See discussions, stats, and author profiles for this publication at: [Link]

net/publication/226242949

Field Modeling Method for Identifying Urban Sphere of Influence: A Case

Study on Central China

Article in Chinese Geographical Science · August 2010

DOI: 10.1007/s11769-010-0408-2

CITATIONS READS

15 4,103

5 authors, including:

Liu Shenghe Li Wang

Chinese Academy of Sciences Dalian Minzhu University
36 PUBLICATIONS 1,419 CITATIONS 1,018 PUBLICATIONS 32,729 CITATIONS

SEE PROFILE SEE PROFILE

Ma Hanqing Jianghao Wang

National Science Library, Chinese Academy of Sciences Chinese Academy of Sciences
17 PUBLICATIONS 331 CITATIONS 97 PUBLICATIONS 2,906 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Liu Shenghe on 05 June 2014.

The user has requested enhancement of the downloaded file.

Chin. Geogra. Sci. 2010 20(4) 353–362
DOI: 10.1007/s11769-010-0408-2

Field Modeling Method for Identifying Urban Sphere of Influence:

A Case Study on Central China

DENG Yu1, 2, LIU Shenghe1, WANG Li2, 3, 4, MA Hanqing1, 2, WANG Jianghao1, 2

(1. Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101,
China; 2. Graduate University of the Chinese Academy of Sciences, Beijing 100049, China; 3. Institute of Policy and
Management, Chinese Academy of Sciences, Beijing 100190, China; 4. Center for Interdisciplinary Studies
of Natural and Social Sciences, Chinese Academy of Sciences, Beijing 100190, China)

Abstract: With rapid development of urbanization and regional interaction and interdependence, regional urban ag-
glomeration planning becomes more and more important in China, in order to promote integrated development of
various cities with close interrelationship. However, it is still arguable academically on how to define the boundary or
which cities to be included for the urban agglomeration of a region. This paper aims to shed lights on how to identify
urban spheres of influence scientifically by introducing field modeling method and by practicing a case study on 168
cities in Central China. In our field modeling method, the influence intensities of cities were measured by a compre-
hensive index and urban spheres of influence were represented spatially by field intensity. Then, their classification
and spatial distribution characteristics of study area in 2007 were identified and explored by using GIS and statistical
methods. The result showed that: 1) Wuhan is the absolute dominant city in Central China; 2) the provincial capital
cities dominate their own provinces and there are no other lower grade agglomeration centers; and 3) the basic types of
organization form of urban sphere of influence are single-polar type, agglomeration type, close-related group type and
loose-related group type.
Keywords: Central China; urban sphere of influence; urban influence index; field intensity; urban influence intensity

1 Introduction ment depends on the attraction and radiation force of the

city (Wang and Zhao, 2000; Kong, 2007; Pan et al.,
At present, with the continuous advancement of urbani- 2008). The concept of urban sphere of influence came
zation and regional integration, metropolitan areas and from the central place theory initiated by Christaller in
urban agglomerations have emerged at different regions. 1933 (Christaller, 1998). City is an important node in
Thus, various types of urban and regional planning be- regional development, in which both node-to-node link
come more and more important and are formulated in and node-to-hinterland link occur (Mulligan, 1984).
China. Among of those, it is critical to identify the urban Urban sphere of influence is the latter situation (in this
spheres of influence scientifically and to promote the case, we call it node center).
integrated development of cities and their hinterlands. In According to indicators, the type of urban sphere of
particular, the identification of urban sphere of influence influence and its different measurements, research meth-
is of great significance to the improvement of central ods can be divided into two types: empirical method and
city′s synthesized competitiveness, the building of a modeling method. Empirical method determines urban
reasonable urban system and the establishment of sup- sphere of influence according to data features, regional
porting policies. The urban sphere of influence refers to characteristics and the comprehensive analysis of exper-
surrounding areas whose social and economic develop- tise on the basis of the selecting and accessing of urban

Received date: 2009-08-24; accepted date: 2010-03-29

Foundation item: Under the auspices of National Natural Science Foundation of China (No. 40971102), Knowledge Innovation Pro-
grams of Chinese Academy of Sciences (No. KZCX2-YW-322), Special Grant for Postgraduates′ Scientific Innovation and Social
Practice in 2008
Corresponding author: LIU Shenghe. E-mail: liush@[Link]
© Science Press, Northeast Institute of Geography and Agroecology, CAS and Springer-Verlag Berlin Heidelberg 2010
354 DENG Yu, LIU Shenghe, WANG Li et al.

and regional flow index (commuting data). This in- and urban agglomerations, grasp the interaction between
cludes air passenger traffic (Taaffe, 1962), newspaper cities, and provide suggestions for regional urban sys-
sales (Berry and Lamb, 1974), telephone calls (Nystuen tem planning, regional economic division and coopera-
and Dacey, 1961), and several composite indicators tion, as well as reasonable institutional arrangements.
(Green, 1955). However, with the improvement of the
convenience of urban transportation and regional 2 Method
reachability, relationships among cities, and of city and
region are becoming extraordinary complex. Kinds of 2.1 Field model
data are not easy to get and are lack of comprehensive- As a regional core, a big city has a strong attraction (ag-
ness, which makes the modeling method become the glomeration) and the radiation force (diffusion) for ad-
most important research tools (Du, 2001; Gu and Pang, jacent regions. Interactions between them occur with
2008). Modeling method describes the interaction be- various kinds of ″fluid″ as carrier, and the furthest in-
tween spaces using theoretical model, grasps the inten- fluenced edge forms the urban sphere of influence. In
sity and pattern of contact between cities, and deter- physics, urban regional system is referred to as city field
mines the urban sphere of influence. For an example, (Okabe et al., 2000; Wang and Chen, 2004). The forma-
gravity model has been maturely applied for identifying tion of field can not be separated from the potential en-
the definition of urban sphere of influence in the United ergy difference between the center and its edge, only
States, Ireland and Ghana (Huff, 1973; Huff and Lutz, when the influence of the center is strong enough, can
1979; 1989; 1995). In recent years, Voronoi diagram is the fluid flow in the field. On the other hand, city field
widely used in the research of urban sphere of influence, decays gradually with the increase of distance from the
which is considered to be a good tool to describe the in- center, until the field action disappears (Gu, 1997).
teraction pattern between cities (Gold, 1992; Edwards, Based on the above analysis, the expression of city field
1993; Okabe et al., 2000). However, it also has draw- was as follows:
backs such as lack of a comprehensive metric only using
Fki = f(Zk, Dki) (1)
population as a measure of urban influence index; the
f′(Zk) > 0
definition of urban sphere based on city classification,
f′(Dki) < 0
while its process more subjective; complicated calcula-
tion process and the result expression being not intuitive. where Fki is the field intensity of city k on point i, Zk is
Based on those mentioned above, this paper intends the urban influence index of city k, Dki is the Euclidean
to introduce the field modeling method to identify the distance from city k to point i, f is the mapping function,
urban sphere of influence and is expected to overcome f ′is the partial derivative.
the abovementioned shortcomings. In that method, field
2.2 Setting and main steps of field model on urban
is used to represent and measure the change of urban
sphere of influence
radiation force from the city centre to the edge, and then
There are several distance attenuation models for field
urban sphere of influence can be identified through
model, such as curve attenuation model, linear attenua-
measuring urban influence index and the field intensity.
tion model, exponential attenuation model and so on.
The result shows spatial characteristics of continuity on
The distance attenuation intensity can be adjusted
the local and partly-covered on the whole, and can rep-
through the set of friction coefficient. In this paper, we
resent the spatial characteristics of interaction between
used the curve attenuation model, the friction coefficient
cities and the hinterland more reasonably. Further, the
was 2 (Wang and Zhao, 2000; Pan et al., 2008), and its
characteristics of their classification and spatial distribu-
formula was as follows:
tion of the urban sphere of influence are identified and
explored by using GIS and statistical methods, in order Fki  Z k / D aki (2)
to provide references for regional urban spatial organi- where Fki is the field intensity of city k on point i, Zk is
zation planning and urban economic divisions. Here the urban influence index of city k, Dki is the Euclidean
Central China was chosen as a case study to get a distance from city k to point i, a is the friction coeffi-
clearer understanding on the development trend of cities cient.
 Field Modeling Method for Identifying Urban Sphere of Influence: A Case Study on Central China 355

The main steps of field model on urban sphere of in- represented according to Equation (2), and its value de-
fluence are the following: 1) measure the urban influence clines from center to edge until reaching zero. In order
index; 2) get urban influence sphere based on Equation to measure and display uniformly, we proposed that the
(2); 3) classify the urban sphere of influence, and explore line connecting spatial points, whose field intensity val-
the composition of each level; 4) study the spatial distri- ues equal to 0.01, constitutes the urban sphere of influ-
bution of urban sphere of influence, and get the concrete ence of the center. In the case of a single city, the border
condition about different provinces; and 5) classify the of the periphery constitutes the urban sphere of influ-
spatial distribution of urban sphere of influence. ence (Fig. 1a).
Any point in the region can accept the radiation from
2.3 Urban influence index all the cities. Field intensity may overlap in space, be-
Urban influence index should reflect the inter-relation- cause of the spatial location of the city and urban influ-
ship between the center and its hinterland all-sidedly, in ence index. On the one hand, we can judge the mem-
view of the comprehensive characteristics of urban influ- bership situation of each point based on the principle of
ence index and the availability of data collection, our at- maximum membership; On the other hand, the boundary
tention is paid to comprehensive economy, market, em- of two different urban spheres of influence (the line
ployment, investment, population, basic education, health connecting two points whose field intensities are the
care and the municipal infrastructure construction. We same) can be combined with the peripheral border of a
initially defined 33 indexes. In order to avoid the impact single urban sphere of influence to constitute a specific
of data units, we standardized all the data. Sifted indexes urban sphere of influence (Fig. 1b). Due to different
of each year were correlatively analyzed to ensure the overlap situations, the interaction between cities and its
briefness of the index system and the system work best. development trend is different too.
In this paper, we calculated urban influence index using
principal component analysis, that is, extracted and ro- 2.5 Spatial distribution of urban sphere of influence
tated factors whose eigenvalues were greater than 1 to We used the semivariograms and crosscovariance cloud
minimize the number of variables with a high load. Th- to further analyze the spatial distribution of urban sphere
erefore, calculation formula of urban influence index is: of influence, and its formula was as follows:
m 33
Z k   [An   Cnj  M kj ] (3) 1 (4)
 (h)  E[ Z ( x)  Z ( x  h)]2
n 1 j 1 2
where Zk is the urban influence index of city k, An is the where γ(h) is the value of semi-variant function; Z(x)
contribution rate of the nth principal component, m is and Z(x + h) are arbitrary urban sphere of influence. The
the number of principal component, Cnj is the load of the smaller value of semi-variant function shows the smaller
nth principal component on the jth variable, Mkj is the difference between Z(x) and Z(x + h), vice versa. In the
standardized city index. cloud chart of covariance function, each point represents
the value of semi-variant function of a pair of cities. If a
2.4 Setting of urban sphere of influence spatial object is spatial self-correlation, the value of
The spatial distribution patterns of field intensity can be semi-variant function will increase with the increase of

Fig. 1 Spatial patterns of urban sphere of influence

356 DENG Yu, LIU Shenghe, WANG Li et al.

distance between sample points, until paralleling with

the horizontal axis.

2.6 Classification of spatial distribution of urban

sphere of influence
The spatial distribution feature of urban sphere of influ-
ence is unique. We could determine the basic type of
organization form of urban sphere of influence. Here,
we measured the interaction strength of urban sphere of
influence by introducing ″residual″ indicators. The con-
crete steps were as follows: 1) process the urban sphere
of influence and urban influence index by regression
analysis; 2) The residual value could be obtained by the
formula: residuals (C) = actual value (W) – fitting value 1. Shanxi Province; 2. Henan Province; 3. Hubei Province;
(w). The size of the residual value is similar to the over- 4. Anhui Province; 5. Hunan Province; 6. Jiangxi Province
lap ratio of cloud, and is superior to the simple calcula- Fig. 2 Location of study area
tion of the overlapping area. Residuals measure the in-
teraction between cities taking the overall situation into of 1000 m × 1000 m in ArcGIS (1 000 m equal to about
account, therefore it is more uniform and lateral compa- 32′′), after transformation, the size of the map of Central
rable. China was 1823 × 1052. According to different indica-
tors, attribute data came from China City Statistical
3 Study Area and Data Yearbook 2008 (National Bureau of Statistics of China,
2008), China Population and Employment Statistical
The six provinces of Central China including Shanxi, Yearbook 2008 (National Bureau of Statistics of China,
Henan, Hubei, Hunan, Anhui and Jiangxi have been ex- 2008), China County Statistical Yearbook 2008 (Na-
tremely important parts in China′s regional development tional Bureau of Statistics of China, 2008), China Urban
(Fig. 2). With the implementation of leading develop- Construction Statistical Yearbook 2007 (Ministry of
ment of the eastern regions, the development of the Housing and Urban-rural Development, 2008), or data
western regions and revitalizing the northeast old indus- calculated from the yearbooks above indirectly.
trial bases, the central region stayed in the ″policy edge″ Data were rejected in two situations: 1) from the per-
position for a long time, and fell into the embarrassing spective of a single measure, some indicators were
situation of ″central collapse″. In order to promote the highly correlated with each other. In this case, we re-
development of the central region, the state raised jected those indicators to ensure the briefness of these
clearly the strategy of ″Rise of Central China″ in 2004 data; 2) indicators, which had low correlations with oth-
(Wen, 2004; Su and Wei, 2006). ers, may be due to the specific research background of
There were 168 cities in Central China at the end of this paper. They departed from regular practice, and
2007, of which the urban population and GDP ac- could not reflect the index of urban influence. After
counted for 41% and 60% of the whole region respec- analysis, those indicators were also rejected.
tively. This signified that the aggregation function of
cities in Central China was strong, and those cities 4 Results
leaded the socio-economic development of the Central
China. Overall, there are lots of small cities compared to 4.1 Urban influence index
metropolis. The spatial data in this paper are based on The Kaiser-Meyer-Olkin value of data of 2007 was 0.872.
the electronic version of the county-level boundary map Therefore, the sample size of original data was quite suf-
of China at the scale of 1︰100 000 000 provided by Na- ficient. Bartlett Sphericity test showed that in approxi-
tional Geomatics Center of China (Albers secant conic mate chi-square value of 2007 was 8 438.662, degree of
projection), and then used grid module to generate grids freedom (F) was 300, and the significance of the test (P)
 Field Modeling Method for Identifying Urban Sphere of Influence: A Case Study on Central China 357

was much less than 0.01. Here, we explored the existing This indicates the absolute leading position of Wuhan in
hypothesis that the correlation matrix was a unit matrix. central China. There are five second-level cities, which
Therefore, factor analysis had a good applicability. On the are the capitals of the remaining five provinces. Their
other hand, the regenerations of 2007 were more than urban spheres of influence were very large and only sec-
50%, and most of them were more than 80%. Therefore, ond to Wuhan City. There are 11 third-level cities, and
information extraction of this analysis was effective. most of them are the regional center of each province. On
Eventually, we selected five kind of factors. Overall, the whole, although the urban sphere of influence of
the results of factor analysis reflect the practical situa- third-level cities was significantly smaller than first-and
tion reasonably. First of all, the first kind of factor which second-level cities′, they still showed a considerable area
makes the greatest contribution to population variance (Fig. 3). The urban sphere of influence of fourth-level
reflects the whole economic strength of the city clearly. cities was slightly smaller than that of the third-level cit-
The second one reflects the aggregation degree of urban ies and they were generally better developed cities within
economy, embedding the economic development mani- the provinces they belong to. The total number of
fested on unit land area. The third one means the public fourth-level cities in Central China was 47, accounting for
services. The fourth one mainly focuses on per capita 28% of the total. A small number of large cities were
indicators, that is, urban per capita economic situation. classified into this class as a result of their urban spheres
The fifth one clearly reflects the municipal infrastruc- of influence being squeezed and so on. But overall, the
ture situation. fourth-level cities all had good economic strength. The
fifth-level cities in Central China were general and the
4.2 Classification of urban sphere of influence number of such cities (a total of 71) was the largest which
The classification of urban sphere of influence is shown accounted for 42.3% of the total cities in Contral China.
in Table 1. In the grade classification of urban sphere of The development of sixth-level cities in Central China
influence in central China, Wuhan—the provincial capi- was on a weak level and their urban spheres of influence
tal of Hubei Province is the only one first-level city. The were the smallest. Such cities had a total of 33 in the
area of its urban sphere of influence reached 5 914 km2. number, accounting for 19.6% of the total.

Table1 Classification of urban sphere of influence

2
Grade Number Criteria (km ) City
First-level 1 S > 3000 Wuhan

Second-level 5 1700 < S < 3000 Changsha, Taiyuan, Hefei, Zhengzhou, Nanchang

Third-level 11 1000 < S < 1700 Luoyang, Datong, Wuhu, Xiangfan, Yichang, Yueyang, Huainan, Hengyang, Pingdingshan, Zhuzhou,
Nanyang

Fourth-level 47 660 < S < 1000 Changde, Xinyu, Huaibei, Shiyan, Maanshan, Bengbu, Jiujiang, Anyang, Puyang, Tongling, Yangquan,
Huangshi, Jiaozuo, Luohe, Xinxiang, Xiangtan, Kaifeng, Anqing, Xinmi, Yuzhou, Changzhi, Jingzhou,
Jincheng, Pingxiang, Xuchang, Shuozhou, Chenzhou, Suzhou, Liuyang, Shangqiu, Jingdezhen, Jiyuan,
Qianjiang, Fuyang, Xinyang, Jingmen, Linzhou, Linfen, Loudi, Xinzheng, Gongyi, Yongcheng, Ganzhou,
Dengfeng, Yiyang, Yongzhou, Xiantao

Fifth-level 71 480 < S < 660 Huaihua, Lu′an, Sanmenxia, Huangshan, Leiyang, Yima, Suizhou, Hebi, Fengcheng, Zhoukou, Hejin,
Hanchuan, Lingbao, Shaoyang, Fuzhou, Zhumadian, Chuzhou, Changning, Liling, Xiangxiang, Qinyang,
Ezhou, Tianmen, Lengshuijiang, Chaohu, Xingyang, Xiaogan, Tongcheng, Bozhou, Ruzhou, Yichun,
Guixi, Xiaoyi, Xinzhou, Zhongxiang, Dengzhou, Chibi, Zhangjiajie, Xiangcheng, Chizhou, Ningguo,
Jishou, Enshi, Jian, Tianchang, Xuancheng, Wuxue, Yidu, Huozhou, Leping, Macheng, Yanshi, Zhang-
shu, Xianning, Shangrao, Houma, Yuncheng, Gaoping, Miluo, Wugang, Yuanjiang, Yingcheng, Lianyuan,
Changge, Zixing, Daye, Zaoyang, Guangshui, Honghu, Yuanping, Danjiangkou

Sixth-level 33 S < 480 Nankang, Jiexiu, Shishou, Dexing, Gao′an, Hongjiang, Dangyang, Jinzhong, Laohekou, Mengzhou,
Songzi, Gujiao, Huixian, Zhijiang, Yingtan, Linxiang, Wugang, Jieshou, Yongji, Mingguang, Luliang,
Anlu, Ruijin, Weihui, Ruichang, Fenyang, Yicheng, Jinshi, Lucheng, Lichuan, Shaoshan, Jinggangshan,
Huanggang
Note: S is the urban sphere of influence
358 DENG Yu, LIU Shenghe, WANG Li et al.

moderate economic strength level of this province. The

number of fourth-level cities was 17 in Henan Province,
accounting for 44.74% of all the cities in that province.
Accordingly, its fifth-level cities accounted for 36.84%,
sixth-level cities only accounted for 7.89%. It could be
seen that the urban spheres of influence in Henan Prov-
ince had an obvious advantage in Central China. This
also reflected the economic development level of that
province was relatively high. The numbers of cities in
Shanxi, Anhui and Jiangxi were not large. The numbers
of fourth-, fifth- and sixth-level cities in Shanxi and Ji-
angxi were almost the same; fourth-level cities ac-
counted for 22.73% and 23.81% respectively; fifth-level
cities accounted for 36.36% and 38.10% respectively;
sixth-level cities accounted for 31.82% and 33.33%.
From the urban sphere of influence, the numbers of low-
and middle-level cities in the two provinces were large,
and the overall economic development levels were weak.
The fourth-level cities accounted for 31.82% in Anhui
Fig. 3 Urban sphere of influence in Central China
Province; fifth-level cities accounted for 45.45%;
As seen in Table 2, the number of the fourth-, fifth- sixth-level cities accounted for 9.09%, which reflected
and sixth-level cities accounted for 89.88% of all the the equilibrate economic strength of this province.
cities in Central China. About 50.00% of all the cities in
Hubei Province were fifth-level cities, 16.67% were 4.3 Spatial distribution of urban sphere of influence
fourth-level cities, and 25.00% were sixth-level cities. It The Fig. 4a is a chart of semivariograms and crossco-
can be seen, although the urban sphere of influence of variance cloud, Fig. 4b, c, d represent the lines connecting
Wuhan was the largest in Central China, urban spheres corresponding point between cities. It can be seen from
of influence of other cities of Hubei Province were the Fig. 4a that the cloud chart is not spatial self-correla-
small. This is mainly due to the weak overall economic ted but the points are divided into four layers which re-
strength of Hubei, which is a typical example of ″weak flects the hierarchical relationships of urban sphere of
province with strong city″. But for Hunan Province, the influence in central regions. The highest layer represents
number of fourth-level cities accounted for 24.14% of that almost all the lines connect Wuhan and the other cit-
all the cities, sixth-level cities accounted for 17.24%, ies of Central China, and the large value of semi-variog-
and fifth-level cities accounted for 44.83%. It showed ram reflects the dominant position of Wuhan City in Cen-
that urban spheres of influence of cities in Hunan were tral China. Wuhan is also the regional heterogeneous
in the middle of Central China, which also reflected the point. The second layer which represents the lines con-

Table 2 Urban sphere of influence divided by province

Hubei Hunan Shanxi Henan Anhui Jiangxi
Grade Number Proportion Number Proportion Number Proportion Number Proportion Number Proportion Number Proportion
(%) (%) (%) (%) (%) (%)

First-level 1 2.77 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00

Second-level 0 0.00 1 3.45 1 4.55 1 2.64 1 4.55 1 4.76
Third-level 2 5.56 3 10.34 1 4.54 3 7.89 2 9.09 0 0.00
Fourth-level 6 16.67 7 24.14 5 22.73 17 44.74 7 31.82 5 23.81
Fifth-level 18 50.00 13 44.83 8 36.36 14 36.84 10 45.45 8 38.10
Sixth-level 9 25.00 5 17.24 7 31.82 3 7.89 2 9.09 7 33.33
Total 36 100.00 29 100.00 22 100.00 38 100.00 22 100.00 21 100.00
 Field Modeling Method for Identifying Urban Sphere of Influence: A Case Study on Central China 359

Fig. 4 Semivariograms and crosscovariance cloud and pair of cities

necting Wuhan and other provincial capital cities is a line, symmetrically distributed, while the points below
transitional layer; the first half of the third layer is the the line were scattered distributed (Fig. 5).
lines connecting the provincial capital city and its sur-
rounding cities. With the increase of distance between
cities, their connecting lines become scattered, instead
of aggregated towards provincial capital cities. This
fully explains the dominant position of the provincial
capital city which is also the local maximum in prov-
inces of Central China; the fourth layer is disturbed
connected point pairs, and there is no lower grade ag-
glomeration centre city.

4.4 Classification of spatial distribution of urban sp-

here of influence Fig. 5 Linear regression between total area of urban
sphere of influence and urban influence index
The result of regression analysis on urban sphere of in-
fluence was satisfied (R2 = 0.990, P < 0.01), and the
The former resulted from the monopolar development
equation was:
of most cities in the central region and there was no di-
Log(S) = 0.96 × Log(Z) + 19.55 (5) rect overlap between cities; the latter was because that
where S is the total area of urban sphere of influence; Z the level of urban spheres of influence and the interac-
is the urban influence index; and the Log transformation tion strength differed between them. As mentioned pre-
can reduce the difference of data scope between indica- viously, the overlapping situation of field intensity
tors. About 80% of the cities were above the regression which contains the relationship type between cities var-
360 DENG Yu, LIU Shenghe, WANG Li et al.

ies due to urban spatial location and urban influence two provinces is also an important reason. Since the
index. Therefore, the recognition of overlapping situa- whole province′s economic strength is weak, economic
tion of field intensity is the premise to further identify development capacity of its capital and other cities is
the relationship type between cities. We used spatial limited. Nanchang is also in a similar situation.
morphological characteristics, urban sphere of influence 2) Agglomeration type. This distribution type which
level and the residual to define the basic type of organi- had high-grade (first- and second-level) cities intermin-
zation form of urban sphere of influence: single-polar gled with low-grade cities mainly concentrated around
type, agglomeration type, closed-related group type and large cities (Fig. 6b). Combination of Taiyuan, Jinzhong
loose-related group type (Table 3). and Gujiao in Shanxi Province formed a big group, be-
1) Single-polar type. About 73% of the cities belonged cause the three cities were close to each other. There
to this distribution type whose urban sphere of influence was no city above the county-level within 50 km around
was broadly similar to a circle (Fig. 6a). For example, Wuhan, but due to its large influence, it could integrate
Hefei and Nanchang are found to belong to this sin- with cities far away from it. Henan Province is one of
gle-polar type rather than the agglomeration type, which the birthplaces of Central Plain culture. Because of its
have been proposed as the Wanjiang urban agglomera- geographical, historical and many other reasons, there
tion and the Poyang Lake urban agglomeration respec- are large numbers of intensive-distributed cities in this
tively by some pervious literatures (Fang, 2008; Wang et province. Thus, eight cities aggregated in the group with
al., 2008). Main reasons of this phenomenon include Zhengzhou at its core, and the area of this group was big.
historical evolution, sparsely distribution of cities in There is potential and tendency of integrated develop-
these provinces, and no county-level or above cities near ment in Changsha-Xiangtan-Zhuzhou region because of
the provincial capital. Such as, there is no countylevel their approximate sizes, the short distance between them
city within 60 km around Hefei, which sub-stantially and continuous policy guidance. As regional core cities,
hinders the group between Hefei and its surrounding cit- the city levels of Wuhan, Taiyuan and Zhengzhou are
ies. In addition, low economic development level of the much higher than other cities. It illustrated their strong

Table 3 Classification of spatial distribution of urban sphere of influence

Type Number Criteria City
Single-polar type 122 Independent distribution Hefei, Nanchang, Luoyang, Datong, Wuhu, Xiangfan, Yichang, Yueyang, Huainan,
Hengyang, Pingdingshan, Nanyang, Changde, Xinyu, Huaibei, Shiyan, Maanshan,
Bengbu, Jiujiang, Puyang, Tongling, Yangquan, Jiaozuo, Luohe, Kaifeng, Anqing, Jing-
zhou, Pingxiang, Shuozhou, Suzhou, Liuyang, Shangqiu, Jingdezhen, Jiyuan, Qianjiang,
Fuyang, Xinyang, Jingmen, Linzhou, Linfen, Loudi, Yongcheng, Ganzhou, Dengfeng,
Yongzhou, Xiantao, Huaihua, Lu′an, Sanmenxia, Huangshan, Leiyang, Yima, Suizhou,
Fengcheng, Zhoukou, Hejin, Lingbao, Shaoyang, Fuzhou, Zhumadian, Chuzhou,
Changning, Liling, Xiangxiang, Qinyang, Tianmen, Chaohu, Tongcheng, Bozhou,
Ruzhou, Yichun, Xinzhou, Zhongxiang, Dengzhou, Chibi, Zhangjiajie, Xiangcheng,
Chizhou, Ningguo, Jishou, Enshi, Ji′an, Tianchang, Xuancheng, Yidu, Huozhou, Leping,
Macheng, Zhangshu, Xianning, Shangrao, Houma, Yuncheng, Gaoping, Miluo, Wugang,
Yingcheng, Zaoyang, Guangshui, Honghu, Yuanping, Nankang, Shishou, Dexing,
Gao′an, Hongjiang, Dangyang, Songzi, Zhijiang, Linxiang, Wugang, Jieshou, Yongji,
Mingguang, Luliang, Anlu, Ruijin, Yicheng, Jinshi, Lichuan, Shaoshan, Jinggangshan

Agglomeration type 4 With high-grade city (first- and Wuhan-Ezhou-Huangshi-Huanggang-Daye-Xiaogan-Hanchuan, Zhengzhou-Xingyang-

second-level), and surrounding Xinmi-Xinzheng-Changge-Yuzhou-Xuchang, Changsha-Xiangtan-Zhuzhou, Taiyuan-
low grade city Gujiao-Jinzhong

Close-related group type 5 Residuals < 0 Fenyang-Xiaoyi-Jiexiu, Xinxiang-Huixian-Weihui, Lucheng-Changzhi, Anyang-Hebi,

Yingtan-Guixi, Yanshi-Mengzhou-Gongyi

Loose-related group type 5 Residuals > 0 Danjiangkou-Laohekou, Wuxue-Ruichang, Lengshuijiang-Lianyuan, Chenzhou-Zixing,

Yuanjiang-Yiyang
 Field Modeling Method for Identifying Urban Sphere of Influence: A Case Study on Central China 361

Fig. 6 Type of urban sphere of influence

accumulation ability and the position of regional devel- spatial distribution of the urban spheres of influence can
opment core. On the other hand, however, it showed the be further explored to provide references for various
excessive concentration of limited development re- types of urban and regional planning by using GIS and
sources. The development prospect of small- and me- statistical methods. A case study was done to identify
dium-sized cities was far from optimistic, and it was not urban spheres of influence of 168 cities in Central China
conducive to the harmonious development of regional in 2007. The results showed:
economy. 1) Wuhan was the absolute dominant city in Central
3) Close-related group type. The residual of this type China. There were a large number of low-grade cities
of distribution was less than 0. The urban sphere of in- with small urban sphere of influence in Central China.
fluence was squeezed like Yingtan and Guixi, Yanshi, Different provinces had different development charac-
Mengyi and Gongyi cities (Fig. 6c). Proximity space teristics: Hubei was ″weak province with strong city″;
and approximate influence index aggravated the compe- medium-sized cities of Hunan and Henan were well de-
tition on resources and space, and substantially reduced veloped; development of the whole Jiangxi and Shanxi
the urban sphere of influence. was backward, while the development of Anhui was
4) Loose-related group type. The residual of this type more balanced.
of distribution was larger than 0. The overlap between 2) The semivariograms and crosscovariance cloud
urban sphere of influence was small like Shaoshan and showed that the spatial structure of urban sphere of in-
Xiangxiang, Lengshuijiang and Lianyuan, and Chen- fluence of Central China was hierarchical. The first-lev-
zhou and Zixing (Fig. 6d). The accumulation ability of el city is correspondly in the first layer. After the filtra-
single city of this type of distribution was not strong, tion of the second layer, the third layer expressed the
and there may exist two cores in this group. dominant position of the second-level cities in the pro-
vincial region. Wuhan City was the leader in the central
5 Conclusions region, provincial capital cities dominated the whole
province, and there was no other lower grade agglom-
This paper proposed that the field modeling method is a eration center.
promising method to overcome the current shortcomings. 3) We identify four basic types of organization form
Field can represent the spatial characteristics of interac- of urban sphere of influence: namely single-polar type,
tion between cities and the hinterland more reasonably, agglomeration type, close-related group type and
and also the characteristics of their classification and loose-related group type, through analyzing their spatial
362 DENG Yu, LIU Shenghe, WANG Li et al.

characteristics, level and residual of urban sphere of Huff D L, Lutz J M, 1979. Ireland′s urban system. Economic
influence. For examples, Wuhan, Zhengzhou, Taiyuan Geography, 55: 196–212.
Huff D L, Lutz J M, 1989. Urban spheres of influence in Ghana.
and Changsha all belonged to the agglomeration type
The Journal of Developing Areas, 23(2): 201–220.
while the number of cities is different among those ur- Huff D L, Lutz J M, 1995. Change and continuity in the Irish
ban agglomerations. Further, Hefei and Nanchang are urban system, 1966–1981. Urban Studies, 32(1): 155–173. DOI:
found to belong to the single-polar type rather than the 10.1080/00420989550013275
agglomeration type, which are named as the Wanjiang Kong Xiangbin, 2007. Study on city′s economic hinterland and its
urban agglomeration and the Poyang Lake urban ag- spatial definition—Take Chengdu and Mianyang as examples.
Chengdu: Southwest Jiaotong University. (in Chinese)
glomeration respectively by some researchers.
Ministry of Housing and Urban-rural Development, 2008. China
Certainly, Euclidean distance in this paper is an ide- Urban Construction Statistical Yearbook 2007. Beijing: China
alized expression of space. Therefore, comprehensive Architecture & Building Press. (in Chinese)
measure of regional accessibility helps to ascertain the Mulligan G F, 1984. Agglomeration and central place theory: A
distribution feature of urban sphere of influence. Inves- review of the literature. International Regional Science Review,
tigation of urban development track based on dynamic 9(1): 1–42. DOI: 10.1177/016001768400900101
National Bureau of Statistics of China, 2008. China City Statisti-
analysis of time series is the focus of future work.
cal Yearbook 2008. Beijing: China Statistics Press. (in Chinese)
National Bureau of Statistics of China, 2008. China County Sta-
References tistical Yearbook 2008. Beijing: China Statistics Press. (in Chi-
nese)
Berry B J, Lamb R F, 1974. The delineation of urban spheres of National Bureau of Statistics of China, 2008. China Population
influence: Evaluation of an interaction model. Regional Studies, and Employment Statistics Yearbook 2008. Beijing: China Sta-
8(2): 185–190. tistics Press. (in Chinese)
Christaller W (Translators: Chang Z W et al.), 1998. Central Nystuen J D, Dacey M F, 1961. A graph theory interpretation of
Places Southern Germany. Beijing: The Commercial Press. (in nodal regions. Papers and Proceedings of the Regional Science
Chinese) Association, 7(1): 29–42. DOI: 10.1007/BF01969070
Du G Q, 2001. Using GIS for analysis of urban system. Geo- Okabe A, Boots B, Sugihara K et al., 2000. Spatial tessellations:
Journal, 52: 13–21. DOI: 10.1023/A:1014268007599 Concepts and applications of Voronoi diagrams. New York:
Edwards G, 1993. The Voronoi model and cultural space: Appli- John Wiley & Sons Ltd.
cations to the social sciences and humanities. Spatial Informa- Pan Jinhu, Shi Peiji, Dong Xiaofeng, 2008. Measurements for
tion Theory A Theoretical Basis for GIS, 716: 202–214. DOI: urban hinterland area of cities at prefecture level or above in
10.1007/3-540-57207-4 14 China. Acta Geographica Sinica, 63(6): 635–645. (in Chinese)
Fang Chuanglin, 2008. Researches on comprehensive measure- Su Changgui, Wei Xiao, 2006. Some considers of rising strategy
ment of compactness of urban agglomerations in China. Acta in center region. Economic Geography, 26(2): 207–210. (in
Geographica Sinica, 63(10): 1011–1021. (in Chinese) Chinese)
Gold C M, 1992. The meaning of ′neighbour′. In: Theories and Taaffe E J, 1962. The urban hierarchy: An air passenger definition.
Methods of Spatio-temporal Reasoning in Geographic Space. Economic Geography, 38(1): 1–14.
Berlin: Springer-Verlag, 220–235. Wang De, Zhao Jinhua, 2000. HAP: A computer assisted hinter-
Green H L, 1955. Hinterland boundaries of New York City and land division system—Its development and application. City
Boston in southern New England. Economic Geography, 31(4): Planning Review, 24(12): 37–41. (in Chinese)
283−300. Wang Guiyuan, Chen Meiwu, 2004. Measurement of urban hin-
Gu Chaolin, 1997. China Urban Geography. Beijing: The Com- terland area based on GIS—A case study of the Yangtze river
mercial Press. (in Chinese) delta. Geography and Geo-Information Science, 20(3): 69–73.
Gu Chaolin, Pang Haifeng, 2008. Study on spatial relations of (in Chinese)
Chinese urban system: Gravity model approach. Geographical Wang Kaiyong, Chen Tian, Dong Mali, 2008. The spatial patterns
Research, 27(1): 2–10. (in Chinese) of urbanization in the middle area of China. Economic Geog-
Huff D L, 1973. The delineation of a national system of planning raphy, 28(3): 353–356. (in Chinese)
regions on the basis of urban spheres of influence. Regional Wen Jiabao, 2004. Government work report. [Link]
Studies, 7(3): 323–329. DOI: 10.1080/09595237300185321 [Link]/GB/shizheng/1024/[Link]. (in Chinese)

View publication stats