W. Zhang, B. Derudder, J. Wang, F. Witlox / Bulletin of Geography. Socio-economic Series / 31 (2016): 145–160
153
geographic concentration of passenger flows along
the Nanjing-Shanghai-Hangzhou-Ningbo belt,
where the main HSR lines lie, i.e. Shanghai-Nan-
jing HSR line, Shanghai-Hangzhou HSR line and
Hangzhou-Ningbo HSR line. In addition, Shanghai,
Nanjing and Hangzhou emerge as the most con-
nected cities in the network of passenger flows; Su-
zhou (one of the most dynamic cities that attract
foreign direct investment in YRD (Zhao, Zhang,
2007)), Ningbo (the main gateway city in the south-
ern part of the YRD) and Hefei (the administra-
tive and economic centre of Anhui province that
has been looking to join the YRD regional collec-
tive) are three sub-centre nodes of the network of
passenger flows.
4.2. Comparison
between the original network generated
by the proxy of the number of daily trains
and the transformed network
Our alternative approach is devised to address the
obstacle of overly flat structures produced by train
schedule-based methods for assessing urban net-
works. Here, we examine the changes put forward
by applying the transformation set out in section 3
by comparing original and transformed networks at
the level of nodes, linkages, and network structures.
We first offer a direct comparison of cities’
degree centralities in both networks (fig. 3). Degree
centrality is a measure of nodes’ position, which
represents the (valued) number of passenger flows
of cities. The first obvious change to note is that
the degree centralities of a range of cities, which
can be separated into two categories, seem lower
in the transformed network. The first category
is Nanjing, the sub-center city within the YRD.
There are 444 HSR trains operating across Nanjing
on a daily basis, which is almost the same as the
number of HSR trains operating across Shanghai
(490 per day). However, part of these trains only
transit across Nanjing, while most of them depart
from or have their final stop at Shanghai. That
means Shanghai contributes most of the passen-
gers, whereas Nanjing only contributes part of the
passengers. In this case, Nanjing’s position in the
original network is obviously overestimated. The
other category includes Suzhou (Jiangsu), Wuxi,
Changzhou, Zhenjiang, Shaoxing, and Xuzhou,
which are transit cities in main corridors: Suzhou
(Jiangsu), Wuxi, Changzhou, and Zhenjiang are on
the Shanghai-Nanjing HSR railway line, Shaoxing
is on the Hangzhou-Ningbo HSR railway line, and
Xuzhou is on the Beijing-Shanghai HSR railway
line (fig. 1). This is consistent with the theoretical
illustration of over-estimations of the position of
transit cities in section 2.2. On the other hand there
are also nodes becoming relatively more important
in the transformed network. The most dramatic
change is the higher rankings of Hefei, Ningbo,
Hangzhou and Wenzhou.
Second, fig. 4, in which edge thickness reflects
the flow strength of city-pairs, maps the 15 most
connected city-dyads in the original network as well
as the transformed network. City-dyads along the
Nanjing-Shanghai HSR line are the most connected
city-dyad—with the exception of Shanghai-Hang-
zhou—in the original network (fig. 4a). This reflects
the fact that any pair of cities along the Shang-
hai-Nanjing HSR line will have similar number of
inter-city trains. Compared with the pattern of con-
centrating on the Shanghai-Nanjing corridor in the
original network, the backbone of the transformed
network (fig. 4b) consists of the key cities along the
Nanjing-Shanghai-Hangzhou-Ningbo belt, which is
more consistent with the central corridor of YRD
urban agglomerations (Gu et al., 2007). More spe-
cifically, the original network tends to overvalue
inter-city connections, such as Nanjing-Wuxi and
Shanghai-Zhenjiang, along the Nanjing-Shanghai
HSR line, but on the other hand there are also in-
ter-city connections that are being undervalued.
These connections can be divided into two simple
categories: the connections between Shanghai and
sub-centers that are not on the Shanghai-Nanjing
corridor (i.e. Hangzhou, Ningbo and Hefei), and the
connections between pairs of proximate sub-centers
(i.e. Nanjing-Hefei, Hangzhou-Ningbo). In the lat-
ter cases, the dense flows of people between Nanjing
and Hefei—the closest pair of provincial capitals
in China—are apparent, especially in the context
of the regional integration of Yangtze Economic
Zone. The Hangzhou-Ningbo corridor, along which
long-running and dynamic peri-urbanization pro-
cess has occured (Webster, Muller, 2002), typifies
the cooperative pattern of core city (Hangzhou) and
sub-centre & port city (Ningbo): Ningbo—Hang-
- 10.1515/bog-2016-0010
Downloaded from De Gruyter Online at 09/23/2016 11:26:34AM
via Universiteit Antwerpen
W. Zhang, B. Derudder, J. Wang, F. Witlox / Bulletin of Geography. Socio-economic Series / 31 (2016): 145–160
154
Fig. 3. Cities’ degree centralities in the original network and the transformed network
Source: Own studies
zhou’s vicinity having more attractive labor, land
and tax costs—attracts many manufacturing func-
tions to moving from Hangzhou with keeping R&D
and sales functions in Hangzhou (Webster et al.,
2003); on the other hand, Ningbo’s deep-sea con-
tainer port provides Hangzhou with more wide in-
ternational market and hinterland. This provides
fundamental bases for the dense inter-city flows be-
tween Hangzhou and Ningbo.
And third and finally, to explore the structural
difference between both networks, we compare the
rank-size distributions of cities’ degree centralities.
The posited flatter structure of the original network
is indeed shown by the much steeper drop-off in the
cities’ degree centralities in the transformed network,
shown in fig. 5. We calculate the integration of rank-
size curve of cities’ degree centralities to measure
the flat degree of both networks. After normalizing
cities’ ranks into the interval [0,1], the flattening
ratio (F) of networks can be calculated as:
1
0
( )
F
L X dX
=
∫
F= (11)
where the function Y = L(X) represents the rank-
size curve. The flattening ratio varies between 0 for
completely even and 1 for completely uneven net-
works. In our measures, the flattening ratio of the
original network (0.39) is much higher than the
flattening ratio of the transformed network (0.23):
more precisely, the flattening ratio of the trans-
formed network has decreased to 60% of the origi-
nal flattening ratio in the case of the HSR network
within the YRD.
- 10.1515/bog-2016-0010
Downloaded from De Gruyter Online at 09/23/2016 11:26:34AM
via Universiteit Antwerpen