Essence Edition
Driven by globalization, industrialization, and urbanization, China's cities have expanded rapidly over the past 20 years. According to data released by China's National Bureau of Statistics, the country's urban built-up area increased from 22,400 km² in 2000 to 62,400 km² in 2021, with an average annual growth rate of about 5%. Under the background of the current slowdown of urbanization, fiscal tightening in some cities, and insufficient economic growth momentum, it is of great significance to scientifically and reasonably understand the spatial growth characteristics of Chinese cities, especially the differences in the spatial growth characteristics of cities of different sizes, for formulating differentiated strategies to promote the sustainable growth of cities.
Spatial density is one of the important dimensions to explore the growth characteristics of urban space, which usually includes indicators such as building density, construction land density and floor area ratio, which usually represent the building area, construction land area, and the ratio of total building area to plot area on a specific plot of land, respectively. In recent years, many scholars have used the circle distribution of construction land density in the city to reflect the "center-periphery" characteristics of urban space growth, but few studies have used the distribution of building density to describe the growth process of urban space from the center to the periphery. In fact, compared with the density of construction land, the building density can capture the impact of existing land renewal projects on urban space to a certain extent. Compared with the plot ratio index, which is difficult to obtain, some high-precision data products reflecting the distribution of building density in cities have emerged in recent years, which provides support for the comparative study of multi-city, large-sample and diachronic urban spatial growth, and is conducive to making up for the lack of attention to small and medium-sized cities in existing studies.
Based on the above background, this paper takes 287 cities at and above the prefecture level in China as the research sample, takes 2000-2010 and 2010-2020 as the two research periods, and uses the global building surface data product released by the "Global Human Settlement Layer" project of the European Union Joint Research Center in 2022 to superimpose the building density data at the 100m grid scale with the 1km circle and the urban land administrative boundary, and apply " Circle gradient analysis method analyzes the centrality of building density distribution in the city (Fig. 1), and constructs the marginal growth length index of the sample cities in two time periods, so as to explore the growth degree of the urban edge and the differences in growth characteristics between cities of different sizes in the past 20 years.
Fig.1 Schematic diagram of the superposition of 100m raster-scale building density data and 1km circle layer in the city (taking Beijing as an example)
The main conclusions of this paper are as follows: (1) the distribution of urban building density gradually decreases from the center to the periphery as a whole, which can be fitted well by the negative exponential function, and (2) most cities show the characteristics of further marginal growth in the two time periods of 2000-2010 and 2010-2020, and the growth range of marginal growth length from high to low is in the order of large cities, medium cities, super large and mega cities, and Small cities, but the growth rate of small cities is not statistically significant (Fig. 2);(3) Natural geographical conditions, population and economic scale, and macro policy and planning guidance are important factors affecting the spatial growth characteristics of cities and the differences between cities of different sizes.
Fig.2 Statistical characteristics of urban marginal growth length of different scales in two time periods
full text
【Abstract】Building density is one of the important indicators to measure urban spatial density, and its distribution pattern in urban areas reflects the spatial growth characteristics of urban horizontal scale to a certain extent. Therefore, it is of great significance to explore the distribution of building density in urban areas to reveal the basic characteristics of urban spatial growth and formulate planning strategies to promote sustainable urban growth. In this paper, 287 cities at and above the prefecture level in China were used as research samples, and the building density data of 100m raster scale in 2000, 2010 and 2020 were used to construct the urban edge growth length index based on building density circle gradient analysis, and the spatial growth characteristics of the sample cities at the horizontal scale were studied. The main conclusions are as follows: (1) the distribution of urban building density gradually decreases from the center to the periphery as a whole, which can be fitted well by the negative exponential function;(2) most cities show the characteristics of further marginal growth in the two time periods of 2000-2010 and 2010-2020, and the growth range of marginal growth length from high to low is large city, medium city, super large and mega city, and small city, but the growth rate of small city is not statistically significant; (3) natural geographical conditions, Population and economic scale, macro policies and planning guidance are important factors affecting the characteristics of urban spatial growth and the differences between cities of different sizes.
introduction
Driven by globalization, industrialization and urbanization, China's cities have expanded rapidly over the past 20 years. According to data released by the National Bureau of Statistics, the built-up area of cities across the country increased from 22,400 km² in 2000 to 62,400 km² in 2021, with an average annual growth rate of about 5%. However, the rapid expansion of cities has also created a number of problems that are not conducive to their sustainable growth, such as the encroachment of forests and farmland, and the energy consumption and pollution caused by long-distance commuting. Therefore, a reasonable understanding of the spatial growth characteristics of Chinese cities is of great significance for formulating planning strategies to promote sustainable urban growth.
Spatial density is one of the important dimensions of urban spatial growth characteristics in existing studies, which usually includes specific indicators such as building density, construction land density and floor area ratio, which usually represent the building area and construction land area on a specific plot respectively [In actual research, most of them identify construction land based on remote sensing data by analyzing the natural ecological attributes of the land, and generally use the "impervious layer" in remote sensing images to characterize it, which is different from the "construction land" standard in the field of urban planning]. The ratio of the total area of the building to the area of the plot. It should be emphasized that for a certain plot, there is not a complete positive correlation between the above three indicators. For example, plots with high building density can have a lower plot ratio (e.g., the buildings are all low-rise residential buildings), and plots with high construction density can also have lower building density (if the area of roads or plazas is large). However, in terms of the distribution characteristics of the circles of these indicators in the city, they generally show a decreasing trend from the center to the outer circles. Among them, the distribution pattern of building density and construction land density in the city can reflect the spatial growth characteristics of the city at the horizontal scale to a certain extent, while the distribution pattern of the floor area ratio in the city can reflect the "three-dimensional growth" characteristics of urban space (including horizontal and vertical scales) to a certain extent.
In recent years, the distribution pattern of spatial density in urban areas and the resulting characteristics of urban spatial growth have gradually attracted academic attention, but related studies are mainly based on the analysis of construction land density data. For example, some studies analyze the decreasing law of construction land density from the center to the edge of the urban circle, and then measure and discuss the "center-periphery" characteristics of typical urban space growth by comparing the difference in the decreasing rate of urban circle density in different years. Although many scholars have pointed out the importance of the floor area ratio index in depicting the spatial characteristics of urban three-dimensional buildings, due to the lack of high-precision, multi-temporal and standardized urban height data, the research on urban spatial growth based on the distribution of floor area ratio mainly focuses on a single city or a small area in Western European countries. In terms of building density, some studies have paid attention to the "center-periphery" distribution characteristics of building density at the plot scale in the central area and the overall city, but few studies have analyzed the evolution mode of the spatial distribution of building density in the city and the resulting differences in the growth characteristics of urban space. In fact, compared with the density of construction land, building density is often more sensitive to the impact of construction projects, and is one of the key indicators that need to be controlled in detailed planning. With the continuous advancement of urban renewal actions in the context of territorial spatial planning, the number of construction projects on existing construction land, such as the renovation of old residential areas and the redevelopment of inefficient land, is increasing. Although these construction projects have a small impact on the change in construction land density, they can bring about significant changes in building density, especially in the urban core area (e.g., the circle 5 km from the city center). In addition, compared with the plot ratio index, which is difficult to obtain, some high-precision data products reflecting the distribution of building density in cities have appeared in recent years, which provides support for the comparative study of multi-city, large-sample and diachronic urban spatial growth, and is conducive to making up for the lack of attention to small and medium-sized cities in existing studies.
Based on the above background, this paper takes 287 cities at and above the prefecture level in China as the research sample, and uses the global built-up area database released by the Global Human Settlement Layer project of the European Union Joint Research Center in 2022 to calculate the building density data at 100m raster scale in 2000, 2010 and 2020, and uses the "circle gradient analysis method" In order to further deepen the theoretical understanding of the relationship between building density and urban space growth, and provide a theoretical basis for formulating relevant policies for the sustainable growth of urban space, this paper analyzes the distribution and evolution characteristics of urban fringe in the past 20 years, and discusses the growth degree of urban fringe and the differences in growth characteristics between cities of different sizes in the past 20 years.
1 Data and Methodology
1.1 Research Samples and Data
In this paper, 287 cities at and above the prefecture level in China were selected as the research sample. According to the 2014 Notice of the State Council on Adjusting the Criteria for Dividing the Scale of Cities, and according to the population size of the municipal districts of the sample cities in 2020 published in the Seventh Population Census by County, the 287 sample cities are divided into 28 megacities (population size greater than or equal to 5 million), 139 large cities (population size of 1 million ~ 5 million, excluding 5 million), and 91 medium-sized cities (population size of 500,000~). 1 million, not including 1 million) and 29 small cities (population size less than 500,000).
In this paper, we use the grid-scale building density index to characterize the spatial growth state of the city, which is calculated based on the global built-up area dataset released by the Global Habitat Layer project of the European Union Joint Research Center in 2022. Based on high-resolution, long-time satellite remote sensing image data such as Landsat and Sentinel-2, as well as building contour data from platforms such as Microsoft and OSM, the dataset uses symbolic machine learning technology to identify building roofs, and then calculates the "ratio of building roof area to grid area" at a grid scale of 100m, that is, the density of grid buildings. The dataset is one of the most commonly used high-resolution global built-up area remote sensing data products, and its accuracy in calculating building density in the United States, European countries and China has been verified by relevant studies [The dataset (2022 version) provides 10m, The 100m resolution data is mainly based on the following considerations: (1) the building density data in this paper needs to be counted within the 1km circle, so it is not appropriate to use the 1km resolution data;(2) the relevant studies show that the dataset (2020 version) can reflect the building density in China at a lower resolution, but it will overestimate the building density in the high-value area at a higher resolution, so the 10m resolution data is not selected to reduce the accuracy error;(3) because the 100m raster building density data is not directly used in this paper to analyze the urban spatial growth, but it is added to the 1km circle, and the error caused by the data accuracy has relatively little impact on the analysis results in this paper. Based on this dataset, the building density of each grid at 100m grid scale in the study area in 2000, 2010 and 2020 was obtained, and the spatial growth characteristics of Chinese cities in the past 20 years were analyzed on this basis.
1.2 Research Methodology
In this paper, we use the "circle gradient analysis method" to add up the roof area of the building at the grid scale to each circle layer that expands outward, and quantitatively analyze the "center-edge" distribution characteristics of urban building density by calculating the proportion of the total roof area of each circle building within the city boundary to the circle area (equivalent to the average building density of all grids in each circle). The theoretical basis of this method is the concentric circle structure of urban element distribution and the "center-periphery" characteristic of spatial growth. In the early studies on urban spatial growth, the correlation studies mainly used the circle gradient analysis method to analyze the distribution characteristics of population density in cities, but the data used were usually of low resolution, resulting in low spatial accuracy of correlation analysis. In recent years, with the continuous emergence of satellite remote sensing data products, some studies have begun to use circle gradient analysis to quantify the "center-periphery" degree of urban spatial growth based on high-resolution data such as construction land and night lights. From the perspective of research objects, the existing studies mainly focus on some typical large cities, megacities or megacities, but some studies have also begun to pay attention to the spatial growth characteristics of small and medium-sized cities and their differences from large cities.
When using the circle gradient method, it is necessary to first determine the circle center and circle radius. Urban growth tends to be centered around the old city, and government offices are generally located in the old part of the city. In recent years, some cities have moved their government locations from the old city to the new city, but the results of the calculation centered on the new city cannot accurately reflect the characteristics of urban growth. Therefore, in this study, the city government station in 2000 was used as the circle center. Referring to the existing research, the radius of the circle is selected as 50km to ensure that the central urban area of the megacity is covered, in fact, the radius of the central urban area of most cities is between 20~30km.
The specific steps of circle gradient analysis are as follows: (1) the government station in 2000 is taken as the center point of the city, and 50 concentric circles are divided outward at 1km intervals, (2) based on the ArcGIS platform, these 50 circles and urban land administrative boundaries are superimposed with the raster-scale building density data (Fig. 1), and ;(3) the building density of each circle layer of each sample city in 2000, 2010 and 2020 is calculated based on the ArcGIS platform, that is" (4) Based on the Python platform, the relationship between the density of the circle building and the distance from the circle to the city center is fitted with a nonlinear least squares curve. The expression is:
Fig.1 Schematic diagram of the superposition of 100m raster-scale building density data and 1km circle layer in the city (taking Beijing as an example)
In Eq. (1), Y represents the building density of a certain circle, X represents the distance from the circle to the city center, E represents the natural constant, A represents the fitting building density of the city center, and B represents the decay rate of the circle building density or the centrality of the building density distribution within the city. Specifically, the larger the b value, the faster the attenuation rate of the circle building density from the center to the periphery, that is, the grids with higher building density are mainly distributed in the city center and its adjacent circles, which also indicates that the building density distribution in the city has a strong centrality.
Due to the upper limit of the building density of a particular plot, the B value of the city with a large construction volume will be higher than that of the city with a small construction volume under the same distribution structure, so the B value of different cities is not comparable. In this paper, this paper uses the rate of change of the b-value of the city in a specific study period to reflect the "center-periphery" characteristics of urban spatial growth, which is recorded as the marginal growth length, and the calculation formula is as follows:
In equation (2), bt and bt+Δt represent the centrality of urban building density distribution at the beginning and end of the study period, respectively. A higher C value indicates a higher marginal growth length of the city, and vice versa indicates a lower marginal growth length of the city.
Fig. 2 takes Langfang City as an example to further illustrate how the marginal growth length reflects the "center-periphery" characteristics of urban spatial growth. The solid black line in the figure is the negative exponential function fitting curve of building density in Langfang City in 2000, and the green and blue dotted lines represent the negative exponential function fitting curves of building density under the "center growth" scenario and the "edge growth" scenario from 2000 to 2010, respectively. In both scenarios, the increase of building area is 20% of that in 2000, but the increase of building area in the "center growth" scenario is evenly distributed in the circle layer of 5~10km from the city center, and the increase of building area in the "edge growth" scenario is distributed in the circle layer of 10~15km in the "edge growth" scenario. It can be seen that the centrality of the distribution of urban building density in both scenarios decreases compared with 2000, but the decline in the "marginal growth" scenario is larger. Correspondingly, the length of urban marginal growth under the "edge growth" scenario (0.679) is greater than that under the "center growth" scenario (0.551), which indicates that the urban edge growth length index can reflect the location difference of the building area increment in the central urban area.
Fig.2 Distribution fitting of urban building density under different spatial growth scenarios (taking Langfang City as an example) It should be emphasized that the marginal growth length based on the measurement of building density distribution characteristics in this paper mainly reflects the edge growth characteristics of the city at the horizontal scale, and cannot represent the marginal growth length of the city in three-dimensional space, which needs to be reflected by the distribution characteristics of the floor area ratio in the city. However, through calculation and comparison, we find that the difference between "floor area ratio" and "building density" is mainly reflected in the plot scale, and at the larger scale of "1km circle", the circle distribution characteristics of "plot ratio" and "building density" have a high consistency [In order to verify this view, this paper calculates the circle distribution of the sample urban plot ratio based on the 2018 building volume data released by the EU Joint Research Center "Human Habitation Layer", and compares it with the circle distribution of building density. The results show that: (1) the circle distribution of floor area ratio is also in line with the negative exponential function distribution characteristics on the whole, (2) for the same city, the volume ratio of each circle has a strong positive correlation with the building density, and (3) the smaller the city scale and the farther away from the city center, the more consistent the circle distribution characteristics of the floor area ratio and building density. Due to space constraints, the results of the analysis are not presented in this article and can be obtained by contacting the corresponding author]. Therefore, the marginal growth length of the city at the horizontal scale calculated based on the circle distribution of building density in this paper can reflect the basic characteristics of urban spatial growth in China to a certain extent. 2 Fitting results of building density distribution in urban interiorAs mentioned above, this paper uses the negative exponential function to fit the relationship between the building density of different circles and the distance from the circle to the city center, and Fig. 3 shows the distribution of goodness-of-fit of the negative exponential function in different years and different scales in the form of box plots. The goodness-of-fit value is between 0~1, and the higher the value, the better the fitting effect on the density distribution of circle buildings. Overall, the mean (large black dot in the figure) and median (horizontal black line in the figure) of the goodness-of-fit of the negative exponential function exceeded 0.9, indicating that the negative exponential function could better fit the spatial distribution of building density in the urban circle. In addition, the negative exponential function fits better for 2000 and 2010 than in 2020, and fits small and medium-sized cities better than for large cities and megacities and megacities. In order to ensure the accuracy of the analysis results, three sample cities with mean goodness-of-fit lower than 0.5 were excluded from the statistical analysis below, including Jiaxing, Zhongshan and Binzhou.
Fig.3 Distribution of goodness-of-fit of negative exponential function in different years (left) and cities of different sizes (right)Fig. 4Taking some sample cities as examples, the distribution of circle building density and its negative exponential function in various regions and cities of different sizes in mainland China are shown [images within 40 km of circle are intercepted for super megacities and large cities, and images within 15 km circle are intercepted for medium cities and small cities]. From the perspective of the shape of the fitting curve, the density of urban buildings in the circle layer and the distance from the circle layer to the center of the city generally obeys the distribution of negative exponential functions, that is, the density of urban buildings decreases rapidly from the center to the outward, but the rate of decline gradually slows down. For super-large, extra-large and some large cities, there is an area with a relatively stable density of circle buildings within 3~5km of the city center. However, for most small and medium-sized cities, the building density decreases rapidly from the center to the edge, and the decline rate gradually decreases, generally reaching a relatively stable state after 5~7km from the center, which is basically consistent with the existing studies on small and medium-sized cities in Africa and India.
Fig.4 Density distribution of circle buildings and the fitting curve of negative exponential function in some sample cities
Figure 5 shows the statistical characteristics of the b-value of the centrality of building density distribution in all sample cities over three different years. It can be seen that the B value of different cities is quite different, and the B value of a few cities is higher, but the B value of most cities is low. Comparing the mean value of the centrality of building density distribution in different years, it can be found (the black dot in Figure 5) that the value decreased from 0.32 in 2000 to 0.27 in 2010 and then to 0.22 in 2020, showing an overall downward trend. In order to further verify whether this downward trend is significant, the distribution of b values in different years was tested by Kruskal-Wallis test. It can be found that the mean change of the centrality of building density distribution from 2000 to 2010 was significant at the 5% significance level, and the mean change of the centrality of building density distribution from 2010 to 2020 was significant at the 0.1% significance level, which further indicates that the centrality of building density distribution in Chinese cities has shown a significant downward trend in the past 20 years.
Fig.5 Statistical characteristics of the b-value of urban building density distribution in different years analyzed by K-Litis test3 Differences in the overall characteristics and scale of urban spatial growth3.1 Overall characteristics of urban spatial growth: The marginal growth trend is obviousBased on the calculation of the centrality of building density distribution based on the negative exponential function, the marginal growth length of each city in the two time periods of 2000-2010 and 2010-2020 is further calculated by using equation (2). Table 1 lists the top 10 cities in terms of marginal growth length in two time periods. Compared with 2000-2010, the maximum growth length of urban fringe increased from 0.46 (Nantong) to 0.85 (Langfang) from 2010 to 2020, an increase of 84.8%. The four cities of Nantong, Dongying, Zhenjiang and Hebi all ranked among the top 10 in terms of marginal growth length in the two time periods, and their respective marginal growth lengths also increased further with time, reflecting the increasingly obvious trend of urban marginal growth to a certain extent. In terms of geographical distribution, most of the top 10 cities in terms of marginal growth length are located in the eastern and central provinces, especially in Jiangsu Province (35%) and Henan Province (25%). It is worth noting that among the top 10 cities in terms of marginal student length from 2010 to 2020, 6 cities are bordered by megacities and megacities, such as Langfang around Beijing, Nantong near Shanghai, and Zhenjiang near Nanjing. Table 1 The top 10 cities in terms of the length of lower marginal growth in the two time periods
Figure 6 further illustrates the geographical distribution of urban marginal growth length over the two time periods. It can be seen that the marginal growth length of most cities is greater than 0, which is consistent with the conclusion that the centrality of building density distribution is declining as a whole, and also indicates that the spatial growth of Chinese cities has been mainly expanding to the periphery in the past 20 years. Comparing the marginal growth length of the two time periods, it can be found that the marginal growth length of most cities in 2010-2020 is greater than that of the marginal growth in the period 2000-2010, indicating that the trend of urban marginal growth in China has become more obvious from 2010 to 2020. From the perspective of geographical distribution, the cities with higher marginal growth length are mainly concentrated in the Beijing-Tianjin-Hebei urban agglomeration, the Central Plains urban agglomeration, the Shandong Peninsula urban agglomeration, the Guanzhong Plain urban agglomeration, the Chengdu-Chongqing urban agglomeration, the urban agglomeration in the middle reaches of the Yangtze River and the Yangtze River Delta urban agglomeration.
Fig. 6 Geographical distribution of urban marginal growth length in two time periods
Source: Based on the map of China GS (2020) 4628, in order to further analyze the relationship between the circle distribution of building density and the growth of urban edges, Figure 7 selects some representative cities to show the spatial distribution of rasters with high growth rate (more than 10 times) of building density from 2010 to 2020. It can be seen that for cities with high marginal growth length, most of the high-growth grids are distributed in the marginal circle layer away from the center, while for cities with low marginal growth length, the high-growth grids are mainly distributed in the circle layer near the city center. Combined with the results of the above multi-scenario analysis of Langfang City, it can be concluded that the urban fringe growth length index constructed in this paper can better reflect the "structural characteristics" of the distance of new buildings from the city center.
Fig. 7 Distribution of high-growth grids of building density in representative cities from 2010 to 2020 3.2 Spatial growth differences in cities of different sizes: the marginal growth trend of large cities and medium-sized cities is the most obviousFig. 8 compares the statistical characteristics of marginal growth length in cities of different sizes in the two time periods of 2000-2010 and 2010-2020. On the whole, the mean value of marginal growth length in cities of different sizes is greater than 0, indicating that the spatial growth of cities of different sizes shows a certain characteristic of "marginal growth", which is consistent with the previous analysis conclusions. Specifically, the mean marginal length of megacities and megacities increased from 0.15 in 2000-2010 to 0.21 in 2010-2020, from 0.15 to 0.25 in large cities, from 0.11 to 0.17 in medium-sized cities, and from 0.10 to 0.12 in small cities. It can be seen that there are certain differences in the growth rate of marginal growth in cities of different sizes in two different time periods. In order to further test whether the change of the mean marginal length of cities of different sizes is significant, the Klinefelter test was performed on the distribution of marginal length in cities of different sizes. The results showed that the changes in the marginal length of megacities and megacities were significant at the 5% significance level, and the changes in the marginal length of large and medium-sized cities were significant at the 0.1% significance level, while the changes in the marginal length of small cities were statistically significant.
Fig.8 Statistical characteristics of urban marginal growth length of different scales in two time periods
Based on the analysis results of the above two aspects, it can be concluded that from the period from 2000 to 2010 to 2010, mega and mega cities, large cities and medium-sized cities as a whole showed a significant trend of further growth to the periphery, and the marginal growth of large cities was the largest, followed by medium-sized cities, and the marginal growth of super-large and mega-cities was lower. In the case of small cities, the trend of further growth to the periphery is not significant, and the growth of the periphery is the lowest. In fact, it can also be seen from Figure 8 that except for a few small cities such as Ulanqab, the marginal growth length of most small cities did not change significantly in the two time periods. 4 Preliminary Discussion on the Influencing Factors of Urban Space Growth Characteristics The marginal growth characteristics of urban space in China and the differences in growth characteristics between cities of different sizes are affected by a variety of factors, and this paper mainly discusses the three aspects of physical geographical conditions, population and economic scale, and macro policy and planning guidance. 4.1 Physical Geographical ConditionsNatural geographical conditions generally include natural geographical factors such as topography, geology, and landform, which have always been the most basic factors affecting the growth and spatial form of urban space. Figure 9 illustrates the natural geographical conditions of the cities with the lowest and highest marginal growth length among cities of different sizes from 2010 to 2020. Nationally, the urban marginal growth length in the North China Plain is generally higher than that in the southeastern coastal hilly areas (Fig. 6). In fact, the marginal growth length of coastal cities such as Dalian, Xiamen, Zhoushan, and Fangchenggang ranked in the bottom 1% of the sample cities. In addition, the distribution of urban marginal length in Zhejiang Province can also reflect the influence of natural geographical conditions to a certain extent, and the marginal length of the southern cities located in the hilly area is generally lower than that of the northern cities located in the plain area (Fig. 6).
Fig.9 Physical geographical conditions of cities with the highest and lowest marginal growth lengths classified by scale (2010-2020) Physical geographical conditions can also affect the spatial growth characteristics of cities of different sizes. Generally speaking, there is a correlation between the size of the city and the surrounding topographical conditions, and small cities are more likely to be located in hilly and mountainous areas than large cities, and the scale of land suitable for building development is often relatively limited. Of the 29 small cities in the sample, 18 have a central urban area surrounded by mountains and hills, and the expansion of their urban areas is also limited to some extent. For example, in Lijiang City, located in the Hengduan Mountains, the marginal growth length decreased from 0.15 in 2000-2010 to 0.04 in 2010-2020. Under the influence of topography, some small cities tend to choose the method of "connotation filling" in the process of spatial growth. For example, in Figure 7, the high-density growth grid of Ulanqab is mainly distributed around the city center. 4.2 Population and economic size
Population growth and economic development are the fundamental driving forces for the expansion of urban land. In the past 20 years of rapid urbanization in mainland China, the housing gap in urban centers caused by population inflow and the demand for low-cost factories due to large-scale manufacturing development have promoted the emergence of residential quarters, industrial plants and other buildings in urban fringe areas. As can be seen from Figure 6, cities with high marginal growth are usually cities with faster population and economic growth, such as many economically strong cities on the southeast coast and some provincial capitals in the central and western regions. On the contrary, cities with low marginal growth length often face more severe problems such as population loss and economic recession, which can be intuitively reflected in the changes of marginal birth length in some cities in Northeast China. For many cities in Northeast China, resource depletion, economic decline of heavy industry, and population migration have inhibited the construction of real estate development and industrial parks in the marginal areas to a certain extent, and the demolition and reconstruction of industrial plant plots with high construction density in the marginal areas may further reduce the overall building density. Therefore, the marginal growth length of these cities in Northeast China is generally low, and the average marginal growth length of Fushun, Yichun, Baishan and other cities is less than 0.05.
In terms of cities of different sizes, the population growth rate of the small cities with the least obvious growth trend was -11% and the GDP growth rate was only 41% between 2010 and 2020, both of which were the lowest among the four types of cities. In fact, many of the 29 small cities in this paper have been identified as shrinking cities by relevant studies, such as Baishan, Shuangyashan, Tonghua, etc., and their marginal growth lengths decreased from 0.07, 0.19, and 0.16 from 2000 to 0.05, 0.14, and 0.06 from 2010 to 2020, respectively. These cities are generally facing problems such as population loss, single type of industry, and backward production capacity. In contrast, megacities and megacities with high marginal growth and a significant trend had a population growth rate of 45% and a GDP growth rate of 164% between 2010 and 2020.
At the same time, the impact of population and economic scale changes on the length of urban fringe students of different sizes is also heterogeneous. Taking the period from 2000 to 2010 as an example, this paper analyzes the correlation between the change of urban economic scale and the length of marginal students. As shown in Figure 10, there is a significant positive correlation between the GDP growth value (logarithm) and the marginal growth length of cities of different sizes during this period. Further comparison shows that the correlation coefficient of small cities (0.68) is significantly larger than that of large cities (0.19) and medium cities (0.26), indicating that the added value of marginal growth per unit of GDP growth in small cities is lower. This further confirms the previous conclusion of this paper, that is, the marginal growth phenomenon of large and medium-sized urban spaces is more prominent than that of small cities.
Fig.10 Relationship between GDP growth and marginal growth in cities of different sizes (2000-2010)4.3 Macroeconomic policy and planning guidance
As an important external intervention factor, macro policy and planning guidance mainly affect the distribution and marginal growth of urban buildings through real estate and land policies. On the one hand, after the housing market-oriented reform in 1994, real estate-related industries have gradually become the pillar industries of many cities, and the construction of new towns and new districts dominated by real estate development has become an important factor in promoting the growth of many urban fringes. On the other hand, strategic planning at the national level is also closely related to the growth of urban space. Some scholars pointed out that in order to promote balanced regional development, the supply of construction land in the country has gradually shifted to the central and western provinces since 2003, which may accelerate the spatial growth of some central and western cities. Relevant studies have also found that the sprawl trend of Henan cities slowed down after 2005, but intensified after 2015, which can be attributed to the favorable policies for Henan's development in the Central Plains Urban Agglomeration Development Plan approved by the state in 2016. This can also be verified in the analysis results of this paper: from 2010 to 2020, the marginal growth length of Henan, Jiangxi, Hubei, Sichuan, Shaanxi and other central and western provinces increased significantly (Fig. 6), and the marginal growth rate was the highest in Xuchang, Hebi, Puyang and other cities in Henan Province.
Of course, there are also differences in the impact of macroeconomic policies and planning guidance on cities of different sizes. Since 2010, the state has tightened restrictions on the supply of land for construction in megacities and megacities, while megacities such as Beijing, Shenzhen and Shanghai also took the lead in introducing strict real estate purchase restrictions in 2010. Under the guidance of a series of policies, the urban construction of mega and mega cities has begun to shift from the land expansion model focusing on the number of buildings to the "connotation filling" and urban renewal mode focusing on the quality of buildings. For example, in 2021, the Ministry of Housing and Urban-Rural Development announced that 10 of the 21 "First Batch of Urban Renewal Pilot Cities" were megacities and megacities. This shift may not only reduce the demand for building land expansion in the marginal areas, but also increase the building density in the urban center area, thereby reducing the marginal growth length of the megacity [according to the calculation results of this paper, the building density of each circle in the central area of almost all sample cities (within 5 km of the city center as a rough standard) showed an upward trend in three years. This may be due to the fact that even if there are urban renewal behaviors that reduce building density, such as shantytown renovation, there are still some "vacant land" or inefficient land in the central area that have not yet been used, and building development on these lands can also significantly increase the density of circle buildings in the central area. In contrast, large and medium-sized cities are somewhere between megacities and small cities, and do not face as strict policy restrictions as megacities and megacities, nor do they face severe population loss and economic recession like some small cities. Thus, while megacities and megacities have the fastest population and economic growth, they tend to grow on the periphery more than in large and medium-sized cities.
In addition, under the guidance of relevant policies and plans to promote the development of urban agglomerations and metropolitan areas, some large and medium-sized cities close to megacities have shown obvious marginal growth trends, such as Langfang near Beijing, Nantong near Shanghai, Zhenjiang near Nanjing, and Ziyang near Chengdu from 2010 to 2020. To a certain extent, this reflects that under the influence of urban agglomeration and metropolitan area policies, the spatial sprawl of megacities is shifting to the surrounding medium-sized cities and large cities, and the formulation of relevant policies and plans in the future needs to pay more attention to the spatial growth of medium-sized cities and large cities, and coordinate the relationship between spatial growth and population and economic development at the regional scale. 5 Conclusion
Under the background of the current slowdown of urbanization, fiscal tightening in some cities, and insufficient economic growth momentum, it is of great significance to scientifically and reasonably understand the spatial growth characteristics of Chinese cities, especially the differences in the spatial growth characteristics of cities of different sizes, for formulating differentiated strategies to promote the sustainable growth of cities. The main conclusions of this study show that most cities in China showed further marginal growth in the two periods of 2000-2010 and 2010-2020, among which the growth of marginal growth in large and medium-sized cities was the most obvious, followed by megacities and megacities, and the marginal growth trend of small cities was relatively insignificant. In order to curb the sprawl of urban space and promote the compact and intensive development of cities, it is necessary to formulate more precise regulation policies and appropriately increase the policy attention to large and medium-sized cities.
Of course, this study also has certain limitations, which need to be further improved in follow-up studies. Firstly, due to the lack of large-sample and multi-time urban building volume data, this paper only discusses the marginal growth characteristics of urban horizontal scale, and then further explores the difference between urban "three-dimensional growth" and "horizontal growth" in combination with typical urban samples. Secondly, only the "center-periphery" dimension is considered in the measurement of urban spatial growth, and other dimensions such as "agglomeration-dispersion" and "single-center-multi-center" can be considered in the future. Thirdly, due to the selection of cities above the prefecture level as the research object, the number of small cities included in the sample is small, and the follow-up study can consider further including county-level cities. Finally, the analysis of the influencing factors of urban spatial growth characteristics is mainly qualitative, and the subsequent quantitative model can be considered to construct a quantitative model to analyze the effect of different factors. UPI
Author: Li Yingcheng, Professor, School of Architecture, Southeast University. [email protected]
Zhong Xiaohan is a master's student in the School of Architecture, Southeast University
Qi Junheng, School of Architecture, Southeast University, Ph.D. candidate
Hu Xingxing (corresponding author), School of Architecture, Southeast University, Professor, Ph.D. supervisor. [email protected]
Edit | Gu Chunxue
Typography | Xu Dudu
This article was originally written by this subscription account