Multimodal transportation network centrality analysis for Belt and Road Initiative

Abstract

As the Belt and Road Initiative (BRI) takes root, the Air Transportation Network (ATN) is gaining prominence, as evidenced by the Air Silk Road Initiative proposed in 2017. However, the ATN of the BRI countries and its integration with maritime and rail transport have not been studied. This paper analyzes the network structure of the ATN of the BRI countries. A multi-layer weighted betweenness metric is used to rank the centrality of the hubs in the network based on the connectivity, distance, and traffic between the airports. Using the centrality ranking of the airports to reflect the distribution of traffic demand in the BRI countries, a gravity model is then applied to quantify the centrality of the maritime and rail hubs. Mapping these hubs to their home cities, this paper assesses the current state of the multimodal transport hubs and recommends new hubs to increase the international trading reach of the BRI.

Full text

1
Multimodal transportation network centrality analysis for Belt and Road Initiative
1
Yaoming Zhou, Tanmoy Kundu*, Mark Goh, Jiuh-Biing Sheu
2
Abstract
3
As the Belt and Road Initiative (BRI) takes root, the Air Transportation Network (ATN) is gaining
4
prominence, as evidenced by the Air Silk Road Initiative proposed in 2017. However, the ATN of the BRI
5
countries and its integration with maritime and rail transport have not been studied. This paper analyzes the
6
network structure of the ATN of the BRI countries. A multi-layer weighted betweenness metric is used to
7
rank the centrality of the hubs in the network based on the connectivity, distance, and traffic between the
8
airports. Using the centrality ranking of the airports to reflect the distribution of traffic demand in the BRI
9
countries, a gravity model is then applied to quantify the centrality of the maritime and rail hubs. Mapping
10
these hubs to their home cities, this paper assesses the current state of the multimodal transport hubs and
11
recommends new hubs to increase the international trading reach of the BRI.
12
Keywords: Belt and Road Initiative; Multimodal transportation; Network centrality; Air transport network;
13
Gravity model
14
1 Introduction
15
In 2013, China, in an ambitious effort to lift the economic, infrastructural, and international trade efficiency,
16
launched the Belt and Road Initiative (BRI) (Sheu and Kundu, 2018). The BRI through its new Silk Route
17
st , seeks
18
to overcome the international logistics and transportation barriers in the BRI network of 78 countries
19
(Kundu and Sheu, 2019a; Sheu and Kundu, 2018). Figure 1 presents the BRI network.
20
The BRI network comprises various economic corridors to promote and enhance international trade.
21
For example, the China-Pakistan Economic Corridor (CPEC) with the port of Gwadar as its hub is an
22
important overland corridor under the BRI due to its significance of securing Chinese oil reserves coming
23
from the Middle East (Kundu and Sheu, 2019b; Sheu and Kundu, 2018; Wen et al., 2019). CPEC is also
24
connected to the maritime corridor that connects China to Europe via Africa and the Middle East. This
25
combination of the overland and maritime transportation network serves a key role in increasing the
26
logistics connectivity between the various transportation hubs (rail and maritime ports) of those countries.
27
Similarly, the New Eurasia Land Bridge Economic Corridor (NELBEC) connects China to Europe via
28
Russia and several central Asian countries. Further, to promote the NELBEC, the Chinese government
29
launched the China-Europe inter-continental freight train service in 2011, known as the China Railway
30
Express (CR Express) (Jiang et al., 2018; Kundu and Sheu, 2019b; Shao et al., 2018). Associated with the
31
CR Express has been the development of many rail hubs along the NELBEC. This suggests that the
32
development of economic corridors can lead to the growth of transportation (maritime or rail) hubs.
33

2
34
Fig. 1. Belt and Road network
1
35
In recent years, the significance of the Air Transportation Network (ATN) has been increasingly
36
recognized with the advancement of BRI. By late July 2019, China has concluded intergovernmental
37
cooperation with 136 countries, which goes well beyond the scope of the traditional BRoad. To
38
increase the trading outreach and connectivity among the BRI countries, the Air Silk Road Initiative was
39
proposed in 2017 by Aviation Industry Corporation of China, Ltd. (AVIC) and the first Air Silk Road
40
International Aviation Cooperation Summit held in 2018
2
. Since then, the construction of the Air Silk Road
41
has been sped up. By late 2019, China has concluded bilateral air transport agreements with 126 countries
42
and regions
3
. Different from maritime and rail transportation, which may be restricted by terrain, every BRI
43
country has airports for international and domestic freight transport. However, the ATN of the BRI
44
countries and its integration with maritime and land transport have not been studied. Thus, this paper
45
attempts to close this gap using a multi-method approach. Figure 2 presents the research framework
46
associated with this study.
47
The contributions of this paper are threefold. First, the ATN composed of the BRI countries is modeled
48
and analyzed using flight data. A novel multi-layer weighted betweenness metric is used to rank the
49
important airports acting as international gateways in the BRI network, with consideration of connectivity,
50
1
Figure 1 is adapted from Kundu and Sheu (2019b).
2
https://www.avic.com/en/aboutus/overview/index.shtml.
(accessed 12th October 2020).
3
https://eng.yidaiyilu.gov.cn/qwyw/rdxw/105854.htm. (accessed 12th October 2020).




 


3
distance, and traffic between the airports. Next, a gravity model is applied to rank the maritime and rail
51
hubs based on their distance to the airport hubs. Third, the hubs in the three transport modes are integrated
52
based on their locations. The intermodal hubs are filtered and the potential multimodal hubs are
53
recommended.
54
In summary, this paper incrementally contributes to (i) the logistics and transportation management
55
literature by investigating a novel transportation problem of identifying the potential transportation hubs
56
that can host multimodal (rail, maritime and air) transportation operations, and (ii) the BRI-related literature
57
by generating important managerial insights for carriers and multimodal transport operators (i.e., potential
58
cities in the BRI network for multimodal transportation operations) and policy implications for governments
59
(i.e., potential cities in the BRI network for multimodal transportation infrastructure development) in the
60
BRI strategic context. Furthermore, this paper also contributes to the practical utilization of the BRI-
61
proposed new trading hubs (such as cities along the China-Europe rail route) and the parallel development
62
of the Air Silk Road through providing developmental suggestions, which benefits both the carriers and
63
governments significantly. More importantly, this paper positions itself as one of the first study which
64
contributes to the BRI-induced logistics and transportation management literature by discussing the
65
importance of the air silk route in the BRI network development. Accordingly, the proposed approach which
66
is based on network centrality measures and gravity model generates insights to suggest potential
67
multimodal hubs (air-rail, rail-maritime, maritime-air, air-rail-maritime) in the BRI network.
68
69
Fig. 2. Research framework of multimodal transport analysis
70
The rest of this paper is organized as follows. Section 2 outlines the extant literature. Section 3 presents
71
an analysis of the BRI air transport network. Section 4 assesses the centrality of the maritime and rail hubs
72
using a gravity model. Section 5 evaluates the criticality of the cities in the BRI multimodal transport
73
network. Section 6 offers some managerial insights and policy implications. Section 7 concludes this study
74
with some future research directions.
75

4
2 Literature review
76
This section gives an overview of the current literature on the transportation network under the BRI. The
77
most appropriate articles published in the transportation-related journals are selected using keyword search
78
and content analysis. These articles are categorized according to the transportation mode being studied, as
79
shown in Table 1. The contribution and methodology of each article is also summarized.
80
Table 1. Literature review on transportation network analysis in the BRI context
81
Authors
Transportation
mode
Summary of contribution
Methodology
(Jiang et al., 2018)
Rail
Analyzing the transportation costs and hub
cities of the China Railway Express
Random utility theory
(Zhao et al., 2018)
Evaluating consolidation centers for China
railway express considering government policy
and operation experience
TOPSIS and mixed integer
programming
(Shao et al., 2018)
Evaluation of large-scale transnational high-
speed railway construction priority in the belt
and road region
Large-scale optimization with exact
solution approach
(Yang et al., 2018)
Reconstructing the shipping service network
between Asia and Europe by improving New
Eurasian Land Bridge rail services and
Budapest- Piraeus railway
Bi-level mixed integer programming
model
(Li et al., 2019)
Predicting the market share rate of the China
Railway Express
Multiple logit model
(Yang et al., 2020)
Estimating the impact of China-Europe
Railway Express on trade transport
accessibility, cargo volumes, and logistics
center formation
Stochastic frontier analysis and
Multifactor dynamic coefficient
generation
(Zhang et al., 2020)
Ranking node importance in the China Railway
Express network
Multiple attribute decision making
model
(Wei and Lee, 2021)
Establishes a coordinated horizontal alliance
system for inland ports with CR Express
platforms in China
Entropy weighted TOPSIS and
Logistics gravity model
(Wang et al., 2018)
Maritime
Analyzing the spatial pattern evolution of the
-first-   
container shipping network
Graph-theoretical tools and
Herfindahl-Hirschman index
(Zeng et al., 2018)
Evaluating the impact of the Carat Canal on the
BRI shipping networks and evolution of hub
ports
Transit-trade prediction model based
on spatial interaction

5
(Ruan et al., 2019)
Analyzing the impacts of the BRI on the Indian
subcontinent ports emphasizing on Colombo
and Hambantota ports in Sri Lanka
Trend impact analysis and Intuitive
logic method
(Kundu and Sheu, 2019b)
Rail-maritime
Investigating the impact of the subsidy on
transportation mode choice (Rail vs. Maritime)
behavior of shippers
Competition model grounded on
game-theoretic technique
(Wang and Yau, 2018)
Rail-road
Tracing the network building of three major
BRI corridors, namely, CPEC, NELBEC, and
PAR (Pan-Asian Railway)
Actor-Network theory
(Wen et al., 2019)
Exploring the reliable advantage of 4 corridors
on improving the connectivity between China
and neighboring countries
BRI enhanced logistics network
theory based on route utility function
(Wang et al., 2020)
Investigating the impact of rail and road
transport on economic growth in the BRI
countries
Spatial autocorrelation test
(Lee et al., 2018)
Rail-road-
maritime
Analyzing the structure of 6 corridors and their
impact on BRI transportation and trade
Network analysis
(Sheu and Kundu, 2018)
Exploring the impact of BRI corridors (CPEC,
Myanmar-China, Strait of Malacca) on
Chinese oil supply chain network
Spatial-temporal logistics interaction
model integrated with Markov chain
(Li et al., 2018)
Rail-road-river
Evaluating the importance of 23 grain
distribution centers in the multimodal transport
network
Complex network theory improved
entropy TOPSIS
(Derudder et al., 2018)
Rail-road-air
Ranking the centrality of cities along 6
corridors considering multimodal connectivity
Composite corridor network method
(Chhetri et al., 2018)
Rail-road-
maritime-air
Assessing logistic cities using country-level
transport indexes
Five-stage indicator-based
assessment method
(Shi et al., 2019)
Rail-road-
water-air
Assessing multimodal transportation
accessibility using gridded output and a global
index
Global transportation accessibility
index model
82
As shown in Table 1, over half of the papers study either the rail or maritime network, which are the
83
most important transport mode in the      st
84
 Among these papers, Jiang et al. (2018),
85

6
Shao et al. (2018), Yang et al. (2018), Zhang et al. (2020), and Wang et al. (2018) have studied the structure
86
of the transportation network, while the others mostly focus on the economic impact of BRI on the countries.
87
The integrated rail and road networks have also been studied; these are the main transportation modes
88
for the 6 corridors in the BRI context, namely, the China-Mongolia-Russia Economic Corridor, the New
89
Eurasian Land Bridge, the China- Central Asia-Western Asia Corridor, the China-Indochina Peninsula, the
90
China-Pakistan Economic Corridor, and the Bangladesh-China-India-Myanmar Corridor. Wang and Yau
91
(2018), and Wen et al. (2019) have studied the effect of the network structure of several corridors on the
92
connectivity of the BRI countries. Wang et al. (2020) have examined the effects of rail and road transport
93
on the economic growth of the BRI countries. Lee et al. (2018) studied the structure and impact of the 6
94
corridors, and proposed the possible connectivity between the railway and maritime networks in Eurasia.
95
Li et al. (2018) have evaluated the importance of 23 grain distribution nodes based on their centrality in the
96
freight railway, highway, and waterway networks, which is then used to assess the distribution capacity of
97
these nodes.
98
Several research pieces involving ATNs in the multimodal transport network analysis exist. Chhetri et
99
al. (2018) have developed an indicator-based assessment framework to evaluate the position of the key
100
global logistics cities. Twenty country-level indicators, such as road density, rail network length, and
101
connectivity of the port and airport, are used to measure the capacity of the four key components of these
102
selected cities, i.e. road, rail, port, and air transport. Only the most important cities are involved in this
103
study, such as Shanghai, Hong Kong, and Bangkok. Shi et al. (2019) proposed a global accessibility index
104
to visualize the transportation accessibility of the BRI region using grid cells. The road density, railway
105
density, and waterway density are computed, which is later aggregated to yield the transportation density
106
index. The global accessibility index is defined as the average of the transportation density index and the
107
transportation convenience index. Both Chhetri et al. (2018) and Shi et al. (2019) have used grid-based or
108
area-based assessment methods, and the connectivity relationships in the transportation networks have not
109
been well investigated. Derudder et al. (2018) studied the centrality of the cities in the six BRI corridors. A
110
multimodal transport network is constructed by aggregating four layers, i.e. rail, road, air, and information
111
technology networks. Three centrality measures including degree centrality, betweenness centrality, and
112
closeness centrality are used to rank the cities from various aspects. In each layer, the strength of the
113
connectivity between two nodes is determined by their size and the Euclidean distance between them. In
114
short, the actual connectivity relationship between the nodes in the transport network is ignored.
115
Although the importance of air transport in the BRI transport network has been realized, the network
116
structure of the air transportation system and its integration with the other transportation modes is under
117
studied. To fill the gap, this paper constructs the structure of the ATN in the BRI region using fight data.
118

7
Then, a multilayer weighted betweenness metric is proposed to assess node centralities in the network based
119
on the connectivity, distance, and traffic between the airports. To evaluate the coordination between ATN
120
and the other transportation modes (rail and maritime), a gravity model is applied to quantify the importance
121
of the maritime and rail hubs. Through this approach, one can assess the current multimodal transport hubs
122
and recommends potential hubs to increase the international trade reach of the BRI.
123
3 BRI air transport network
124
3.1 Data
125
There are 98 countries in this study, as found in Table A1 of the Appendix. The data set is obtained
126
from OAG database (https://analytics.oag.com/analyser-client/home), which provides the information of
127
all the commercial flights globally. We use the flights in one whole week from November 21 to November
128
27, 2019 to build the BRI network. In this week, there are 209,811 flights operated by the airlines on 7,380
129
flight connections among 1193 airports in the countries of interest, as shown in Fig. 3. The average number
130
of flights in one week on these connections is 14.28.
131
132
Fig 3. BRI air transport network
133
3.2 Airport centrality
134

8
Let
 
1 2 3
, , ,..., ,...,
in
A A A A A A=
denote the set of airports. The centrality of an airport
i
A
denoted by
135
i
A
C
is evaluated by its contribution to the connectivity of the network. This paper will assess the centrality
4
136
of the airports from two perspectives, i.e., local centrality
5
, and global centrality
6
. The local centrality of
137
an airport is quantified by its node degree, which is the number of direct connections to the other airports.
138
In an unweighted network, it is equal to the number of airports directly connected to this airport. An airport
139
with a larger node degree has more airports directly connected to it by flights and therefore contributes
140
more to the local connectivity of the overall network. The global centrality (
GC
) of an airport
i
A
is given
141
by its betweenness (Freeman, 1977; Guimerà et al., 2005; Wang et al., 2019), namely,
142
i
jk
ijk
i j k
A
AA
AAA
A A A A
GC


  
=
, (1)
143
where
j
A
and
k
A
are airports in the set of airports
A
,
jk
AA

is the number of shortest paths from airport
144
j
A
to airport
k
A
, and
i
AA
jk
A

is the number of such paths traversing through an airport
i
A
. An airport with
145
a larger betweenness centrality acts as a bridge for more airport pairs, and therefore, contributes more to
146
the global connectivity of the network.
147
The centralities and ranking of the top 20 airports in the BRI network are listed in Table 2. The complete
148
table containing the top 100 airports is shown in Appendix Table A2. In Table 2, the local centrality of the
149
top-ranked airport IST (Istanbul) is 189, i.e., 189 of 1193 airports in the BRI ATN have direct flights to
150
IST. The global centrality of IST is 128,992, which informs that IST is the best transit airport for 128,992
151
out of (1193
2) = 711,028 pairs of airports based on the shortest path.
152
Comparing the ranking of the airports using the two centrality metrics suggests that the top 1 ranked
153
airport by local connectivity is the same as that in terms of global connectivity, i.e., IST (Istanbul). On one
154
hand, IST is the largest hub airport in Turkey, connecting the other smaller airports in Turkey and some
155
neighboring countries. On the other hand, IST is also the transit for connecting to the airports from other
156
regions in the BRI network. However, the situation is quite different for the other airports. For example,
157
the 3rd ranked airport in terms of local connectivity, CTU (Chengdu), is ranked 9th on global connectivity.
158
The reason is that most of the airports connected to CTU are inside China and these airports are
159
simultaneously connected to some other larger airports such as CKG (Chongqing). Thus, CTU may not be
160
4
The approach of using centrality measures in network analysis is adapted from Freeman (1978), and Wang and Cullinane (2016).
5
Local centrality measures the importance of an airport to the local connectivity of the network.
6
Global centrality measures the importance of an airport to the global connectivity of the network.

9
a hub or the only hub for the neighboring smaller airports, notably on international freight. Similarly, DXB
161
(Dubai), the 2nd ranked airport on global connectivity, is ranked 8th on local connectivity. Here, while there
162
are fewer airports directly linked to DXB, compared with the other 7 airports; DXB acts as a very important
163
bridge for the airports from different regions in the BRI network. We also find that most of the top 20
164
airports by local connectivity are from China, because China is currently the largest member country in
165
BRI strategic context. However, since a majority of these connections are domestic connections inside
166
China, their ranks on global connectivity are lower. The top 20 airports by global connectivity are all
167
international hubs for connecting their home countries to the others in the BRI network, such as IST
168
(Istanbul), CGK (Jakarta), and PEK (Beijing), or those acting as international hubs for the other countries,
169
such as DXB (Dubai) and SIN (Singapore).
170
Table 2. Top 20 airports by local and global centralities
171
Rank
Local centrality
Global centrality
IATA code (City)
Degree
Country
IATA code (City)
Betweenness
Country
1
IST (Istanbul)
189
Turkey
IST (Istanbul)
128992
Turkey
2
PEK (Beijing)
164
China
DXB (Dubai)
89295
United Arab
Emirates
3
CTU (Chengdu)
159
China
CGK (Jakarta)
49193
Indonesia
4
XIY (Xi'an)
150
China
PEK (Beijing)
46328
China
5
CAN
(Guangzhou)
146
China
SVO (Moscow)
43492
Russia
6
PVG (Shanghai)
138
China
ADD (Addis
Ababa)
39332
Ethiopia
7
KMG (Kunming)
134
China
ATH (Athens)
38990
Greece
8
DXB (Dubai)
131
United Arab
Emirates
DOH (Doha)
37356
Qatar
9
SZX (Shenzhen)
131
China
CTU (Chengdu)
31413
China
10
CKG
(Chongqing)
130
China
DME (Moscow)
31202
Russia
11
SVO (Moscow)
121
Russia
SIN (Singapore)
29578
Singapore
12
BKK (Bangkok)
108
Thailand
KUL (Kuala
Lumpur)
29100
Malaysia
13
DME (Moscow)
107
Russia
JNB
(Johannesburg)
26912
South Africa
14
HGH (Hangzhou)
104
China
CAN (Guangzhou)
25644
China
15
VKO (Moscow)
102
Russia
JED (Jeddah)
25484
Saudi Arabia
16
LED (St.
Petersburg)
100
Russia
ALG (Algier)
23497
Algeria
17
CSX (Changcha)
98
China
BKK (Bangkok)
22764
Thailand
18
TSN (Tianjin)
98
China
THR (Teheran)
20895
Iran
19
HAK (Haikou)
95
China
LED (St.
Petersburg)
20660
Russia
20
KUL (Kuala
Lumpur)
95
Malaysia
VKO (Moscow)
19981
Russia
The local centrality was plotted against their global centrality of the top 100 BRI airports and is
172
illustrated in Fig. 4. It is important to note that the ratio between the global and local centrality of most of
173

10
(PEK, CTU, CAN, XIY, PVG, KMG, SZX, CKG   
174
LED, VKO) are smaller than that of the airports in the Middle East (DXB, DOH, JED). Interestingly, the
175
airports in the African countries (ALG, JNB, ADD) have the largest ratio between global and local centrality.
176
This suggests that the ATN of China and Russia can be improved by introducing more international flights
177
to other regions to enhance their pivotal role in the BRI network and the overall global connectivity of the
178
BRI network.
179
180
Fig 4. Local vs. global centrality of top 100 airports in BRI network
181
3.3 Airport centrality considering link capacity and node distance
182
The ATNs are weighted networks, meaning that there are different number of flights operating on the
183
airport connections (Zhou et al., 2019). The number of flights can be regarded as the link capacity.
184
Accordingly, the concept of a weighted degree is employed to assess the local centrality of the airports in
185
the weighted BRI ATN (Opsahl et al., 2010). The weighted degree of an airport is defined as the sum of
186

11
the number of scheduled direct flights to the other airports in the ATN. As to global connectivity, there is
187
no existing definition of weighted betweenness appropriate for assessing the airports in the weighted ATNs.
188
Therefore, this paper proposes a new metric of weighted betweenness by dividing the weighted ATN into
189
layers. As shown in Fig. 5, the links in the weighted network have non-uniform weights to represent the
190
number of flights. For each connected airport pair in the weighted network, one unit of connection between
191
them is removed and taken to construct the first layer. This process is repeated on the residual weighted
192
network until no more connections remain in the weighted network. Therefore, the number of layers equals
193
the largest link weight in the weighted network. Then, the betweenness of the airports in each layer is
194
computed individually. Finally, the betweenness of each airport in the layers is aggregated and the sum is
195
taken as the weighted betweenness of each airport in the ATN. Furthermore, the spherical distance between
196
different airport pairs is considered when determining the shortest paths to obtain the betweenness in the
197
different layers using
198
( )
1
cos sin sin cos cos cos
i j i j i j i j
A A A A A A A A
DR
     
−
= + −


, (2)
199
where
ij
AA
D
is the spherical distance between airport
i
A
and airport
j
A
, (
i
A

,
i
A

) and (
j
A

,
j
A

) are the
200
longitude and latitude of airport
i
A
and airport
j
A
, and
R
is the radius of the Earth respectively
201
(Bhaskaran and Turnquist, 1990; Liu et al., 2016).
202
203
Fig. 5. Illustration of dividing weighted ATN into layers
204
Table 3 lists the centralities and ranking of the top 20 airports in the weighted BRI ATN. The complete
205
table containing the top 100 airports is shown in Appendix Table A3. In Table 3, the weighted degree of
206
the top-ranked airport PEK (Beijing) is 4,816, i.e., there are 4,816 direct flights from PEK to the other
207
airports in the BRI network during the studied week. Although weighted betweenness has no exact physical
208
meaning, it measures the frequency of an airport acting as the hub for other airport pairs in the weighted
209
BRI ATN.
210

12
Comparing Table 3 with Table 2, notice that while IST (Istanbul) has more direct connections than
211
PEK (Beijing), the direct connections of PEK (Beijing) are stronger. On the weighted betweenness, DXB
212
(Dubai) replaces IST (Istanbul) as the top 1 airport, as the geographic position of DXB in the BRI network
213
is more central than that of IST, thus, DXB is more possibly to be the international transit for other airports.
214
BKK (Bangkok) is the 3rd ranked airport but it is the 17th ranked on betweenness in Table 1. The reason is
215
that BKK is the hub for some other smaller airports in Southeast Asia. In the unweighted network, BKK,
216
together with some other hubs, serves as the bridge between these small airports and the other airports. In
217
the weighted network, when the distance between airports is considered, BKK becomes the highest priority
218
for these small airports. In short, the weighted degree and weighted betweenness indicate local and global
219
centrality better, respectively.
220
Table 3. Top 20 airports by local and global centralities in weighted BRI ATN
221
Rank
Local centrality
Global centrality
IATA code (City)
Weighted
degree
Country
IATA code (City)
Weighted
betweenness
Country
1
PEK (Beijing)
4816
China
DXB (Dubai)
1498738
United Arab
Emirates
2
CGK (Jakarta)
4385
Indonesia
IST (Istanbul)
1345114
Turkey
3
CAN
(Guangzhou)
4278
China
BKK (Bangkok)
1021471
Thailand
4
KUL (Kuala
Lumpur)
3303
Malaysia
KUL (Kuala
Lumpur)
931838
Malaysia
5
CTU (Chengdu)
3276
China
SVO (Moscow)
887144
Russia
6
SZX (Shenzhen)
3258
China
PEK (Beijing)
835737
China
7
PVG (Shanghai)
3090
China
ADD (Addis
Ababa)
627801
Ethiopia
8
XIY (Xi'an)
3042
China
KTM
(Kathmandu)
613484
Nepal
9
CKG
(Chongqing)
2979
China
SIN (Singapore)
605307
Singapore
10
KMG (Kunming)
2886
China
KMG (Kunming)
561946
China
11
IST (Istanbul)
2839
Turkey
CAN
(Guangzhou)
503341
China
12
SHA (Shanghai)
2593
China
JNB
(Johannesburg)
470314
South Africa
13
BKK (Bangkok)
2495
Thailand
URC (Urumqi)
423810
China
14
SVO (Moscow)
2425
Russia
CGK (Jakarta)
423801
Indonesia
15
HGH (Hangzhou)
2387
China
NBO (Nairobi)
406891
Kenya
16
SIN (Singapore)
2310
Singapore
ATH (Athens)
390666
Greece
17
DMK (Bangkok)
2223
Thailand
CAI (Cairo)
363431
Egypt
18
SGN (Ho Chi
Minh City)
2194
Vietnam
CTU (Chengdu)
351784
China
19
DXB (Dubai)
2161
United
Arab
Emirates
DMK (Bangkok)
349148
Thailand
20
NKG (Nanjing)
1994
China
UPG (Ujung
Pandang)
307774
Indonesia

13
The top 20 airports in terms of the two metrics are presented in Fig. 6 and Fig. 7, respectively. The size
222
of the airport corresponds to the centrality measure value. In Fig. 6, the top 20 airports can be classified
223
into five groups. The largest group has eleven airports (PEK, CAN, CTU, SZX, PVG, XIY, CKG, KMG,
224
SHA, HGH, NKG) in China. The second group consists of the big airports from Southeast Asia (CGK,
225
CKL, BKK, SIN, DMK, SGN). The last three airports are respectively the local hub from Russia (SVO),
226
South Europe (IST), and the Middle East (DXB). There is no local hub from Africa in the top 20 airports.
227
Hence, developing potential airports with large local centrality in Africa can enhance the outreach of the
228
BRI-based international trading network.
229
230
Fig. 6. Top 20 airports in weighted BRI ATN by local centrality (airport size corresponds to its weighted
231
degree value).
232
Comparing Fig. 7 with Fig. 6, the top 20 airports in terms of global connectivity are more dispersed
233
geographically in the BRI network. There are six dominant transit hubs in the ATN, i.e., DXB, IST, BKK,
234
KUL, SVO, and PEK. While these six airports are primary domestic or regional hubs for the Middle East,
235
South Europe, Southeast Asia, Russia, and China, they are also spokes for connecting their corresponding
236
sub-networks to the other parts of the BRI ATN.
237


 











 











14
238
Fig. 7. Top 20 airports in weighted BRI ATN by global centrality (airport size corresponds to its weighted
239
betweenness value).
240
4 Maritime and rail hub centrality
241
A multimodal transport network under the BRI has three modes, i.e., ATN, maritime transport network,
242
and rail transport network. Among the three transportation modes in the BRI network, the ATN is the most
243
complex as all the countries under the BRI are either directly or indirectly linked to the others via the airport
244
network. Only a few countries have maritime ports, and though most countries have railways, they are
245
currently not connected as a network due to the different gauge sizes. Thus, for simplicity and fairness, this
246
work treats the ATN as the primary mode when analyzing the multimodal transport network under the BRI.
247
In the last section, the centrality of airports in the BRI-linked countries has been assessed based on their
248
contributions to the connectivity of the entire ATN. Next, this work assesses the importance of the maritime
249
ports and rail ports based on their distance to all the airports under study, using a gravity model. The
250
centrality of the maritime ports and rail ports are evaluated based on two information attributes: (1) the
251
distribution of the airport centralities, and (2) the distance between the maritime/rail ports and all the
252
airports.
253
Similar to the representation of the centrality of the airports, let
i
M
C
and
i
R
C
be the centrality of a
254
maritime port and rail port, respectively. Note that an airport has a certain level of connectivity
255




 






















15
attractiveness, which is quantified by its centrality. The centrality of a maritime port is the combination of
256
that attractiveness obtained from all the airports divided by the distance between them, i.e.
257
j
iij
j
A
MMA
AA
C
CD

=
, (3)
258
( )
1
cos sin sin cos cos cos
i j i j i j i j
M A M A M A M A
DR
     
−
= + −


, (4)
259
where
ij
MA
D
is the spherical distance between maritime port
i
M
and airport
j
A
, (
i
M

,
i
M

) and (
j
A

,
260
j
A

) are the longitude and latitude of maritime port
i
M
and airport
j
A
,
j
A
C
is the centrality of airport
261
j
A
, and
R
is the radius of the Earth. As we only care about the relative centrality of these hubs, the
262
coefficient of Equation (3) is set to be 1, and all the centrality values for the maritime and rail ports are
263
normalized in later analysis. The centrality of the rail ports is found in the same way.
264
Recall that this work chose the ATN as the focal transport mode since air transportation is the most
265
comprehensive and connects all cities in the BRI network. To an extent, the ranking of the airports reflects
266
the ranking of the cities under the BRI. To integrate the maritime transport network with the rail network,
267
this work assesses the centrality of the nodes (that reflect a transportation hub, i.e., either rail or maritime
268
associated with a city) on the single-mode transportation networks based on their distance to the airport
269
under study.
270
4.1 Maritime port centrality
271
Taking the gravity model approach to determine the centrality of the maritime ports in the BRI network,
272
the centralities of maritime ports are determined based on their distance to the airports. The centrality and
273
ranks of the top 20 maritime ports in the BRI network in terms of the local and global centralities are listed
274
in Table 4. Appendix Table A4 contains the 72 maritime ports. The data on the 72 maritime ports are
275
obtained from the Mercator Institute for China Studies (https://www.merics.org/en/bri-tracker).
276
From Table 4, 14 of the top 20 maritime ports are from China based on local centrality, with Shanghai
277
being the top 1. The other six maritime ports are all in Southeast Asia, namely Saigon and Danang in
278
Vietnam, Jakarta and Surabaya in Indonesia, Singapore, and Laem Chabang in Thailand. This ranking result
279
concurs with the fact that China is the driver of the BRI, and, together with the other countries in Southeast
280
Asia, contributes most to the BRI.
281
The rankings of the maritime ports on global centrality yield a different outcome from that on local
282
centrality. First, the rankings of the maritime ports outside of China are now higher, such as Singapore, Dar
283

16
es salaam (Tanzania), Abu Dhabi (United Arab Emirates), Kuantan (Malaysia), Karachi (Pakistan), and
284
Suez (Egypt). This is mainly because the contributions of the ports of the connecting countries in the BRI
285
network are considered. Second, the ranking of the maritime ports in China has changed considerably. For
286
example, Guangzhou has replaced Shanghai as the top-ranked maritime port in China, as Guangzhou is
287
more important given its location in the BRI network.
288
Table 4. Top 20 maritime ports using the gravity model in terms of local and global centralities
289
Rank
Local centrality
Global centrality
Hub
Normalized
centrality
Country
Hub
Normalized
centrality
Country
1
Shanghai
1.0000
China
Singapore
1.0000
Singapore
2
Saigon
0.8449
Vietnam
Saigon
0.8026
Vietnam
3
Xiamen
0.8099
China
Jakarta
0.5831
Indonesia
4
Da Nang
0.7983
Vietnam
Guangzhou
0.5687
China
5
Jakarta
0.7888
Indonesia
Laem
Chabang
0.5432
Thailand
6
Shenzhen
0.7789
China
Shenzhen
0.4705
China
7
Guangzhou
0.7504
China
Hong Kong
0.4507
China
8
Hong Kong
0.6346
China
Malacca
0.4461
Malaysia
9
Ningbo
0.5832
China
Dar es salaam
0.3698
Tanzania
10
Dalian
0.5743
China
Surabaya
0.3653
Indonesia
11
Singapore
0.4986
Singapore
Dalian
0.3423
China
12
Tianjin
0.4705
China
Abu Dhabi
0.3093
United Arab
Emirates
13
Quanzhou
0.4414
China
Kuantan
0.2954
Malaysia
14
Qingdao
0.4211
China
Tianjin
0.2901
China
15
Haikou
0.3806
China
Karachi
0.2568
Pakistan
16
Fuzhou
0.3349
China
Suez
0.2440
Egypt
17
Surabaya
0.3184
Indonesia
Sihanoukville
0.2359
Cambodia
18
Laem
Chabang
0.3040
Thailand
Hai Phong
0.2172
Vietnam
19
Zhanjiang
0.2869
China
Shanghai
0.2152
China
20
Beihai
0.2736
China
Lagos
0.2078
Nigeria
290
4.2 Rail port centrality
291
The local and global centralities of the rail ports in the BRI network are thus evaluated as listed in Table
292
5, and the complete table containing all 76 rail ports are found in Appendix Table A5. The data on the 76
293
rail ports are taken from the Mercator Institute for China Studies (https://www.merics.org/en/bri-tracker).
294
In terms of local centrality, eight of the top 10 rail ports are from China, including Chengdu, Beijing,
295
Hong KongNingbo, Kunming, Wuhan, and Nanjing. Nevertheless, from the perspective of global
296
centrality, the rankings of many ports outside of China are higher, such as Addis Ababa (Ethiopia), and
297
Bangkok (Thailand).
298

17
The rankings of the maritime and rail hubs are found using the great circle distance between the ports
299
and the evaluated airports. These rankings have already considered the integration of the ATN and the
300
maritime network or rail network.
301
Table 5. Top 20 rail ports using gravity model in terms of local and global centralities
302
Rank
Local centrality
Global centrality
Hub
Normalized
centrality
Country
Hub
Normalized
centrality
Country
1
Chengdu
1.0000
China
Addis Ababa
1.0000
Ethiopia
2
Beijing
0.8139
China
Bangkok
0.5377
Thailand
3
Bangkok
0.6748
Thailand
Moscow
0.4597
Russia
4
Hong Kong
0.6714
China
Warsaw
0.3969
Poland
5
Xian
0.6286
China
Tehran
0.3962
Iran
6
Ningbo
0.6170
China
Beijing
0.3616
China
7
Moscow
0.5094
Russia
Chengdu
0.3551
China
8
Kunming
0.5045
China
Nairobi
0.3329
Kenya
9
Wuhan
0.5035
China
Urumqi
0.3194
China
10
Nanjing
0.4731
China
Tashkent
0.2860
Uzbekistan
11
Addis Ababa
0.4569
Ethiopia
Kuala
Lumpur
0.2751
Malaysia
12
Chongqing
0.4478
China
Kunming
0.2683
China
13
Kuala
Lumpur
0.4432
Malaysia
Mashhad
0.2440
Iran
14
Changsha
0.4423
China
Hong Kong
0.1993
China
15
Zhengzhou
0.4248
China
Dar es salaam
0.1891
Tanzania
16
Warsaw
0.3885
Poland
Irkutsk
0.1746
Russia
17
Urumqi
0.3875
China
Xian
0.1638
China
18
Tehran
0.3658
Iran
Novosibirsk
0.1556
Russia
19
Fuzhou
0.3543
China
Chongqing
0.1115
China
20
Hefei
0.3487
China
Hat Yai
0.1055
Thailand
303
5. Ranking of cities based on multimodal transportation networks
304
After obtaining the centralities of the airports, maritime ports, and rail ports, this work maps them to
305
their home cities and assess the criticality of the cities in the BRI multimodal transport network. Doing so
306
allows us to identify the potential cities under the BRI as new transport hubs or to enhance the current ones.
307
In what follows, the ranking of the cities is based on the ranks of their maritime or rail ports. The six
308
identified cities which have both maritime and rail hubs, hold their rank orders for the maritime ports and
309
rail ports.
310
5.1 Ranking of cities having both maritime and rail ports
311
As shown in Table 6, seven cities in the BRI network have a maritime port, rail port, and airport in one
312
location. Based on local centrality, Hong Kong and Ningbo are the top 2 cities. On global centrality, Hong
313

18
Kong and Dar es salaam are the top 2 cities. The ranks of the airports in these cities are however not high,
314
except Hong Kong. Fig. 8 shows the locations of these cities.
315
Table 6. Ranking of cities with maritime and rail ports by local and global centrality
316
Rank
Local centrality
Global centrality
City
Country
Airport Ranking
(out of 1193)
City
Country
Airport Ranking
(out of 1193)
1
Hong Kong
China
24
Hong Kong
China
23
2
Ningbo
China
69
Dar es salaam
Tanzania
57
3
Fuzhou
China
56
Fuzhou
China
144
4
Dar es salaam
Tanzania
86
Ningbo
China
365
5
Djibouti
Djibouti
475
Djibouti
Djibouti
228
6
Mombasa
Kenya
156
Mombasa
Kenya
163
7
Mtwara
Tanzania
924
Mtwara
Tanzania
922
317
318
Fig. 8. Cities with maritime and rail ports
319
5.2 Ranking of cities having both airport and maritime port
320
The cities with maritime ports and lie in the top 100 airports list are ranked and listed in Table 6. As
321
only the top 100 airports of the 1193 airports are analyzed, this explains why the number of cities in terms
322
of local centrality is different from that in terms of global centrality. Our focus is on the hub cities.
323
From Table 7, eight of the top 10 cities by local centrality are from China, which is consistent with
324
Table 4. In contrast, the cities having both maritime port and airport in terms of global centrality are better
325
distributed, albeit with fewer cities (see Fig. 9). In terms of global centrality, thirteen cities have maritime
326
ports and lie in the top 100 airports which are well distributed in the BRI network: Singapore, Jakarta,
327




 









19
Surabaya (Indonesia), Guangzhou, Shenzhen, Hong Kong, Dalian, and Shanghai (China), Dar es salaam
328
(Tanzania), Karachi (Pakistan), Lagos (Nigeria), Casablanca (Morocco), and Vladivostok (Russia).
329
Table 7. Ranking of cities with top 100 airports and maritime port by local and global centralities
330
Rank
Local centrality
Global centrality
City
Country
City
Country
1
Shanghai
China
Singapore
Singapore
2
Xiamen
China
Jakarta
Indonesia
3
Jakarta
Indonesia
Guangzhou
China
4
Shenzhen
China
Shenzhen
China
5
Guangzhou
China
Hong Kong
China
6
Hong Kong
China
Dar es salaam
Tanzania
7
Dalian
China
Surabaya
Indonesia
8
Singapore
Singapore
Dalian
China
9
Tianjin
China
Karachi
Pakistan
10
Quanzhou
China
Shanghai
China
11
Qingdao
China
Lagos
Nigeria
12
Haikou
China
Casablanca
Morocco
13
Fuzhou
China
Vladivostok
Russia
14
Surabaya
Indonesia
-
-
15
Lagos
Nigeria
-
-
16
Dar es salaam
Tanzania
-
-
17
Abu Dhabi
United Arab Emirates
-
-
331
332
Fig. 9. Cities having maritime ports and top 100 airports in BRI network by local and global centralities
333




 

























20
5.3 Ranking of cities with both airport and rail port
334
Table 8 lists the ranks of the cities with rail ports and is part of the top 100 airports. Compared with
335
Table 7, more cities are from outside China. We can find in Fig. 10 that most of the rail ports lie along the
336
New Silk Route Economic Belt.
337
Table 8. Ranking of cities with top 100 airports and rail port by local and global centralities
338
Rank
Local centrality
Global centrality
City
Country
City
Country
1
Chengdu
China
Addis Ababa
Ethiopia
2
Beijing
China
Bangkok
Thailand
3
Bangkok
Thailand
Moscow
Russia
4
Hong Kong
China
Warsaw
Poland
5
Moscow
Russia
Tehran
Iran
6
Kunming
China
Beijing
China
7
Wuhan
China
Chengdu
China
8
Nanjing
China
Nairobi
Kenya
9
Addis Ababa
Ethiopia
Urumqi
China
10
Chongqing
China
Tashkent
Uzbekistan
11
Kuala Lumpur
Malaysia
Kuala Lumpur
Malaysia
12
Zhengzhou
China
Kunming
China
13
Warsaw
Poland
Mashhad
Iran
14
Urumqi
China
Hong Kong
China
15
Fuzhou
China
Dar es salaam
Tanzania
16
Hefei
China
Irkutsk
Russia
17
Harbin
China
Novosibirsk
Russia
18
Lanzhou
China
Chongqing
China
19
Dar es salaam
Tanzania
Wuhan
China
20
Nairobi
Kenya
Lanzhou
China
21
Abuja
Nigeria
Abuja
Nigeria
22
-
-
Harbin
China
339
A summary of observation concerning the above analysis is further presented in a tabular form (Table
340
9) to suggest major prospective multimodal (rail-air, maritime-air, rail-maritime-air) transportation hubs.
341
Table 9. Summary of Analysis suggesting major prospective multimodal hubs
342
Major Hubs (Cities) in the
BRI network
Region
Transportation mode
Rail-Air
Maritime-Air
Rail-Maritime-Air
Hong Kong
China
✓
Ningbo
China
✓
Dar es salaam
Africa
✓
Fuzhou
China
✓
Shanghai
China
✓
Xiamen
China
✓
Singapore
Asia
✓

21
Karachi
Asia
✓
Lagos
Africa
✓
Chengdu
China
✓
Addis Ababa
Africa
✓
Moscow
Russia
✓
Tehran
Middle East
✓
343
344
Fig. 10. Cities with rail ports and in top 100 airports in BRI network by local and global centrality
345
6 Insights and recommendations
346
Two insights on enlarging the BRI transportation network development follow from the above findings.
347
6.1 Developing multimodal hubs
348
The results of this work suggest that several cities offer good multimodal transport options and could
349
act as key hubs or gateways in the BRI network. To validate this observation, this work categorizes the
350
cities as 
351
for a city connected through more than one mode of transport. Some cities have maritime
352
ports, but not airports, with high centrality. These cities, as shown in Table 10, have the top 20 ranked
353
maritime ports but not the top 100 ranked airports. Such cities are good candidates for improving airport
354
centrality.
355
Table 10. Candidates for improving airport centrality for cities with top 20 maritime ports
356




 




















 









22
Rank
Local centrality
Global centrality
City
Country
City
Country
1
Saigon
Vietnam
Saigon
Vietnam
2
Da Nang
Da Nang
Laem Chabang
Thailand
3
Ningbo
China
Malacca
Malaysia
4
Laem Chabang
Thailand
Abu Dhabi
United Arab Emirates
5
Zhanjiang
China
Kuantan
Malaysia
6
Beihai
China
Tianjin
China
7
-
-
Suez
Egypt
8
-
-
Sihanoukville
Cambodia
9
-
-
Hai Phong
Vietnam
Similarly, the cities, as shown in Table 11, have the top 20 ranked rail ports but not the top 100 ranked
357
airports. Such cities are potential candidates for improving the airport centrality.
358
Table 11. Candidates for improving airport centrality for cities with top 20 rail ports
359
Rank
Local centrality
Global centrality
City
Country
City
Country
1
Xian
China
Xian
China
2
Ningbo
China
Hat Yai
Thailand
3
Changsha
China
-
-
4
Tehran
Iran
-
-
Comparing Table 3 with Table A4, the candidate cities for introducing the maritime ports into the BRI
360
network, with top-ranked airports, are found in Table 12. Most of these cities already have maritime ports.
361
Similarly, Table 13 lists the candidate cities for introducing rail ports into the BRI network. However, these
362
ports have not yet been given proper consideration under the BRI context. Therefore, including these ports
363
can reap positive externalities from the BRI, tempered with considerations such as the terrain 
364
relative position in the current maritime or rail network.
365
Table 12. Candidate cities for maritime ports (within top 20 airports) in BRI network
366
Rank
Local centrality
Global centrality
City
Country
City
Country
1
Istanbul
Turkey
Athens
Greece
2
Hangzhou
China
Ujung Pandang
Indonesia
3
Ho Chi Minh City
Vietnam
-
-
4
Dubai
United Arab Emirates
-
-
Table 13. Candidate cities for rail ports (within top 20 airports) in BRI network
367
Rank
Local centrality
Global centrality
City
Country
City
Country
1
Guangzhou
China
Dubai
The United Arab
Emirates
2
Shenzhen
China
Kathmandu
Nepal

23
3
Shanghai
China
Johannesburg
South Africa
4
Xi'an
China
Athens
Greece
5
Istanbul
Turkey
Cairo
Egypt
6
Hangzhou
China
-
-
7
Ho Chi Minh City
Vietnam
-
-
368
6.2 Improving airport centrality of current multimodal cities
369
From Section 5.1, seven cities (Hong Kong, Ningbo, Fuzhou, Djibouti, Mombasa, Dar es salaam, and
370
Mtwara) have all three types of ports. However, the centrality of the airports belonging to these cities is not
371
high, except Hong Kong. By increasing the centrality of these airports, the integration between the modes
372
of transportation networks can be enhanced. For example, BRI policymakers may consider launching more
373
direct flights between NGB (Ningbo) and other airports, especially those airports outside of China, noting
374
that the NGB is not as well connected to the other countries in the BRI network. Conversely, Dar es salaam
375
(Tanzania) has high global centrality as compared to its local centrality implying that its domestic
376
connectivity is low. Hence, Dar es salaam can be promoted by introducing more direct flights to the other
377
African cities, thereby enlarging the transportation network of BRI.
378
7 Conclusion
379
This paper presented a study on the structural analysis of the multimodal transportation network in the BRI
380
context, with a focus on the ATN. This paper has proposed a hybrid method by combining the weighted
381
centrality measures and a gravity model for multimodal transportation network analysis. This method is
382
then applied to the BRI network to evaluate the rankings of the current multimodal hubs and identify the
383
cities that have the potential to become multimodal transport hubs. The hub centrality analysis is conducted
384
from two perspectives, i.e., local centrality and global centrality. The first step involved measuring the
385
centrality of the BRI ATN involving 1193 airports in 98 countries. The centrality analysis of the BRI ATN
386
suggests top 100 transit airports (based on degree centrality for local connectivity) and hub airports (based
387
on betweenness centrality for global connectivity). The gravity model is applied to rank the maritime and
388
rail ports of the BRI network based on the great circle distance to the airport network. The analysis suggests
389
some major hubs (cities in the BRI network) that have the potential to become multimodal (rail-air, air-
390
maritime, maritime-rail, or rail-air-maritime) transportation gateways to improve the connectivity of the
391
BRI network. Several multimodal transportation hub ports could be established to increase the international
392
trade outreach of the BRI. Given the level of development in Africa, it is useful to develop the transportation
393
hub ports in Africa to lift the connectivity of the BRI for better international trade efficiency.
394
This paper has attempted to provide evidence using a multi-data analysis to validate the effectiveness
395
of the proposed model and thus has presented its practical significance. Importantly, this work can generate
396

24
further discussion on multimodal transportation hub locations in not only the BRI network but also the
397
global transportation network. Future studies could include more practical factors that can influence
398
network analysis decision makings, such as regional trading blocs, geopolitical disputes, political conflicts
399
and the recent COVID-19 pandemic-induced disruptions. The post COVID-19 economic operational plan
400
may have some significant implications to the BRI transportation infrastructure projects and thus demands
401
further research. Besides, various constraints such as capacity, frequency and other operational parameters
402
(such a customs procedure, volume and price) of rail, maritime and air transport modes should also be
403
considered in the model development to justify an efficient integrated multimodal transportation systems
404
in the BRI strategic context. Finally, the infrastructural health of the primary maritime ports, rail ports and
405
airports could also be incorporated in the model development and thus warrants further research.
406
Appendix
407
Table A1. List of 98 countries studied in BRI network
408
Afghanistan
Albania
Algeria
Angola
Armenia
Austria
Azerbaijan
Bahrain
Bangladesh
Belarus
Bosnia
Bulgaria
Burundi
Cambodia
Cameroon
Chad
China
Congo
Côte d'Ivoire
Croatia
Czech
Republic
Djibouti
Egypt
Equatorial
Guinea
Estonia
Ethiopia
Gabon
Gambia
Georgia
Ghana
Greece
Guinea
Hungary
Indonesia
Iran
Iraq
Israel
Jordan
Kazkhtstan
Kenya
Krygystan
Kuwait
Lao PDR
Latvia
Lebanon
Lesotho
Liberia
Libya
Lithuania
Luxembourg
Malaysia
Mali
Mauritiiana
Moldova
Mongolia
Montenegro
Morocco
Mozambique
Myanmar
Namibia
Nepal
Niger
Nigeria
North
Macedonia
Oman
Pakistan
Poland
Portugal
Qatar
Romania
Russia
Rwanda
Saudi
Arabia
Senegal
Serbia
Sierra
Leone
Singapore
Slovakia
Slovenia
Somalia
South
Africa
South Sudan
Sri Lanka
Sudan
Tajikistan
Tanzania
Thailand
Togo
Tunisia
Turkey
Turkmenistan
Uganda
Ukraine
United Arab
Emirates
Uzbekistan
Vietnam
Zambia
Zimbabwe
409
Table A2. Top 100 airports in terms of global and local centrality
410
Rank
Local centrality
Global centrality
IATA code (City)
Degree
Country
IATA code (City)
Betweenness
Country
1
IST (Istanbul)
189
Turkey
IST (Istanbul)
128992
Turkey
2
PEK (Beijing)
164
China
DXB (Dubai)
89295
United Arab
Emirates
3
CTU (Chengdu)
159
China
CGK (Jakarta)
49193
Indonesia
4
XIY (Xi'an)
150
China
PEK (Beijing)
46328
China
5
CAN (Guangzhou)
146
China
SVO (Moscow)
43492
Russia
6
PVG (Shanghai)
138
China
ADD (Addis Ababa)
39332
Ethiopia
7
KMG (Kunming)
134
China
ATH (Athens)
38990
Greece
8
DXB (Dubai)
131
United Arab
Emirates
DOH (Doha)
37356
Qatar
9
SZX (Shenzhen)
131
China
CTU (Chengdu)
31413
China
10
CKG (Chongqing)
130
China
DME (Moscow)
31202
Russia
11
SVO (Moscow)
121
Russia
SIN (Singapore)
29578
Singapore

25
12
BKK (Bangkok)
108
Thailand
KUL (Kuala Lumpur)
29100
Malaysia
13
DME (Moscow)
107
Russia
JNB (Johannesburg)
26912
South Africa
14
HGH (Hangzhou)
104
China
CAN (Guangzhou)
25644
China
15
VKO (Moscow)
102
Russia
JED (Jeddah)
25484
Saudi Arabia
16
LED (St.
Petersburg)
100
Russia
ALG (Algier)
23497
Algeria
17
CSX (Changcha)
98
China
BKK (Bangkok)
22764
Thailand
18
TSN (Tianjin)
98
China
THR (Teheran)
20895
Iran
19
HAK (Haikou)
95
China
LED (St. Petersburg)
20660
Russia
20
KUL (Kuala
Lumpur)
95
Malaysia
VKO (Moscow)
19981
Russia
21
NKG (Nanjing)
93
China
YKS (Yakutsk)
19619
Russia
22
JED (Jeddah)
92
Saudi Arabia
PVG (Shanghai)
19452
China
23
PKX (Beijing)
92
China
MHD (Mashhad)
18581
Iran
24
CGO (Zhengzhou)
91
China
KTM (Kathmandu)
18493
Nepal
25
KWE (Guiyang)
91
China
DMK (Bangkok)
18062
Thailand
26
XMN (Xiamen)
89
China
XIY (Xi'an)
17116
China
27
DOH (Doha)
88
Qatar
NBO (Nairobi)
16886
Kenya
28
WUH (Wuhan)
88
China
DPS (Denpasar)
16206
Indonesia
29
NNG (Nanning)
85
China
VIE (Vienna)
16101
Austria
30
HRB (Harbin)
83
China
CMN (Casablanca)
15675
Morocco
31
SYX (Sanya)
82
China
CAI (Cairo)
14487
Egypt
32
TAO (Qingdao)
82
China
LIS (Lisbon)
14146
Portugal
33
LHW (Lanzhou)
80
China
LAD (Luanda)
13969
Angola
34
SIN (Singapore)
80
Singapore
CPT (Cape Town)
13005
South Africa
35
SAW (Istanbul)
78
Turkey
SZX (Shenzhen)
12693
China
36
HKG (Hong Kong)
77
China
HKG (Hong Kong)
12684
China
37
DMK (Bangkok)
75
Thailand
KMG (Kunming)
11515
China
38
VIE (Vienna)
75
Austria
UPG (Ujung Pandang)
11278
Indonesia
39
ATH (Athens)
74
Greece
SYZ (Shiraz)
10728
Iran
40
CGK (Jakarta)
74
Indonesia
DAR (Dar es salaam)
10672
Tanzania
41
SHE (Shenyang)
74
China
HRB (Harbin)
10550
China
42
TNA (Jinan)
74
China
KBL (Kabul)
10408
Afghanistan
43
ADD (Addis Ababa)
72
Ethiopia
WAW (Warsaw)
10280
Poland
44
DLC (Dalian)
72
China
URC (Urumqi)
10224
China
45
FOC (Fuzhou)
71
China
SGN (HCM City)
10148
Vietnam
46
OVB (Novosibirsk)
70
Russia
LOS (Lagos)
10043
Nigeria
47
URC (Urumqi)
68
China
ABV (Abuja)
9752
Nigeria
48
TPE (Taipei)
67
Taiwan
TLV (Tel-aviv)
9649
Israel
49
CAI (Cairo)
66
Egypt
CMB (Colombo)
9639
Sri Lanka
50
HKT (Phuket)
63
Thailand
CKG (Chongqing)
9278
China
51
KHN (Nanchang)
63
China
IKT (Irkutsk)
8616
Russia
52
NGB (Ninbo)
63
China
MCT (Muscat)
8559
Oman
53
TYN (Taiyuan)
63
China
MED (Madinah)
8346
Saudi Arabia
54
SHA (Shanghai)
61
China
HKT (Phuket)
8024
Thailand
55
SHJ (Sharjah)
61
United Arab
Emirates
OVB (Novosibirsk)
7523
Russia
56
SVX
(Yekaterinburg)
61
Russia
KRT (Khartoum)
7403
Sudan
57
YNT (Yantai)
61
China
TAS (Tashkent)
7375
Uzbekistan
58
WAW (Warsaw)
60
Poland
MYY (Miri)
7228
Malaysia
59
WNZ (Wenzhou)
60
China
RUH (Riyadh)
6752
Saudi Arabia
60
HET (Hohhot)
59
China
SHJ (Sharjah)
6694
United Arab
Emirates
61
SJW (Shijiazhuang)
59
China
SVX (Yekaterinburg)
6616
Russia
62
TLV (Tel-aviv)
59
Israel
DAC (Dhaka)
6548
Bangladesh
63
CXR (Nha Trang)
58
Vietnam
BKI (Kota Kinabalu)
6477
Malaysia
64
HFE (Hefei)
58
China
SUB (Surabaya)
6361
Indonesia
65
KWL (Guilin)
56
China
ZNZ (Zanzibar)
6151
Tanzania

26
66
MED (Madinah)
56
Saudi Arabia
WUH (Wuhan)
6145
China
67
RUH (Riyadh)
56
Saudi Arabia
HAN (Hanoi)
5977
Vietnam
68
ZUH (Zhuhai)
56
China
IFN (Esfahan)
5957
Iran
69
JNB (Johannesburg)
54
South Africa
OMD (Oranjemund)
5890
Namibia
70
KJA (Krasnoyarsk)
52
Russia
KJA (Krasnoyarsk)
5854
Russia
71
SGN (HCM City)
52
Vietnam
SAW (Istanbul)
5820
Turkey
72
CMN (Casablanca)
51
Morocco
AMQ (Ambon)
5702
Indonesia
73
DPS (Denpasar)
51
Indonesia
MDC (Manado)
5621
Indonesia
74
JJN (Quanzhou)
50
China
OTP (Bucharest)
5617
Romania
75
ESB (Ankara)
48
Turkey
ISB (Islamabad)
5512
Pakistan
76
KZN (Kazan)
48
Russia
AUH (Abu Dhabi)
5476
United Arab
Emirates
77
SWA (Shantou)
48
China
SYX (Sanya)
5323
China
78
KBP (Kiev)
47
Ukraine
KNO (Medan)
5106
Indonesia
79
NBO (Nairobi)
47
Kenya
DJJ (Jayapura)
5103
Indonesia
80
TAS (Tashkent)
47
Uzbekistan
WIL (Nairobi)
5073
Kenya
81
CGQ (Changchun)
46
China
KBP (Kiev)
5059
Ukraine
82
KRR (Krasnodar)
46
Russia
JRO (Kilimanjaro)
4985
Tanzania
83
MCT (Muscat)
46
Oman
CSX (Changcha)
4974
China
84
AMM (Amman)
44
Jordan
HGH (Hangzhou)
4791
China
85
GYD (Baku)
44
Azerbaijan
NDJ (N'djamena)
4786
Chad
86
LJG (Lijiang)
44
China
SCW (Syktyvkar)
4777
Russia
87
MFM (Macau)
44
Macau
UUS (Yuzhno-
sakhalinsk)
4729
Russia
88
ALG (Algier)
43
Algeria
MPM (Maputo)
4528
Mozambique
89
AUH (Abu Dhabi)
43
United Arab
Emirates
KHI (Karachi)
4416
Pakistan
90
IKT (Irkutsk)
43
Russia
ZAG (Zagreb)
4357
Croatia
91
PRG (Prague)
43
Czech Republic
ACC (Accra)
4337
Ghana
92
SUB (Surabaya)
43
Indonesia
TSA (Taipei)
4294
Taiwan
93
AYT (Antalya)
42
Turkey
SKG (Thessaloniki)
4194
Greece
94
HAN (Hanoi)
41
Vietnam
EBB (Entebbe)
4022
Uganda
95
KWI (Kuwait)
40
Kuwait
PRG (Prague)
4000
Czech Republic
96
THR (Teheran)
40
Iran
XNN (Xining)
3895
China
97
BUD (Budapest)
39
Hungary
CXR (Nha Trang)
3829
Vietnam
98
MSQ (Minsk 2)
39
Belarus
PKX (Beijing)
3781
China
99
UPG (Ujung
Pandang)
39
Indonesia
MBA (Mombasa)
3749
Kenya
100
BAH (Bahrain)
38
Bahrain
TJM (Tyumen)
3740
Russia
411
Table A3. Top 100 airports in terms of global and local centrality in the weighted BRI ATN
412
Rank
Local centrality
Global centrality
IATA code (City)
Weighted
degree
Country
IATA code (City)
Weighted
betweenness
Country
1
PEK (Beijing)
4816
China
DXB (Dubai)
1498738
United Arab
Emirates
2
CGK (Jakarta)
4385
Indonesia
IST (Istanbul)
1345114
Turkey
3
CAN (Guangzhou)
4278
China
BKK (Bangkok)
1021471
Thailand
4
KUL (Kuala
Lumpur)
3303
Malaysia
KUL (Kuala
Lumpur)
931838
Malaysia
5
CTU (Chengdu)
3276
China
SVO (Moscow)
887144
Russia
6
SZX (Shenzhen)
3258
China
PEK (Beijing)
835737
China
7
PVG (Shanghai)
3090
China
ADD (Addis
Ababa)
627801
Ethiopia
8
XIY (Xi'an)
3042
China
KTM (Kathmandu)
613484
Nepal
9
CKG (Chongqing)
2979
China
SIN (Singapore)
605307
Singapore
10
KMG (Kunming)
2886
China
KMG (Kunming)
561946
China
11
IST (Istanbul)
2839
Turkey
CAN (Guangzhou)
503341
China

27
12
SHA (Shanghai)
2593
China
JNB
(Johannesburg)
470314
South Africa
13
BKK (Bangkok)
2495
Thailand
URC (Urumqi)
423810
China
14
SVO (Moscow)
2425
Russia
CGK (Jakarta)
423801
Indonesia
15
HGH (Hangzhou)
2387
China
NBO (Nairobi)
406891
Kenya
16
SIN (Singapore)
2310
Singapore
ATH (Athens)
390666
Greece
17
DMK (Bangkok)
2223
Thailand
CAI (Cairo)
363431
Egypt
18
SGN (Ho Chi Minh
City)
2194
Vietnam
CTU (Chengdu)
351784
China
19
DXB (Dubai)
2161
United Arab
Emirates
DMK (Bangkok)
349148
Thailand
20
NKG (Nanjing)
1994
China
UPG (Ujung
Pandang)
307774
Indonesia
21
JED (Jeddah)
1947
Saudi Arabia
VIE (Vienna)
289750
Austria
22
CGO (Zhengzhou)
1846
China
WAW (Warsaw)
288871
Poland
23
WUH (Wuhan)
1829
China
HKG (Hong Kong)
278647
China
24
HKG (Hong Kong)
1759
China
MCT (Muscat)
260633
Oman
25
RUH (Riyadh)
1731
Saudi Arabia
THR (Teheran)
255191
Iran
26
XMN (Xiamen)
1708
China
SGN (Ho Chi Minh
City)
247584
Vietnam
27
SAW (Istanbul)
1698
Turkey
PNK (Pontianak)
242425
Indonesia
28
JNB
(Johannesburg)
1656
South Africa
DME (Moscow)
238115
Russia
29
HAN (Hanoi)
1636
Vietnam
KWI (Kuwait)
232508
Kuwait
30
CSX (Changcha)
1624
China
TLV (Tel-aviv)
226612
Israel
31
HAK (Haikou)
1596
China
SAW (Istanbul)
225106
Turkey
32
KWE (Guiyang)
1541
China
VKO (Moscow)
214349
Russia
33
SUB (Surabaya)
1523
Indonesia
SUB (Surabaya)
212255
Indonesia
34
URC (Urumqi)
1494
China
XIY (Xi'an)
205087
China
35
TAO (Qingdao)
1468
China
OVB (Novosibirsk)
199884
Russia
36
DME (Moscow)
1397
Russia
DAC (Dhaka)
198241
Bangladesh
37
DPS (Denpasar)
1351
Indonesia
ALG (Algier)
197054
Algeria
38
SYX (Sanya)
1309
China
DOH (Doha)
196819
Qatar
39
HRB (Harbin)
1283
China
RUH (Riyadh)
188574
Saudi Arabia
40
DOH (Doha)
1258
Qatar
SHJ (Sharjah)
177955
United Arab
Emirates
41
SHE (Shenyang)
1213
China
HET (Hohhot)
177565
China
42
TSN (Tianjin)
1200
China
BGW (Baghdad)
177265
Iraq
43
TPE (Taipei)
1178
Taiwan
BKI (Kota
Kinabalu)
173075
Malaysia
44
DLC (Dalian)
1176
China
OTP (Bucharest)
164485
Romania
45
UPG (Ujung
Pandang)
1175
Indonesia
KRT (Khartoum)
158123
Sudan
46
VKO (Moscow)
1142
Russia
DPS (Denpasar)
157127
Indonesia
47
CAI (Cairo)
1132
Egypt
LOS (Lagos)
152869
Nigeria
48
LED (St.
Petersburg)
1112
Russia
HKT (Phuket)
151373
Thailand
49
NNG (Nanning)
1061
China
TAS (Tashkent)
149106
Uzbekistan
50
TNA (Jinan)
1060
China
SYZ (Shiraz)
148959
Iran
51
LHW (Lanzhou)
1053
China
BEY (Beirut)
144378
Lebanon
52
HKT (Phuket)
1036
Thailand
IKA (Tehran)
143534
Iran
53
ATH (Athens)
978
Greece
JED (Jeddah)
142887
Saudi Arabia
54
KHN (Nanchang)
941
China
MHD (Mashhad)
141564
Iran
55
KTM (Kathmandu)
941
Nepal
ESB (Ankara)
140141
Turkey
56
TYN (Taiyuan)
932
China
LIS (Lisbon)
138517
Portugal
57
FOC (Fuzhou)
925
China
DAR (Dar es
salaam)
137974
Tanzania
58
CGQ (Changchun)
909
China
PEN (Penang)
134983
Malaysia
59
VIE (Vienna)
907
Austria
BPN (Balikpapan)
130856
Indonesia
60
HFE (Hefei)
892
China
CMN (Casablanca)
127923
Morocco

28
61
PKX (Beijing)
890
China
GYD (Baku)
125988
Azerbaijan
62
WAW (Warsaw)
890
Poland
JRO (Kilimanjaro)
124782
Tanzania
63
KWI (Kuwait)
879
Kuwait
PVG (Shanghai)
121114
China
64
HET (Hohhot)
871
China
YKS (Yakutsk)
119735
Russia
65
WNZ (Wenzhou)
858
China
KCH (Kuching)
115571
Malaysia
66
ADD (Addis
Ababa)
852
Ethiopia
BTH (Batam)
113850
Indonesia
67
MCT (Muscat)
800
Oman
PKU (Pekanbaru)
111256
Indonesia
68
THR (Teheran)
800
Iran
ABV (Abuja)
110971
Nigeria
69
NGB (Ninbo)
791
China
HAN (Hanoi)
109663
Vietnam
70
NBO (Nairobi)
789
Kenya
MDC (Manado)
107958
Indonesia
71
SJW
(Shijiazhuang)
783
China
CMB (Colombo)
106647
Sri Lanka
72
DAC (Dhaka)
778
Bangladesh
CKG (Chongqing)
106615
China
73
CPT (Cape Town)
768
South Africa
IKT (Irkutsk)
105160
Russia
74
ZUH (Zhuhai)
764
China
ZNZ (Zanzibar)
99871
Tanzania
75
YNT (Yantai)
753
China
KNO (Medan)
98554
Indonesia
76
CNX (Chiang Mai)
749
Thailand
AMM (Amman)
96889
Jordan
77
PEN (Penang)
730
Malaysia
LHW (Lanzhou)
92655
China
78
DMM (Dammam)
727
Saudi Arabia
MSQ (Minsk 2)
91143
Belarus
79
ESB (Ankara)
724
Turkey
ISB (Islamabad)
90385
Pakistan
80
DAD (Danang)
704
Vietnam
LAD (Luanda)
89120
Angola
81
KNO (Medan)
680
Indonesia
TSA (Taipei)
88792
Taiwan
82
LOS (Lagos)
673
Nigeria
MYY (Miri)
87217
Malaysia
83
TLV (Tel-aviv)
670
Israel
REP (Siem-reap)
83803
Cambodia
84
BKI (Kota
Kinabalu)
640
Malaysia
KWE (Guiyang)
83096
China
85
BPN (Balikpapan)
631
Indonesia
BDJ (Banjarmasin)
83022
Indonesia
86
DAR (Dar es
salaam)
613
Tanzania
KBP (Kiev)
81757
Ukraine
87
BAH (Bahrain)
607
Bahrain
DLC (Dalian)
81365
China
88
JJN (Quanzhou)
597
China
LXA (Lhasa)
76997
China
89
JOG (Yogyakarta)
590
Indonesia
KBL (Kabul)
76903
Afghanistan
90
MFM (Macau)
572
Macau
WUH (Wuhan)
71702
China
91
AUH (Abu Dhabi)
564
United Arab
Emirates
SZX (Shenzhen)
70024
China
92
HLP (Jakarta)
554
Indonesia
KHH (Kaohsiung)
68186
Taiwan
93
ADB (Izmir)
553
Turkey
LED (St.
Petersburg)
68085
Russia
94
SHJ (Sharjah)
551
United Arab
Emirates
KHI (Karachi)
67366
Pakistan
95
SZB (Kuala
Lumpur)
551
Malaysia
TUN (Tunis)
67307
Tunisia
96
ZNZ (Zanzibar)
537
Tanzania
HRB (Harbin)
66074
China
97
AMM (Amman)
534
Jordan
DJJ (Jayapura)
66047
Indonesia
98
WUX (Wuxi)
534
China
VVO
(Vladivostok)
65748
Russia
99
ABV (Abuja)
526
Nigeria
XNN (Xining)
65649
China
100
PNH (Phnom penh)
521
Cambodia
ASB (Ashkhabad)
65375
Turkmenistan
413
Table A4. Ranking of maritime ports by gravity model on global and local centrality
414
Rank
Local centrality
Global centrality
Hub
Normalized
centrality
Country
Hub
Normalized
centrality
Country
1
Shanghai
1.0000
China
Singapore
1.0000
Singapore
2
Saigon
0.8449
Vietnam
Saigon
0.8026
Vietnam
3
Xiamen
0.8099
China
Jakarta
0.5831
Indonesia

29
4
Da Nang
0.7983
Vietnam
Guangzhou
0.5687
China
5
Jakarta
0.7888
Indonesia
Laem Chabang
0.5432
Thailand
6
Shenzhen
0.7789
China
Shenzhen
0.4705
China
7
Guangzhou
0.7504
China
Hong Kong
0.4507
China
8
Hong Kong
0.6346
China
Malacca
0.4461
Malaysia
9
Ningbo
0.5832
China
Dar es salaam
0.3698
Tanzania
10
Dalian
0.5743
China
Surabaya
0.3653
Indonesia
11
Singapore
0.4986
Singapore
Dalian
0.3423
China
12
Tianjin
0.4705
China
Abu Dhabi
0.3093
United Arab
Emirates
13
Quanzhou
0.4414
China
Kuantan
0.2954
Malaysia
14
Qingdao
0.4211
China
Tianjin
0.2901
China
15
Haikou
0.3806
China
Karachi
0.2568
Pakistan
16
Fuzhou
0.3349
China
Suez
0.2440
Egypt
17
Surabaya
0.3184
Indonesia
Sihanoukville
0.2359
Cambodia
18
Laem Chabang
0.3040
Thailand
Hai Phong
0.2172
Vietnam
19
Zhanjiang
0.2869
China
Shanghai
0.2152
China
20
Beihai
0.2736
China
Lagos
0.2078
Nigeria
21
Hai Phong
0.2669
Vietnam
Zhanjiang
0.2050
China
22
Malacca
0.2608
Malaysia
Beihai
0.2025
China
23
Lagos
0.2100
Nigeria
Yuzhne
0.1995
Ukraine
24
Dar es salaam
0.2069
Tanzania
Haikou
0.1965
China
25
Kuantan
0.1950
Malaysia
Ashdad
0.1939
Canada
26
Sihanoukville
0.1895
Cambodia
Da Nang
0.1929
Vietnam
27
Abu Dhabi
0.1455
United Arab
Emirates
Ambarli
0.1897
Turkey
28
Sittwe
0.1415
Burma
Gwadar
0.1884
Pakistan
29
Vladivostok
0.1305
Russia
Casablanca
0.1883
Morocco
30
Suez
0.1159
Egypt
Sittwe
0.1757
Burma
31
Ashdad
0.1076
Canada
Xiamen
0.1734
China
32
Gwadar
0.0984
Pakistan
Quanzhou
0.1689
China
33
Yuzhne
0.0956
Ukraine
Fuzhou
0.1634
China
34
Ambarli
0.0940
Turkey
Qingdao
0.1580
China
35
Karachi
0.0921
Pakistan
Ningbo
0.1580
China
36
Hambantota
0.0893
Sri Lanka
Bagamoyo
0.1568
Tanzania
37
Bagamoyo
0.0851
Tanzania
Klaipeda
0.1529
Lithuania
38
Klaipeda
0.0807
Lithuania
Massawa
0.1470
Eritrea
39
Massawa
0.0785
Eritrea
Cherchell
0.1443
Algeria
40
Aden
0.0762
Yemen
Djibouti
0.1441
Djibouti
41
Djibouti
0.0745
Djibouti
Vladivostok
0.1415
Russia
42
Venice
0.0706
Italy
Aden
0.1415
Yemen
43
Mombasa
0.0685
Kenya
Venice
0.1411
Italy
44
Malta
0.0668
Malta
Hambantota
0.1362
Sri Lanka
45
Lamu
0.0652
Kenya
Malta
0.1357
Malta
46
Rotterdam
0.0600
Netherlands
Mombasa
0.1353
Kenya
47
Antwerp
0.0599
Belgium
Lamu
0.1275
Kenya
48
Marseille
0.0590
France
Marseille
0.1168
France
49
Mtwara
0.0585
Tanzania
Antwerp
0.1117
Belgium
50
Dunkirk
0.0579
United States
Rotterdam
0.1113
Netherlands

30
51
Montoire-de-
Bretgne
0.0570
France
Montoire-de-
Bretgne
0.1083
France
52
Le Havre
0.0556
France
Valencia
0.1078
Spain
53
Cherchell
0.0530
Algeria
Dunkirk
0.1073
United States
54
Valencia
0.0521
Spain
Le Havre
0.1034
France
55
Maputo
0.0517
Mozambique
Mtwara
0.1032
Tanzania
56
Bilbao
0.0515
Spain
Tanger
0.1006
Morocco
57
Ndiago
0.0513
Mauritania
Bilbao
0.1004
Spain
58
Vado
0.0502
Italy
Vado
0.0984
Italy
59
Beira
0.0495
Mozambique
Ndiago
0.0913
Mauritania
60
Lome
0.0489
Togo
Maputo
0.0882
Mozambique
61
Tanger
0.0470
Morocco
Lome
0.0865
Togo
62
Kribi
0.0469
Cameroun
Beira
0.0837
Mozambique
63
Casablanca
0.0453
Morocco
Kribi
0.0836
Cameroun
64
Tema
0.0452
Ghana
Libreville
0.0803
Gabon
65
Libreville
0.0452
Gabon
Tema
0.0796
Ghana
66
Aboadze
0.0426
Ghana
Aboadze
0.0746
Ghana
67
Walvis Bay
0.0408
Namibia
Abijan
0.0709
Côte d'Ivoire
68
Abijan
0.0406
Côte d'Ivoire
Walvis Bay
0.0674
Namibia
69
Noukchott
0.0373
Mauritania
Noukchott
0.0656
Mauritania
70
Conakry
0.0364
Guinea
Conakry
0.0630
Guinea
71
Piraeus
0.0316
Greece
Piraeus
0.0441
Greece
72
Sao Tome
0.0257
Sao Tome and
Principe
Sao Tome
0.0411
Sao Tome and
Principe
415
Table A5. Ranking of rail ports by gravity model in terms of global and local centrality
416
Rank
Local centrality
Global centrality
Hub
Normalized centrality
Country
Hub
Normalized centrality
Country
1
Chengdu
1.0000
China
Addis Ababa
1.0000
Ethiopia
2
Beijing
0.8139
China
Bangkok
0.5377
Thailand
3
Bangkok
0.6748
Thailand
Moscow
0.4597
Russia
4
Hong Kong
0.6714
China
Warsaw
0.3969
Poland
5
Xian
0.6286
China
Tehran
0.3962
Iran
6
Ningbo
0.6170
China
Beijing
0.3616
China
7
Moscow
0.5094
Russia
Chengdu
0.3551
China
8
Kunming
0.5045
China
Nairobi
0.3329
Kenya
9
Wuhan
0.5035
China
Urumqi
0.3194
China
10
Nanjing
0.4731
China
Tashkent
0.2860
Uzbekistan
11
Addis Ababa
0.4569
Ethiopia
Kuala Lumpur
0.2751
Malaysia
12
Chongqing
0.4478
China
Kunming
0.2683
China
13
Kuala Lumpur
0.4432
Malaysia
Mashhad
0.2440
Iran
14
Changsha
0.4423
China
Hong Kong
0.1993
China
15
Zhengzhou
0.4248
China
Dar es salaam
0.1891
Tanzania
16
Warsaw
0.3885
Poland
Irkutsk
0.1746
Russia
17
Urumqi
0.3875
China
Xian
0.1638
China
18
Tehran
0.3658
Iran
Novosibirsk
0.1556
Russia
19
Fuzhou
0.3543
China
Chongqing
0.1115
China

31
20
Hefei
0.3487
China
Hat Yai
0.1055
Thailand
21
Jingdezhen
0.3304
China
Minsk
0.1027
Belarus
22
Lianyungang
0.3023
China
Wuhan
0.1024
China
23
Harbin
0.2643
China
Vientiane
0.0956
Laos
24
Lanzhou
0.2534
China
Isfahan
0.0955
Iran
25
Dar es salaam
0.2530
Tanzania
Sofia
0.0933
Bulgaria
26
Nairobi
0.2274
Kenya
Lanzhou
0.0931
China
27
Vientiane
0.2094
Laos
Naivasha
0.0914
Kenya
28
Erenhot
0.1887
China
Sivas
0.0882
Turkey
29
Hat Yai
0.1858
Thailand
Mecca
0.0845
Saudi Arabia
30
Mecca
0.1540
Saudi Arabia
Changsha
0.0819
China
31
Hunchun
0.1485
China
Kars
0.0796
Turkey
32
Ulan bator
0.1370
Mongolia
Tbilisi
0.0781
Georgia
33
Irkutsk
0.1146
Russia
Zhengzhou
0.0774
China
34
Alashankou
0.1112
China
Abuja
0.0763
Nigeria
35
Isfahan
0.1102
Iran
Medina
0.0756
United States
36
Medina
0.1055
United States
Aktau port
0.0739
Kazakhstan
37
Abuja
0.1053
Nigeria
Hefei
0.0738
China
38
Sivas
0.1048
Turkey
Khargos
0.0738
China
39
Almaty
0.1020
Kazakhstan
Jingdezhen
0.0734
China
40
Sofia
0.1018
Bulgaria
Fuzhou
0.0723
China
41
Khargos
0.0983
China
Harbin
0.0709
China
42
Krasnoyarsk
0.0972
Russia
Erenhot
0.0708
China
43
Mashhad
0.0972
Iran
Nanjing
0.0705
China
44
Kars
0.0962
Turkey
Lianyungang
0.0702
China
45
Minsk
0.0961
Belarus
Ningbo
0.0699
China
46
Tashkent
0.0957
Uzbekistan
Kazan
0.0676
Russia
47
Tbilisi
0.0949
Georgia
Alashankou
0.0670
China
48
Novosibirsk
0.0933
Russia
Almaty
0.0648
Kazakhstan
49
Kazan
0.0925
Russia
Djibouti
0.0637
Djibouti
50
Aktau port
0.0923
Kazakhstan
Berlin
0.0613
Germany
51
Uzen
0.0916
Kazakhstan
Nuremburg
0.0601
Germany
52
Astana
0.0911
Kazakhstan
Mombasa
0.0598
Kenya
53
Naivasha
0.0867
Kenya
Ulan bator
0.0593
Mongolia
54
Yekaterinburg
0.0851
Russia
Astana
0.0593
Kazakhstan
55
Djibouti
0.0788
Djibouti
Yekaterinburg
0.0576
Russia
56
Berlin
0.0763
Germany
Uzen
0.0558
Kazakhstan
57
Nuremburg
0.0738
Germany
Hamburg
0.0547
Germany
58
Mombasa
0.0725
Kenya
Strasbourg
0.0543
France
59
Hamburg
0.0700
Germany
Krasnoyarsk
0.0537
Russia
60
Strasbourg
0.0676
France
Dodoma
0.0531
Tanzania
61
Dodoma
0.0654
Tanzania
Hunchun
0.0527
China
62
Juba
0.0638
South Sudan
Juba
0.0520
South Sudan
63
Mtwara
0.0619
Tanzania
Tabora
0.0488
Tanzania
64
Paris
0.0609
France
Paris
0.0479
France
65
Tabora
0.0607
Tanzania
Mtwara
0.0456
Tanzania
66
Njombe
0.0598
Tanzania
Njombe
0.0455
Tanzania
67
London
0.0590
Canada
Kano
0.0454
Nigeria

32
68
Kano
0.0583
Nigeria
London
0.0452
Canada
69
Bujumbura
0.0574
Burundi
Bujumbura
0.0450
Burundi
70
Madrid
0.0529
Spain
Madrid
0.0447
Spain
71
Kapiri mposhi
0.0518
Zambia
Kapiri mposhi
0.0378
Zambia
72
Luau
0.0493
United States
Luau
0.0364
United States
73
Lobito
0.0445
Angola
Lobito
0.0335
Angola
74
Bamako
0.0420
Mali
Bamako
0.0310
Mali
75
kigoma
0.0389
Tanzania
kigoma
0.0278
Tanzania
76
Dakar
0.0381
Senegal
Dakar
0.0276
Senegal
417
References
418
Bhaskaran, S., Turnquist, M.A., 1990. Multiobjective transportation considerations in multiple facility
419
location. Transp. Res. Part A Gen. 24, 139148.
420
Chhetri, P., Nkhoma, M., Peszynski, K., Chhetri, A., Lee, P.T.W., 2018. Global logistics city concept: a
421
cluster-led strategy under the belt and road initiative. Marit. Policy Manag. 45, 319335.
422
https://doi.org/10.1080/03088839.2017.1400700
423
Derudder, B., Liu, X., Kunaka, C., 2018. Connectivity Along Overland Corridors of the Belt and Road
424
Initiative, Connectivity Along Overland Corridors of the Belt and Road Initiative. World Bank.
425
https://doi.org/10.1596/30609
426
Freeman, L.C., 1977. A set of measures of centrality based on betweenness. Sociometry 3541.
427
Guimerà, R., Mossa, S., Turtschi, A., Amaral, L.A.N., 2005. The worldwide air transportation network:
428

429
102, 77947799. https://doi.org/10.1073/pnas.0407994102
430
Jiang, Y., Sheu, J.B., Peng, Z., Yu, B., 2018. Hinterland patterns of China Railway (CR) express in China
431
under the Belt and Road Initiative: A preliminary analysis. Transp. Res. Part E Logist. Transp. Rev.
432
https://doi.org/10.1016/j.tre.2018.10.002
433
Kundu, T., Sheu, J.-B., 2019a. Analysing the effect of government policy intervention on cross-border
434
freight transportation flows: The Belt and Road perspective. Transp. A Transp. Sci. 15, 13601381.
435
https://doi.org/10.1080/23249935.2019.1594448
436
Kundu, T., Sheu, J.-
437
behavior in the Belt and Road strategic context. Transp. Res. Part E Logist. Transp. Rev. 129, 175
438
202. https://doi.org/10.1016/j.tre.2019.08.007
439
Lee, P.T.W., Hu, Z.H., Lee, S.J., Choi, K.S., Shin, S.H., 2018. Research trends and agenda on the Belt
440
and Road (B&R) initiative with a focus on maritime transport. Marit. Policy Manag. 45, 282300.
441
https://doi.org/10.1080/03088839.2017.1400189
442
Li, D., Zhao, L., Wang, C., 
443
centers in the context of the Belt and Road initiative. Transp. Res. Part E Logist. Transp. Rev.
444
https://doi.org/10.1016/j.tre.2018.10.007
445
Li, S., Lang, M., Yu, X., Zhang, M., Jiang, M., Tsai, S., Wang, C.K., Bian, F., 2019. A sustainable
446
transport competitiveness analysis of the China Railway Express in the context of the belt and road
447
initiative. Sustain. 11. https://doi.org/10.3390/su11102896
448

33
Liu, H., Stockwell, N., Rodgers, M.O., Guensler, R., 2016. A comparative life-cycle energy and
449
emissions analysis for intercity passenger transportation in the US by aviation, intercity bus, and
450
automobile. Transp. Res. part d Transp. Environ. 48, 267283.
451
Opsahl, T., Agneessens, F., Skvoretz, J., 2010. Node centrality in weighted networks: Generalizing
452
degree and shortest paths. Soc. Networks 32, 245251. https://doi.org/10.1016/j.socnet.2010.03.006
453
Ruan, X., Bandara, Y.M., Lee, J.Y., Lee, P.T.W., Chhetri, P., 2019. Impacts of the Belt and Road
454
Initiative in the Indian subcontinent under future port development scenarios. Marit. Policy Manag.
455
46, 905919. https://doi.org/10.1080/03088839.2019.1594425
456
Shao, Z.-Z., Ma, Z.-J., Sheu, J.-B., Gao, H.O., 2018. Evaluation of large-scale transnational high-speed
457
railway construction priority in the belt and road region. Transp. Res. Part E Logist. Transp. Rev.
458
117, 4057. https://doi.org/10.1016/J.TRE.2017.07.007
459
Shao, Z.Z., Ma, Z.J., Sheu, J.B., Gao, H.O., 2018. Evaluation of large-scale transnational high-speed
460
railway construction priority in the belt and road region. Transp. Res. Part E Logist. Transp. Rev.
461
117, 4057. https://doi.org/10.1016/j.tre.2017.07.007
462
Sheu, J.-B., Kundu, T., 2018. Forecasting time-varying logistics distribution flows in the One Belt-One
463
Road strategic context. Transp. Res. Part E Logist. Transp. Rev. 117, 522.
464
https://doi.org/10.1016/J.TRE.2017.03.003
465
Shi, H., You, Z., Feng, Z., Yang, Y., 2019. Numerical simulation and spatial distribution of transportation
466
accessibility in the regions involved in the belt and road initiative. Sustain. 11.
467
https://doi.org/10.3390/su11226187
468
Wang, C., Lim, M.K., Zhang, X., Zhao, L., Lee, P.T.W., 2020. Railway and road infrastructure in the Belt
469
and Road Initiative countries: Estimating the impact of transport infrastructure on economic growth.
470
Transp. Res. Part A Policy Pract. https://doi.org/10.1016/j.tra.2020.02.009
471
Wang, J., Zhou, Y., Huang, G.Q., 2019. Alternative pair in the airport network. Transp. Res. Part A
472
Policy Pract. 124, 408418.
473
WANG, J.J., YAU, S., 2018. Case studies on transport infrastructure projects in belt and road initiative:
474
An actor network theory perspective. J. Transp. Geogr. 71, 213223.
475
https://doi.org/10.1016/j.jtrangeo.2018.01.007
476
Wang, L., Zhu, Y., Ducruet, C., Bunel, M., Lau, Y. yip, 2018. From hierarchy to networking: the
477
-first-
478
38, 416435. https://doi.org/10.1080/01441647.2018.1441923
479
Wei, H., Lee, P.T.W., 2021. Designing a 
480
with China railway express platforms along the Silk Road Economic Belt. Transp. Res. Part E
481
Logist. Transp. Rev. 147, 102238. https://doi.org/10.1016/j.tre.2021.102238
482
Wen, X., Ma, H.-L., Choi, T.-M., Sheu, J.-B., 2019. Impacts of the Belt and Road Initiative on the China-
483
Europe trading route selections. Transp. Res. Part E Logist. Transp. Rev. 122, 581604.
484
https://doi.org/10.1016/J.TRE.2019.01.006
485
Yang, D., Pan, K., Wang, S., 2018. On service network improvement for shipping lines under the one belt
486
one road initiative of China. Transp. Res. Part E Logist. Transp. Rev. 117, 8295.
487
https://doi.org/10.1016/j.tre.2017.07.003
488

34
Yang, Z., Sun, Y., Lee, P.T.W., 2020. Impact of the development of the China-Europe Railway Express 
489
A case on the Chongqing international logistics center. Transp. Res. Part A Policy Pract. 136, 244
490
261. https://doi.org/10.1016/j.tra.2020.03.022
491
Zeng, Q., Wang, G.W.Y., Qu, C., Li, K.X., 2018. Impact of the Carat Canal on the evolution of hub ports
492

493
https://doi.org/10.1016/j.tre.2017.05.009
494
Zhang, X., Zhang, W., Lee, P.T.W., 2020. Importance rankings of nodes in the China Railway Express
495
network under the Belt and Road Initiative. Transp. Res. Part A Policy Pract.
496
https://doi.org/10.1016/j.tra.2020.07.003
497
Zhao, L., Zhao, Y., Hu, Q., Li, H., Stoeter, J., 2018. Evaluation of consolidation center cargo capacity and
498
loctions for China railway express. Transp. Res. Part E Logist. Transp. Rev. 117, 5881.
499
https://doi.org/10.1016/j.tre.2017.09.007
500
Zhou, Y., Wang, J., Huang, G.Q., 2019. Efficiency and robustness of weighted air transport networks.
501
Transp. Res. Part E Logist. Transp. Rev. 122, 1426. https://doi.org/10.1016/J.TRE.2018.11.008
502
503