Research on the Effects of Spatial Forms in Residential Blocks on Road Traffic Noise Distribution in Typical City of China

Abstract

Integrating the block system into construction is the current trend in the development of residential areas in China. Road traffic noise is the major noise source in residential blocks, and its relationship with spatial forms of blocks remains unclear. In this study, 852 block models (258 block road network models and 594 models with buildings inside) were established and simulated with SoundPLAN 8.2 software, in order to reveal the impact of spatial forms (road network morphology, neighborhood scale, and architectural texture) on the road traffic noise distribution in residential blocks. Meantime, a prediction model based on spatial morphology parameters is proposed. It was found that (1) Without considering the impact of buildings, both the road network morphology and neighborhood scale parameters have significant effects on the distribution of road traffic noise, but road network morphology has a larger effect than neighborhood scale. (2) In the presence of buildings within the block, architectural texture parameters have effects on the distribution of road traffic noise, but to a lesser extent than road network morphology and neighborhood scale parameters. (3) This research employs principal component analysis to reduce the dimensionality of urban spatial form parameters. Subsequently, a model was developed to predict overall noise exposure levels in residential areas, which was validated by example. This model can be used as a tool for rapid prediction and diagnosis of the block acoustic environment. These findings offer insights for the planning and design of residential blocks from the perspective of optimizing the acoustic environment.

1. Introduction

As urbanization rapidly advances in China with a burgeoning urban population, issues such as road congestion and traffic noise have been exacerbated, making the living environment less comfortable. For the foreseeable future, a significant portion of urban residents will reside in high-rise residential blocks. The outdoor acoustic environment, as a critical part of a neighborhood’s physical environment, plays an undeniable role in how favorable the living environment is. Road traffic noise is a primary source of noise in the outdoor acoustic environment of residential blocks [1,2,3,4], and long-term exposure to it negatively affects both the physical and mental well-being of people [5,6,7,8,9,10,11,12].

In urban development in Western countries, the block system typically follows the form of “small blocks with a dense road network”, in which residential and commercial functions, as well as some other public business types, are integrated to create open blocks toward the city. Assessing the adaptability of the block system in China calls for a reexamination of the essence of the relationship between buildings, roads, and urban spaces. The scale of blocks and the layout of internal road networks directly determine the basic urban morphology and influence residents’ daily travel patterns. Since residential buildings were commercialized in China, gated residential communities have been more welcomed than open blocks as the latter are thought to compromise privacy and create more traffic noise. Previous studies largely focus on completed residential areas. For example, a study compared enclosed and open residential communities and found that traffic noise is a key factor influencing soundscape evaluation [13]; assessments of noise in open residential areas located along main roads have shown severe disruptions caused by traffic noise to residents [14]. However, enclosed and open blocks with the same size and traffic volume, with the influence of buildings excluded, have been insufficiently studied. Additionally, significant changes have emerged in the form, structure, and layout of high-rise residential buildings in China over the past year due to constraints imposed by fire safety and other regulations. The real estate market is placing greater importance on high-quality residential construction. How new high-rise residential buildings can be arranged in open blocks and how spatial forms can be adjusted to mitigate the impact of road traffic noise are pressing issues to be addressed.

As exploring the relationship between architecture and space is the key to grasping the spatial characteristics of blocks, it is necessary to analyze the composition of their spatial forms. In this study, residential blocks refer to urban residential units of a certain size enclosed by road networks. Residential neighborhoods refer to residential land enclosed by urban roads such as branch roads or land boundaries in a residential block. They are basic residential units composed of residential buildings and open spaces [15]. As a block is divided into neighborhoods with different sizes by road networks, its spatial form is constrained by that of road networks. A neighborhood consists of buildings and open spaces, and its spatial form depends mainly on the volume and layout of buildings. Therefore, the impact of the spatial forms of open blocks on acoustic environments should be examined at two levels, the block and the neighborhood.

Researchers have investigated the relationship between urban spatial forms and acoustic environments from different angles and made significant achievements [16]. For instance, a study used the SoundPLAN simulation software to analyze the relationship between urban shapes and environmental noise in the city of Aracaju (Brazil). The results reveal that factors such as construction density, open spaces, and the shape and physical position of buildings greatly influence environmental noise [17]. Another study assessed the noise levels of 212 residential complexes in Hong Kong using the noise mapping method and found a strong correlation between the noise characteristics and morphological indicators at the urban level [18]. Jian Kang analyzed the characteristics of the sound fields in urban streets and squares in the UK and the effectiveness of architectural changes and urban design options, with factors such as the shape of squares or streets and relevant architectural parameters taken into account [3]. Further studies have proven that building-related factors, such as building density, height, and the length of openings along street-facing interfaces, can impact traffic noise distribution [17,19]. Lengths of road segments, the arrangement of intersection nodes, areas of streets, road network density, road area density, road coverage, and other road-related factors also influence the distribution of urban traffic noise in urban spaces and on building facades [20,21,22]. Measured data from different neighborhoods proved that architectural layouts and building dimensions influence the acoustic environment [23]. The finite-difference time-domain (FDTD) method was used to find that street shapes and building facades can mitigate the impact of road traffic noise [24]. A study on the morphology of Macao’s historical districts indicates that narrower roads, complex road networks, and higher intersection densities can reduce traffic volume, thus abating noise pollution [25]. These studies represent significant explorations that integrate acoustic environments with planning and design, demonstrating the correlation between acoustic environments and spatial forms of blocks. However, most studies focus solely on building and road factors in these spatial forms, targeting specific regions. In addition, spatial models are often idealized and lack integration with actual design standards and construction regulations. As a result, this study categorizes the urban spatial form parameter system into three distinct categories: road network morphology, neighborhood scale, and architectural texture. It then combines the empirical values of the classical block spatial morphology parameters from both domestic and international contexts with the actual specifications for modeling, with the aim of elucidating the influence mechanism of each type of parameter on the acoustic environment and clarifying the threshold values of some parameters.

In recent years, computer simulation and analysis technology has been widely used to predict and evaluate urban acoustic environments [26,27,28,29,30]. Aletta et al. proposed predictive soundscape models based on existing acoustic indices and soundscape descriptors [31]. Kang et al. developed models for mapping soundscapes in smart cities and monitoring and managing the acoustic environment [32]. Machine learning methods have also been employed to build models for predicting acoustic environments and soundscapes [33,34], and complex networks have been used to analyze the spatial and temporal characteristics of acoustic environments [35]. Most of these models rely on physically measured auditory parameters [36]. These models are not built specifically for residential blocks despite their broad application. Therefore, this article will form a prediction model for the acoustic environment of residential blocks, so as to play a practical application value for the construction of residential blocks.

Residential blocks are typically adjacent to or enclosed by urban roads, with branch roads intersecting within. These blocks, compared to other urban areas, have denser populations and buildings. Intensive resident activity leads to high levels of road traffic noise, and denser urban road networks inevitably channel more traffic into blocks. While this increase in internal traffic within the blocks helps alleviate traffic pressure on surrounding roads, acoustic environments in these blocks may be influenced. Previous studies have demonstrated the significant impact of diverse spatial morphology parameters on the urban sound environment from multiple perspectives [17,18,19,20,21,22,23,24,25]. Nevertheless, existing research has yet to systematically categorize urban spatial form parameters in order to more closely align them with the work habits of design practitioners. Furthermore, there has been comparatively little research conducted with the aim of translating the impacts of spatial form on the acoustic environment into planning and design methods. To achieve rapid prediction of the acoustic environment in new settlements through spatial morphology and to propose targeted adjustments in the early stages of design by planners, this study aims to elucidate the mechanisms by which the spatial forms of residential blocks affect the distribution of road traffic noise. To address these issues, the following questions need to be answered: (1) What is the relationship between the road network morphology, neighborhood scale of blocks, and road traffic noise distribution without considering the impact of buildings? (2) What is the impact of the architectural texture parameters on road traffic noise distribution when there are high-rise residential buildings in the block? (3) How can spatial form parameters be used to quickly predict the quality of the acoustic environment in residential blocks during the initial planning and design?

2. Materials and Methods

2.1. Classification of Spatial Form Parameters of Residential Blocks

Previous studies have proven that people’s subjective evaluations of acoustic environments are strongly connected to the levels of noise exposure and the spatial characteristics of noise sources, all of which are influenced by changes in the spatial forms of blocks. The spatial form, as an outcome of residential block planning and design, impacts the quality of the acoustic environment in a block. Therefore, it is essential to further analyze, summarize, and categorize spatial form characteristics as a basis for investigating the mechanism through which they affect the acoustic environment. The Conzenian school first proposed that buildings, plots, and streets are the fundamental elements of urban morphology research [37]; building on this study, the basic elements of block spatial form have been expanded to include land use, plot composition, road composition, buildings, and public spaces [38,39]. In this ranking system, spatial form parameters of residential blocks can be classified into three categories from the meso-scale level to the micro-scale level: road network morphology, neighborhood scale, and architectural texture. In the selection of road network morphology parameters, it is of particular importance to consider those that can demonstrate the characteristics of road network division, such as road network density, road intersection density, road boundary line widths, and so forth. These parameters directly influence traffic flow, which in turn affects road traffic noise. The selection of neighborhood scale parameters encompasses those that influence the dimensions of individual neighborhoods, including such factors as neighborhood area, perimeter, and side length. These parameters may be employed to examine the influence of road traffic noise on diverse neighborhood scales. In the selection of architectural texture parameters, the majority of spatial morphology alterations resulting from modifications to the configuration of buildings, including those pertaining to building density, floor area ratio, and building height, can be employed to ascertain the optimal arrangement for the acoustic environment when road traffic noise remains constant, through alterations to the configuration of buildings. The selection of the aforementioned three types of parameters essentially encompasses those that regulate the spatial pattern modifications of blocks; a block is divided into neighborhoods with different sizes by road networks, and a neighborhood is composed of buildings and open spaces, as shown in

Table 1. Classification of spatial form parameters of Residential Blocks.

Category of Parameters	Meaning of Parameters	Specific Parameters
Road network morphology	Morphological features of road networks	The number of internal roads (NR), Road network density (DR), Road intersection density (DI), Road boundary line widths (WR), Average road boundary line width (WRavg), etc.
Neighborhood scale	External spatial form characteristics of a neighborhood	Neighborhood length (NL), Neighborhood perimeter (NP), Neighborhood area (NA), The difference between the maximum and minimum neighborhood areas (NAdif), etc.
Architectural texture	Internal architectural layout parameters of a neighborhood	Building density (BD), Open space ratio (OSR), Building height (BH), Floor area ratio (FAR), Degree of enclosure (DE), Number of openings along street-facing interfaces (NOSI), Length of openings along street-facing interfaces (LOSI), Maximum continuous width of street-facing buildings (WSBmax), Total length of building reflective surfaces within the neighborhood (LBRS), etc.

.

2.2. Collection and Analysis of Spatial Forms of Residential Blocks

In order to obtain empirical values for each spatial morphology parameter to construct more realistic models, spatial form data of 302 residential blocks from 38 big cities worldwide, including open, enclosed, and mixed blocks, was scraped using Open StreetMap and Google Earth. The majority of Chinese cities included in the study are selected from a list of municipalities, provincial capitals, and cities in East China. The selection of cities in other countries is primarily based on the choice of central cities from a range of countries, as well as cities that demonstrate a more comprehensive approach to urban planning. In order to ensure greater alignment between the study’s findings and the prevailing construction norms in China, the architectural texture parameters have been temporarily excluded from the database. IBM SPSS Statistics 26 is a statistical software that serves as a convenient tool for data management, visualization, and advanced statistical analysis functions. This paper utilizes SPSS for descriptive statistics, linear regression, bivariate statistics, etc. The results of the data analysis are presented in the form of figure or table, which serve to visualize the trends in the distribution of the data. The parameter value ranges for the research models to be simulated were obtained by comparing the values of the indicators with China’s regulatory standards.

Specifically, road network density ranged from 15.1 to 21.2 km/km², conforming to the recommended standard of over 8 km/km². Intersection density ranged from 5 to 15 intersections/ha. Roads with boundary line widths of 12 m–20 m accounted for 55.8% of the total, while those with widths of 20 m–35 m and 40 m–50 m comprised 34.4% and 8.2%, respectively (Figure 1). According to the recommendations for urban road boundary line widths in the Code for Design of Urban Road Engineering (CJJ37-2012) [40], residential blocks are primarily enclosed by secondary and branch roads. A statistical analysis of the grades and layouts of roads in four directions in the sample blocks revealed three common situations: 30.4% of blocks are enclosed by four branch roads, 27.2% by one secondary road and three branch roads, and 14.1% by two secondary roads and two branch roads. These three situations were adopted in the subsequent establishment of block road network models.

Neighborhood length (short side) ranged from 70 m to 150 m, while the long side ranged from 90 m to 150 m. The perimeter and area of a neighborhood fell between 310 m to 600 m and 0.6 ha–2.5 ha, respectively (Figure 2). Typical neighborhood scales in different countries are as follows: 130 m × 130 m grids in Barcelona, 75 m × 115 m grids in Berlin, 65 m × 200 m rectangular road networks used to divide blocks in Manhattan, and a scale of 70 m × 100 m used in Hong Kong. Based on the actual situations and construction requirements in China, along with the regulations stated in the Standard for Urban Residential Area Planning and Design (GB50180-2018) regarding different levels of control in residential areas and the distance between roads (not exceeding 300 m) [15], a five-minute living circle (i.e., a 300 m × 300 m area) was selected as the range for block road network models. A block is divided into neighborhoods, and the ranges of neighborhood lengths, perimeters, and areas were set based on the results of the case analysis.

2.3. Design of the Block Models

2.3.1. Block Road Network Models

The parameter values for road network morphology and neighborhood scale under the block road network models were designed according to the established value ranges. Due to the close relationship between road traffic noise and factors such as the proportion of road area [41], traffic flow, number of lanes, and vehicle speed, and according to the recommendations given in the Code for Design of Urban Road Engineering (CJJ37-2012), the design speeds for secondary and branch roads were set to 40 km/h, with a basic traffic capacity of 1650 pcu/h (passenger car unit/hour) and a minimum width for motor vehicles of 3.5 m [40]. To examine the impact of different road layouts (with the same traffic volume) on the acoustic environment in a block, the number of roads in four directions and the number of lanes in all blocks were both set to 16. Specifically, there are four lanes for each secondary road and two for each branch road, and there is only one grade of road in each neighborhood (i.e., urban branch road). The road layout for block road network models obtained is presented in

Table 2. Road Layout for Block Road Network Models.

External Road Layout	Number of External Motor Vehicle Lanes	Number of Internal Branch Roads	Number of Internal Motor Vehicle Lanes	Total Number of Motor Vehicle Lanes
Enclosed by four secondary roads	16	0	0	16
Enclosed by two secondary roads and two branch roads	12	2	4	16
Enclosed by one secondary road and three branch roads	10	3	6	16
Enclosed by four branch roads	8	4	8	16

.

The neighborhood lengths were varied in increments of 10 m in the modeling phase. Due to changes in the positions of internal roads and constrained by the values of neighborhood scales, a total of 258 valid block road network models were obtained. The neighborhood scales divided by branch roads ranged from 70 m × 90 m to 150 m × 150 m. This section examines the impact of 11 road network patterns and neighborhood scale parameters on the acoustic environment (NR, DR, DI, WR_avg, NL_max, NL_min, NA_max, NA_min, NP_max, NP_min, NA_dif). The optimal acoustic network model is then selected for integration with the building layout.

2.3.2. Block Models with Buildings

In the design of block models with buildings, attention was mainly paid to the volume and layout of buildings, with the minimum spacing for sunlight taken into account. Sunlight spacing refers to the minimum distance maintained between two rows of southbound houses in the back row to ensure that the bottom floor of the back row houses receives no less than two hours of full window sunlight (sunshine) on the winter solstice (or severe cold days). In architectural design, especially in the layout of residential buildings in China, it is essential to integrate the principle of land conservation to ensure the minimum solar spacing required in building layouts. As insolation standards in China vary by region, the sunlight requirements in Nanjing, a city located in the middle of China’s latitude and climate zone ranges, were chosen as the standard for modeling to better align the model with real construction conditions. The sunshine simulation analysis in Figure 3 demonstrates compliance of building combinations with 18, 25, and 32 floors to the sunshine spacing regulations in Nanjing, ensuring their adherence to the local building code requirements.

The statistical analysis of high-rise residential buildings, in terms of the number of floors and layout, in new residential blocks in Nanjing from January to June 2023 revealed that most new residential buildings are largely slab-type high-rises with popular floor numbers of 18, 25, and 32. The most prevalent housing sizes in the market were 89 m², 110 m², and 143 m² (Figure 4). Therefore, these floor numbers and housing sizes were used in the design of block models with buildings.

The three types of high-rise residential buildings were integrated according to the current conventions of high-rise residential design and fire safety regulations in China (Figure 5), obtaining the following four slab types for residential buildings, with building widths ranging from 27 m to 80 m.

Based on these four planes and the road network forms selected after the simulations of block road network models, a total of 594 block models with buildings were established through variations in parameters such as building density, height, and degree of enclosure. Sunlight analysis and simulations ensured that all architectural layouts met the sunlight requirements. This section examines the impact of nine building texture parameters (BD, OSR, BH, FAR, DE, NOSI, LOSI, WSB_max, LBRS) on the acoustic environment, modeling their effects in conjunction with the optimal road network layout for the acoustic environment.

2.4. Acoustic Environment Simulation

SoundPLAN is a dedicated software for performing noise modeling and environmental noise assessment. Released in 1986 with built-in specifications based on the German RLS-90 standard, this model uses the equivalent continuous A sound level as the evaluation index and includes a sound source model and a sound propagation model. The noise impact factors take into account the volume and type of vehicles, vehicle speed, the natural condition of the road (longitudinal gradient, type of pavement), the environmental conditions (air absorption, atmospheric effects), and the type of obstacles in the propagation process (distances, buildings), and are applied to simple interruptions of the traffic flow and to unknown traffic flows. It has evolved into the world’s leading tool for noise prediction, mapping, and evaluation. It can be widely used for noise prediction, mapping, and evaluation tasks in different regions [17,42,43,44]. The software contains a variety of parameter settings: for example, defining the width of the road red line, vehicle speed, average daily traffic (ADT, refers to the total traffic volume during a certain period divided by the total number of days during that period), type of ground material, and the length from the road radial line; defining the location of the sound source and its point and surface shapes, etc.

Simulations were conducted using the acoustic simulation software SoundPLAN 8.2. The process involved importing model files drawn in AutoCAD 2018 into SoundPLAN. A secondary road was defined with lane widths of 7 m on both sides, with a distance of 5.25 m to the radiating lines on each side, no central green belt, a height difference of 0 m, an average daily traffic (ADT) of 39,600 vehicles/24 h, a vehicle speed of 40 km/h, and a road surface of smooth asphalt and asphalt concrete. A branch road had lane widths of 3.5 m on both sides, with a distance of 1.75 m to the radiating lines on each side, an ADT of 19,800 vehicles/24 h, and other settings identical to those of a secondary road. All roads and buildings are in an ideal state, and the facade materials and road materials remain the same. The reflection loss of all buildings in a neighborhood was set to 1 dBA. Factors such as air and plant absorption and additional losses caused by ground interference, temperature, or wind speed gradient-induced refraction effects were not considered. The calculation grid was set at 1 m, with calculations performed on a plane approximately 1.5 m above the ground—close to the height of ears when human standing [45,46], which is the international harmonized standard for environmental acoustic measurements. (The scope of this paper is the acoustic environment in the outdoor public space of a residential neighborhood, not the acoustic environment inside a house, so the calculation surface is 1.5 m above the ground, which is approximately the height of the human ear.) The simulations showed the values as time-averaged sound pressure level isosurfaces, with A-weighting used for assessment, and the evaluation measure as L_Aeq.

For larger blocks or neighborhoods, the acoustic environment evaluation indicators cannot be replaced by the mean value of the noise exposure levels at certain points. Instead, it is necessary to set specific indicators in this article to describe the distribution of noise exposure levels in a block or neighborhood. Each 5 dBA interval was defined, and the median noise exposure level in each interval was multiplied by the percentage of that interval’s area relative to the total area, summing up to define the overall noise exposure level (L_o). As the noise exposure levels for all models ranged between 45 dBA and 75 dBA, the equation is as follows:

L_o = (47.5 dBA × S_45–50/S_t) + (52.5 dBA × S_50–55/S_t) + (57.5 dBA × S_55–60/S_t) + (62.5 dBA × S_60–65/S_t) + (67.5 dBA × S_65–70/S_t) + (72.5 dBA × S_70–75/S_t)

(1)

where L_o represents the overall noise exposure level, S_t denotes the total area of the block, and S_x–y indicates the area occupied by the segment from x to y (x is the minimum value of each interval range, and y is the maximum value of each interval range).

The pixel method was employed to quantify the distribution of noise exposure levels. Specifically, the noise isosurface map in Photoshop was used to match the colors of the isosurfaces to the corresponding noise exposure levels within each interval, and the pixel values of the intervals and the total pixel value were obtained through histogram information.

3. Results

3.1. Relationship between the Road Network Morphology, Neighborhood Scale Parameters, and Road Traffic Noise Distribution

3.1.1. Correlation Analysis Results

A correlation analysis was conducted between the road network morphology parameters and neighborhood scale parameters of 258 block road network models and the overall noise exposure level (

Table 3. Pearson’s Correlation Analysis of Road Network Morphology Parameters, Neighborhood Scale Parameters, and the Overall Noise Exposure Level.

Category of Parameters	Spatial Form Parameters	Sig.	Pearson Correlation Coefficient
Road network morphology	The number of internal roads	0.000	0.940 **
	Road network density	0.000	0.940 **
	Road intersection density	0.000	0.922 **
	Average road boundary line width	0.000	−0.943 **
Neighborhood scale	The maximum neighborhood length	0.000	−0.768 **
	The minimum neighborhood length	0.000	−0.307 **
	The maximum neighborhood perimeter	0.000	−0.835 **
	The minimum neighborhood perimeter	0.000	−0.870 **
	The maximum neighborhood area	0.000	−0.918 **
	The minimum neighborhood area	0.000	−0.847 **
	The difference between the maximum and minimum neighborhood areas	0.001	−0.202 **

** denotes a significant correlation at the 0.01 level (two-tailed).

). Significance (Sig.), is a statistical value, the value after Sig. is the calculated p, and significance testing is performed based on the p-value. If 0.01 < p < 0.05, the difference is significant; if p < 0.01, the difference is extremely significant. The results revealed that 11 variables were significantly correlated with the overall noise exposure level. Specifically, the number of internal roads, road network density, and intersection density were positively correlated with the overall noise exposure level, respectively (p < 0.01). The maximum neighborhood length, minimum neighborhood length, maximum neighborhood perimeter, minimum neighborhood perimeter, maximum neighborhood area, minimum neighborhood area, and the difference between the maximum and minimum neighborhood areas were negatively correlated with the overall noise exposure level, respectively (p < 0.01). It is worth noting that as the number of internal roads, the road network density, and the intersection density increase, more traffic noise is brought to the block, leading to a significant rise in the overall noise exposure level in the block, with noticeable differences among different groups.

The correlation analysis results in Table 3 show that the overall noise exposure level decreases as the number of internal roads decreases, and when there are no roads inside the block (the number of internal roads = 0), the block becomes an enclosed block. Therefore, four enclosed block models without internal branch roads were established for a thorough comparison with the model with the optimal acoustic environment in each group of block road network models (Figure 6).

Vertical comparisons revealed that when the external road layouts of the two models were the same, the acoustic environment worsened as the number of internal roads increased. Horizontal comparisons of the four models in the first group demonstrated that when the area of the block or neighborhood remained constant, the acoustic environment in a block was quieter as the grade of external roads lowered. Horizontal comparisons of the four models in the second group showed that with a constant total road traffic volume in the block, the acoustic environment worsened as external traffic flow was distributed to the internal roads, and the acoustic environment in the block could not be improved by lowering the grade of each neighborhood’s external roads. Therefore, enclosed blocks have significantly better acoustic environment quality than open blocks. In urban renewal projects, it is not advisable to simply divide large blocks into smaller ones to alleviate traffic congestion. The following is a discussion of simulation results for blocks with different numbers of internal roads. Figure 7 shows that when there are two roads in each block, due to the limitation of the value range, the difference between the situation obtained by changing the division method is small, and the overall noise exposure levels of the four models show no significant differences. The four scenarios with two internal roads demonstrate that a cross-equalized road network form is an effective solution for a block of 300 m × 300 m, resulting in an optimal acoustic environment. The division of the road network is constrained by a reasonable maximum neighborhood area, resulting in nearly balanced divisions.

A total of 54 effective models with three internal roads in the block were built, the three best-performing models in terms of acoustic environment quality (Figure 8) shared the following characteristics: the maximum neighborhood area, length, and perimeter were the maximum values of their respective group while the minimum neighborhood area, length, and perimeter were the maximum values of their respective group, and the largest neighborhood was not adjacent to the secondary road. The three worst-performing models (Figure 8) all had blocks divided by roads into relatively homogeneous neighborhoods, with minor differences between the forms of the largest and smallest neighborhoods.

A total of 200 effective models with four internal roads in the block were built, and the interactive patterns between the overall noise exposure level and road network morphology/neighborhood scale parameters were consistent with those when each block had three internal roads (Figure 9). Further comparison of the noise isosurface maps in the group revealed that, when the grade of roads in four directions, the number of lanes, and traffic volume are the same across blocks, the location of the largest neighborhood has a relatively minor impact on the overall noise exposure level.

In general, the fewer roads and the lower the density of the road network in a block, the lower the intersection density and the smaller the average boundary line width. When a block is divided into neighborhoods with large differences in area by the internal road network and large neighborhoods are not adjacent to high-grade roads, its overall acoustic environment is better than a block divided by the internal road network into relatively homogeneous neighborhoods; the larger the area and longer the sides of a neighborhood, the better the acoustic environment. The road networks corresponding to the three models with the most optimal acoustic environments in the groups of block models (i.e., 2-a, 3-a, and 4-a) were applied to the subsequent modeling for block models with buildings.

3.1.2. Threshold Analysis Results

After the impact of road network morphology parameters on the overall noise exposure level had been verified, a scatter plot analysis was conducted to investigate the relationship between neighborhood scale parameters and the overall noise exposure level and further establish threshold values for these parameters (Figure 10). The results indicated that as each neighborhood scale parameter increased, the overall noise exposure level decreased, a trend primarily controlled by road network morphology parameters. Further analysis of the various neighborhood scale parameters revealed that the overall noise exposure level hit its lowest value when the area, length, and perimeter of a neighborhood stood between 20,000 m² and 22,000 m², 140 m and 150 m, and 550 m and 600 m, respectively.

Therefore, to ensure good acoustic environment quality in residential areas while adhering to the block system, in road planning, efforts should be taken to design the scale of neighborhoods according to the upper limit of the design and ensure the largest neighborhoods are far from high-grade roads. In new residential blocks, it is recommended that neighborhood lengths be at least 140 m to 150 m, neighborhood perimeters be at least 550 m to 600 m, and neighborhood areas be at least 20,000 m² to 22,000 m². Additionally, public service facilities with relatively low requirements for good acoustic environments can be installed in smaller neighborhoods close to high-grade roads to reduce noise interference.

3.2. Relationship between the Architectural Texture Parameters and Road Traffic Noise Distribution

3.2.1. Correlation Analysis Results

To explore the mechanism through which architectural texture parameters influence the overall noise exposure level while excluding the effects of road network morphology and neighborhood scale parameters, models with varying numbers of internal roads were separately discussed. Modeling was performed based on the road network models selected above and architectural layouts, and the models were divided into Groups A, B, C, and D for further discussion, within Group A, there is no road; within Group B, there are two roads; within Group C, there are three roads; and within Group D, there are four roads. The models in group C were divided by internal road networks into neighborhoods with greatly different sizes while models in groups A, B, and D were relatively equally divided. The division pattern of group C is not a common planning pattern in the design of residential areas, and this kind of parcel division with large differences is often used in the case of new roads added in the process of urban renewal. In addition, the calculation of the parameters is mostly based on the mean and median, which cannot fully reflect the layout characteristics in some special cases, so further discussions on neighborhoods are needed for group C in subsequent research.

Pearson correlation analyses were conducted between the architectural texture parameters of the four model groups (all parameters underwent a normality test) and the overall noise exposure level (

Table 4. Pearson’s Correlation Analysis of Architectural Texture Parameters and the Overall Noise Exposure Level.

Architectural Texture Parameters	Group A		Group B		Group C		Group D
	Sig.	PPMCC	Sig.	PPMCC	Sig.	PPMCC	Sig.	PPMCC
Building density	0.000	−0.710 **	0.000	−0.401 **	0.512	0.056	0.000	0.399 **
Open space ratio	0.000	0.710 **	0.000	0.401 **	0.512	−0.056	0.000	−0.399 **
Building height	0.000	0.438 **	0.000	0.441 **	0.177	0.115	0.000	0.435 **
Floor area ratio	0.000	−0.401 **	0.000	0.081 **	0.003	0.246 **	0.000	0.613 **
Degree of enclosure	0.000	−0.790 **	0.000	−0.746 **	0.228	−0.102	0.000	0.446 **
Number of openings along street-facing interfaces	0.000	−0.468 **	0.000	−0.771 **	0.003	−0.250 **	0.000	−0.603 **
Length of openings along street-facing interfaces	0.000	0.790 **	0.000	0.746 **	0.228	0.102	0.000	−0.446 **
Maximum continuous width of street-facing buildings	0.000	−0.327 **	0.000	−0.355 **	0.006	0.229 **	0.000	0.506 **
Total length of building reflective surfaces within the neighborhood	0.000	−0.770 **	0.000	−0.648 **	0.126	−0.130	0.000	0.314 **

** denotes a significant correlation at the 0.01 level (two-tailed).

).

The results suggested that in groups A, B, and D, all architectural texture parameters were significantly correlated with the overall noise exposure level, respectively (p < 0.01). In group C, only the floor area ratio, number of openings along street-facing interfaces, and maximum continuous width of street-facing buildings showed significant correlations with the overall noise exposure level. Since models in group C were divided by internal road networks into neighborhoods with greatly different sizes while models in groups A, B, and D were relatively equally divided, further discussions on neighborhoods are needed for group C in subsequent research.

Specifically, for the two groups of models with identical indicators except for the number of openings (Figure 11), the overall noise exposure level increased when the number of openings along street-facing interfaces dropped; models with shorter openings, compared with those with longer openings, enjoyed relatively low noise exposure levels. Therefore, when the length of openings along street-facing interfaces is fixed, adding more openings and making each opening smaller can optimize acoustic environment quality.

Similarly, for models with identical indicators except for the degree of enclosure and the length of openings along street-facing interfaces (Figure 12), when the openings along street-facing interfaces were longer, the degree of enclosure lowered, and the overall noise exposure level in the block rose. This reason is that the roads in four directions in a block exert a weaker influence with shorter openings along street-facing interfaces and a higher degree of enclosure. Therefore, it is advisable to raise the degree of enclosure in planning to reduce the impact of noise from surrounding roads.

3.2.2. Threshold Analysis Results

Preliminary analysis of the neighborhoods with the lowest overall noise exposure levels in each group revealed that the building density of the models ranged from 10% to 19%, the open space ratio fell between 81% and 90%, the average numbers of floors were largely 18, and the degree of enclosure ranged from 40% to 50% (Figure 13).

A scatter plot was made to demonstrate the relationship between each architectural texture parameter and the overall noise exposure level and explore threshold values of these parameters. Figure 14 shows a linear relationship between the architectural texture parameters of the block models with buildings and the noise exposure level. There is also a hierarchical clustering trend due to different numbers of internal roads. The overall noise exposure level in a block hits its lowest value with a building density of over 18%, open space ratio lower than 80%, plot ratio between 2.0 and 3.5, degree of enclosure between 40% and 50%, length of openings along street-facing interfaces between 600 m and 800 m, maximum continuous width of street-facing buildings between 60 m and 70 m, and total length of building reflective surfaces within the neighborhood between 3200 m and 3500 m. These values are recommended for new residential blocks to achieve better acoustic environment quality.

3.3. Prediction Model Construction Based on Multiple Linear Regression

3.3.1. Principal Component Analysis of Variables

Given the typical characteristics of urban spatial forms, spatial form parameters are inevitably correlated with each other. In order to identify the most significant correlations, only four types of road networks were selected for the 594 block models with buildings established. This resulted in insignificant differences in spatial forms at the neighborhood level. Therefore, a correlation analysis was conducted using architectural texture parameters.

Table 5. Correlation of Architectural Texture Parameters.

	BD	OSR	BH	FAR	DE	NOSI	LOSI	WSBmax	LBRS
BD	1.000	−1.000 **	−0.431 **	0.673 **	0.455 **	0.323 **	−0.012	0.232 **	0.920 **
OSR	−1.000 **	1.000	0.431 **	−0.673 **	−0.455 **	−0.323 **	0.012	−0.232 **	−0.920 **
BH	−0.431 **	0.431 **	1.000	0.354 **	−0.162 **	−0.344 **	0.046	0.030	−0.509 **
FAR	0.673 **	−0.673 **	0.354 **	1.000	0.343 **	0.036	0.020	0.256 **	0.524 **
DE	0.455 **	−0.455 **	−0.162 **	0.343 **	1.000	−0.142 **	−0.637 **	0.603 **	0.543 **
NOSI	0.323 **	−0.323 **	−0.344 **	0.036	−0.142 **	1.000	0.607 **	−0.441 **	0.362 **
LOSI	−0.012	0.012	0.046	0.020	−0.637 **	0.607 **	1.000	−0.488 **	−0.100 *
WSBmax	0.232 **	−0.232 **	0.030	0.256 **	0.603 **	−0.441 **	−0.488 **	1.000	0.249 **
LBRS	0.920 **	−0.920 **	−0.509 **	0.524 **	0.543 **	0.362 **	−0.100 *	0.249 **	1.000

** denotes a significant correlation at the 0.01 level (two-tailed). * denotes a significant correlation at the 0.05 level (two-tailed).

indicates that, with the exception of the length of openings along street-facing interfaces, which is not significantly correlated with other architectural texture parameters, the remaining parameters exhibit some degree of correlation. Due to the intricate interrelationships between the parameters, analyzing the effect of one parameter on sound propagation in a residential block alone cannot truly reflect the correlation between the two. The correlation analysis of the data using SPSS software has been carried out in the previous section and a clear correlation between the selected influencing factors and the total noise exposure level has been obtained. Therefore, the choice of regression analysis is reasonable. Nevertheless, it is evident that the correlation between the remaining independent variables is exceedingly high, which also indicates the presence of multicollinearity between the respective variables within the model.

The presence of multicollinearity between independent variables affects the accuracy of the regression model. In order to address the issue of multicollinearity, this paper employs Principal Component Analysis (PCA) as a means of eliminating the covariance. The method transforms multiple independent variables into a few uncorrelated principal components, thus avoiding multicollinearity and making the prediction model more scientifically accurate. Principal component regression is the application of principal component analysis to reduce the dimensionality of the independent variables and extract individual principal components, which are capable of explaining the overall variables. These extracted principal components are then subjected to regression analysis, which serves to circumvent the impact of the linear relationship between the individual independent variables (multicollinearity) on the overall linear relationship. Firstly, the Kaiser–Meyer–Olkin (KMO) test and Bartlett’s test are performed on the independent variables, the KMO value is 0.762, which is greater than 0.7, indicating that the correlation between the original variables is strong. The correlation between the variables is strong, as indicated by the value of p = 0.000 for Bartlett’s spherical test of the chi-square statistic. This demonstrates that the independent variables possess structural validity, allowing for the implementation of principal component analysis. The respective variables were standardized and subjected to principal component analysis. In the table of the variance of the common factor, the extracted values indicate the degree to which each variable is expressed by the common factor. It is generally accepted that a value greater than 0.7 indicates that the variable is well expressed by the common factor.

Table 6. Common Factor Variance (statistics).

Specific Parameters	Initial Value	Extraction Value
The number of internal roads (NR)	1.000	0.987
Road network density (DR)	1.000	1.000
Road intersection density (DI)	1.000	0.876
Average road boundary line width (WRavg)	1.000	0.924
The maximum of neighborhood length (NLmax)	1.000	0.996
The minimum of neighborhood length (NLmin)	1.000	0.995
The maximum of neighborhood area (NAmax)	1.000	0.912
The minimum of neighborhood area (NAmin)	1.000	0.989
The maximum of neighborhood perimeter (NPmax)	1.000	0.986
The minimum of neighborhood perimeter (NPmin)	1.000	0.998
The difference between the maximum and minimum neighborhood areas (NAdif)	1.000	1.000
Building density (BD)	1.000	0.967
Open space ratio (OSR)	1.000	0.967
Building height (BH)	1.000	0.950
Floor area ratio (FAR)	1.000	0.969
Degree of enclosure (DE)	1.000	0.750
Number of openings along street-facing interfaces (NOSI)	1.000	0.790
Length of openings along street-facing interfaces (LOSI)	1.000	0.950
Maximum continuous width of street-facing buildings (WSBmax)	1.000	0.628
Total length of building reflective surfaces within the neighborhood (LBRS)	1.000	0.936

reveals that the majority of variables exhibit an extracted value exceeding 0.85, indicating a high degree of expression by the common factor.

The contribution rate of the first four principal components reached 92.845%. Scree Plot is a Principal Component Analysis (PCA) visualization tool that helps us decide how many principal components (PCs) it is appropriate to keep in a data set. According to Figure 15, it can be learned that the fifth component’s eigenvalue trend has begun to flatten. Therefore, it can be concluded that selecting four principal components is more appropriate for determining the number of principal components for four.

The contribution of each independent variable to the extracted principal components can be observed in the rotated component matrix presented in

Table 7. Rotated component matrix ^a.

Specific Parameters	Ingredient
	1	2	3	4
The number of internal roads (NR)	−0.947
Road network density (DR)	−0.997
Road intersection density (DI)	−0.802
Average road boundary line width (WRavg)	0.672		−0.681
The maximum of neighborhood length (NLmax)	0.996
The minimum of neighborhood length (NLmin)	0.997
The maximum of neighborhood area (NAmax)	−0.667		0.669
The minimum of neighborhood area (NAmin)	0.988
The maximum of neighborhood perimeter (NPmax)	0.973
The minimum of neighborhood perimeter (NPmin)	0.998
The difference between the maximum and minimum neighborhood areas (NAdif)	0.991
Building density (BD)		0.981
Open space ratio (OSR)		−0.981
Building height (BH)				0.899
Floor area ratio (FAR)		0.704		0.685
Degree of enclosure (DE)		0.553	0.643
Number of openings along street-facing interfaces (NOSI)			−0.764
Length of openings along street-facing interfaces (LOSI)			−0.949
Maximum continuous width of street-facing buildings (WSBmax)			0.658
Total length of building reflective surfaces within the neighborhood (LBRS)		0.945

^a. The rotation has converged after 5 iterations.

. The urban spatial morphology parameters that play a significant role in the first principal component include the number of internal roads (NR), road network density (DR), road intersection density (DI), average road boundary line width (WR_avg), the maximum of neighborhood length (NL_max), the minimum of neighborhood length (NL_min), the maximum of neighborhood area (NA_max), the minimum of neighborhood area (NA_min), the maximum of neighborhood perimeter (NP_max), the minimum of neighborhood perimeter (NP_min), and the difference between the maximum and minimum neighborhood areas (NA_dif). The urban spatial morphology parameters that play a significant role in the second principal component include building density (BD), open space ratio (OSR), floor area ratio (FAR), degree of enclosure (DE), and total length of building reflective surfaces within the neighborhood (LBRS). The urban spatial morphology parameters that play a significant role in the third principal component include average road boundary line width (WR_avg), the maximum neighborhood area (NA_max), degree of enclosure (DE), number of openings along street-facing interfaces (NOSI), length of openings along street-facing interfaces (LOSI), maximum continuous width of street-facing buildings (WSB_max). The urban spatial form parameters that play a significant role in the fourth principal component include building height (BH) and floor area ratio (FAR). In summary, the first four principal components encompass all 20 independent variables, representing the full range of urban spatial form parameters selected in this paper. Among these, the road network form parameters and neighborhood scale parameters play a significant role in determining the overall noise exposure level.

The matrix of component score coefficients in

Table 8. Matrix of component score coefficients.

Specific Parameters		Ingredient
		1	2	3	4
X1	The number of internal roads (NR)	−0.093	−0.006	0.033	−0.011
X2	Road network density (DR)	−0.114	0.004	−0.040	0.012
X3	Road intersection density (DI)	−0.064	−0.011	0.098	−0.028
X4	Average road boundary line width (WRavg)	0.035	0.026	−0.171	0.059
X5	The maximum of neighborhood length (NLmax)	0.122	−0.011	0.081	−0.026
X6	The minimum of neighborhood length (NLmin)	0.116	−0.005	0.052	−0.015
X7	The maximum of neighborhood area (NAmax)	−0.035	−0.037	0.168	−0.073
X8	The minimum of neighborhood area (NAmin)	0.125	−0.013	0.097	−0.031
X9	The maximum of neighborhood perimeter (NPmax)	0.103	0.005	0.001	0.006
X10	The minimum of neighborhood perimeter (NPmin)	0.121	−0.010	0.072	−0.024
X11	The difference between the maximum and minimum neighborhood areas (NAdif)	0.109	−0.001	0.022	−0.006
X12	Building density (BD)	0.002	0.246	−0.016	0.003
X13	Open space ratio (OSR)	−0.002	−0.246	0.016	−0.003
X14	Building height (BH)	−0.006	−0.064	0.000	0.604
X15	Floor area ratio (FAR)	−0.004	0.201	−0.015	0.494
X16	Degree of enclosure (DE)	0.020	0.130	0.182	−0.030
X17	Number of openings along street-facing interfaces (NOSI)	−0.047	0.091	−0.235	−0.160
X18	Length of openings along street-facing interfaces (LOSI)	−0.083	0.010	−0.308	0.119
X19	Maximum continuous width of street-facing buildings (WSBmax)	0.006	0.073	0.178	0.120
X20	Total length of building reflective surfaces within the neighborhood (LBRS)	−0.002	0.231	0.003	−0.107

was employed to calculate the formulae for the four principal components.

3.3.2. Modeling Multiple Linear Regression

Four common factors, which can summarize all the information of the original variables, were obtained by principal component analysis. Using SPSS software, multiple linear regression analysis was performed with total noise exposure as the dependent variable and the four principal components as the independent variables to simulate the model of factors influencing total noise exposure:

Y = β₀ + β₁F₁ + β₂F₂ + β₃F₃ + β₄F₄

(2)

where Y represents the overall noise exposure level, β_j (j = 1,2,…k) denotes the regression coefficient, and F indicates principal components.

R² in

Table 9. Multiple linear regression summary.

Implicit Variable	Independent Variable	Non-Standardized Coefficient		Standardized Coefficient	t	Sig.	VIF
		B	Standard Error	Beta
Overall noise exposure level	β0	61.636	0.010		6407.268	0.000
	F1	−2.726	0.010	−0.987	−283.145	0.000	1.000
	F2	−0.115	0.010	−0.042	−11.956	0.000	1.000
	F3	0.352	0.010	0.128	36.609	0.000	1.000
	F4	0.054	0.010	0.020	5.631	0.000	1.000
R2 = 0.993, Adjusted R2 = 0.995, F = 24.764 ** (p = 0.000), Durbin–Watson test statistic = 1.973

** denotes a significant correlation at the 0.01 level (two-tailed).

represents the degree of fit of the model, which reflects the extent to which the independent variable explains the dependent variable. The value of R² ranges between 0 and 1, and the closer it is to 1, the better the model fit is. From Table 9, it can be found that R² = 0.995, which indicates that the independent variable can explain 99.5% of the cause of change in the dependent variable, i.e., 99.5% of Y is caused by F₁, F₂, F₃, and F₄, which means that the urban spatial morphology parameters included in this study are more systematic and comprehensive. f-test is used to determine whether the existence of the regression model has significance or not, and if it presents as significant, it means that x will have an impact on Y. From Table 9, F = 24.764 (p < 0.01) indicates that the model is significant, i.e., the model is constructed in a meaningful way, which means that at least one of the independent variables will have an effect on the dependent variable.

Multiple linear regression equations for rapid prediction of total noise exposure in residential neighborhoods are given in Table 9:

Y = 61.636 − 2.726F₁ − 0.115F₂ + 0.352F₃ + 0.054F₄

(3)

In the regression model, each urban spatial morphology parameter has a significant effect on the Y variable, in which the larger the first and second principal components are, the lower the overall noise exposure level and the better the acoustic environment, and the larger the third and fourth principal components are, the higher the overall noise exposure level and the worse the acoustic environment. Specifically, the first principal component has the most significant influence on the urban spatial form factors, which mainly include road network morphology and neighborhood scale parameters.

3.4. Example Verification of Prediction Model

In order to verify the accuracy of the prediction model in judging the differences in the acoustic environment in residential blocks, this study selected two high-rise residential areas with similar surrounding traffic conditions in Nanjing for field measurement of the acoustic environment. The Phoenix Harmony Community is situated on Yuhua Road in the Yuhua District. It was completed in 2012 and encompasses an area of 6.4 ha, with a total building area of 249,600 m². The floor area ratio is 3.5, the greening rate is 35%, and there are nine high-rise residential buildings, comprising 1602 households. The majority of these households reside in slab-type high-rise buildings. The settlement features a row-type layout with no basement shop enclosure (Figure 16).

The New City Holiday Community is situated on Jiangdong North Road in Gulou District. Construction was completed in 2005, with an area of 4.1 ha and a building area of 99,580 m². The floor area ratio is 2.4, the greening rate is 50%, and the settlement comprises seven blocks of high-rise residential buildings combining slabs and towers, with a total of 528 households (Figure 17).

Both two communities are surrounded by three urban branch roads and one urban expressway, while access to external vehicles is prohibited within the block. The study identified nine locations in the community as measurement points, which were named from 1 to 9. The distribution of these locations is designed to reflect the typical acoustic environment of outdoor public spaces within the community. The primary source of noise is road traffic noise, with no interference from other noise sources. Photos of the measurement points are shown in Figure 16 and Figure 17. The greenery in these two communities is predominantly grass with small trees, so there is limited absorption by the greenery at 1.5 m (the height of the average human ear when standing), which has a negligible effect in this test.

The acoustic measurements were taken on weekdays. These measurements were taken from 6:00 am to 22:00 pm, with the objective of encompassing the periods during which residents engage in their daily activities within the public space of the community. The night-time acoustic environment was not included in the measurements, given that fewer residents were observed to be active within the public space during this time. The field measurement and recording of spatial form information and environmental information of the selected settlements was conducted with the use of a Leica DISTO laser rangefinder D510 (

Table 10. The Measurement Information of the Selected Communities.

Selected Communities	Spatial Form Information	Measurement Time	Road Conditions around the Community	Type of Vehicle
The Phoenix Harmony Community	NR = 0, DR = 0, DI = 0, WRavg = 26.25 m, NLmax = 280 m, NLmin = 240 m, NAmax = 64,000 m2, NAmin = 64,000 m2, NPmax = 1020 m, NPmin = 1020 m, NAdif = 0, BD = 25.0%, OSR = 75.0%, BH = 84 m, FAR = 3.5, DE = 36.3%, NOSI = 9, LOSI = 370 m, WSBmax = 75 m, LBRS = 1440 m	6:00 am to 22:00 pm	Three urban branch roads and one urban expressway	Motorbikes, cars, buses
The New City Holiday Community	NR = 0, DR = 0, DI = 0, WRavg = 24.25 m, NLmax = 240 m, NLmin = 190 m, NAmax = 41,000 m2, NAmin = 41,000 m2, NPmax = 900 m, NPmin = 900 m, NAdif = 0, BD = 25.0%, OSR = 75.0%, BH = 50.4 m, FAR = 2.4, DE = 19.8%, NOSI = 8, LOSI = 178 m, WSBmax = 70 m, LBRS = 1140 m	6:00 am to 22:00 pm	Three urban branch roads and one urban expressway	Motorbikes, cars, buses, minivans

). Considering the personnel and time allocation of the actual test, the community and the surrounding neighborhood were surveyed before the test. The survey results found that there is no construction site near the community, and the noise source is mainly traffic noise and other living noise. Because the living noise is relatively not prominent and is affected by a variety of different factors, it is difficult to make a stable evaluation; therefore, this study only considers the traffic noise, and the measurement work is mainly around the change in environmental noise in the residential area. On the whole, the terrain around the community and the sections are relatively flat, which can be regarded as a road surface without slope. The daytime of 6:00 am–22:00 pm was evenly divided into eight time periods (6:00–8:00, 8:00–10:00, 10:00–12:00, 12:00–14:00, 14:00–16:00, 16:00–18:00, 18:00–20:00, 20:00–22:00), and it is considered that the data measured in the 10 min in each period can represent the environmental noise of the community during this period. The surveyors were divided into two groups and performed measurement simultaneously within the two communities, the A-weighted continuous equivalent sound pressure level (L_Aeq) was measured for a 10 min period at each measurement point with a multifunction sound level meter (type AWA6228+, accurate to type 1) during each time period. In accordance with the ISO 1996-2 method of measuring ambient noise [47], this measurement was carried out in conditions free from rain, snow, thunder, and lightning, and at a wind speed of 5 m/s or less. The measurements were taken during a period that was representative of typical conditions, excluding periods of public holidays and inclement weather. Prior to and following the measurements, a sound calibrator (AWA6223+F) was employed to ensure the accuracy of the data obtained from the sound level meter. The objective of the data collection was to ascertain whether discrepancies between projected and actual noise exposure levels in the two plots could be discerned, thereby validating the efficacy of the rapid prediction model developed in this study for comparing the acoustic environments of scenarios with similar road conditions.

The statistical analysis of noise levels reveals that the highest peaks occur between 8:00 and 10:00 am and between 12:00 and 14:00 pm. The highest peak in noise level occurs at 18:00–20:00, representing the period of the day when the noise level is the highest overall. The mean LAeq values for each community measurement point were calculated for the 16 h daytime period on each measurement day. The resulting daily mean LAeq was then derived and recorded in

Table 11. Spatial Form Information of the Selected Communities.

Selected Communities	Arithmetic Mean of 16-h Daytime LAeq from 6:00 AM to 22:00 PM (dBA)					The Measurement Day LAeq Average (dBA)
	7.31	8.1	8.2	8.5	8.6
The Phoenix Harmony Community	54.3	52.8	55.2	53.6	53.1	53.8
The New City Holiday Community	55.6	57.4	53.0	54.5	54.2	54.9

. The measurement day LAeq average for the Phoenix Harmony Community was 53.8 dBA, while the measurement day LAeq average for the New City Holiday Community was 54.9 dBA.

The spatial morphological information data of the two communities were standardized and incorporated into the formula for calculating the principal components in the previous paper. According to the calculation, the principal components of the Phoenix Harmony Community are F1 = 1.73, F2 = 2.04, F3 = −1.49, and F4 = 2.82; the principal components of the New City Holiday Community are F1 = 1.33, F2 = 1.38, F3 = −1.59, and F4 = −0.05. These values were subsequently incorporated into the prediction model (Equation (3)). The predicted L_o was approximately 56.0 dBA for the Phoenix Harmony Community and 57.0 dBA for the New City Holiday Community. Therefore, in the actual measurement and prediction, the environmental noise of The New City Holiday Community is higher than The Phoenix Harmony Community by about 1 dBA. The results demonstrate that the model can be used to rapidly assess the acoustic environment between different schemes through the spatial form at the design stage. However, in the actual case selection, although two communities with the same road class and width are selected, the actual traffic flow and road traffic noise caused by the different roads will not be identical due to the varying roads and other factors. Consequently, a larger sample size is required for further validation.

4. Discussion

Similar to most research, this study demonstrates the substantial influence of urban spatial form on the acoustic environment. Previous studies have demonstrated the relationship between the morphological indicators at the urban level and environmental noise through various methods [16,17,18,19,20,21,22,23,24,25]. For instance, it is demonstrated that the lengths of road segments, the arrangement of intersection nodes, areas of streets, road network density, road area density, road coverage, and other road-related factors all have an impact on the urban acoustic environment [20,21,22]. Additionally, research at the level of architectural texture has demonstrated that factors such as construction density, open spaces, shape and physical position of buildings, building density, building height, and length of openings along street-facing interfaces can all affect the level of noise exposure [17,18,19,23,24,25]. However, most previous studies have only considered these factors individually without systematically sorting and considering spatial form parameters. Therefore, this study categorized spatial form parameters of blocks into road network morphology, neighborhood scale, and architectural texture and examined the impacts of different categories of parameters on the acoustic environment from three perspectives, with a view to obtaining more realistic application results.

In addition to the same conclusion, this study has made further contributions. At the level of the impact of urban spatial form parameters on the acoustic environment, previous studies are largely focused on specific blocks; for example, noise maps of representative areas in Manchester, UK, and Wuhan, China, were produced and simulated using Cadna/A. It was concluded that urban form indicators generally have a significant effect on noise levels [21]. A study combining objective measurements of noise pollution with surveys of public perception of noise in urban areas in Spain revealed a significant correlation between the impact of noise on the city and two key factors: the height of buildings and the type of road [48]. Nevertheless, the building volumes and neighborhood scales in other countries are distinct from those in China, and the findings are not wholly applicable to Chinese planning and construction. Using the method of soundscape evaluation to compare specific gated and open communities, it was found that the noise annoyance of gated communities is significantly lower than that of open communities, and that traffic noise should be mainly considered in open communities [13]. In comparison with the aforementioned results, this study additionally analyses the comparative acoustic environments of open communities that have different distributions of road traffic noise formed through the road network division. In addition to the fact that the majority of existing studies are based on field data, the present study also considers the actual design codes in order to propose a customized model for Chinese settlements.

A summary of the characteristics of existing noise prediction models reveals that the most frequently utilized traffic noise prediction models are the RLS-90 and the NMPB, while the most commonly employed mapping programs are SoundPLAN and ArcGIS. The majority of measurements were conducted over a 15 min period at an elevation of 1.5 m above ground level [46]. Accordingly, the model developed in this study is consistent with the prevailing approach and methodology in the international scientific community and is founded on rigorous scientific principles. When comparing with existing sound environment prediction models, for instance, Aletta et al. proposed predictive soundscape models [31], and other models utilizing machine learning have been developed for predicting acoustic environments and soundscapes [33,34]. This study, by summarizing the spatial form data from multiple typical urban residential blocks, obtained a road network division method that better aligns with actual construction situations and reflects real-world road layouts more accurately. Architectural layouts in all models in this study meet sunlight standards. Additionally, layouts and designs based on the latest fire regulations and the most recent products in the real estate market were used, in line with the new developments and trends in urban residential construction. This approach enhances the applicability of the impact mechanisms and threshold recommendations proposed in this study. Compared to previous predictive models, previous studies mainly focused on measurements and statistics in local urban areas to establish models [49,50,51], or combined acoustic indices for acoustic environment monitoring [52]. These models are not built specifically for residential blocks despite their broad application. This study revolves around residential blocks and pays much attention to the combined impact of more parameters. It integrates spatial form parameters derived from real data with simulation to explore the role of design elements in adjusting predictive models and their practical application. The urban spatial morphology parameters in the first and second principal components can be prioritized according to the ordering of the four principal components in the planning and scale control of roads in new residential blocks. This approach makes it possible to quickly optimize the acoustic environment in residential areas.

However, it is worth noting that residential blocks have complex acoustic environments and traffic conditions, and the environments are typically dominated by biophonic and traffic sounds [53]; in cities, there is the noise of trains, airplanes, and nightlife [54], whereas this study only considers road traffic noise brought by cars. Furthermore, the scope of this paper is the acoustic environment of outdoor public spaces in residential blocks, which lacks attention to the acoustic environment within residential dwellings. Additionally, focusing solely on objective spatial form parameters may be one-sided; it is also important to consider residents’ perceptions of acoustic environments [55]. Lastly, this study primarily focuses on newly developed residential blocks in cities, and the results are only applicable to adjustments in new residential areas, with some methods potentially unsuitable for urban renewal in older areas.

5. Conclusions

Based on the analysis and summary of the spatial forms of residential blocks, this study classified the spatial form parameters into three categories, road network morphology, neighborhood scale, and architectural texture; explored the impact of the spatial forms of residential blocks on the distribution of road traffic noise exposure levels; put forward the proposed thresholds of some parameters; and then constructed a prediction model for the overall road traffic noise exposure levels of blocks based on the spatial form parameters. The following conclusions have been drawn:

(1)

Without considering the impact of buildings, two categories of the parameters, road network morphology and neighborhood scale, significantly affect the distribution of road traffic noise levels, and road network morphology has a larger effect than neighborhood scale. Specifically, the number of internal roads, road network density, and intersection density were positively correlated with the overall noise exposure level; the maximum neighborhood length, minimum neighborhood length, maximum neighborhood perimeter, minimum neighborhood perimeter, maximum neighborhood area, minimum neighborhood area, and the difference between the maximum and minimum neighborhood areas were negatively correlated with the overall noise exposure level. The acoustic environment of enclosed blocks is overall better than open blocks, under the same area and the same traffic conditions. In new residential blocks, it is recommended that neighborhood lengths be at least 140 m to 150 m, neighborhood perimeters be at least 550 m to 600 m, and neighborhood areas be at least 20,000 m² to 22,000 m².

(2)

When there are high-rise residential buildings in the block, all architectural texture parameters have an impact on road traffic noise distribution, showing different correlation trends according to different road network morphology divisions. Based on the results of the data analysis, and with reference to the planning and design specifications and design principles for residential blocks, it is recommended that the building density should be controlled at 18% or above, the open space ratio should be less than 80%, the floor area ratio should be set at 2.0–3.5, the degree of enclosure should be set at 40–50%, the length of openings along street-facing interfaces should be controlled at 600–800 m, and the maximum continuous width of street-facing buildings should be set at 60 m–70 m; the total length of building reflective surfaces within the neighborhood should reach 3200 m–3500 m.

(3)

This research employs principal component analysis to reduce the dimensionality of urban spatial form parameters, resulting in the identification of four principal components. Subsequently, a model was developed to predict overall noise exposure levels in residential areas, which was validated by example. This model can serve as an effective tool for predicting and diagnosing the quality of acoustic environments in urban residential blocks. It can be used for the rapid assessment of acoustic environment quality during the initial stages of block design or for evaluating acoustic environments after blocks have been completed.

This study responds to the reality of low acoustic environment quality in the construction of residential blocks in China and the needs of the residents. It follows the development trend of ‘small neighborhoods and dense road network’ and explores the impact of the spatial forms of residential blocks on the acoustic environment. It provides a new idea of acoustic environment for the urban planning and design level and proposes a predictive model for the acoustic environment of residential blocks, which is proposed as a rapid assessment tool for the acoustic quality of urban residential blocks. It can be used for the preliminary evaluation of the acoustic environment at the early stage of a block’s design or for the assessment of the acoustic environment after it has been constructed. Nevertheless, the study is not without limitations. It considers only road traffic noise from cars and focuses exclusively on the acoustic environment of outdoor public spaces in residential blocks. In reality, the acoustic environment in residential blocks is complex and diverse. The relationship between noise from the facade of residential blocks and the parameters of urban spatial morphology will be further explored in future research.