonstrates the existence of a tripartite relationship between morphological characteristics, wind environment, and pollutant dispersion.

The SMPs showed different influent rules on the PM_2.5 diffusion. It revealed a positive and cosine curve trend of BD and BED with PM_2.5. PM_2.5 is lowest when BD is around 10% and BED around 3 and continues to rise when BD is above 10%, reaching a maximum of PM_2.5 when BD is at around 20% and BED is at 5. GCR was significant to dust retention along with vertical canopy height, with a most noticeable effect in the vertical zone beneath the canopy and a progressive decrease in the zone above the canopy. When ABV = 40,000 m³ and ABF = 20F, the lowest PM_2.5 concentration was observed. Increased SDH could promote airflow and enhance the capacity of PM_2.5 diffusion.

To optimize the circumstances of pollutant dispersion, three residential planning strategies were proposed. First, the BED of residential buildings along the street should be minimized while commercial buildings along the street should be expanded, and BD should be reasonably managed. Second, vegetation species of appropriate height should be considered, and GCR should be increased. Third, building height should be increased appropriately, as should a proper division of building height disparities in neighborhoods. Furthermore, acceptable control of individual building mass and the total number of buildings in the neighborhood should be considered.