Journal of Transport Geography 86 (2020) 102741

Contents lists available at ScienceDirect

Journal of Transport Geography

journal homepage: www.elsevier.com/locate/jtrangeo

Time efficiency assessment of ship movements in maritime ports: A case T

study of two ports based on AIS data
Mingxiang Fenga,b, Shih-Lung Shawc, , Guojun Pengc, Zhixiang Fangb,d
⁎

a
Wuhan Transportation Development Strategy Institute, 6 Siwei Road, Wuhan 430017, Hubei, PR China
b
State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, 129 Luoyu Road, Wuhan 430079, Hubei, PR China
c
Department of Geography, University of Tennessee, Knoxville, TN 37996, USA
d
Collaborative Innovation Center of Geospatial Technology, 129 LuoyuRoad, Wuhan 430079, PR China

ARTICLE INFO ABSTRACT

Keywords: Operational efficiency of a maritime port is an important issue for shipping lines and port authorities. It is
Automatic identification system desirable to move ships in and out of a maritime port as efficiently as possible. Automatic identification system
Space-time trajectory (AIS) data recording the trajectory of ship movements allow us to assess the operational efficiency as ships move
Time efficiency in and out of a port. This study proposes a time efficiency assessment framework that evaluates the amount of
Port performance
time each ship spends in the different areas within a port (i.e., berth, anchorage, and fairway) based on the
space-time trajectories of ship movements derived from AIS data. According to the statistical distributions of
time spent by different types of ship in each port area, the proposed framework can compare time efficiency
across different zones within a port and between different ports. This study uses AIS data of four types of ships
(i.e., container, cargo, tanker, and passenger ships) at two selected ports in China, Shanghai Yangshan Port and
Xiamen Port, to demonstrate how the proposed approach can effectively assess and compare time efficiency
levels of ship movements between the times entering and leaving the vessel traffic service (VTS) lines of a port.
This study demonstrates the value of deriving space-time trajectories from AIS data based on the concepts of time
geography to assess time efficiency levels of maritime ports and monitor their performance over time, which
offer useful information to both shipping lines and port authorities for operations such as efficient scheduling
and logistic support.

1. Introduction (UNCTAD, 2019). This large scale of maritime transportation con-

tributes significantly to the development of world trade, global supply
Maritime transportation is an important part of international lo- chain, and international economy. Efficiency of maritime port opera-
gistics. World seaborne trade volume expanded by 2.7% and the total tion therefore is a core issue of maritime transportation in support of
volume grew to 11 billion tons in 2018 (UNCTAD, 2019). Cargo ships, global trade flows.
tanker ships, and container ships ranked the top three among all ship Previous studies have focused mainly on operation efficiency at
deliveries in 2018, with cargo ships accounting for 26.7% of total gross berth, while studies of the efficiency of other phases in a port (e.g.,
tonnage, followed by tanker ships (25%) and container ships (23.5%) vessel traffic services (VTS) line-to-berth, berth-to-VTS line, and an-
(UNCTAD, 2019). An estimate of 793.26 million twenty-foot equivalent chorage) are limited. Based on space-time trajectories of different types
units (TEUs) were moved at container ports around the world and of ships derived from Automatic Identification System (AIS) data, it is
global container port traffic increased by 4.7% in 2018 (UNCTAD, possible to analyze the efficiency of ships undergoing various phases in
2019). Furthermore, containerized trade increased by 2.6%, dry bulk a port. In addition, areas with low time efficiency of ship movements in
trade increased by 2.6%, and global tanker trade increased by 1.5% in a port can be identified, which enables port authorities to improve ef-
2018 (UNCTAD, 2019). Although the main purpose of passenger ships ficiency and provide better transportation services. This study uses AIS
is not for transporting goods and their share in dead-weight tonnage is data of four types of ships (i.e., container, cargo, tanker, and passenger
negligible compared to container, cargo and tanker ships, passenger ships) at two selected ports, which are Shanghai Yangshan Port and
ships accounted for more than 11.9% of the fleet's market value in 2018 Xiamen Port in China, to demonstrate how the proposed approach can

⁎
Corresponding author.
E-mail address: [email protected] (S.-L. Shaw).

https://doi.org/10.1016/j.jtrangeo.2020.102741
Received 30 November 2018; Received in revised form 9 May 2020; Accepted 10 May 2020
0966-6923/ © 2020 Elsevier Ltd. All rights reserved.
M. Feng, et al. Journal of Transport Geography 86 (2020) 102741

effectively assess and compare time efficiency levels of container, efficiency across ports (USDOT, 2019). On the other hand, some re-
cargo, tanker, and passenger ships from the point when they enter the searchers introduce efficiency evaluation indicators by measuring the
vessel traffic service (VTS) line to the point when they leave the VTS economic output, for instance, tonnage worked, berth occupancy rev-
line of each port. Major contributions of this paper are as follows: enue per ton of cargo, cargo handling revenue per ton of cargo, labor
expenditure, capital equipment expenditure per ton of cargo, con-
1) A space-time trajectory framework based on the concepts of time tribution per ton of cargo, total contribution, and monetary and added
geography (Hägerstraand, 1970) is developed to assess the time value of cargo throughput (UNCTAD, 1976) for comparison of port
efficiency performance of ship activities as they move through dif- efficiency around the world. The above indicators have been frequently
ferent zones from the point they enter the VTS line to the point they used to assess port efficiency from facility utilization and economic
leave the VTS line in each port; output perspectives with a focus on the terminals. Efficiency in other
2) A set of AIS-based time efficiency indicators is introduced to assess parts of a port has received less attention in the literature.
the operational performance of different types of ships (i.e., con- Data collected from the Automatic Identification System (AIS) now
tainer ships, cargo ships, tanker ships, and passenger ships) that provide rich vessel movement information. AIS data have been widely
exhibit varying distribution curves of time efficiency; and employed in various research, such as maritime data mining, navigation
3) A case study of two major ports in China, which are Shanghai safety, ship behavior analysis, environmental evaluation, trade analysis,
Yangshan Port and Xiamen Port, demonstrates how the proposed and ship and port performance (Yang et al., 2019). In studies of port
AIS-based approach can effectively assess and compare the time performance, port performance indicators derived from AIS data in-
efficiency patterns of different types of ships and of different ship clude ship traffic, container throughput, berth utilization, and terminal
sizes as they pass through different zones in each port. productivity (Chen et al., 2016). Freight fluidity is also proposed to
evaluate the performance of trade corridors and multimodal supply
The remaining parts of this article are organized as follows. The next chains by measuring total transit time and travel time reliability using
section reviews previous research related to AIS data analysis and port AIS data (Kruse et al., 2018). Other researchers compute travel time
efficiency assessment. The third section presents a proposed time effi- statistics from AIS data of various vessels using different maritime
ciency assessment framework based on AIS data and the related transportation systems (Mitchell and Scully, 2014). Vessel performance
methods. The fourth section discusses the two selected ports and the within a waterway also has been studied with AIS data. For instance,
data used in this study as well as the analysis results derived from the transit durations of inbound cargo vessels, outbound tankers, etc. are
proposed time efficiency assessment framework. The final section compared temporally and spatially to investigate performance trends
summarizes this study and offers some future research directions. (Scully and Mitchell, 2015). In addition, some researchers use AIS data
to assess the resilience of port operations following major disasters and
2. Literature review other disruptive events (Farhadi et al., 2016). From the perspective of
transportation networks, researchers have developed a port con-
Analysis of time efficiency of vessel activities in ports is a vital part nectivity index to evaluate a port's connectivity based on AIS data of
of port efficiency research. Research of port efficiency can help us multiple vessel types (Jia et al., 2017). A port's connectivity in a
better understand the operational and traffic situations in a port as well transportation network can help port authorities and policy makers in
as the management efficiency of a port. Previous studies have focused port planning and infrastructure investment.
on economic efficiency (Coto-Millán et al., 2000; Estache et al., 2004;
Cheon et al., 2010; Nuñez-Sánchez and Coto-Millán, 2012), regulation 2.2. Port efficiency evaluation methods
efficiency (Niavis and Tsekeris, 2012; Coto-Millán et al., 2016;
Serebrisky et al., 2016), terminal operation efficiency (Yuen et al., Researchers have developed various methods to assess port effi-
2013), environmental efficiency (Song, 2014; Moon and Woo, 2014; ciency, such as data envelopment analysis (DEA), stochastic frontier
Tichavska and Tovar, 2015; Coello et al., 2015), and energy con- analysis (SFA), ordinary least squares (OLS), multi-agent-based simu-
sumption efficiency (Johnson and Styhre, 2015). This section will re- lation model, analytic hierarchy process (AHP), and directional dis-
view the efficiency evaluation indicators of a port, and the related Port tance function (DDF) (Irannezhad et al., 2017; Liu, 2010; Suárez-
efficiency evaluation methods. Alemán et al., 2016; Tovar and Wall, 2017; Vaggelas and Pallis, 2010;
Zahran et al., 2017). These methods in general attempt to determine
2.1. Efficiency evaluation indicators of a port relationships between inputs and outputs of port operation using either
a parametric or non-parametric approach.
From an economic perspective, a variety of indicators have been Data envelopment analysis (DEA) is a popular non-parametric
proposed in the literature to evaluate the efficiency level. For example, method for estimating port efficiency (Dyck, 2015). Using a program-
some studies use traffic volumes to characterize and analyze opera- ming approach, efficiency in different units can be evaluated and
tional efficiency of ports. Estrada et al. (2017) propose to use total compared (Zhou et al., 2008). For example, the operational efficiency
volume of export/import operations, port cargo growth rate, and total of container vessels in the main ports of China and South Korea is
factor productivity to characterize port openness, productivity level, analyzed and compared through DEA model based on the length of
cargo expansion, and adaptability to technological changes. Le-Griffin berth, yard area, number of quay cranes, and number of yard cranes
and Murphy (2006) use traffic volume per total length of quay to (Zheng and Park, 2016). The results indicate that major terminals in
measure service level of different elements in ports, such as cranes, South Korea and China have similar efficiency levels. Cullinane et al.
berths, yards, gates, and gang levels at Los Angeles and Long Beach and (2005) use DEA method to assess the efficiency of 30 ports and find no
find that they are underperforming relative to other leading container detectable relationship between the degree of privatization and port
ports. Furthermore, Ducruet et al. (2014) investigate the utilization efficiency.
rates of berths, cranes and ground areas based on the twenty-foot Stochastic frontier analysis (SFA) acts as a parametric and stochastic
equivalent unit (TEU). Their measures include berth length utilization approach to assess port efficiency, which improves over DEA by in-
rate (TEUs per foot of container quay), crane utilization rate (TEUs per troducing distribution assumptions about inefficiency of economic
container gantry crane), crane productivity (TEUs per container gantry agents and random errors of uncontrolled factors. Tongzon and Wu
crane-hour), and land area utilization rate (TEUs per acre of terminal (2005) analyze the relationship between container port ownership
area). In addition, for liquid bulk vessels (such as tankers), the average structure and port efficiency and find that privatization can, to some
dwell time in port terminal boundaries is used to evaluate tanker's extent, improve the efficiency of a container port. The relative

2
M. Feng, et al. Journal of Transport Geography 86 (2020) 102741

Fig. 1. AIS-based time efficiency assessment framework.

efficiency of major container terminals in Asia are assessed by SFA and sub-problems such that each sub-problem can be addressed in-
the results suggest a terminal's size is closely correlated with its effi- dependently. Vaggelas and Pallis (2010) apply a modified AHP to
ciency (Cullinane et al., 2002). Wiegmans and Witte (2017) focus on evaluate port services at twenty major European passenger ports. Their
analyzing the influence of terminals' design on efficiency and find the study confirms that the provision of 17 core port services results in a
most important factors for inland waterway container terminals are mixture of public and private benefits.
yard and crane based on a stochastic frontier analysis. Directional distance function (DDF) requires the specification of a
Ordinary least squares (OLS) is a regression method that fits an direction vector to measure efficiency (Shephard, 1981). Tovar and
equation to reflect the degree of influence on efficiency based on se- Wall (2017) estimate an input-oriented directional distance to measure
lected independent variables. Using variables of distance, product dynamic technical efficiency for a set of 26 Spanish port authorities
weight, product type and difference between imports and exports, over the period of 1993–2012. Ports in this study are assumed to invest
Blonigen and Wilson (2006) examine the efficiency among ports in the with an attempt of minimizing the present value of future production
United States and other countries. The results indicate that the Gulf costs.
coast and west coast ports in the United States as well as European and In short, the above methods are useful in tackling various efficiency
Japanese ports are more efficient than other ports around the world. issues related to port operation. However, very few of them address the
Multi-agent-based simulation model is used to evaluate the colla- assessment of time efficiency as ships sail through different zones in a
boration efficiency among different departments in a port (Irannezhad port between the time a ship first enters the VTS line and the time when
et al., 2017). It evaluates port efficiency by examining the gap between the same ship leaves the VTS line of a port. Different from variables
actual behavior and optimal behavior of each department using a re- such as capital, equipment, labor force, energy, or fuel resources, time
inforcement learning algorithm. It reveals that collaboration can de- is an exogenous variable that usually depends on how the other vari-
crease total logistics cost and improve vehicle utilization. ables perform (de Rus Mendoza et al., 2003). Time can reflect the in-
Analytic hierarchy process (AHP) is an approach that decomposes a fluence of various factors on efficiency (Suárez-Alemán et al., 2014).
complex decision problem into a hierarchy of more comprehensible This study extends previous studies to develop a time efficiency

3
M. Feng, et al. Journal of Transport Geography 86 (2020) 102741

Fig. 2. Different statuses associated with the space-time trajectory of a ship.

assessment framework that uses AIS data to derive space-time trajec- represents the nth record of a ship's location; tn is the time stamp of
tories of ships in a port such that it is feasible to quantitatively assess the nth record; vn is the speed of the nth record, and sn is the ship
the time efficiency level as ships go through different statuses (i.e., VTS status of the nth record. While a ship's location, time stamp and
Line-to-Berth status, Anchored at Anchorage Zone status, Moored at speed can be derived from AIS data directly, a ship's status must be
Berth status, and Berth-to-VTS Line status) in a port. derived from both AIS data and different geographic zones in a port
together.
3. An AIS-based time efficiency assessment framework
Trj = { (x 0 , y0 , t 0 , v0 , s0),…(xn , yn , tn , vn , sn) } (1)
This study employs the concept of space-time path in time geo-
graphy (Hägerstraand, 1970) to generate space-time trajectories of 2. Geographic zone identification. Each port in this study is classified
ships from AIS data. Based on this concept, a ship in a port is considered into three different geographic zones, which are “within the VTS
as a moving object whose location changes over time in a port. This report line,” “within the anchorage zone,” and “within the berth
study defines the boundary of a port by its vessel traffic services (VTS) zone.” The “within VTS report line zone” is to help us identify if a
line within which ships are monitored and provided with maritime ship is inside or outside a port area. If it is outside a port area, it is
transportation navigation advices. In addition, this study divides ship excluded from this study. For those space-time trajectories that are
movements in a port into three major phases, which are “from VTS line inside a port area, this study further identifies if it is anchored in an
to berth”, “at berth”, and “from berth to VTS line”. At the individual “anchorage zone” or docked within a “berth zone”. A “berth zone” is
level, the space-time trajectory of a ship can be used to analyze its defined by extending 100 m from the berth line into the adjacent
operational characteristics in a space-time context, such as the dwell water. The parameter of 100 m is selected based on the width of
time or speed at different locations in a port to reflect the activity ships, because a ship needs to be moored along a berth line for
patterns in different phases. By extracting the space-time trajectory of loading and unloading goods. Each space-time trajectory is then
each ship from the AIS data, this study analyzes the time component segmented based on its location with respect to these three zones.
associated with activities in each phase to assess the time efficiency of This study uses spatial overlay operation in GIS to identify space-
four specific types of ships (i.e., container, cargo, tanker, and pas- time trajectory segments that fall within each of the three zones
senger) in each port. An AIS trajectory-based time efficiency assessment based on the boundary lines shown on official nautical charts.
framework based on space-time paths is proposed below (Fig. 1). 3. Status classification. This study classifies the movements of each
This framework consists of five steps, namely, trajectory construc- ship within a port area into six different statuses, which include
tion, geographic zone identification, status classification, time effi- Inside Port Zone, Moored at Berth, VTS Line-to-Berth, Berth-to-VTS
ciency statistics, and time efficiency assessment. Data processing per- Line, Anchored in Anchorage Zone during VTS Line-to-Berth, and
formed at each of these five steps is discussed below: Anchored in Anchorage Zone during Berth-to-VTS Line, to reflect the
status of a ship in a port area. Fig. 2 (a) illustrates these six statuses
1. Trajectory construction. AIS data of each ship are first sorted by the in a port.
time stamp in AIS records. AIS data records with obvious errors such
as abnormal coordinates or velocity are removed from further The definition of each of the six statuses is provided below:
analysis. The cleaned AIS data sorted by time are used to construct Definition 1. Inside Port Zone refers to the status from the point
the space-time trajectories of all ships. Eq. (1) shows the structure of when a ship crosses the VTS line to enter a port area to the point when
a ship's space-time trajectory derived from AIS data, where (xn, yn) the same ship moves outside the VTS line and leaves a port. It should be

4
M. Feng, et al. Journal of Transport Geography 86 (2020) 102741

Table 1
Time efficiency indicators of ship movements in a port.
Status Indicator Meanings

Inside port zone (inside VTS report line) Port time (PT) Total time that a ship takes to complete all activities in a port
Moored at Berth Berth time (BT) Time that a ship is moored at a berth
VTS Line-to-Berth VTS Line-to-Berth time (VBT) Time that a ship takes to move from the VTS line to the status of Moored at Berth
Berth-to-VTS Line Berth-to-VTS Line time (BVT) Time that a ship takes to move from the end of Moored at Berth status to the point it
crosses the VTS line to leave a port
Anchored in anchorage zone during VTS Line-to- Anchorage time in VBT (ATVB) Time that a ship is anchored in an anchorage zone during VTS Line-to-Berth
Berth
Anchored in anchorage zone during Berth-to-VTS Anchorage time in BVT (ATBV) Time that a ship is anchored in an anchorage zone during Berth-to-VTS Line
Line

noted that this phase includes all activities of a ship inside a port, coefficient of variation (CV), and confidence interval (CI)) are used as
including time spent at berth and/or anchorage. This definition allows indicators to assess time efficiency as ships go through different
researchers to assess the total time spent in a port area by each ship statuses in a port.
which is one indicator of time efficiency at a port. 5. Time efficiency assessment. The proposed time efficiency assess-
ment approach can assess time efficiency in each port as well as
Definition 2. VTS Line-to-Berth refers to the status from the point
compare time efficiency between different ports to identify which
when a vessel crosses the VTS line of a port until the same ship docks at
segments in a port that can move ships efficiently and find the port
a berth, including anchorage time if any. If a ship moves around a port's
areas with room for improvements.
VTS line and stays within the VTS line less than five minutes, it is
excluded from this study. This status allows researchers to assess how
fast a ship can move from the VTS line to a berth within a port area. 3.1. Selected time efficiency indicators
Definition 3. Berth-to-VTS Line refers to the status from the point
when a ship departs a berth to the point the same ship crosses the VTS Fig. 2(b) shows the time components associated with a ship's
line to leave a port. A comparison of the time taken for ships in the “VTS movements within a port area. A ship passes the VTS report line (i.e.,
Line-to-Berth” status versus the time in the “Berth-to-VTS Line” status the dashed magenta lines) into a port area and starts the status of VTS
can shed light on factors affecting the time efficiency of port operations. Line-to-Berth. Before the ship moors at berth, this ship maintains its VTS
Line-to-Berth status as shown by the orange segment in Fig. 2(b). Under
Definition 4. Moored at Berth refers to the status when a ship is the status of VTS Line-to-Berth, a ship may be Anchored in Anchorage
moored at a berth. Since AIS data can track very low speed movements Zone during VTS Line-to-Berth (i.e., the purple stripes segment). This ship
caused by slight movements of a ship at berth, we consider a ship is at then arrives at a berth and transitions into the “Moored at Berth” status
berth only if it is located within a berth zone and its speed recorded in (i.e., the red segment). After completing unloading/loading operations
the AIS data is less than 1 knot (Chen et al., 2016). at berth, this ship begins to leave the berth area and changes into the
Definition 5. Anchored in Anchorage Zone during VTS Line-to- Berth-to-VTS Line status (i.e., the blue segment). A ship may be in An-
Berth refers to the status that covers the period from the point a ship is chored in Anchorage Zone during VTS Line-to-Berth status (i.e., the green
anchored in an anchorage zone to the point the same ship leaves the stripes segment) during the Berth-to-VTS Line status. Finally, as the
anchorage zone to sail to a berth area. An anchorage zone is delineated vessel crosses the VTS line and leaves the port, the Berth-to-VTS Line
as a polygon object on a nautical chart. Due to limited berth space, status ends. We then can compute the total time under the “Inside Port
ships sometimes need to be anchored in an anchorage zone and wait for Zone” status. Based on the time span of these 6 statuses in a port, several
available berth space when sailing towards a berth area. time efficiency indicators can be derived from the defined space-time
trajectory of each vessel (Table 1).
Definition 6. Anchored in Anchorage Zone during Berth-to-VTS For each of the indicators in Table 1, we compute the following
Line refers to the status that covers the period from the point a ship is statistics to compare the time efficiency between different ports based
anchored in an anchorage zone to the point the same ship sails out of on these six indicators:
the anchorage zone to leave a port. Min/Max (SizeTypeMinpS and SizeTypeMeanpS) are the minimum and
After the status classification step, the space-time AIS trajectory of maximum time of the ships with a particular ship type (i.e., Type), a
each ship should have at least the Inside Port Zone, VTS Line-to-Berth, particular class of ship size (i.e., Length or Draught), and a particular status
Moored at Berth, and Berth-to-VTS Line statuses with or without an- (i.e., S) in a particular port (i.e., p). The difference between the max time
chorage status. Various segments of the space-time trajectory of each and the min time (i.e.,SizeTypeRangepS = SizeTypeMaxpS − SizeTypeMinpS)
ship (including container, cargo, tanker, and passenger ships) now are shows the time range which is one indicator of time efficiency. A smaller
identified with a status code value, s, as shown in time range suggests that a port can handle ships with a smaller time
Trj = {(x0, y0, t0, v0, s0), …(xn, yn, tn, vn, sn)} that can be used to assess fluctuation. In other words, it indicates that time efficiency is more
time efficiency. Ships which do not have all four of the basic statuses consistent and reliable.
suggest that they have missing data in the AIS dataset and are removed Mean (SizeTypeMeanpS) is the average time of ships in each status. A
from further analyses in this study. smaller mean value implies less time, which could suggest either a more
efficient operation or a lower quantity of goods is handled at berth.
4. Time efficiency statistics. Based on the trajectory segments of each Median (SizeTypeMedianpS) is the time that separates the upper half
status derived from the above steps, this study computes the time from the lower half of ships in each status. A smaller median value
spent in Inside Port Zone, Moored at Berth, Anchored in Anchorage suggests less time, which could indicate either a more efficient opera-
Zone, VTS Line-to-Berth, and Berth-to-VTS Line, Anchored in Anchorage tion or a lower quantity of goods is handled at berth. Compared to the
Zone during VTS Line-to-Berth, and Anchored in Anchorage Zone during mean, the median is less skewed when there are extreme values in a
Berth-to-VTS Line, respectively, to analyze time efficiency. Six sta- dataset.
tistics (i.e., max/min, median, mean, standard deviation (SD), Standard deviation (SizeTypeSDpS) reflects how far a set of time spans is
dispersed around the mean. This statistic also can be used to analyze the

5
M. Feng, et al. Journal of Transport Geography 86 (2020) 102741

consistency of port operations with respect to time efficiency. 4. Data and analysis results
Coefficient of variation (SizeTypeCVpS) is computed as a ratio of the
standard deviation to the mean. A smaller CV value indicates that the 4.1. Study area and data
variance is relatively small compared to the mean. Since CV is a stan-
dardized measure, it is a better indicator for comparisons between ships In this study, Shanghai Yangshan Port and Xiamen Port in China are
of different types or between different ports. selected as the case study ports. These ports are large seaports that
Confidence interval (SizeTypeCIpS) is a type of interval estimate that is service a large number of vessels every month. This study collects AIS
computed from the observed data of ships. It consists of an interval data of these two ports for the month of May 2014 and the nautical
(SizeTypeCI1pS, SizeTypeCI2pS) that acts as an estimate of the unknown po- chart maps to assess the time efficiency of these two ports based on the
pulation parameter. Since the distribution of time spans does not satisfy proposed framework.
a normal distribution and the data covers all ships, the confidence in- AIS is a tracking system that reports specific information of vessels for
tervals are estimated in this study by removing the data at the top 5% vessel traffic services. It is required by the International Maritime
and the bottom 5%. The larger the gap between SizeTypeCI1pS and Organization's International Convention for the Safety of Life at Sea that
Type
Size CI2pS, the less consistent service is provided by a port. all international voyaging ships with 300 or more gross tonnage (GT), and
all passenger ships regardless of size, should be equipped with AIS devices
3.2. Time efficiency evaluation methods (SOLAS, 1974). Integrated with a positioning system such as a GPS re-
ceiver and other electronic navigation sensors such as a gyrocompass, AIS
This study aims to assess: (1) port efficiency of moving container, data include the following information (Fiorini et al., 2016):
cargo, tanker, and passenger ships in two selected ports, (2) the effects
of different types of ships on port efficiency, and (3) the effects of ship's • Maritime Mobile Service Identity (MMSI): a unique nine-digit
size on time efficiency. We expect that a more efficient operation cor- identification number;
responds to a shorter time and a higher consistency of the total time • Navigation status: “0: under way using engine”, “1: at anchor”, “2:
spent in Inside Port Zone, Moored at Berth, Anchored in Anchorage Zone not under command”, etc.;
during VTS Line-to-Berth, and Anchored in Anchorage Zone during Berth-to- • Rate of turn: right or left (0 to 720 degrees per minute);
VTS Line. • Speed over ground: 0–102 knots or 0–189 km/h (0.1-knot or
Since the time spent at a berth is affected by the carrying capacity of 0.19 km/h resolution);
ships that require different amounts of unloading/loading time, we • Position coordinates: Latitude/Longitude (up to 0.0001 min accuracy);
therefore also assess the time spent during the status of Moored at Berth • Course over ground: up to 0.1 degree relative to the true north;
based on four categories of ship size, which are (0,100] m, (100,200] m, • Heading: 0 to 359 degree;
(200,300] m, and (300, +∞) m in length, as well as (0,5] m, (5, 10] m, • Bearing at own position: 0 to 359 degree;
(10,15] m, and (15, +∞) m in draught according to the industry • UTC seconds: the second field of the UTC time.
practice. In addition, different ports are likely to have different spatial
configurations and different sizes. It is unfair to simply compare the AIS transponders transmit the above information at specific time
time spent across different ports because a port may have a much larger intervals. For instance, AIS report interval is 10 s for not changing
physical size and a longer fairway from the VTS line to the berth area course and 3 1/3 s for changing course when ship speed is between 0
than other ports. This study therefore also computes the time spent per and 14 knots (7.194 m/s), 6 s for not changing course and 2 s for
unit distance to compare the time efficiency for the VTS Line-to-Berth changing course when ship speed is from 14 knots (7.194 m/s) to 23
status and the Berth-to-VTS Line status between different ports. These knots (11.822 m/s). When ship speed is over 23 knots (11.822 m/s),
performance indicators allow researchers to conduct a consistent as- report interval for both not changing course and changing course is 2 s
sessment of time efficiency of ship activities in different statuses at each (Sang et al., 2015). Additional information of AIS report intervals for
port and then compare the time efficiency performance across different different types of AIS transponders is available at HPVUMT (2018).
ports. Specifically, the following indicators are computed to perform Vessels' space-time trajectories can be constructed from the AIS data.
the time efficiency assessment: Nautical chart maps record basic port information, such as an-
chorages, fairways, and VTS report lines, as geographical features that
• Six statistics of time at the status of Moored at Berth, abbreviated as contain their coordinates for geographic zone identification. This study
Berth Time (BT), are computed to assess the time efficiency of also collects the detailed berth locations from the port authorities se-
Moored at Berth. parately when the detailed berth locations are not available on nautical
• Six statistics of the time per kilometer spent during the VTS Line-to- chart maps. Fig. 3 shows the nautical charts of Xiamen Port and
Berth status (abbreviated as VBT) and the Berth-to-VTS Line status Shanghai Yangshan Port. The geographic zone identification and status
(abbreviated as BVT) are computed to assess the time efficiency of classification are completed in this study based on the vessels' space-
these two statuses, respectively. time trajectories and port information in nautical charts.
• The different between VBT and BTV is computed to indicate the time This study further identifies the type, length, and draught of each
efficiency difference between the VTS Line-to-Berth status and the ship based on the MMSI. A ship is removed from this study if it misses
Berth-to-VTS Line status. the key information such as ship type. There are twenty-four types of
• Counts of the number of stops and total time length in Anchored in ships reported in AIS data. This paper focuses on four types of ships
Anchorage Zone during VTS Line-to-Berth and Anchored in Anchorage (i.e., container, cargo, tanker, and passenger), which are selected as
Zone during Berth-to-VTS Line statuses are computed to assess the examples to illustrate the time efficiency assessment based on AIS data.
delay of moving ships between the VTS line and berth area. The counts reported in Table 2 are the number of complete ship visits
that cover the entire process from a ship entering a port to the same
The above indicators also are computed to compare the time effi- ship leaving the port. Both Shanghai Yangshan Port and Xiamen Port
ciency between the four types of ships (i.e., container, cargo, tanker, have a sufficient sample size to carry out this study.
and passenger ships) included in this study. Furthermore, this study
groups each of these four types of ships by their ship size to take into 4.2. Time efficiency analysis of the four statuses in two ports
account the effect of carrying capacity on unloading/loading time. Time
efficiency assessment based on ship size is performed for the status of This section discusses the time efficiency assessment results by each
Moored at Berth only. of the five statuses (i.e., Moored at Berth status, Berth-to-VTS Line status,

6
M. Feng, et al. Journal of Transport Geography 86 (2020) 102741

Fig. 3. Nautical charts of (a) Xiamen Port and (b) Shanghai Yangshan Port in China.

Table 2 Table 3
Number of complete ship visits by each of the four selected ship types in Statistics of Moored at Berth time for container, cargo, tanker, and passenger
Yangshan Port and Xiamen Port. ships in Yangshan Port and Xiamen Port (unit: hour).
Ship type Yangshan port Xiamen port Port Min Median Mean Max SD CV CI1 CI2

Container 199 264 Part I: container ships

Cargo 75 257 Yangshan 5.75 15.91 18.39 46.02 7.92 43.07 10.09 27.79
Tanker 21 28 Xiamen 1.62 11.25 12.40 48.26 6.08 49.03 6.81 19.78
Passenger 27 120
Part II: cargo ships
Total number of all four types 322 669
Yangshan 1.55 12.22 14.85 62.35 10.48 70.61 3.32 27.92
Xiamen 1.50 13.64 24.31 154.22 29.09 119.63 3.20 63.11

Part III: tanker ships

Anchored at Anchorage Zone during VTS Line-to-Berth status, and
Yangshan 3.44 10.52 12.66 47.84 10.72 84.62 3.87 25.29
Anchored at Anchorage Zone during Berth-to-VTS Line status) for each Xiamen 1.59 8.60 16.01 137.89 26.36 108.43 2.12 26.65
ship type (i.e., container, cargo, tanker, and passenger) in Shanghai
Part IV: passenger ships
Yangshan Port and Xiamen Port, respectively. First of all, a time effi- Yangshan 2.26 16.22 16.01 20.14 3.01 18.80 15.68 18.54
ciency assessment of the Moored at Berth status for each ship type be- Xiamen 1.52 8.01 11.31 44.97 11.90 105.27 1.88 18.08
tween Shanghai Yangshan Port and Xiamen Port is reported in Table 3.
Figs. 4-7 illustrate the distributions of berth time for each of the four
ship types at the two ports, respectively. The horizontal axis is the time words, although Shanghai Yangshan Port has a larger mean, median,
spent during the Moored at Berth status in hours, and the vertical axis is and standard deviation than Xiamen Port, variation of the time spent at
the percentage of ship visits corresponding to each time amount. the Moored at Berth status at Shanghai Yangshan Port is a little smaller
For container ships, Fig. 4 shows that Shanghai Yangshan Port has a than Xiamen Port with respect to their mean. This suggests that
more dispersed distribution pattern with a two-peak distribution rather Shanghai Yangshan Port is slightly more consistent in the operation
than a single-peak distribution of Xiamen Port in terms of the amount of time at the Moored at Berth status than that of Xiamen Port.
time spent at the Moored at Berth status. Table 3 also indicates that For cargo ships, Fig. 5 indicates that Xiamen Port has a longer tail of
Shanghai Yangshan Port has a larger mean, median, and standard de- its distribution than Shanghai Yangshan Port. Statistics in Table 3 also
viation (SD) than Xiamen Port for container ships. The larger mean and show that Xiamen Port has a larger mean, median, SD, and CV than
median of Moored at Berth time at Shanghai Yangshan Port could be a those of Shanghai Yangshan Port in terms of the amount of time spent
reflection of larger amounts of containers unloaded and loaded at berth at the Moored at Berth status. They reflect that Xiamen Port is less time
due to larger container ships at Shanghai Yangshan Port than those at efficient than Shanghai Yangshan Port in handling cargo ships at the
Xiamen Port. According to UNCTAD (2014), Shanghai Yangshan Port Moored at Berth status. However, we should notice in Fig. 5 that Xiamen
handled 35.30 million TEUs in 2014 while Xiamen Port's container Port has a general decreasing distribution pattern as time increases
throughput was at 8.57 million TEUs in the same year. On the other while Shanghai Yangshan Port exhibits multiple peaks in its distribu-
hand, the coefficient of variation (CV) of Moored at Berth time in tion. This reminds us the importance of assessing not only the statistics
Shanghai Yangshan Port is smaller than that of Xiamen Port. In other but also the distribution curves to gain a more complete picture when

7
M. Feng, et al. Journal of Transport Geography 86 (2020) 102741

Fig. 4. Container ships' distribution of berth time in Yangshan Port and Xiamen Port.

comparing time efficiency across different ports. indicate that Shanghai Yangshan Port has a smaller mean, SD, and CV,
Due to a smaller sample size of tanker ships, less details are shown yet a larger median, than Xiamen Port at the Moored at Berth status. A
in Fig. 6. A significant peak at 2–3 h stands out for Xiamen Port that is smaller median at Xiamen Port is clearly influenced by the high per-
worthy looking into it in the future to investigate if it is due to a special centage of 2–3 h at Xiamen Port.
tanker service at the port or data errors. The statistics in Table 3 For passenger ships, Fig. 7 shows one concentrated peak at Shanghai

Fig. 5. Cargo ships' distribution of berth time in Yangshan Port and Xiamen Port.

8
M. Feng, et al. Journal of Transport Geography 86 (2020) 102741

Fig. 6. Tanker ships' distribution of berth time in Yangshan Port and Xiamen Port.

Yangshan Port and a dispersed distribution at Xiamen Port. Again, due efficiency of handling passenger ships than Shanghai Yangshan Port at
to a smaller sample size for passenger ships, less details are shown in the Moored at Berth status. However, the consistency of time efficiency
Fig. 7. The statistics in Table 3 indicate that Shanghai Yangshan Port at Xiamen Port is lower than the consistency at Shanghai Yangshan
has a larger mean and median, yet a much smaller SD and CV, than Port.
Xiamen Port. They suggest Xiamen Port has a better overall time The above analysis results also reflect some key differences between

Fig. 7. Passenger ships' distribution of berth time in Yangshan Port and Xiamen Port.

9
M. Feng, et al. Journal of Transport Geography 86 (2020) 102741

Table 4
Statistics of Anchored at Anchorage Zone during VTS Line-to-Berth (ATVB) time and Anchored at Anchorage Zone during Berth-to-VTS Line (ATBV) time of cargo and
tanker ships in Xiamen Port (unit: hour).
Type Min Median Mean Max SD CV CI1 CI2 Count

Part I: ATVB
Cargo 3.01 162.98 193.88 506.60 138.56 71.46 54.14 406.01 34
Tanker 8.95 57.57 57.57 106.20 68.76 119.43 18.67 96.47 2

Part II: ATBV

Cargo 107.96 272.65 277.20 451.00 171.56 61.88 140.90 415.33 3
Tanker 125.68 263.81 263.81 401.95 195.35 74.04 153.31 374.32 2

Shanghai Yangshan Port and Xiamen Port. Shanghai Yangshan Port is Shanghai Yangshan Port has a smaller CV than Xiamen Port except for
mainly a container port with a large amount of transshipment of TEUs passenger ships. This indicates that Shanghai Yangshan Port has more
to/from cargo ships that sail in inland waterways. On the other hand, consistent time efficiency performance in both VBT and BVT for con-
Xiamen Port is a comprehensive seaport that services a wide variety of tainer, cargo, and tanker ships, but not for passenger ships.
ships. The types of goods shipped through Xiamen Port also is much The analysis results presented in this section indicate that we can
more diverse than those through Shanghai Yangshan Port. In addition, use the space-time trajectories derived from AIS data to assess time
Shanghai Yangshan Port does not provide passenger services and most efficiency at the Moored at Berth status, the Anchored at Anchorage Zone
passenger ships at this port are used for transporting port staff. Xiamen during VTS Line-to-Berth status, the Anchored at Anchorage Zone during
Port however provides several types of passenger services (e.g., local Berth-to-VTS Line status, the VTS Line-to-Berth status, and the Berth-to-
transportation, recreation). These contrasts between Shanghai VTS Line status for different types of ships across different ports. This
Yangshan Port and Xiamen Port can help explain the differences in time allows us to assess time efficiency in different geographic zones in a
efficiency at the Moored at Berth status for container ships, cargo ships, port and gain deeper insight on potential operation improvements for
tanker ships, and passenger ships between these two ports. different types of ships in different geographic zones of a port. It also
Next, this study assesses time efficiency at the Anchored at Anchorage provides a method of comparing time efficiency across different ports.
Zone during VTS Line-to-Berth status and the Anchored at Anchorage Zone
during Berth-to-VTS Line status. Anchored at anchorage is normally
needed only when the time in anchorage is longer than one hour. Based 4.3. Time efficiency analysis based on categories of ship size
on the space-time trajectories derived from the AIS data in this study,
we find only some cargo ships and tanker ships at Xiamen Port meet the Since ship size could have impacts on time efficiency especially at
one-hour requirement for the Anchored at Anchorage Zone during VTS the Moored at Berth status to unload/load goods, this section further
Line-to-Berth status and the Anchored at Anchorage Zone during Berth-to- assesses the effects of ship size (based on ship length and draught) on
VTS Line status at Xiamen Port (Table 4). Ships are at the Anchored at time efficiency at the Moored at Berth status. Ships that do not have
Anchorage Zone during VTS Line-to-Berth status mainly due to no avail- information about their size are removed from this assessment. The
able berth space for loading/unloading goods or waiting for port in- remaining ships are grouped into four categories based on their size:
structions, while ships are Anchored at Anchorage Zone during Berth-to-
VTS Line mainly due to ship schedules or customs checks. Table 4 in- Table 5
dicates a relatively large mean, median, SD, and CV for cargo ships and Statistics of VTS Line-to-Berth time (VBT) and Berth-to-VTS Line time (BVT) of
container, cargo, tanker, and passenger ships in Yangshan Port and Xiamen Port
tanker ships at Xiamen Port for both Anchored at Anchorage Zone during
(unit: hour/km).
VTS Line-to-Berth status and Anchored at Anchorage Zone during Berth-to-
VTS Line status. This is one area that Xiamen Port can seek improve- Port Min Median Mean Max SD CV CI1 CI2
ments.
Part I: VBT of container ships
The third time efficiency assessment performed in this study focuses Yangshan 0.08 0.13 0.14 0.26 0.04 30.06 0.09 0.20
on the VTS Line-to-Berth time (VBT) and the Berth-to-VTS Line time Xiamen 0.04 0.05 0.07 0.27 0.04 59.38 0.04 0.12
(BVT). Table 5 shows the statistics of time per kilometer spent at the Part II: BVT of container ships
VTS Line-to-Berth status and the Berth-to-VTS Line status for container, Yangshan 0.05 0.09 0.10 0.23 0.03 28.85 0.07 0.14
cargo, tanker, and passenger ships in Shanghai Yangshan Port and Xiamen 0.03 0.04 0.04 0.27 0.01 33.03 0.04 0.05
Xiamen Port. Fig. 8 illustrates the distributions of VBT and BVT for the Part III: VBT of cargo ships
four types of ships in Shanghai Yangshan Port and Xiamen Port. Yangshan 0.05 0.13 0.14 0.28 0.06 41.26 0.07 0.23
Several patterns are observed from comparisons of VTS Line-to-Berth Xiamen 0.04 0.07 0.10 0.27 0.05 53.65 0.06 0.20
time (VBT) and the Berth-to-VTS Line time (BVT). First of all, the mean, Part IV: BVT of cargo ships
median, SD, and CV of VBT are always greater than or equal to those of Yangshan 0.05 0.09 0.10 0.27 0.03 36.84 0.06 0.14
VBT across all four types of ships in each of the two ports (see Table 5). Xiamen 0.04 0.06 0.07 0.27 0.03 43.43 0.05 0.10

This indicates that it usually takes longer for ships to travel from the Part V: VBT of tanker ships
VTS line to a berth than from a berth to the VTS line in both ports. This Yangshan 0.07 0.15 0.16 0.27 0.07 43.40 0.09 0.26
Xiamen 0.03 0.06 0.07 0.28 0.05 58.04 0.03 0.12
observation is not surprising because incoming traffic usually needs to
wait for available berths and/or other instructions from the port au- Part VI: BVT of tanker ships
Yangshan 0.06 0.10 0.13 0.27 0.05 44.86 0.08 0.22
thority. Second, Xiamen Port has more concentrated distribution pat-
Xiamen 0.03 0.05 0.06 0.22 0.03 53.34 0.03 0.10
terns on both VBT and BVT across all four types of ship than those of
Part VII: VBT of passenger ships
Shanghai Yangshan Port (see Fig. 8). This leads to a smaller mean and
Yangshan 0.03 0.08 0.09 0.23 0.04 49.57 0.06 0.16
median of both VBT and BVT for all four types of ships in Xiamen Port Xiamen 0.01 0.03 0.03 0.10 0.01 41.95 0.02 0.04
than those in Shanghai Yangshan Port (see Table 5), which suggests
Part VIII: BVT of passenger ships
that it takes less time to move between the VTS line and a berth in Yangshan 0.03 0.07 0.08 0.23 0.04 51.52 0.06 0.08
Xiamen Port than in Shanghai Yangshan Port. In the meantime, Xiamen 0.01 0.03 0.03 0.10 0.01 38.35 0.02 0.03

10
M. Feng, et al. Journal of Transport Geography 86 (2020) 102741

Fig. 8. Distribution of BVT and VBT in Yangshan Port and Xiamen Port.

11
M. Feng, et al. Journal of Transport Geography 86 (2020) 102741

Table 6
Statistics of time spent at the Moored at Berth status for container, cargo, tanker, and passenger ships based on ship length in Yangshan Port and Xiamen Port (unit:
hour).
Port Min Median Mean Max SD CV CI1 CI2 Length Count

Part I: container ships

Yangshan port NA NA NA NA NA NA NA NA (0,100] 1
7.10 12.64 13.82 21.26 4.66 33.72 9.40 19.63 (100,200] 9
5.75 12.95 14.49 28.99 5.56 38.38 9.42 23.41 (200,300] 84
9.32 22.70 21.94 46.02 8.12 36.99 12.39 35.05 (300,+∞) 105
Xiamen port NA NA NA NA NA NA NA NA (0,100] 0
4.07 9.79 11.69 38.42 5.81 49.70 6.51 18.13 (100,200] 104
3.02 11.04 12.28 48.26 6.26 51.00 7.01 18.92 (200,300] 109
1.62 13.39 14.09 26.97 6.01 42.66 7.36 22.10 (300,+∞) 51

Part II: cargo ships

Yangshan port 1.55 10.80 12.71 32.16 8.48 66.71 2.55 24.16 (0,100] 53
6.60 15.24 20.00 62.35 13.04 65.21 8.81 34.12 (100,200] 22
NA NA NA NA NA NA NA NA (200,300] 0
NA NA NA NA NA NA NA NA (300,+∞) 0
Xiamen port 1.50 10.83 17.96 148.11 23.71 132.01 2.61 41.50 (0,100] 149
1.51 21.60 32.70 154.22 32.49 99.37 5.51 73.40 (100,200] 89
6.26 66.29 59.57 134.14 43.87 73.64 7.97 99.77 (200,300] 9
NA NA NA NA NA NA NA NA (300,+∞) 0

Part III: tanker ships

Yangshan port 3.44 6.79 11.29 47.84 11.33 100.34 4.01 19.12 (0,100] 15
3.45 14.28 16.09 27.37 8.95 55.61 7.67 26.33 (100,200] 6
NA NA NA NA NA NA NA NA (200,300] 0
NA NA NA NA NA NA NA NA (300,+∞) 0
Xiamen port 2.05 2.39 8.54 46.86 12.07 141.33 2.12 21.62 (0,100] 17
1.59 18.35 28.83 137.89 39.21 135.99 4.04 37.98 (100,200] 10
NA NA NA NA NA NA NA NA (200,300] 0
NA NA NA NA NA NA NA NA (300,+∞) 1

Part IV: passenger ships

Yangshan port 2.26 16.22 16.01 20.14 3.01 18.80 15.68 18.54 (0,100] 27
NA NA NA NA NA NA NA NA (100,200] 0
NA NA NA NA NA NA NA NA (200,300] 0
NA NA NA NA NA NA NA NA (300,+∞) 0
Xiamen port 1.52 8.01 11.31 44.97 11.90 105.27 1.88 18.08 (0,100] 120
NA NA NA NA NA NA NA NA (100,200] 0
NA NA NA NA NA NA NA NA (200,300] 0
NA NA NA NA NA NA NA NA (300,+∞) 0

(0,100] m, (100,200] m, (200,300] m, and (300, +∞) m in length as Port. In other words, container ships of (200,300] m in length or
well as (0,5] m, (5, 10] m, (10,15] m, and (15, +∞) m in draught, (10,15] m in draught tend to have a larger variation of time efficiency
respectively. Table 6 lists the statistics of time spent at the Moored at at the Moored at Berth status than container ships of other sizes in both
Berth status based on the four categories of ship length for all four types ports. This is an observation worthy further investigation in a future
of ships in Shanghai Yangshan Port and Xiamen Port, while Table 7 study.
shows the same statistics based on the four categories of ship draught.
Tables 6 and 7 show that the mean and median of time spent at the
5. Conclusion and future work
Moored at Berth status in general increases with either ship length or
draught for all types of ships. In other words, as ship size gets bigger, it
This study develops a framework and related methods for assessing
takes longer to unload/load goods at berth. However, due to either a
time efficiency of ship activities within different geographic zones of a
very small sample size or no data in some categories of ship length or
port based on space-time trajectories derived from automatic identifi-
draught for cargo, tanker, and passenger ships, it is not feasible to
cation system (AIS) data. AIS data are routinely used and collected for
compare these ship types between Shanghai Yangshan Port and Xiamen
vessel traffic services (VTS). This study applies the space-time path
Port. Time efficiency assessment in this section therefore focuses on
concept from time geography to generate space-time trajectories from
container ships only.
AIS data. This proposed time efficiency assessment framework based on
It is clear that Xiamen Port consistently has a smaller mean and
time geography and AIS data enables researchers and port authorities to
median of time spent at the Moored at Berth status across different ca-
analyze time efficiency of various types of ships (e.g., container, cargo,
tegories of ship length (Table 6) and across different categories of
tanker, and passenger ships) at various statuses in a port such as the
draught (Table 7) than those of Shanghai Yangshan Port. In the
VTS Line-to-Berth status, Moored at Berth status, Berth-to-VTS Line status,
meantime, Shanghai Yangshan Port has a smaller CV of time spent at
Anchored at Anchorage Zone during VTS Line-to-Berth status, and
the Moored at Berth status across different categories of ship length
Anchored at Anchorage Zone during Berth-to-VTS Line status. This pro-
(Table 6) and across different categories of draught (Table 7) than those
posed approach also makes it easy to compare time efficiency of
of Shanghai Yangshan Port, except for the category of (15,+∞) m in
handling different types of ships at various statuses across different
draught that has a very small sample size. These results are in agree-
ports using a set of simple statistics. Major contributions of this study
ment with the findings in Section 4.2 (see Table 3). However, one in-
include: (1) an innovative idea of using time-geographic concepts to
teresting observation is that the category of (200,300] m in ship length
develop a space-time trajectory framework for assessing time efficiency
and the category of (10,15] m in draught has the largest CV than other
performance of ship activities in a port, and (2) a set of AIS-based time
categories of ship size in both Shanghai Yangshan Port and Xiamen
efficiency indicators to assess the operational performance of various

12
M. Feng, et al. Journal of Transport Geography 86 (2020) 102741

Table 7
Statistics of time spent at the Moored at Berth status for container, cargo, tanker, and passenger ships based on draught in Yangshan Port and Xiamen Port (unit: hour).
Port Min Median Mean Max SD CV CI1 CI2 Draught Count

Part I: container ships

Yangshan port 12.73 12.73 12.73 12.73 NA NA 12.73 12.73 (0,5] 1
6.11 14.01 16.19 38.93 6.78 41.89 9.68 24.40 (5,10] 33
5.75 16.06 18.85 46.02 8.11 43.03 10.10 28.33 (10,15] 161
12.38 17.41 19.07 29.07 7.83 41.06 12.66 26.80 (15,+∞) 4
Xiamen port NA NA NA NA NA NA NA NA (0,5] 0
4.07 11.04 11.92 38.42 5.70 47.80 6.85 18.37 (5,10] 121
3.02 11.57 12.95 48.26 6.22 48.08 7.09 20.52 (10,15] 137
13.31 13.69 17.01 24.04 6.08 35.77 13.38 21.97 (15,+∞) 3

Part II: cargo ships

Yangshan port 1.55 12.29 15.26 62.35 10.79 70.68 3.24 28.31 (0,5] 63
5.55 23.12 19.70 26.19 8.40 42.65 10.81 25.70 (5,10] 51
NA NA NA NA NA NA NA NA (10,15] 0
NA NA NA NA NA NA NA NA (15,+∞) 0
Xiamen port 1.50 11.26 15.37 148.11 17.25 112.24 2.78 25.29 (0,5] 125
1.51 18.69 30.84 140.38 28.70 93.04 5.71 70.63 (5,10] 86
1.50 26.16 51.07 154.22 50.84 99.54 6.46 136.99 (10,15] 23
6.26 8.40 23.23 55.03 27.56 118.60 6.69 45.71 (15,+∞) 3

Part III: tanker ships

Yangshan port 3.44 7.00 11.39 47.84 10.62 93.31 3.92 17.75 (0,5] 17
15.86 21.61 21.61 27.37 8.13 37.65 17.01 26.22 (5,10] 2
3.87 14.58 14.58 25.29 15.15 103.88 6.01 23.15 (10,15] 2
NA NA NA NA NA NA NA NA (15,+∞) 0
Xiamen port 1.59 13.29 23.81 137.89 38.28 160.74 2.07 44.47 (0,5] 12
4.31 17.24 18.49 26.87 7.85 42.49 10.67 26.68 (5,10] 7
14.88 14.88 14.88 14.88 NA NA 14.88 14.88 (10,15] 1
2.09 2.34 2.30 2.46 0.13 5.66 2.12 2.41 (15,+∞) 8

Part IV: passenger ships

Yangshan port 2.26 16.22 16.01 20.14 3.01 18.80 15.68 18.54 (0,5] 27
NA NA NA NA NA NA NA NA (5,10] 0
NA NA NA NA NA NA NA NA (10,15] 0
NA NA NA NA NA NA NA NA (15,+∞) 0
Xiamen port 1.52 12.79 11.78 44.97 12.03 102.14 1.88 18.10 (0,5] 114
NA NA NA NA NA NA NA NA (5,10] 0
NA NA NA NA NA NA NA NA (10,15] 0
NA NA NA NA NA NA NA NA (15,+∞) 0

types of ship within different zones of a port as well as their perfor- either ship length or draught for all types of ships in both Shanghai
mance between different ports to enable useful analyses and applica- Yangshan Port and Xiamen Port.
tions of AIS data. 6) Xiamen Port consistently has a smaller mean and median of time
A case study using AIS data collected at two major ports in China, spent at the Moored at Berth status across different categories of ship
which are Shanghai Yangshan Port and Xiamen Port, is conduced to length and across different categories of draught than those of
illustrate how the proposed framework and methods can help us Shanghai Yangshan Port. On the other hand, Shanghai Yangshan
quantitatively assess time efficiency as various types of ships moving Port generally has a smaller coefficient of variation of time spent at
through different areas of a port from the time they enter the VTS line the Moored at Berth status across different categories of ship length
to the time they leave the VTS line of a port. For example, this case and across different categories of draught than those of Shanghai
study identifies some useful results from time efficiency assessments Yangshan Port.
based on the AIS data.
Findings from this case study should not be interpreted as the cur-
1) Xiamen Port is more time efficient in handling container ships and rent situations at Shanghai Yangshan Port or Xiamen Port. Instead, they
passenger ships in terms of the average time spent at the Moored at are presented to illustrate the insights that can be derived from the
Berth status than Shanghai Yangshan Port, while Shanghai Yangshan proposed time efficiency assessment approach using AIS data. Although
Port is more time efficient in handling cargo ships and tanker ships the proposed time efficiency assessment framework demonstrates how
in terms of the average time spent at the Moored at Berth status than it can enable researchers and port authorities to assess port efficiency in
Xiamen Port. different geographic zones and across different types of ships to gain
2) Shanghai Yangshan Port has a more consistent time efficiency per- insights for potential improvements and assist port management, there
formance in handling container, cargo, tanker, and passenger ships are many other potential applications and some challenges of using AIS
in terms of time spent at the Moored at Berth status than Xiamen data for maritime transportation studies.
Port. AIS is still a relatively new “big data” source for maritime transport
3) Only some cargo and tanker ships at Xiamen Port experienced long studies. AIS data can be considered as tracking data of individual ves-
anchorage time, especially at the Anchored at Anchorage Zone during sels like Global Positioning System (GPS) tracking data of cars, trucks,
VTS Line-to-Berth status. and airplanes that have been widely used to tackle a diverse range of
4) It generally takes longer for all types of ships to move from the VTS transportation and geographic problems. Since AIS data include spatial,
line to a berth than from a berth to the VTS line in both Shanghai temporal, and other data of vessels, it is a useful data source for
Yangshan Port and Xiamen Port. studying maritime transportation from a geographic perspective. This
5) Time spent at the Moored at Berth status in general increases with study illustrates one particular application area using AIS data. Other

13
M. Feng, et al. Journal of Transport Geography 86 (2020) 102741

potential applications areas of AIS data include maritime environ- Estrada, M.A.R., Jenatabadi, H.S., Chin, A.T.H., 2017. Measuring ports efficiency under the
mental monitoring, collision warning and avoidance, maritime logis- application of PEP-model. Procedia Comp. Sci. 104 (C), 205–212.
Farhadi, N., Parr, S.A., Mitchell, K.N., Wolshon, B., 2016. Use of nationwide automatic iden-
tics, and many other intelligent applications. With satellite-based AIS tification system data to quantify resiliency of marine transportation systems. Transp. Res.
data, study areas are no longer limited to coastal zones and can be Rec. 2549, 9–18.
Fiorini, M., Capata, A., Bloisi, D.D., 2016. AIS data visualization for maritime spatial planning
extended to a global scale. (MSP). Int. J. e-Navigation Maritime Econ. 5, 45–60.
However, AIS data is not perfect. It often requires preprocessing of Hägerstraand, T., 1970. What about people in regional science? Pap. Reg. Sci. 24 (1), 7–24.
AIS data to deal with missing data and data noises. With large data HPVUMT, 2018. How often do the positions of the vessels get updated on MarineTraffic
(HPVUMT). available at. https://help.marinetraffic.com/hc/en-us/articles/217631867-
volume and spatiotemporal nature of AIS data, it also requires some How-often-do-the-positions-of-the-vessels-get-updated-on-MarineTraffic last accessed on
specific knowledge and skills to work with AIS data. Furthermore, like February 26, 2020.
Irannezhad, E., Hickman, M., Prato, C.G., 2017. Modeling the efficiency of a port community
many other big data application areas, it sometimes is challenging to
system as an agent-based process. Procedia Comp. Sci. 109, 917–922.
validate the patterns derived from AIS data since we may not have Jia, H., Daae Lampe, O., Solteszova, V., Strandenes, S.P., 2017. Norwegian port connectivity
other better data to validate the derived patterns. For the time effi- and its policy implications. Marit. Policy Manag. 44, 956–966.
Johnson, H., Styhre, L., 2015. Increased energy efficiency in short sea shipping through de-
ciency assessment framework proposed in this study, we also face some creased time in port. Transp. Res. A Policy Pract. 71, 167–178.
challenges that need to be addressed in future research. One challenge Kruse, C.J., Mitchell, K.N., DiJoseph, P.K., Kang, D.H., Schrank, D.L., Eisele, W.L., 2018.
resides in the integration of AIS data with other relevant ship in- Developing and implementing a port fluidity performance measurement methodology
using automatic identification system data. Transp. Res. Rec. 2672, 30–40.
formation (e.g., cargo volume, cargo type) and port operation data Le-Griffin, H.D., Murphy, M., 2006. Container Terminal Productivity: Experiences at the Ports
(e.g., operation schedule, loading/unloading capacity) that can further of Los Angeles and Long Beach. In NUF Conference (pp. 1-21).
Liu, Q., 2010. Efficiency Analysis of Container Ports and Terminals. Doctoral dissertation. UCL
enhance the proposed framework to conduct more comprehensive ef- (University College London).
ficiency assessments of port operation. In addition, when we work with Mitchell, K.N., Scully, B., 2014. Waterway performance monitoring via automatic identification
“big” AIS data that cover multiple years and many ports, we need to system (AIS) data. Transport. Res. Rec.: J.Transport. Res. Board. 2426, 20–26.
Moon, D.S.H., Woo, J.K., 2014. The impact of port operations on efficient ship operation: from
develop more efficient data management and analysis methods to fa- both economic and environmental perspectives. Marit. Policy Manag. 41 (5), 444–461.
cilitate spatiotemporal efficiency analysis of maritime transport. This Niavis, S., Tsekeris, T., 2012. Ranking and causes of inefficiency of container seaportsin South-
Eastern Europe. Eur. Transp. Res. Rev. 4 (4), 235–244.
study is the first step of developing an approach of integrating space-
Nuñez-Sánchez, R., Coto-Millán, P., 2012. The impact of public reforms on the productivity of
time concepts in time geography with AIS data to assess time efficiency the Spanish ports: a parametric distance function approach. Transp. Policy 24, 653–666.
of different types of ships across different zones in a port. We hope it Sang, L.Z., Wall, A., Mao, Z., Yan, X.P., Wang, J., 2015. A novel method for restoring the
trajectory of the inland waterway ship by using AIS data. Ocean Eng. 110, 183–194.
serves as an example to encourage more transportation researchers and Scully, B., Mitchell, K.N., 2015. Archival Automatic Identification System (AIS) Data for
geographers to pursue other exciting research topics and applications Navigation Project Performance Evaluation. ERDC/CHL CHETN-IX-40. US Army Corps of
that can benefit from the spatial and temporal information embedded in Engineers.
Serebrisky, T., Sarriera, J.M., Suárez-Alemán, A., Araya, G., Briceño-Garmendía, C., Schwartz,
AIS data. J., 2016. Exploring the drivers of port efficiency in Latin America and the Caribbean.
Transp. Policy 45, 31–45.
Shephard, R.W., 1981. Cost and Production Functions, Lecture Notes in Economics and
Acknowledgement
Mathematical Systems. Springer, Berlin Heidelberg, Berlin, Heidelberg.
SOLAS, 1974. International convention for the safety of life at sea (SOLAS). available at.
This research was supported in part by the National Natural Science http://www.imo.org/en/About/conventions/listofconventions/pages/international-
convention-for-the-safety-of-life-at-sea-(solas),-1974.aspx last accessed on February 22,
Foundation of China (Grants 41231171, 41771473), the Fundamental 2020.
Research Funds for the Central Universities, China (Grant Song, S., 2014. Ship emissions inventory, social cost and eco-efficiency in Shanghai Yangshan
2042020kfxg24), and the Alvin and Sally Beaman Professorship and the port. Atmos. Environ. 82, 288–297.
Suárez-Alemán, A., Trujillo, L., Cullinane, K.P.B., 2014. Time at ports in short sea shipping:
Arts and Sciences Excellence Professorship at the University of when timing is crucial. Maritime Econ. Logist. 16, 399–417.
Tennessee, USA. Suárez-Alemán, A., Sarriera, J.M., Serebrisky, T., Trujillo, L., 2016. When it comes to container
port efficiency, are all developing regions equal? Transp. Res. A Policy Pract. 86, 56–77.
Tichavska, M., Tovar, B., 2015. Environmental cost and eco-efficiency from vessel emissions in
References Las Palmas port. Transport. Res. E: Logist. Transport. Rev. 83, 126–140.
Tongzon, J., Wu, H., 2005. Port privatization, efficiency and competitiveness: some empirical
evidence from container ports (terminals). Transp. Res. A Policy Pract. 39 (5), 405–424.
Blonigen, B.A., Wilson, W.W., 2006. Port efficiency and trade flows. Rev. Int. Econ. 16 (1),
Tovar, B., Wall, A., 2017. Dynamic cost efficiency in port infrastructure using a directional
21–36.
distance function: accounting for the adjustment of quasi-fixed inputs over time. Transp.
Chen, L., Zhang, D., Ma, X., Wang, L., Li, S., Wu, Z., Pan, G., 2016. Container port performance
Sci. 51, 296–304.
measurement and comparison leveraging ship gps traces and maritime open data. IEEE
U.S. Department of Transportation (USDOT), 2019. Bureau of Transportation Statistics, Port
Trans. Intell. Transport. Syst. 17, 1227–1242.
Performance Freight Statistics in 2018.
Cheon, S., Dowall, D.E., Song, D.W., 2010. Evaluating impacts of institutional reforms on port
United Nations Conference on Trade and Development (UNCTAD), 1976. Port Performance
efficiency changes: ownership, corporate structure, and total factor productivity changes of
Indicators. United Nations Publications, New York.
world container ports. Transport. Res. E: Logist. Transport. Rev. 46 (4), 546–561.
United Nations Conference on Trade and Development (UNCTAD), 2014. Review of Maritime
Coello, J., Williams, I., Hudson, D.A., Kemp, S., 2015. An AIS-based approach to calculate at-
Transport 2014. United Nations Publications, New York.
mospheric emissions from the UK fishing fleet. Atmos. Environ. 114, 1–7.
United Nations Conference on Trade and Development (UNCTAD), 2019. Review of Maritime
Coto-Millán, P., Banos-Pino, J., Rodriguez-Alvarez, A., 2000. Economic efficiency in Spanish
Transport 2019. United Nations Publications, New York.
ports: some empirical evidence. Marit. Policy Manag. 27 (2), 169–174.
Vaggelas, G.K., Pallis, A.A., 2010. Passenger ports: services provision and their benefits. Marit.
Coto-Millán, P., Fernández, X.L., Hidalgo, S., Pesquera, M.Á., 2016. Public regulation and
Policy Manag. 37, 73–89.
technical efficiency in the Spanish port authorities: 1986–2012. Transp. Policy 47,
Wiegmans, B., Witte, P., 2017. Efficiency of inland waterway container terminals: stochastic
139–148.
frontier and data envelopment analysis to analyze the capacity design- and throughput
Cullinane, K., Song, D.-W., Gray, R., 2002. A stochastic frontier model of the efficiency of major
efficiency. Transp. Res. A Policy Pract. 106, 12–21.
container terminals in Asia: assessing the influence of administrative and ownership
Yang, D., Wu, L., Wang, S., Jia, H., Li, K.X., 2019. How big data enriches maritime research – a
structures. Transp. Res. A Policy Pract. 36, 743–762.
critical review of automatic identification system (AIS) data applications. Transp. Rev. 39,
Cullinane, K., Ji, P., Wang, T.F., 2005. The relationship between privatization and DEA esti-
755–773.
mates of efficiency in the container port industry. J. Econ. Bus. 57 (5), 433–462.
Yuen, C.L., Zhang, A., Cheung, W., 2013. Foreign participation and competition: a way to
de Rus Mendoza, G., Campos, J., Nombela, G., 2003. Economía del transporte. (Antoni Bosch
improve the container port efficiency in China? Transp. Res. A Policy Pract. 49, 220–231.
editor).
Zahran, S.Z., Alam, J.B., Al-Zahrani, A.H., Smirlis, Y., Papadimitriou, S., Tsioumas, V., 2017.
Ducruet, C., Itoh, H., Merk, O., 2014. Time efficiency at world container ports. Discussion paper
Analysis of port efficiency using imprecise and incomplete data. Oper. Res. 1–28.
no. 2014-08. http://www.itf-oecd.org/time-efficiency-world-container-ports (Accessed on
Zheng, X.B., Park, N.K., 2016. A study on the efficiency of container terminals in Korea and
Jul 6, 2016).
China. Asian J. Shipp. Logist. 32 (4), 213–220.
Dyck, G.K.V., 2015. Assessment of port efficiency in West Africa using data envelopment
Zhou, P., Ang, B.W., Poh, K.L., 2008. A survey of data envelopment analysis in energy and
analysis. Am. J. Ind. Bus. Manag. 5 (4), 208–218.
environmental studies. Eur. J. Oper. Res. 189 (1), 1–18.
Estache, A., Tovar, B., Trujillo, L., 2004. Sources of efficiency gains in port reform: a DEA
decomposition of a Malmquist TFP index for Mexico. Util. Policy 12 (4), 221–230.

14