The Relationship between Urban Expansion and Air Pollution Depends on Urban Spatial Morphology in the “2 + 26” Cities of Beijing-Tianjin-Hebei, China

Abstract

Urban spatial morphology reflects the processes of external urban expansion and the internal spatial structure of the human economy, and exploring urban spatial morphology is important for regional management of PM_2.5 pollutants in the air environment. Whether urban spatial morphology can influence air pollution is a question that has been explored and evaluated. Therefore, we constructed a spatial-temporal driving mechanism framework and explored the interactions among urban expansion and PM_2.5 pollutants in the “2 + 26” cities of Beijing-Tianjin-Hebei (BTH). Based on 2010, 2015, and 2020 urban expansion factors and PM_2.5 pollutants data from the “2 + 26” cities of BTH, the results of the Geo-Detector medal imply that: (1) The spatial distribution characteristics of areas with high PM_2.5 pollutants in the “2 + 26” cities of BTH were affected by the PM_2.5 concentration value, and the high-value phenomenon appeared to decrease with the improvement of PM_2.5 pollutants. (2) The level of urbanization in the “2 + 26” cities of BTH had been increasing annually, with the most significant growth in construction land area located in the northeastern coastal cities. (3) The compactness index, was the most critical influencing urban expansion factor, explaining 0.63, 0.64, and 0.66 of PM_2.5 pollutants in 2010, 2015, and 2020. (4) In factorial interactions, the compactness index and the centrality ratio explained spatial and temporal analysis characteristics of PM_2.5 pollutants, and the interaction factor enhanced the explanation for the spatial-temporal characteristics. This study reveals the coupled spatial-temporal driving mechanism of air pollution in the “2 + 26” cities of BTH and also shows that reasonable urban planning has a positive impact on the regional ecological environment in the future.

1. Introduction

Since the beginning of the 21st century, urban problems such as the expansion of urban construction land, the loss of agricultural resources, and the concentration of industrial areas have become increasingly serious. Rapidly increasing concentrations of atmospheric pollutants from human activities are leading to serious urban ecological environmental issues [1]. Globally, the majority of the population lives in urban agglomerations of middle or low-latitude plains and cities in coastal areas [2]. It is expected that by the middle of this century, more than 70% of the world’s cities will have completed the process of urbanizing their populations [3]. By the next century, the global urban population will increase 1.8 times globally [4]. Rapidly increasing amounts of atmospheric pollutants from human activities are leading to negative environmental effects on citizens’ health [5,6,7]. Worldwide, more than one million people will be threatened with premature death due to air pollutants each year [8]. In addition to increasing the incidence of respiratory and cardiovascular diseases [9], severe air pollutants can impede the rate of photosynthetic reactions and affect the normal growth of plants [10,11]. There is a global consensus that we must control the intensity of air pollution emissions [12]. Asian cities, which are the fastest-growing urban areas in the world, will face serious air pollution challenges.

PM_2.5 refers to atmospheric suspended particulate matter with an aerodynamic particle size of less or equal to 2.5 μm, which characterizes a long suspension time and transmission distance [13]. Rapid urban socioeconomic development and dramatic development have led to more extensive human modification of the ecological environment. PM_2.5 pollutant concentration values in cities are affected by the doubly by the impact of urbanization and industrialization [14]. Nevertheless, under the influence of similar natural and weather conditions, reducing regional energy emissions and decreasing traffic commuting would improve urban air quality [15]. Furthermore, the level of economic development within industrial cities leads to the phenomenon of PM_2.5 pollution in urban agglomerations, while terrain elevation and wind speed affect air mobility and dispersion, which results in the formation of spatial heterogeneity characteristics [16]. A global multi-scale urban study showed that within Asia, China, and India are the countries where urban population change contributes to PM_2.5 concentration changes [17]. Although a consensus has been reached regarding the impact of urbanization on PM_2.5 pollutants [18,19], there is no conclusive evidence to quantify the relationship between urban expansion and PM_2.5 pollutants.

More and more scholars are quantifying the process of urban expansion by exploring the characteristics of urban spatial morphology in terms of the external shape of the city and its internal socioeconomic structure [20]. For example, Bugess [21] defined a circular spatial morphology with a core, a transition zone, and a distribution of residential and commuter zones in a future high-level city; Dempsey and Jenks [22] pointed out that an intensive and compact urban spatial morphology would help improve transport and promote sustainable urban development; and Harris and Ullman [23] argued that as a city develops, it develops from a single nucleus to multicore cities. The influence mechanism of urban spatial morphology on air pollutants includes three aspects: emission, purification, and diffusion [24]. The point (factor) and non-point (locomotives) sources can easily lead to air pollution in the radiation range [25]; urban forests and green areas have a closer relationship with pollutant purification, and the larger the proportion of green areas, the better the relationship for the improvement of air quality [26]; finally, the air quality of the compact urban spatial morphology is often better than that of the traditional urban spatial morphology [27]. Therefore, the urban spatial morphology index can be used to quantify the process of urban expansion and analyze the relationship between urban spatial morphology and PM_2.5 pollutants based on the “source-sink” theory.

The “source-sink” theory involves the components of landscape components and landscape configurations in the subsurface of urban space [28]. The urban landscape component emphasizes the impact of land use types such as urban construction land, urban green space, and water columns on air pollutants through physical properties [29]. Landscape configurations emphasize the impact of landscape interactions on the atmospheric environment [30]. Under the “source-sink” theory, certain urban spatial landscape types, such as roads, building land, and urban socio-economic assemblages within the urban space, are applied to calculate the urban spatial morphology index [31]. Based on remote sensing to extract the urban construction land and the main urban area quantitatively described the urban centrality, outer network contour, and building density [32]. In the existing analyses of urban spatial morphology indices and air quality influencing factors, the urban spatial morphology indices constructed with the landscape of urban construction land are often affected by covariance, which interferes with the results of the research [33,34]. The Geo-Detector model is a model that quantifies causal relationships between impact factors by constructing spatial stratification and calculating within-stratum variance [35]. The Geo-Detector model has became an important statistical method in the field of ecology [36,37], which preserves the original input variables and greatly reducing covariance.

This study was conducted to investigate the sustained influence of urban spatial morphology characteristics on the spatial-temporal of PM_2.5 pollutants in the “2 + 26” cities of BTH. The recommended and concentration values of PM_2.5 pollutants were analyzed by statistical analysis and visualized in ArcGIS 10.2. The land-use-change status was determined, and the urban expansion factor included urban spatial morphology and urban meteorological data. With the assistance of the Geo-Detector model, comprehensive descriptions of the contribution and influence of the important driving factors of spatial-temporal characteristics of urban expansion factors and PM_2.5 pollutants are summarized. Our study would provide recommendations about urban development and governance that inform the building of green and sustainable cities.

2. Materials and Methods

2.1. Study Area

Asia is one of the fastest-growing regions for urban agglomerations worldwide, with more than 800 cities with populations greater than 100,000, and China is one of the most attractive Asian countries for urban development [38]. The “2 + 26” cities of BTH comprise one of the fastest-growing urban agglomerations in China [39]. The “2 + 26” cities of BTH border the Bohai Sea between Liaoning and Shandong, with a geographical range of 34°16′′ N to 41°36′ and 111°30′ to 119°50′ (Figure 1). The “2 + 26” cities of BTH have a strong economy, and their total GDP ranks among the top city clusters in the country. In 2019, the “2 + 26” cities of BTH’s total GDP exceeded CNY 14 trillion, accounting for about 14.3 percent of the country’s total. The “2 + 26” cities of BTH are not only economically powerful but also an important center of urbanization and regional development in China [40]. The “2 + 26” cities of BTH are located along the Taihang Mountains as an air pollution transmission corridor. Cities such as Shijiazhuang, Taiyuan, and Anyang are densely populated, comprising industrial interest areas. The enhancement of the emission factor of civil heating in the “2 + 26” cities of BTH will lead to the rapid accumulation of urban emissions in the winter. Therefore, the “2 + 26” cities of BTH are an area in urgent need of air pollution control.

2.2. Study Data

2.2.1. PM_2.5 Concentration Data

The 2010, 2015, and 2020 PM_2.5 pollutant data in the “2 + 26” cities of BTH were obtained from the National Urban Air Quality Real-time Distribution Platform (https://air.cnemc.cn:18007/; accessed on 17 September 2023). These data comprised municipal PM_2.5 pollutant concentration data values from the hourly monitoring data from each ground station. The PM_2.5 annual average concentration values (μg/m³) in the “2 + 26” cities of BTH were calculated using the daily average concentration values of each prefecture-level city; the missing values were spatially interpolated with inverse distance weighting using ArcGIS 10.2 software.

2.2.2. Remote Sensing, Meteorological, and Socioeconomic Data

We selected the 2010, 2015, and 2020 Landsat land cover and population density data to map the spatial-temporal characteristics in the “2 + 26” cities of BTH. The urban construction land data were obtained from the 30 m resolution China Land Cover Dataset from the Resource and Environmental Data of the Chinese Academy of Sciences (www.resdc.cn; accessed on 17 September 2023). The population density data were obtained from the 1000 m × 1000 m resolution LandScan population density dataset (https://landscan.ornl.gov/; accessed on 17 September 2023).

Considering the socioeconomic causes of air pollutants [41], the population density (P (person/km²)), gross domestic product (GDP), industry structure (Ind (%)), and bus ownership (Bus (units)) were selected as the socioeconomic data, which were obtained from the 2011, 2016, and 2021 China Urban Statistical Yearbook.

Unfavorable meteorological conditions are detrimental to the dispersion of atmospheric pollutants [42,43]. Therefore, the meteorological data of temperature (T (°C)), precipitation (PCP (mm)), relative humidity (RH (%)), and wind speed (WS (m/s)) were picked as the dependent variables. These data were obtained from the hour-by-hour National Meteorological Science Data Center database from 2010 to 2020 (http://data.cma.cn/; accessed on 17 September 2023). In this study,

Table 1. Data sources.

Data Name	Abbreviated Variable and Unit	Data Format	Data Source
population density	P/(Per person/km2)	Raster/1 km	LandScan population density dataset (https://landscan.ornl.gov/; accessed on 17 September 2023)
Land use	/	Raster/30 m	Resource and Environment Science and Data Center, Chinese Academy of Sciences (www.resdc.cn; accessed on 17 September 2023)
the gross domestic product	GDP/(Yuan/Per person)	Excel	The China Statistical Yearbook 2011, 2016, and 2020 (https://www.stats.gov.cn/sj/ndsj/; accessed on 17 September 2023)
the industry structure	Ind/(%)	Excel
bus ownership	Bus/(units)	Excel
Temperature	T/(°C)	Excel	The National Meteorological Science Data Center (http://data.cma.cn/; accessed on 17 September 2023)
Precipitation	PCP/(mm)	Excel
Wind speed	WS/(m/s)	Excel
Relative humidity	RH/(%)	Excel
PM2.5 pollutants	PM2.5/(μg/m3)	Excel	The National Urban Air Quality Real-time Distribution Platform (https://air.cnemc.cn:18007/; accessed on 17 September 2023)

Note: We use acronyms for variables to define them: P–population density; GDP–gross domestic product; Ind–industry structure; Bus–bus ownership; T–temperature; PCP–precipitation; RH–relative humidity; WS–windspeed.

shows that China’s land use remote sensing data, China’s socio-economic data, air pollutants, and meteorological datasets are used.

2.3. Models and Methods

This study analyzed the degree of urban expansion and the PM_2.5 pollution problem in the “2 + 26” cities of BTH. We took the PM_2.5 pollutants as the dependent variable and the urban spatial morphology index and socio-economic and meteorological data as the independent variables to construct a dataset. We calculated the urban construction land and analyzed the spatial-temporal aggregation characteristics; specifically, we combined our method to analyze the influence of PM_2.5 pollutants by the factorial detector and recognized the effects of interactions through interaction detection in the Geo-Detector (Figure 2).

2.3.1. Spatial Autocorrelation Analysis of Urban Construction Land

Spatial autocorrelation followed Tobler’s First Law [41] to analyze the spatial autocorrelation of adjacent urban construction land in the “2 + 26” cities of BTH. It was used to measure the spatial agglomeration of urban construction land. The Lisa (Local indicators of spatial association) index is a typical local spatial autocorrelation statistical method that was used to test the spatial hot and cold spots among prefecture-level cities within the 2 + 26 cities. The calculation formula for the urban spatial autocorrelation is obtained using the following formula:

$$I=\frac{n}{S_0}\times \frac{{\mathit{\Sigma }}_{i=1}^n{\mathit{\Sigma }}_{j=1}^n{w}_{ij}{Z}_i{Z}_j}{{\mathit{\Sigma }}_{i=1}^n{z}_i^2}$$

(1)

$$S_0={\mathit{\Sigma }}_{i=1}^n{\mathit{\Sigma }}_{j=1}^n{w}_{ij}$$

(2)

$$I_i=\frac{Z_i}{s^2}{\mathit{\Sigma }}_{\dot{J}\ne 1}^n{w}_{ij}{Z}_i$$

(3)

where $Z_i$ and $Z_j$ are the urban construction land of ${(x}_i-\overline{x})$ and ${(x}_j-\overline{x})$ in the regions;$w$ is the spatial matrix data; $S_0$ is the aggregation of spatial weights of the region; $s^2$ is the variance of aggregation; $n$ is the number of urban space units;$I$ and $I_i$ represents global and local Moran’s I index of the urban construction land in the “2 + 26” cities of BTH; When 0 < Moran’s I index ≤ 1, there was a positive spatial correlation with the 2 + 26 urban construction land. When Moran’s I index was < 1, there was a negative spatial correlation. When Moran’s I index = 0, the correlation was randomly distributed in space.

2.3.2. Meaning and Characterization of Urban Spatial Morphology Indices

Urban spatial morphology reflects the processes of external urban expansion and the internal spatial structure of the human economy in this study. In urban planning and urban ecology, urban spatial morphology includes three levels: micro, middle, and macro [44]. The microscale represents the functional space within the city [45] and the middle- and the macroscales are structures of the city cluster system [46]. The urban structure indicators (population density, industrial structure, traffic access, and nighttime remote sensing index) and the urban shape indicators (compactness, fractal dimension, centrality ratio, and circularity ratio) are defined on the middle- and macroscales [47]. Globally, cities, as sites of human activities, export anthropogenic air pollutants annually [48]. Therefore, urban spatial morphology and air pollutants are correlated with emission, dispersion, and purification. The flow of urban spatial morphology and PM_2.5 pollutant-driven impact mechanisms is established in Figure 3.

In our study, indices such as fractal dimension (FRD), compactness index (CI), circularity ratio (CR), and centrality ratio (CEN) were applied to define urban spatial morphology. FRD reflected the external profile of prefecture-level cities in the “2 + 26” cities of BTH. The theoretical range of FRD is 1–2, and the higher the calculated FRD was, the closer the cities in the “2 + 26” cities of BTH were to the traditional urban spatial morphology [49]. CI was the degree of concentration of urban spatial patterns outside the cities of the “2 + 26” cities of BTH. The CI was conducive to controlling the rate of urban land expansion and improving public transportation [50]. The circularity ratio, CR, precisely reflected the degree of discrete regional and urban development and measured the circular cities [51]. The centrality ratio or CEN, was the degree of spatial concentration of elements such as population or economic activities in the central area within a city, which showed the grid of the city [49]. Urban transportation is an influential factor that cannot be ignored in the process of urban expansion [49]; this has an impact on the spatial form of cities. Transportation for commuting is an important part of sustainable urban development. The calculation formula for the urban spatial morphology index is obtained using the following formula:

$$FRD=\frac{2{\log }_{10}(\frac{P}{4})}{{\log }_{10}A}$$

(4)

$$CompactIndex=\frac{2\sqrt[]{\mathit{\pi }A}}{P}$$

(5)

$$CR=\frac{4A}{P^2}$$

(6)

$$CEN=\frac{\mathit{\sigma }\left(POP\right)}{ARV\left(POP\right)}$$

(7)

$$TC=\frac{BUS}{A}$$

(8)

where A is the total area of urban construction land; P is the perimeter of urban construction land; $\mathit{\sigma }\left(POP\right)$ is the standard deviation of the population density; $ARV\left(POP\right)$ is the average value of the population density; and $BUS$ denotes the number of buses in the “2 + 26” cities of BTH.

2.3.3. Geo-Detector Modeling to Analyze the Effect of Urban Expansion Factors and PM_2.5 Pollutants

Based on the core concept of geospatial heterogeneity [52], we used Geo-Detector to construct spatial matrices of urban spatial morphology and PM_2.5 pollutants. Then, we used Geo-Detector to calculate their spatial correlations, which could better reflect the causes and mechanisms behind geographic phenomena than traditional models [53]. Spatial heterogeneity included a significant change in the relationship between certain variables in space with spatial changes; secondly, the statistics (e.g., mean and variance) of certain variables in space changed smoothly or not smoothly with the spatial changes [54]. Our study focused more on the spatial variation of geospatial elements.

The Geo-Detector data preprocessing included extracting all kinds of urban spatial morphology and meteorological data and transforming them into raster data; constructing 1000 m × 1000 m resolution grid data and extracting all kinds of raster data into grid data through ArcGIS 10.2 software; and, thirdly, extracting points according to the spatial analysis and statistical tools and exporting them to the Geo-Detector modeling algorithms. The flowchart of the Geo-Detector method in the “2 + 26” cities of BTH is shown in Figure 4.

The factorial detection of the geographic probe explores the urban spatial morphology and PM_2.5 pollutant spatial heterogeneity of the 2 + 26 cities by calculating the q-value. The specific equation is as follows:

$$q=1-\frac{1}{n{\mathit{\sigma }}^2}{\sum }_{h=1}^L{n}_h{\mathit{\sigma }}_h^2$$

(9)

where $q$ is the detection intensity of the PM_2.5 pollutants in the 2 + 26 cities; $n$ is the total number of samples from the 2 + 26 cities; L is the Stata of prefecture-level city variables in the 2 + 26 cities ($h=$ 1,2,3…$h$); and ${\mathit{\sigma }}^2$ is the variance of PM_2.5 pollutants in the 2 + 26 cities. The size of $q$ can be used as an explanation for the PM_2.5 pollutants in the 2 + 26 cities, and its value range is generally [0, 1]. The larger the value of $q$, the higher the explanatory value of the driver for the PM_2.5 pollutants.

The interaction detector is used to identify the interaction between the urban spatial morphological index and the effect of the main factors $x1$ and $x2$ together with the variation in the interpretation rate of PM_2.5 pollutant characteristics in the 2 + 26 cities. The assessment method is shown in

Table 2. Geo-Detector interactive detection.

Basis of Judgment	Interactive Effect of Factors
q(x1∩x2) < Min(q(x1), q(x2))	Nonlinear weaken
Min(q(x1), q(x2)) < q(x1∩x2) < Max(q(x1), q(x2))	Uni-enhance/weaken
q(x1∩x2) > Max(q(x1), q(x2))	Bi-enhance
q(x1∩x2) = q(x1) + q(x2)	Independent
q(x1∩x2) > q(x1) + q(x2)	Nonlinear enhance

.

3. Results

3.1. The Changes in PM_2.5 Pollutant Concentration and Recommended Values in the “2 + 26” Cities of BTH

The concentration of PM_2.5 pollutants in the “2 + 26” cities of BTH increased by 12.86% from 2010 to 2015. After the development of the city in 2020, Figure 5a shows that PM_2.5 pollutant concentrations significantly shifted, from 82.62 μg/m³ to 47.05 μg/m³. However, the average concentration of PM_2.5 pollutants in the last ten years was 67.63 μg/m³. Considering the WHO-recommended level [55], the PM_2.5 pollutants in the “2 + 26” cities of BTH were classified into AQG (Air Quality Guidelines) and IT (Interim Targets). We classified the PM_2.5 pollutants’ annual mean concentration values into five classes: AQG (0.00–15.00 μg/m³), IT-4 (25.00–37.50 μg/m³), IT-3 (37.50–50.00 μg/m³), IT-2 (50.00–75.00 μg/m³), and IT-1 (>75 μg/m³) (Figure 5b). In 2020, the AQG and IT-4 percentages increased by 17.13% and 20.09%, respectively, compared with the 2010 recommended PM_2.5 pollutant levels.

The spatial change characteristics of PM_2.5 pollutants in the “2 + 26” cities of BTH show evident spatial differentiation characteristics (Figure 6). In 2010, Kaifeng City (86.22 μg/m³), Dezhou City (85.59 μg/m³), Hengshui City (84.85 μg/m³), and other prefecture-level cities in the center and the south of the study area had high values of PM_2.5 pollutants; in 2015, Baoding (105.69 μg/m³) had the highest value of PM_2.5 pollutants of all the “2 + 26” cities of BTH; in 2020, the area with high values of PM_2.5 pollutants showed a significant reduction, but Hengshui City (78.90 μg/m³), Dezhou (78.38 μg/m³), and Liaocheng City (78.38 μg/m³) were still above the IT-1-recommended guideline of 75 μg/m³.

3.2. The Land-Use Change in the “2 + 26” Cities of BTH

Figure 7 shows that cropland, as the most dominant land-use type in the “2 + 26” cities of BTH, accounts for more than 55 percent of the total land area (2.7 × 10⁴ km²). From the analysis of land-use-change trends from 2010 to 2020, forest land increased the most of all the land-use types, from 12.36% to 13.08%, followed by a 0.52% growth in urban construction land. Grassland saw an increase of 0.44%. Water and other land-use types accounted for less than 1%. In the past ten years, the total amount of cropland in the “2 + 26” cities of BTH decreased; the proportion of forest land, grassland, and urban construction land-use types increased significantly; and water bodies and other land-use types leveled off. At this stage, while the rapid urbanization of the “2 + 26” cities of BTH has led to the loss of cropland, the protection of ecological land, such as grassland and woodland, has been better overall.

3.3. The Spatial Autocorrelation Analysis of Urban Construction Land in the “2 + 26” Cities of BTH

Table 3. The temporal characteristics of Moran’s I index for construction land in the “2 + 26” cities of BTH.

Year	Moran’s I	Z-Value	p-Value
2010	0.36	14.89	0.00
2015	0.52	13.82	0.00
2020	0.63	13.15	0.00

indicates that there was obvious spatial autocorrelation in the expansion of the “2 + 26” cities of BTH. The Moran’s I index value of the construction land was 0.63 (2020) > 0.52 (2015) > 0.36 (2010), and the spatial agglomeration distribution strengthened annually.

Figure 8 shows the local Lisa index of the changes in the area of construction land in the “2 + 26” cities of BTH from 2010, 2015, and 2020. Among them, the northern part, the eastern coast, and the southwestern region of the “2 + 26” cities of BTH were the high–high aggregation areas with high values of construction land in 2010; Puyang City, Anyang City, and Heze City had relatively low levels of urbanization from 2010 to 2015; and the development of cities in the central and western parts of the “2 + 26” cities of BTH were relatively stable, and their interannual spatial changes were not obvious in 2015. The trend in the development of cities in Shandong, such as Jinan City in the east, from 2015 to 2020. The spatial agglomeration characteristics of the northern cities, including Beijing City, Tianjin City, and Langfang City, gradually shifted towards the eastern coast.

3.4. The Spatial-Temporal Characteristics of Urban Expansion Factors in the “2 + 26” Cities of BTH

There is a clear time-divergence feature of the influence factor expansion from 2010 to 2020 (Figure 9) as the “2 + 26” cities of BTH continued to transform into green, sustainable, and high-quality cities. Except for FRD, which showed a decreasing trend, the spatial structure influence factors of the cities were enhanced annually. With the acceleration of the urbanization process in the “2 + 26” cities of BTH, the CI, CR, and CEN increased by 0.04, 0.55, and 0.03, respectively. The increasing intensity of human activities in the process of rapid urbanization led to a continuous upward trend in the spatial structure from 2010 to 2020. Compared with the factors affecting urban spatial patterns, meteorological factors were less affected by urban expansion and human activities. There were obvious nonlinear differences in the interannual trends in PCP among various meteorological factors.

There are obvious spatial differentiation characteristics of the influence factor of urban expansion in the “2 + 26” cities of BTH (Figure 10). In addition to FRD, the rest of the cities with high values of urban spatial shape and urban spatial structure influence factors were located in Beijing, Jinan, Zhengzhou, and other cities with high urbanization levels. Secondly, the spatial distribution characteristics of the FRD influence factors showed an opposite trend to other related factors. Cities with stable urbanization levels, such as Changzhi City, Cangzhou City, and Jincheng City, had average FRD values of 1.74, 1.73, and 1.70. Overall, the influence factors of various urban spatial morphologies showed spatial changes from the periphery to the middle of the city. In addition, the meteorological factors were high in the north and low in the south, decreasing from coastal to central cities.

3.5. Analysis of the Spatial-Temporal Characteristics between Urban Expansion and PM_2.5 Pollutants in the “2 + 26” Cities of BTH

CI, Ind, and T were the most significant main factors among the urban expansion influencing factors on the spatial and temporal characteristics of PM_2.5 pollutants in the “2 + 26” cities of BTH. With the development of the “2 + 26” cities of BTH, the construction of compact cities improved. Figure 11 shows that the explanation rate of the CI index differentiation characteristics of PM_2.5 pollutants increased from 0.63 to 0.66. Due to the transformation and adjustment of the industrial structure, the explanation rate of the Ind for PM_2.5 pollutants decreased annually. In this context, in 2020, the impact of pandemic prevention and control policies caused a substantial decline, to 0.26. In controlling the spread of the COVID-19 outbreak through regional vehicle restrictions, shutdowns, and production stoppages, the government reduced the intensity of pollutant emissions and mitigated air pollution in urban areas. Meteorological elements were less affected by human activities, and there was a nonlinear trend in interannual changes. Temperature was the most obvious interpreted meteorological data for PM_2.5 pollutants, with an explanation rate of more than 0.40, while RH was the meteorological factor with the least influence on PM_2.5 pollutants, with an explanation rate of less than 0.10.

The interaction values of the geo-detectors explain the spatial heterogeneity characteristics of PM_2.5 pollutants in the “2 + 26” cities of BTH. When the interaction value between factors is higher, it indicates a greater influence on the spatial variation characteristics of PM_2.5 pollutants, and vice versa, it has less influence on PM_2.5 pollutants, and the results are shown in Figure 12. The interaction effects of the factors appeared to change dynamically with time. In terms of urban expansion factors, CI ∩ CEN (0.92), Ind ∩ Green (0.78), and PCP ∩ T (0.88) were the strongest urban spatial shape, structure, and meteorological factors affecting the spatial-temporal characteristics of PM_2.5 pollutants. Additionally, in terms of the types of urban expansion factors, the strongest interaction among the urban spatial shapes between the CI and CEN remained unchanged, while Ind ∩ TC and T ∩ WS were added for urban spatial structure and urban meteorological factors, respectively.

Judging from the most important urban expansion influencing factors on PM_2.5 pollution in the “2 + 26” cities of BTH, all showed an increasing trend (Figure 13). Among them, the proportion of nonlinear enhancement trend in the overall enhancement trend in PM_2.5 pollutants was 66.67%. In terms of spatial differentiation characteristics, the explanation rates of urban spatial shape, urban spatial structure, and urban meteorological factors all exceeded 0.50. Specifically, the most important combinations of urban sprawl influencing factors affecting the 2 + 26 cities were CI ∩ CEN (0.91), Ind ∩ Green (0.78), and PCP ∩ T (0.88). In terms of the temporal divergence characteristics, the explanation rate of each type of interaction combination showed a significant decline. Among them, the urban spatial structure declined most significantly, with the decline in the interaction combination of the Ind∩Green exceeding 0.19.

4. Discussion

4.1. Urban Expansion Exacerbates the Harmful Effects of PM_2.5 Pollutants

The study of the urbanization process is one of the hot topics in human geography and urban ecology research. Urbanization includes the transformation of a rural population into an urban population, socioeconomic restructuring, and land-type changes in peri-urban construction land within cities [56]. Worldwide, large cities and urban agglomerations are generally located along economically developed regions in middle and low latitudes [56,57]. The “2 + 26” cities of BTH, as economically developed and densely populated cities in Asia, are representatives of typical high urbanization. In the process of population urbanization, uncontrolled urban expansion increases the probability of the emergence of urban ecological and environmental problems [58]. Increased energy consumption by urban industries and motor vehicles is a potential “source” of near-surface pollution [59]. Zhou et al. [60] observed the dust aerosol layer between the surface and 4000 m, and combined with the trajectory analysis, found that the air pollution events in the key cities in China were affected by the long-distance transport of the north-westerly winds. The compactness of cities can improve the urban climate by encouraging green and low-consumption energy and by promoting public transport [61]. We found that changing the near-surface atmospheric circulation contributes to improved atmospheric quality and a reduced spatial concentration of pollutants in cities. Following the “smart growth model” will effectively manage urban planning and the environment [62]. Although the environmental hazards associated with increased construction land in megacity clusters cannot fully explain the environmental impacts of urban expansion, they reveal the hazardous effects of urban expansion in the “2 + 26” cities of BTH on the atmospheric environment.

This study analysis provided theoretical knowledge for exploring the mechanisms for the environmental impact of rapid urban development. However, the following shortcomings still exist: (1) The high-resolution, continuous-time-series remote sensing data of land use type used lacks a continuous and long-time series. (2) In this study, we extracted motor vehicle and bus data in the “2 + 26” cities of BTH based on the statistical yearbook and calculated the urban traffic commuting degree with ArcGIS 10.2 software. Due to the lack of lorry and motorbike data, there is still some deviation from the real results. Therefore, in future research, we would improve the impact of regional urbanization on the peri-urban atmospheric environment by combining atmospheric pollution components.

4.2. Definition of the Urban Spatial Morphology Index

The process of urbanization has increased the density and height of urban buildings. The transformation of the urban environment affects the transformation of the natural environment through urban physical space morphology [63]. Urban spatial shape was calculated based on the external contour, perimeter, and total area of the studied cities. Urban spatial structure, which combined social, humanistic, economic, and landscape indices, was often used in cross-disciplinary research [64]. The urban spatial morphology indices included the frontal area index calculated based on the longest axis of the region and the area of the region [65]; the ellipticity ratio obtained by Stoddart based on the long and short axes of the city [66]; and the radial shape index calculated by Boyce and Clark based on the distances from the center to the various districts of the city [67]. We took the circular area of the main city as the standard unit of the study, which not only integrated the multifaceted characteristics of the external shape but also optimized the problem of the cumbersome calculation of the traditional urban spatial morphology index. Urban spatial morphology is a result of the interaction between the urban natural landscape, urban social development dynamics, urban structure, and functional factors [68]. In recent studies, other socioeconomic and urban landscape indicators, such as the intensity of human activities, have been applied to the study of urban spatial morphology [69]. Compact cities reduce line-source traffic pollution by rationally regulating the road network [70], increasing the area of urban green space to reduce the urban heat island effect to reduce air pollutants [71,72], and improving urban ventilation to weaken the spatial concentration of pollutants by changing the spacing arrangement of buildings. The “2 + 26” cities of BTH are one of the urban agglomerations with the highest levels of urbanization development within Asia. We found that PM_2.5 pollutants were relatively high with high urbanization levels, such as those in the “2 + 26” cities of BTH. Therefore, urban expansion enhances human activities within cities, which in turn affects urban spatial patterns and influences PM_2.5 pollutant concentration values.

Urban spatial morphology is affected by urban physical shape elements such as natural geographic location, urban ventilation, and slope [73]. Since the “2 + 26” cities of BTH have similar altitudes and natural climate conditions due to geographical spans, and due to the data sources and research equipment, we did not study the urban boundary layer, urban crown layer, or urban street canyon. Therefore, we would focus on the morphology of the texture and the structure of the urban building components in the urban spatial morphology in future research.

4.3. Effects of the Major Urban Expansion Index on Spatial and Temporal Characteristics of PM_2.5 Pollutants

The main factors influencing urban expansion are characterized as urban construction land, population density, and economic development level in the process of urbanization. The industrial structure and urban compactness can support the explanation of the expansion of urbanization [74]. As the quantification of urban expansion exists in a variety of expressions, there were differences in the urban PM_2.5 pollutants evaluated in the “2 + 26” cities of BTH. The Geo-Detector serves as a model for overlaying spatial variables within a regional unit by reclassifying them, establishing spatial relationships, and re-evaluating spatial differences between variables [75]. The Geo-Detector showed that the same dominant driver had different spatial-temporal explanatory powers for PM_2.5 pollutants. As the industrial structure of the dominant driver increases, the emission of pollutants based on human activities also increases [76]. The change in the share of the industrial structure manifests in the PM_2.5 pollutants due to the different rates of their impacts on air pollutants during the physical and chemical transport of exhaust matter emitted from factories [77]. As an important indicator of spatial shape connectivity, the compactness index is one of the important influencing factors for air pollutants [78]. In addition, urban construction land is a “source” landscape, where the meteorological conditions in “source” landscape centers with high temperatures and low wind speeds could influence air pollutant concentration values [79]. Therefore, encouraging the construction of reasonable urban spatial morphology, promoting the industrial structure, and rationally planning the process of urban expansion will contribute to regional air pollution control.

The main influencing factors of the spatial-temporal characteristics of PM_2.5 pollutants in the “2 + 26” cities of BTH were analyzed using the Geo-Detector model. The improvement of PM_2.5 pollutants in the context of urbanization was discussed. The Environmental Protection Agency reduces air pollution hazards by setting emission standards for new-energy vehicles and weakening pollutant emissions per kilometer traveled [80]. In addition, measures such as increasing the area of urban green space, improving the urban public transport road network, and adjusting the industrial layout also improve the local urban ecological environment. Therefore, in the future, the “2 + 26” cities of BTH should mitigate air pollution hazards through rational urban planning and the promotion of new-energy-vehicle standards.

5. Conclusions

We aimed to address the lack of knowledge on the relationship between the mechanisms of urban expansion and PM_2.5 pollutants from a multi-scale and macro perspective. This paper, based on data on land use, population density, socioeconomic indicators, and environmental monitoring stations, revealed the influence of the coupled driving mechanism of spatial-temporal urban expansion indices on PM_2.5 pollutants. The main conclusions are as follows:

(1)

From 2010, 2015, and 2020, the concentration of PM_2.5 pollutants decreased by 26.15 μg/m³ in the “2 + 26” cities of BTH. Among them, the trend in the spatial characteristics of PM_2.5 pollutants decreased from the central–northern part of the country to the surrounding areas.

(2)

The Moran I Index showed that 0.36, 0.52, and 0.63 of urban construction land were located in “hot-spot areas” with respect to the city area in 2010, 2015, and 2020, respectively.

(3)

There were significant spatial heterogeneity characteristics in the urban expansion index. In addition to the fractal dimension, the values of the other urban spatial morphology factors all increased with urban expansion. The influencing factors of meteorology were relatively less affected by urban expansion.

(4)

The compactness index, the industrial structure, and temperature were the main urban expansion factors affecting PM_2.5 pollutants in the “2 + 26” cities of BTH. Among them, the compactness index and the centrality ratio, as the most important interaction combinations, explained 0.92 of PM_2.5 pollutants.

Urban expansion, human activities, and meteorological conditions were important factors in the variation of PM_2.5 pollutants. At the same time, remote sensing data images as well as Geo-Detector models had strong geospatial analysis capabilities. Our findings will help achieve the environmental eco-city goal of promoting the improvement of the urban atmospheric environment.