1The large scale of Chinese urbanization captivates the observers, largely because of its remarkable speed, the three dimensional aspect of its landscape and its accompanying societal and environmental upheavals. The Chinese multi-millionaire cities such as Shanghai, Beijing or Guangzhou are emblematic signals of this giant urbanization. But beyond these well observed and documented large cities, China offers a much larger variety of urban forms and processes. Their analysis is strongly dependent on the way the urban population is counted.
2In China, the measurement of the cities’ population is a good example of the links between the counting of urban population and the way that cities are defined. Indeed, Chinese cities are defined in an administrative way (Ma, 2005). The administrative boundaries do not delineate a “strictly urban” unit (Gipouloux, 2006) but include a territory composed both of urban and rural territories which are under the jurisdiction of a main city (Figure 1).
Figure 1: Chinese cities’ definitions are embedded one in another
- 1 The lower administrative ranks of urban units, called “Zhen”, are considered by the census as urban (...)
3This nesting of urban territories exists because there are different kinds of urban units and various “Shi” at different administrative levels from the lowest to the highest administrative level: the Zhen (镇), commonly translated as “Town”, the Xianjishi (县级市) or district level cities, the Dijishi (地级市) or prefecture level cities, the Zhijiashi (直辖市) or province level cities. The cities of the lower administrative levels are embedded in cities of higher levels1 (Figure 2).
Figure 2: Chinese cities are defined at all administrative levels
4Because official Chinese Cities do not encompass a strictly urban area, are defined at different administrative levels and are embedded one in another, “The lack of well-defined and standardized terms for urban settlements in China has created much confusion among Chinese as well as Western scholars regarding the size of China’s Urban Population and the nation’s urbanization level” (Ma & Cui, 1987). As a consequence, the use of the census data in statistical analysis leads to two types of misunderstandings:
-
If the total population of all the Shi is considered, meaning at district, prefecture and province level, the population of district level cities is counted twice. It is once isolated, once counted with the total urban population of the Prefecture city to which the district level city belongs.
-
If the urban population of the city center (or “city proper” in the literature) is considered, meaning the population of all the urban districts (Shiqu) for province and prefecture level cities, and the population of the county for the district level cities, the population of the largest cities can be underestimated, insofar as the cities dense urban built up area often extends beyond the boundaries delimited by the urban districts. At the opposite, because the administrative limits of the district level city can be larger than the actual boundaries of the urban built up area the population of the small district level cities could be over-estimated.
5To describe the Chinese urbanization process as precisely as possible and to enable comparison in space and time, it is necessary to harmonize city data series, whatever the recognition and delineation of localities as urban have been officially defined during the evolution in time. With this aim, we have constructed a database called ChinaCities (Swerts, 2013) based on a definition of cities that are morphological agglomerations larger than 10,000 inhabitants (Moriconi-Ebrard, 1993). The principle is to delineate the cities’ boundaries according to the spatial extension of the continuously built-up area regardless of the official definition. The method is to delineate the perimeter of the cities from satellite images and to aggregate information relative to the population contained in elementary administrative units within this perimeter.
Figure 3: Map of the ChinaCities urban agglomerations in 2010
Data : ChinaCities, Swerts, 2015 - Map: Cura, 2017
6The delineation of elementary units (county and township) that were aggregated within the perimeters of urban agglomerations, were purchased from the Data China Center of the University of Michigan (http://chinadatacenter.org/).
7A first original data source describes demographic and socio-economic population figures at district level from the 1953, 1964, 1982, 1990, 2000 and 2010 Chinese Censuses. They are displayed in a GIS map in Shape file format (Figure 3, city sizes in 2010). The data for years 1953 and 1964 were too fragmentary and the counties in those years had surfaces too large for being used as elementary units. Thus they are not integrated into the final database.
8Two additional complementary databases were purchased at the Township level. The first describes the population of Towns (Zhen) for the year 2000 (Figure 4) and the second the population of Towns (Zhen) and townships (Xiang) in year 2010 (Figure 5). These two bases are also displayed in GIS maps in Shape file format.
Figure 4: Delineation of the district level units in China (2010)
Figure 5: Urban and rural townships in China (2010)
9Following this principle of construction, the ChinaCities database was built by Elfie Swerts (2013) according to four steps:
-
In a first step, the continuous urban built up area separated by less than 200 meters was delineated using Google Earth images from the year 2000 with a resolution of 7,000 feet (~ 2,134 m). This step was carried out within the framework of a Master project (Swerts, 2009) and the project e-Geopolis (http://www.agence-nationale-recherche.fr/?Projet=ANR-07-CORP-0019).
-
This perimeter was georeferenced and integrated in a Geographic Information System (GIS) and adjusted to entire district level units where the urban zone was dominant, whatever their urban/rural type (Qu/urban districts, Xian/rural counties and Xianjishi/district level cities).
In a fourth step, when city the population of the small district level cities could be over-estimated were too large to be entirely contained in this agglomeration, the smaller town’s level units (Zhen/towns and Xiang/villages) were used to more precisely delineate the agglomeration’s perimeter. The data on Zhen were available for 2000 and 2010 but not for previous periods. As a consequence, the urban areas are defined with more precision in 2000 and 2010. For years 1982 and 1990, the delineation of the agglomerations remains constrained by the limits of the counties. Therefore, the population of urban agglomerations located in a county whose boundaries are much more extended that those of its built up area were excluded from the data base. For avoiding any ambiguity, we name this part of the ChinaCities Database “ChinaCities Database adapted to counties” (Tab.1).
-
Because some rural areas can be very dense, both in terms of population and urban buildings, the number of towns could be overestimated, some “morphological agglomerations” possess a workforce overwhelmingly engaged in primary sector activities (agriculture, forestry and fishing). To bring a correction, an economic criterion has been added that refines the urban character of these observed morphological settlements. It allows the exclusion of all morphological settlements in agricultural counties from the database. This was made possible by collecting the demographic and economic data from official censuses of 1982, 1990, 2000 and 2010 at district level and associating these data to the zones defined from the continuous urban built area perimeters.
10The ChinaCities database contains 9,525 cities with their population in 2010 and 2000 (Table 1). This is large number compared to the 657 “cities” that are officially defined by the Chinese government. Our definition enables better international comparisons because it includes as well many smaller towns that do not have the status of “city” in China.
- 2 The system of control of the internal migrations means the Hukou system, which is a household regis (...)
11For 897 units of the ChinaCities database, mainly cities larger than 100,000 inhabitants, population at earlier dates is also provided. The population of these cities in 1982 and 1990 was calculated by aggregating the urban or total population of entire districts, because it was impossible to know the detail of elementary units below the district level. Although most of the small towns and a few cities of Western China are not included in this ChinaCities Database adapted to counties, the time depth of this database provides a good coverage of the period of Deng Xiaoping’s “4 modernizations” of 1978 that initiated the strong urban and economic growth of China, the 1990s opening of the country and the 2000s relative liberalization of the system of control on the internal migrations2 (Cheng & Selden, 1994; Chan & Zhang, 1999; Liu, 2005).
12The official statistics which are based on an official urban status tend to underestimate the total number of urban centers: 657 cities only are acknowledged, while they overestimate the urban population counts by including non-urban areas (and possible double counting) in the largest administrative delineations (Table 1).
Table 1: Number of cities and urban population by size class in 2010: comparison of the ChinaCities databases with the Official Chinese Statistics
Source : ChinaCities database, Swerts, 2013 and China statistical yearbooks
13The ChinaCities database allows the drawing of safer figures on the size of urban agglomerations and avoids big mistakes such as considering Chongqing as the largest “city” in the world (Table 2). Indeed, the 32 to 34 millions of inhabitants credited to Chongqing are those of the “municipality” that may indeed denote a “province” (that cover 82,300 km2, approximately the size of Austria) whereas the population of the agglomeration, oscillate between 7 or 10 millions according to different sources (Swerts, 2016a).
Table 2: The 10 most populated Chinese cities in 2010 according to ChinaCities and Census
Nom
|
Chinacities
|
Census
(Adminitrative area)
|
Shanghai
|
23,670,975
|
23,019,148
|
Beijing
|
21,869,769
|
19,612,368
|
Guangzhou
|
19,167,406
|
12,700,800
|
Tianjin
|
13,718,810
|
12,937,954
|
Shenzhen
|
10,358,381
|
10,357,938
|
Wuhan
|
8,376,498
|
9,785,392
|
Chongqing
|
7,337,767
|
28,846,170
|
Chengdu
|
6,432,129
|
14,047,625
|
Nanjing
|
5,748,334
|
8,004,680
|
Shenyang
|
5,685,899
|
8,106,171
|
Source : ChinaCities database, Swerts, 2013
14The ChinaCities harmonized database gives a conclusion on several points where the scientific literature was still very hesitant, such as the adequacy of the Chinese city system to the Zipf’s law (see for instance Anderson and Ge 2005, Zhang et al., 2005, Gangopadhyay and Basu, 2009, Schaffar, 2009). We have tested the adequacy of Zipf's distribution of hierarchical distributions on Chinese cities using ChinaCities including all cities larger than 10,000 inhabitants (Swerts 2013, Pumain et al. 2015, Cura et al. 2017), and varying the minimum size threshold of cities between 10,000 (Figure 6a) and 100 000 (Figure 6b) inhabitants (Malecki, 1980 and Guerin-Pace, 1995).
Figure 6: Rank-size curve for the Chinese Urban system
a. Rank-size curve over a size threshold of 10,000 inhabitants (2000 and 2010)
b. Rank-size curve with a threshold of 100,000 inhabitants (1982, 1990, 2000 and 2010)
Source: ChinaCities database, Swerts, 2013
15The analyses of the official data show a slow growth of large cities between 1990 and 2000, and relatively high population of medium-sized cities in comparison with both small towns and large cities as illustrated in Figure 6b. Conversely, the ChinaCities harmonized database demonstrates both the stability of the slope of the rank-size curve between 1982 and 2000 and a vigorous growth of large cities, confirmed by an increase in the slope of the curve between 2000 and 2010 when the full set of cities is considered (Figure 6a). When a minimum threshold of 100 000 inhabitants is used, the curves start to diverge above a threshold of 500 000 inhabitants, which means that the sizes of large cities is much more differentiated in 2000 than in 1980 (Swerts, 2013). The apparent stability of the total rank-size curve would thus be the result of an increase in the number of small towns and their sustained growth, rather than the slow growth of large cities. These differences in interpretation of Chinese urbanization trends according to the sources are indeed illustrated in the very distinct shapes of the corresponding rank-size curves: concave with the official data and much more linear according to the data of ChinaCities (Figure 7).
Figure 7: Comparison of the rank-size distribution of Chinese cities in 2010, according to official data (total population and urban population) and ChinaCities
16The ChinaCities database also provides socio-economic indicators for the 9,525 urban agglomerations in 2010 and for the 897 ChinaCities’ adapted to districts’ urban agglomerations. These data have already been processed to draw the first images of functional specialization among Chinese urban agglomerations in 2000 (Swerts, 2015) and 2010 (Swerts 2016b; Swerts and Liao, 2017). These data will be presented in a next data paper.
17Spatial coverage
Actual territory of China
18Temporal coverage
1982, 1990, 2000, 2010
19Format name and version:
One csv file
20The ChinaCities Database is organized as follow:
-
The ChinaCities Database is a matrix of eight columns that indicate the name of the cities in English and Chinese, and the population of these cities in 1982, 1990, 2000 and 2010.
Database columns
|
Identifier
|
Name (English)
|
Name (Chinese)
|
Province code
|
Latitude
|
Longitude
|
Population in 1982, 1990, 2000 and 2010 (4 columns)
|
Description
|
9 digit code
|
Name of the ChinaCities urban agglomeration in English
|
Name of the ChinaCities urban agglomeration in Chinese
|
Code of the province where ChinaCities urban agglomeration are located in 2010
|
Latitude of the ChinaCities urban agglomeration in decimal degrees
|
Longitude of the ChinaCities urban agglomeration in decimal degrees
|
Population of the ChinaCities urban agglomeration at each date
|
Example
|
CCSha0001
|
SHANGHAI
|
|
31
|
31.230393
|
121.473704
|
23,267,975
|
21Creation dates:
The data set was created between 2011 and 2016
22Dataset creator:
Elfie Swerts, PhD, University Paris I (associate researcher at Lausanne University)
23Data cost:
Worked time: 36 months shared by three persons
Data acquisition: 25,000 Euros
24Language:
English and Chinese
25Repository location:
On site: http://chinacitiesv1.parisgeo.cnrs.fr/
26License:
Data is made available under the creative commons license of type CC BY-NC-ND 3.0:
27Reuse potential:
To our knowledge this is the only comprehensive data base delineating urban agglomerations of all sizes above 10 000 inhabitants in a systematic way and enabling their comparison in China at different dates and with other countries.
28Acknowledgements:
The dataset was created as part of the doctoral research of Elfie Swerts with a support from Veolia (Environnement Recherche et Innovation) and the ERC grant GeoDiverCity (PI Denise Pumain), and with the help of Alvaro Alvarez Osorio and Lis Obut.