Content uploaded by Patrick Moriarty
Author content
All content in this area was uploaded by Patrick Moriarty on Apr 29, 2018
Content may be subject to copyright.
Final draft of Ch 5 of book by Stephen Jia Wang and Patrick Moriarty Big data for urban
sustainability, Springer, 2018
Chapter 5. Big Data for Sustainable Urban Transport
Abstract
This chapter re-examines the general solutions proposed to improve the environmental sustainability of transport
discussed in Section 1.3.2, with a view to understanding the potential for big data in each of these approaches.
How can big data be used to reduce transport energy and emissions in cities? Specifically, how can big data
encourage modal shift from cars to more environmentally friendly modes, and reduce vehicular transport overall
through better trip planning? The chapter also includes a case study of a ‘personal transport planner’ designed
for use in Beijing, based on the idea of a monthly personal transport energy quota.
5.1. Introduction
As we have discussed in the previous chapters, we are now living in a world where for the
first time in human history, more than half of the world's population lives in an urban
environment. In the next few decades (if present trends continue) perhaps 70% of people will
live in cities, where people will require more mobility than ever. Big data empowered by
cloud computing could make it possible for researchers and planners to deal with the growing
challenges of urban transportation. The challenges we are facing are also one of the most
important in this century: how can we enable urban dwellers in tomorrow’s cities to travel
more efficiently and sustainably, and provide services to the inhabitants, while maintaining
comfort, pleasure, and safety during their journey? In this introductory section we will first
give some detail on the types of big data potentially available in cities, then review the
various options for sustainable transport.
5.1.1. The Portrait of a City from a Transport Viewpoint
As well as understanding personal activities, achieving real intelligence in a Personal Travel
Assistant (PTA) needs a multi-disciplinary approach to gain an overall portrait of the city and
its transport system. Also, data gathering should not be limited to a single source of
information—the fusion of public heterogeneous data sources is becoming more important
than ever. The data within the urban environment is all-encompassing, with data from
different sources portraying the various aspects of the urban environment. For instance:
• The urban Points of Interest (POI) technology gives valuable information on urban
functions, such as hospitals and residential areas
• Urban residents’ activities in the social network not only present an urban crowd
relationship map but also indirectly reflect the city's emotional and dynamic situation
• CCTV cameras collecting traffic image data etc reflect the ongoing activities of the
city.
Government and enterprises have realised the application potential of multi-source
heterogeneous data fusion. For instance, the Shanghai government held the Shanghai Open
Data Apps in 2015, which included the following data:
• Shanghai municipal Road Traffic Index
• Metro operation data
• Passenger card data
• Pudong bus real-time data
• air quality status data
• weather data
• road accident data
• elevated off-ramp data
• Sina microblogging traffic data.
The different types of urban data from all these multiple sources could help improve the
accuracy of transport-related trajectory prediction. However, the multi-source heterogeneous
data vary in attributes: for example, trajectory data is time and space data, monitoring and
acquisition data are from images. How to manage and integrate large-scale multi-source
heterogeneous data is the challenge of trajectory prediction. In the process of constructing a
smart urban environment, the city's information infrastructure will be designed to provide
much more and better information-enriched services. At the same time, this process
accumulates a vast array of complex urban activity data.
Here, we briefly introduce the types of urban data that are commonly considered to be
crucial for realising the goal of sustainable urban transport.
Where are you travelling now? As the premise for any location related discussions, until
now GPS (Global Positioning System) provides the most crucial information for position
recognition. A mobile device with a GPS receiver chip can collect information about the
movement of objects in real-time such as the positions of people and cars in the city.
Is position the only information we can get from GPS? Currently there are several
applications based on GPS in everyday use. For example, the most common one is Floating
Car Data (FCD) technology to detect the traffic flow speed on the road. FCD uses the real-
time data from the mobile phone in a vehicle being driven on the road network [53]. The
mobilephones provide a data set which includes travel speed and direction as well as basic
location data. FCD technology works as a sensing node for sampling the city's overall traffic
situation.
The mobile phone is becoming an indispensable communication tool for daily life. It can
provide many types of data: address book, call records, GPS positioning information,
signalling records with the base station, internet records, and app use records. This data
reflects the interest of residents in the city's activities, the activities themselves, the frequency
of interactions, social relations, and other content—it has a vast application potential. A
smartphone with a GPS receiver chip can also be used as a collection device for personal
trajectories. However, due to privacy, security and many other issues (see Chapter 4) large-
scale mobile GPS data collection is restricted, and current applications rely on volunteers for
small-scale data collection and research. In the case study ISUNS project (see Section 5.6),
the GPS data can be obtained from the mobile phone and navigation devices of participants.
Location Based Service (LBS) is an emerging network service mode in the mobile internet
era. The data collected by LBS applications has clear geographical coordinates and combines
the information features of traditional web services. It can be used for locating persons,
navigation, and providing information on the nearest service, for example, an ATM. It thus
can provide large amounts of information which can help planners more deeply understand
the dynamics of managing a city.
One of the most important data types is maps integrated with POI data. Streets and
buildings are the core framework of the city. The map is a basic way to describe urban
structure, and the POI data is the core information of the city's functional units. Therefore, the
city map and POI data is the smart city's most basic raw material, reflecting people’s basic
activities in the city.
Why are you travelling? The various means of transport for daily commute all generate
passenger data. Taxi passenger data can be obtained using the FCD GPS data with the
passenger status of the taxi meter. Bus and subway passenger data can use the municipal
traffic card credit card records. Passenger data contains very informative information on
urban activities, which can be used in urban area function analysis, population flow detection,
urban traffic system assessment, multi-vehicle human behavior research, urban traffic
economics research and other fields.
Finally, Pan and colleagues [39] have reviewed how trace analysis and data mining from
various sources can help improve both urban transport and urban planning. Using data from
mobile devices, vehicles, smart cards and floating sensors (RFID tags) can help gather
detailed transport information to help both travellers (including urban visitors) and transport
planners. Data on frequency of visits to various locations, for example, is not only useful for
transport planning, but also to urban land use planners.
5.1.2. General Approaches to Sustainable Tranport
Some of the embryonic existing uses for big data discussed in Chapter 3 could be rendered
obsolete, and seen as merely stop-gap solutions for urban transport. Still they could prove
important during the transition to a more sustainable system. If, for example, large cuts in
vehicle-km of travel were made in order to lower air pollution, oil use, and CO2 emissions,
the need for both traffic congestion management and parking spot information, as discussed
in Chapter 3, would become largely irrelevant. However, a prerequisite for such smart and
smooth transport transition is a comprehensive understanding of the overall urban transport
system. Hence this chapter will mainly concentrate on the other aspects of smart transport,
examining how big data can reduce car travel demand itself by either encouraging a shift
travel to other modes, or by reducing the demand for vehicular travel overall.
In this chapter, we will look at how big data could be applied far more widely, with far
greater effect, to one of our main themes, urban transport. Figure 5.1 summarises the general
approaches possible for both reducing the environmental impact of urban travel, while at the
same time enhancing the welfare of urban travellers:
• Travelling less, enabled by ICT communications
• Multi-mode travelling, enabled by navigation systems
• Smart trip planning, enabled by mobile devices with real time update
• Trajectory-based [40, 43] trip planning, enabled by combination of mobile and IoT
technologies
• City profile-based trip planning, enabled by a comprehensive usage of big data from
other sectors as well as transport, and covering the entire city.
Figure 5.1. Methods for more sustainable travel.
As a first task, how can big data help cut the energy consumption of travel in cities? Funk
[14] provided an insight into why big data (or IT generally) can help reduce both the carbon
emissions and resources used in transport. He pointed out that the annual present gains in
energy efficiency are far lower than the efficiency gains experiencied in IT (as enshrined in
‘Moore’s Law’), suggesting that more use of IT in transport will pay large dividends in
sustainability.
Chapter 1 introduced the several means for reducing urban travel energy (and
consequently GHG, air, and noise emissions), which are repeated here:
• by shifting to more energy efficient modes, namely public and non-motorised
transport
• by reducing the demand for urban transport, using IT to either combine previously
separate trips, or to substitute for the trip itself
• by increasing the occupancy rates on both public and private vehicular transport
• by increasing the vehicle efficiencies (vehicle-km per primary MJ) for both public and
private vehicular transport.
The rest of this chapter will build on the discussion in Section 1.3.2, and will examine the
potential role of big data in each of these approaches (Sections 5.2-5.4). Since the potential
for big data is much greater for the first two approaches, they are discussed in greater detail.
In Section 5.5 we discuss traveller comfort on public transport, which is a significant problem
for all high-density cities in Asia, and in low-income cities generally. Section 5.6 describes a
personal travel assistant developed by the authors and their colleagues [51] for use in Beijing.
As we have already stressed, most of big data’s potential lies in the future. The concluding
section (Section 5.7) accordingly looks at future urban transport and summarises the potential
role of big data in meeting sustainability goals for transport.
5.2. Shifting to More Energy Efficient Modes
Urban public transport modes are usually more energy efficient than car travel. Reasons
include higher occupancy rates, particularly at peak travel times, lower rolling friction for
steel rail transport, and the greater values of seat-km per MJ of primary energy that larger
vehicles enjoy. Electric public transport also has the potential for regenerative braking—as of
course do hybrid and battery electric vehicles. Non-motorised transport is even more energy
efficient as well as being superior to vehicular transport in reducing land use, air and noise
pollution, traffic casualties, and GHG emissions [28]. It is also far cheaper, and available
without the need for a driver’s licence. At present, transport priorities are car first, then public
transport, and finally non-motorised travel—if the latter is recognised at all. For sustainable
and healthy urban futures, we need to reverse this order and give priority to non-motorised
transport.
5.2.1. Public Transport
A number of cities have already implemented smart cards for automatically debiting fares on
their public transport systems, such as the Myki card in Melbourne, Australia, and the Oyster
card in London. They are well-accepted by users. Michael Batty [3] has discussed the
promise and difficulties of using data from the Oyster card system for transport analysis.
Over six months in 2011-2012, researchers at University College, London collected around
one billion records of travellers who ‘touched’ on and off. Unfortunately, despite the
enormous amount of data available, only about 85% of the customers of London public
transport use the Oyster card, which raised questions of how representative even this huge
sample was. Of these 85%, 10% did not touch off due to open barriers. He pointed out that
several assumptions must, therefore, be made to make the data useful to transport planners.
Nevertheless, such data has great potential in helping planners understand travel patterns for
regular users of London’s public transport (the 85%), and, as more data accumulates over
time, to see what changes in travel patterns are occurring.
For Beijing City, Long [55] suggested that it is possible to analyze the bus cardholder's
place of residence, employment and commute travel characteristics in Beijing, based on a
week of credit card records combined with the 2005 Beijing residents travel survey and land
use map data from Beijing’s 8.5 million bus cardholders in 2008.
In Brisbane, Australia, Tao et al [47] have used smart travel cards to evaluate the spatial-
temporal dynamics of Bus Rapid Transit (BRT) trips in comparison with trip making on the
non-BRT bus services. Travel data were collected for various days representing different
travel patterns: a workday, a Saturday, a Sunday, a school holiday, and a public holiday. As
with the smart card systems already discussed, the Brisbane card provides both user details,
as well as specific trip details such as the boarding stop, time of touch on of the smart card,
and bus route number. Marked differences were found between BRT and ordinary bus travel.
The authors concluded that: ‘The results offered more detailed and reliable insights into BRT
use than had been previously attained. From this, important implications for evidence-based
BRT policies were identified to inform service management (e.g., service monitoring on
critical pathways, establishing flexible route allocation mechanism) and infrastructure
provision (e.g., enhancing the utility of current busway, selecting potential BRT corridors).’
Two further examples come from Boston, USA. In 2015, city officials started receiving
quarterly reports from Uber, the car booking company, on the origin and destination of all
Uber trips, the time of day and date when they were made, and the distance travelled [20]. It
was hoped this data would help planners better understand the transport flows in that city.
This knowledge would also help public transport network planning. In a second Boston study
based on the effects of widespread ride-sharing, Alexander and González [2] tried to answer
the question: Does ride-sharing increase or decrease urban traffic congestion? They showed
that ride-sharing would generally decrease congestion—in effect, that ride-sharing would not
be at the expense of public transport or non-motorised trips.
One possible reason why public transport patronage is low in many OECD cities is that
many travellers, especially regular motorists, are unfamiliar with both the services available
and the timetables. Also, at present urban travellers largely rely on travel routines to make
decisions about which route to choose and which mode to use [18]. Particularly for car
travellers, the existing routines will have to change if transport is to become more sustainable.
Smart phone apps are increasingly making real-time data for all modes available to travellers
and informing them of any delays. In Melbourne, for example, the available public transport
app gives the times of the next five arrivals at the public transport stop at which the traveller
is waiting.
Miller [27] has outlined an ambitious plan for the use of big data for urban passenger
transport. He pointed out that, at present, particularly in the US, there is a transport
monoculture: it is assumed that all trip types, to all destinations, at all times, are best done by
car. Instead, he urged a transport polyculture, in which the various forms of non-motorised
transport, and public transport, are seen as vital parts of the transport mix. His idea was that
the various transport modes should cooperate to achieve environmental and other aims, rather
than be in conflict, as at present. Innovations such as Seoul’s Personal Travel Assistant (PTA)
are a step in this direction, presenting all modal choices for each trip undertaken. This
journey planner allows travellers to select for any trip the travel mode which will variously
give the fastest trip, the cheapest trip, or the trip with the lowest GHG emissions [4].
5.2.2. Non-Motorised Transport
Non-motorised transport—walking and cycling—will have to play a far larger role in urban
transport if transport GHGs are to be significantly reduced. In an urban context, these modes
have great potential for expansion, given the significant share of urban trips which are less
than two km. In the UK, at least, the share of trips by non-motorised transport has fallen in
recent decades [13]. Increasing the use of these modes can help cut urban transport air
pollution, noise, GHG emissions, and energy use. Their increased use can also reduce the
‘serve passenger’ trips associated with chauffeuring children to school or activities by car.
The existing use of non-motorised travel varies greatly between the world’s cities, with the
greatest use being in African and Asian cities, and the lowest in those of the US; it is
accordingly often stigmatised as the poor person’s mode of choice. Still it can also be high in
wealthier cities, too, suggesting that there is potential for improvement. For the year 1995,
Cameron et al [8] estimated that 34% of all trips in Hong Kong were by walk/cycle,
compared with only 5% in Phoenix in the US. The share of non-motorised travel in the
lower-income cities of Asia would be similar to, or greater than, the share in Hong Kong. In
the cities of Africa and much of Asia, where national car ownership rates can be as low as 10-
20 per 1000 residents (it was only 15 per 1000 in India in 2015 [38]), most trips are probably
still by walking and cycling.
The other major benefit from far more use of non-motorised modes is for urban health, a
topic dealt with in more detail in Chapter 7. Exercise has been termed the ‘wonder drug’ by
Andy Coughlan [11], because of its variety of health benefits. He wrote: ‘A plethora of recent
studies shows that exercise protects us from heart attacks, strokes, diabetes, obesity, cancer,
Alzheimer’s disease and depression. It even boosts memory. And it has the potential to
prevent more premature deaths than any other single treatment, with none of the side effects
of actual medication.’ Given its benefits in so many areas, it is surprising that it still has had
so little real support from transport or health policy makers.
How could big data encourage more non-motorised travel? One way could be through a
PTA. For example, travellers could be advised how long a given trip could take by walking or
cycling, based on the traveller’s average walking/cycling speed as calculated from personal
smart phones. The PTA could also give the estimated personal energy expenditure for the
proposed trip, useful if the traveller had a daily exercise budget, and for the environmentally
conscious, the CO2 saved by forgoing the car option. It could also assess whether the
estimated trip time would fit in with the time available, such as the lunch hour, or before an
appointment, etc. Further, based on data from street and pavement sensors, at traffic lights,
the green time for pedestrians could be extended if the volumes of pedestrian traffic
warranted it.
5.3. Reducing the Demand for Urban Transport
Big data-supported PTAs could support modal shifts, which would help ease the urban
transport sustainability challenge to some degree. However, large cuts in urban transport
energy use will most probably require corresponding cuts in vehicular mobility itself—not an
easy task [29, 34]. One reason why reducing vehicular transport is so difficult is that its
present cost structure in most cities is not conducive to energy conservation. In a paper
comparing Tokyo and Shanghai, Luo et al [23] showed the higher parking restrictions and
costs resulted in far fewer daily trips per car in Tokyo. Perceived costs of car travel, mainly
fuel costs, are a small part of the total costs of owning and operating a car, since purchase,
annual insurance, and registration costs are independent of vehicle kilometres. For urban
freight vehicles, especially smaller ones, driver costs make fuel costs an even smaller fraction
of total costs. With data collected on distance travelled by class of road and time of day for
each vehicle, it would be possible to move to a ‘pay as you drive’ scheme for both passenger
and freight road vehicles, which would make vehicle owners much more aware of the real
costs of road transport. Presently unpaid environmental costs of vehicles—their negative
externalities—could also be included, for example, by a carbon tax based on fuel
consumption.
In imposing such a scheme, it would be important to avoid increasing inequality in urban
areas where car travel is dominant, since in many OECD cities at least, low-income
households often live in outer areas, where they not only have greater travel needs, but also
have lower quality public transport services compared with inner urban households [29]. (In
contrast, in the low-income, low car ownership cities of many industrialising countries,
raising the costs of private motoring could enhance equity.) Nevertheless, the huge subsidies
fossil fuel energy presently receives [32], including those for transport, will eventually have
to be greatly reduced. Already, OECD countries in Europe and particularly Asia which have
low domestic fossil fuel reserves, such as Sweden, Finland, Japan and South Korea, tend to
have higher transport fuel costs and less fuel consumption than the US or Australia, with their
vast fossil fuel reserves [5, 31]. Concerns about supply security, together with lower fuel use,
seems to make higher prices more acceptable.
Car insurance and registration rates could be based on not only vehicle-km but also on
time and location of travel (with higher rates for travel in congested inner city zones, as is
already the case for the London and other road pricing schemes [52]). But with the detailed
data on driving speeds and acceleration increasingly available from vehicles, it would even be
possible to incorporate driving practices into car insurance rates.
5.3.1. Reducing Travel with IT
Over the last decade or so, many OECD cities have seen a fall in average vehicular
passenger-km per capita [29]. At the same time, there has been an explosive growth in both
smart phone ownership and social networking. Some researchers (eg Lyons [24]) have
therefore considered whether the rise in new forms of IT have at least partly caused this
documented decline in personal urban travel. The question arises: could big data facilitate
urban travel reductions through IT substituting for travel? This idea goes back at least four
decades, with a book-length publication on tele-work (also called tele-commuting) in the US
by Jack Nilles [37]. The aim of tele-working in the 1970s, with its oil crises, was to reduce
both traffic congestion and transport energy consumption. Subsequently, the concept was
extended to tele-medicine, tele-education (and cyber universities), tele-shopping and so on.
Although today, IT has been applied in all these areas, it is fair to say that its impact on urban
travel so far has been negligible. Historically, of course, telecommunications and vehicular
travel have both grown in step, prompting some researchers to argue that the two are
complementary rather than competitors [10, 33].
One view is that although use of the latest ICT, specifically mobile phones, may reduce
travel in some cases, it could increase travel in other cases. Dal Fiore et al [12] argued that
‘mobile technology might offer people numerous new reasons to be mobile: by making them
more informed; more capable of using a larger variety of physical spaces and re-negotiating
obligations in real-time; and potentially more efficient in the allocation of their travel time
and resources.’ However, they also added that the technology could in some cases work to
make travel less appealing.
What we also have to take into account is that ICT itself will change the travel patterns it
is supposed to replace. Think of how ICT has the potential to change the nature and location
of work/employment, health, education, shopping, entertainment. So far, we have only
considered ICT effects on existing trip patterns, destinations, etc. Hence the past, minor,
impact of ICT on urban travel volumes, patterns, and modes may be a poor guide to the
future.
As already mentioned, we consider the application of big data to urban sustainability as a
necessary but not sufficient condition: policy support is also needed. For large urban car
travel reductions, it will also be necessary to increase both the monetary costs of road travel
by application of carbon taxes (discussed in more detail in Section 6.1), and increased tolls
and parking charges, and to reduce the convenience of car travel [29]. The convenience of
using the car in urban areas can be lowered by measures such as reducing speed limits and
inner urban parking spaces, ending arterial road building, and road closures in the central
business district (CBD). The main effect of these latter measures will be to raise travel times
for trips by car. Raising both monetary and time costs can be justified by the accompanying
reduction in collisions, traffic noise, community severance, air pollution, oil dependency and
GHG emissions. Although all these measures could be quickly implemented, for equity
reasons raising the travel time costs should be implemented first, as all travellers have 24
hours per day, but incomes are most unevenly distributed, and as discussed above, lower
income households often have greater travel needs in OECD cities.
Importantly, big data facilitates analysis of many natural experiments in urban transport.
The effects of changes in road speed limits, parking restrictions, road closures, road
charges—all regardless of whether temporary or permanent—can be analysed, as well as the
effects of introducing new bus services, bus-only lanes, and high vehicle occupancy lanes.
Unlike domestic energy use, per capita transport use (even after adjustment for income and
other household characteristics) also varies spatially across the city, with generally higher
levels of per capita car travel in outer areas of OECD cities, and lower public transport use. It
also varies by income level, by age group, and by employment status. Trying to incorporate
all these variables into conventional sample surveys is difficult because obtaining reliable
data for sub-groups (for example, low-income elderly householders in the outer suburbs)
requires very large (and thus very expensive) sample sizes.
Nevertheless, some present applications of smart transport can be counterproductive from
an ecological sustainability viewpoint, in that they can lead to increased urban transport
energy use and GHG emissions. This counter-intuitive result arises from feedback effects.
Very congested cities such as Tokyo and Hong Kong have low average travel speeds for road
traffic, with much stop-start driving. But, on the other hand, they have very well-patronised
rail systems, and far greater use of non-motorised transport than in North American or
Australian cities (See Section 5.2.2). High levels of rail patronage (and high rail seat
occupancy rates) result in public transport being several times as energy efficient as car travel
on a passenger-km per primary energy basis [36]. Overall, car energy use and GHG
emissions per vehicle-km might be higher than in lower density cities but are more than
compensated by far lower car vehicle-km per resident.
The following is a speculative view of how big data could help reduce vehicular travel in
the presently car-dominated cities of the OECD. Travel reductions can occur in several
possible ways: some trips can be foregone; some can be combined with another (non-
discretionary) trip; and some can be made to closer destinations. Only the last two are of
interest here. For combining two or more previously separate trips, the intending traveller
could input the required information into their PTA app, such as the start and finish times for
work and its location, or the time and location for picking up children after school, and any
other constraints. These constraints could be designated as constant for every weekday. For
trips that are discretionary for both their timing and location (eg shopping trips), the user
could input various acceptable options. The PTA could then advise the traveller on ways of
minimising total travel time, money cost, or CO2 emissions for that day.
A suitably designed PTA could also help with shortening individual trips. Trips can be
divided into two types: discretionary and non-discretionary. Discretionary trips into trips like
shopping, recreation and social visiting for which time and/or location can be varied. Non-
discretionary trips include work and education trips, which, in the short term at least, must be
made at a definite time to a definite place. Especially in the cities of North America and
Australasia, the rise of the car has brought about a suburbanisation of trip destinations such as
shopping, entertainment, and workplaces. Paradoxically, this increased opportunity for
localising activities was accompanied by a steady rise of per capita vehicular travel in most
cities [29].
Evidently, the perceived low time and money costs of car travel made such extra travel
acceptable. But as the need for travel reductions is increasingly recognised by the urban
public, and supported by policies such as carbon pricing, travellers will increasingly prefer
local destinations—for shopping and social activities, and over time for work as well. As
shown, the PTA will be valuable in the perhaps long transition period, but after individuals
(or households) have established a new travel routine, it will be less necessary for day to day
travel.
5.3.2. Reducing Freight Transport
Reductions in kilometres travelled are also needed for freight transport vehicles. Often this is
possible by eliminating ‘wasted travel’, the unnecessary km of travel caused by sub-optimal
routes. This is especially a problem when a freight vehicle has to make multiple deliveries on
the one trip, with locations that vary each day. Victor Mayer-Schӧnberger and Kenneth
Cukier [25] have described how in 2011 the US freight company UPS was able to cut the
travel of its vehicles by about 48 million km, with attendant fuel savings of over 11 million
litres. The reduced km of travel and fuel translated into lower costs and fewer accidents. One
reason for these savings was that the algorithm used ‘compiles routes with fewer turns that
must cross traffic at intersections’, which in turn reduces the time taken, fuel use, and
accidents.
On-line shopping is often viewed as a way of cutting vehicular travel to shops. One
problem is that even if passenger vehicle-km is thereby reduced, the reduction will be offset
by a rise in freight transport deliveries. What is more, these deliveries will usually be in low
load capacity trucks, which have low energy efficiency per tonne-km of freight carried. Can
IT help solve this problem it has created?
Taniguchi et al [46] have argued that there is a variety of ways in which the new
technology can help reduce the environmental costs of freight travel. They pointed out the
limitations of existing road pricing schemes: cordon-based pricing schemes encourage
speeding to minimise time in the cordon. There is no one objective for freight vehicle pricing,
which could include cost recovery for infrastructure, reducing peak hour travel by trucks, and
ensuring high vehicle loading rates. However, while road pricing schemes might be the best
way forward in the transition period to more sustainable cities, in the longer term reducing
energy, GHGs and air pollution from freight will become most important. Charges would
then be based on the environmental costs of freight, with the levy perhaps added to fuel costs.
One suggestion for achieving reductions in the environmental burden of the urban freight
system is ‘co-modality’—using passenger public transport vehicles at off-peak times for
freight delivery [46]. The present distribution of warehouses is presumably optimal for
distributors but would change if the external environmental damages had to be internalised.
Environmental taxes for freight would, ceteris paribus, raise freight costs, but if these and
other initiatives just discussed were also introduced, overall freight costs in a given city might
even fall because of reduced freight vehicle-km and reduced driver costs.
5.4. Raising the Energy and GHG Efficiencies of Urban Transport
Measures to improve energy and/or GHG efficiency in the transport or other urban energy use
sectors can only take us so far and are in any case subject to the rebound effect [30]. Rebound
in energy use occurs because improving the energy efficiency of, for example, cars, lowers
the per kilometre costs of motoring, leading to more kilometres being travelled. In cases
where demand is inelastic, indirect rebound can still occur, because the money saved from
higher energy efficiency and resulting lower fuel costs can now be diverted to the purchase of
other energy-using goods or services. As stressed in Chapter 1, the urgency of climate change
will probably require reductions in the use of energy-using devices, as well as efficiency
improvements. Raising the price of transport energy (by either changing the structure of
motoring costs or imposing a carbon tax, as discussed in Chapter 6) can also prevent energy
rebound. In this section, we first consider the potential for automated vehicles, then look at
approaches to raising the energy efficiency of vehicles by reducing wasted travel, congestion,
etc.
5.4.1. Automated Vehicles: A Possible Solution for Improving Efficiency?
The idea of fully automated, driverless vehicles has a long history: General Motors trialled
one such vehicle in 1935 [41]. In the 1990s, vehicle automation took the form of Intelligent
Highway Vehicle Systems (IHVS). In IHVS, groups of instrumented vehicles would move in
close formation on instrumented freeways. Trials were conducted in the US, as well as in
Europe and Japan. Proponents claimed that IHVS would both save both fuel and increase
road capacity because the close spacing would reduce air friction on all following vehicles in
the ‘platoon’ and reduce road space per vehicle. On non-instrumented roads, vehicles would
need to revert to manual control.
Today, fully automated vehicles (AVs), such as the Google car, have been licensed to
travel on roads in some US states, and millions of vehicle-km of travel have been logged. The
justifications for widespread AV introduction include those for IHVS, but some new
advantages are claimed. Fuel efficiency would be enhanced not only because of closer
spacing on all roads (not only freeways) but because vehicles could now be redesigned. In the
best case of vehicles never having to revert to manual control, the driver-controlled steering
and braking systems could be dispensed with, as well as the need for forward-facing front
seats. Advocates also claim that AVs would be far safer than driver-controlled vehicles,
because in 90% of today’s collisions, the driver is at least partly in error [50]. With this
higher safety margin, even further weight reductions would be possible, because the vehicles
will not need to be as crash-resistant as present vehicles. AVs would thus be both safer and
more energy and GHG efficient.
Lawrence Burns [6] has stressed another possible advantage of AVs—shared car
ownership. Although shared ownership has always been possible, driverless vehicles could
come to be regarded as taxis or hire vehicles. Privately owned cars are presently parked 90%
of the time, allowing up to an 80% reduction in vehicle numbers, thus greatly reducing both
the parking space needed and the embodied energy costs of car manufacturing. He also
envisioned AVs as being electric drive, but, of course, this is possible without AVs.
But would AVs deliver the above-mentioned energy savings, or will AVs instead greatly
increase vehicle-km of road travel, swamping any possible savings from more efficient
vehicles? For a start, fully-automatic driving on all roads under all weather conditions will
not happen any time soon—perhaps not even before 2030 [17]. Until then even AV-capable
cars will need to be designed to enable functioning in both driver-controlled and automatic
modes, making them heavier, and so less fuel efficient, than existing vehicles.
Even if the world switches completely to AVs and they do prove safer than existing
vehicles, energy reductions are not assured. We have already seen that, the perceived costs of
travel include both a time and a money component. But what if AVs allow both to be cut?
Nikolas Thomopoulos and Moshe Givoni [48] have presented data for the US showing that
total time spent driving, both passenger and freight vehicles, in the US amounts to some 75
billion hours per year. Driverless vehicles could thus generate annual productivity gains
which they estimated at $507 billion. Cost savings would also result from fuel savings, as
already discussed. The resulting perceived drop in both the monetary and perceived time
costs of travel could greatly increase private vehicular travel and undercut any existing
advantages of public or non-motorised urban transport.
Further, car ownership and sales could rise instead of fall, for two reasons. First, there
would be no need for a driver’s licence, so even young children—with wealthy parents—
could now have a car. Second, because car ownership and operation costs would fall,
insurance premiums, fuel costs, and car purchase costs should all be much lower.
Nevertheless, given the likely safety problems AVs will face—particularly for Asian cities
were much urban travel is on foot, or by pedal cycle, electric cycle or motorcycle, modes that
cannot be automated, it seems unlikely that AVs can contribute to the sustainability of future
cities. As Eric Bruun and Moshe Givoni [7] have stressed, getting people to give up
driverless vehicles may be even harder than getting them out of conventional cars. Above all,
AVs may never happen, as there are so many social, legal and technical problems still to be
resolved [16].
5.4.2. Improving Vehicle Efficiency
As seen in Section 5.5.1, some have thought that switching to AVs, by enabling a radical
redesign of the vehicle, would lead to a much lighter vehicle and thus less fuel use. However,
if the driving task is dispensed with, seats could face each other, and cars could now be seen
as mobile offices or entertainment centres. In such cases, vehicles could become larger, with
even less fuel economy than conventional vehicles. Here we see that not all proposed uses of
big data will automatically further the aims of urban sustainability. In some cases, at least, we
will need to turn our backs on the possibilities opened up by big data in the interests of
sustainability.
Fuel efficiency in vehicles depends not only on the physical characteristics of the vehicle,
such as weight and engine efficiency, but also on how they are driven, and the traffic
conditions under which they operate. Big data can help the motorist choose a route, based on
current traffic conditions, that will minimise fuel loss caused by congestion, and wasted
travel caused by non-optimal routes. It can also advise the motorist on driving in an ‘eco-
friendly’ manner. Such fuel savings, while useful, will only be of temporary benefit if we
must transition in future to a passenger transport system based largely on public and non-
motorised transport.
Nevertheless, we should not expect too much from energy efficiency gains. Historically,
there have been continuing efficiency gains for each vehicular mode, but at the same time a
progressive shift to faster and more energy intensive modes. Thus, at the global level, public
transport largely replaced non-motorised travel, car travel replaced most public transport, and
now air travel has made inroads into long-distance car travel.
5.5. Beyond Energy Efficiency: Traveller Well-being and Comfort
Reducing the environmental, resource and health-related negative impacts of urban transport
is vital, but is not the full story. Urban travellers must also be able to make their journeys in
comfort and safety, and in a reasonably timely manner. It is also desirable that public
transport services be relatively frequent to avoid long waiting times, and that interchanges
from one public transport service to another if needed, are convenient and fast. On a
passenger-km basis, public transport modes are in general superior to car travel for both
primary energy consumption and GHG (and air pollution) emissions. The energy efficiency
of public transport varies greatly from city to city, but in general is far higher in Asian
cities—regardless of their average per capita income level—than in the cities of North
America, Australasia, or even Europe. The main reason is the far higher urban density of
Asian cities, and their resulting lower levels of both car ownership and car share of urban
travel, both of which drive up patronage—and occupancy rates—on public transport services.
In the cities of Australasia and the US, urban public transport seldom carries more than
10% of all vehicular travel. In contrast, even in the high-income Asian cities such as Hong
Kong and Tokyo, public transport can account for over half of vehicular passenger-km [36].
Very often, the public transport system is already over-loaded—in Tokyo, the number of total
annual vehicular trips, whether by public transport or car, has not risen for decades [45].
Unlike US/Australasian cities, the present problem is not encouraging more travel by public
transport, but getting more value from the existing system. In the longer term, all cities—at
least those in high and middle-income countries—will need to reduce overall vehicular
passenger-km. However, even in the cities of China and India, there is room for further
expansion of public transport systems if they can help reduce burgeoning car travel.
In OECD countries at least, including high-income Japan, average car seat occupancy
rates are around 30%, or 1.5 persons for a five-seater car, with occupancy at peak hours being
lower than at off-peak times. In marked contrast, Asian public transport services—and even
peak hour services on public transport in non-Asian OECD cities—very often have seat
occupancy rates far exceeding 100%. In other, words many passengers are standing, often at
crush volumes. This crowding results in very high energy efficiencies, but at the expense of
passenger comfort, particularly in hot climates. ‘In Beijing in1995, public transport energy
efficiency (passenger-km per MJ of primary energy) averaged over all modes, ‘was about six
times that of private car travel’ [51].
This overcrowding is exacerbated by prolonged travel times—for all modes, including
non-motorised ones. Bangladesh, for example, has one of the lowest car ownership levels in
the world. Nevertheless, its megacity capital, Dhaka, is one of the most traffic congested
cities on the planet (see Figure 5.2). Even in this low-income city, information technology is
being implemented to assist traffic flow and control, and smart cards have been introduced on
some bus services [21].
[‘World Class Traffic Jam’ by b k available at http://bit.ly/2uOhpU1 under a Creative Commons licence
Attribution-ShareAlike 2.0 Generic (CC BY-SA 2.0). Full terms at http://creativecommons.org/licenses/by/2.0.]
Figure 5.2. Traffic congestion in Dhaka, Bangladesh.
5.6. Case Study of a Personal Travel Assistant for Beijing
5.6.1. Background
In Section 5.2.1 we discussed a personal travel assistant (PTA), in the context of increasing
public transport ridership by providing relevant travel time, energy use and carbon emissions
for the various competing modes. On their own, such approaches are unlikely to give the
reductions needed. For example, a small pilot study in Switzerland showed that modest
increases (14%) in ‘sustainable transport choices’ were possible with a mobile phone app that
test subjects used over one month [15]. However, the authors of the study warned that longer
interventions would be needed for more significant increases in the use of sustainable
transport.
In this section, we describe a PTA developed for Beijing, but with a difference. The aim is
to reduce energy (and hence CO2 emissions) from Beijing passenger travel by using the idea
of a transport energy quota for each resident. An analogous, but more inclusive proposal was
considered for the UK: The Personal Carbon Trading scheme. As described by Richard
Starkey [44], this proposal would ‘allocate equal tradeable carbon credits to all eligible
individuals (who may simply be taken here as all adults), with the aim of reducing carbon
emissions, as part of an emissions trading scheme’. Our proposal would be restricted to one
city and passenger transport, but would likewise allow each ‘eligible person’ (however
defined) a monthly transport energy quota in MJ. Although we use energy for the quota, the
scheme could just as easily use kg of CO2 equivalent. China already has in place several
regional pilot emission trading schemes which could eventually form the basis of a nation-
wide scheme [22]. The idea of a personal transport carbon allowance should be a good fit to
national carbon pricing.
In our paper on a PTA for Beijing [51], we gave details of the algorithm used to compute
energy costs for each trip. Briefly, the algorithm calculates the energy cost of each trip, taking
into account the mode used, and for car travel, the traffic conditions likely to be encountered
for that route at the time of travel. Non-motorised modes are assigned zero energy cost. At
any time during the month, the user can calculate the remaining energy quota, and vary the
travel mode for the remaining days to stay within the quota.
In Section 5.6.2 we describe the application of such an energy credit system through a
series of simulations. The ultimate aim would be to implement this system on mobile phone
based navigation software which is supported by the Urban Transport Energy Saver (UTES)
database on the Azure platform. We also give a rough estimate of the data needs of such a
system, assuming a modest level of use.
5.6.2. Description of the Application*
The UTES tool suite discussed here aims for general usability by providing a broad set of
interface options, but with a particular focus on pre-trip information as opposed to in-trip
guidance. The ultimate goal for UTES is to develop a platform whereby every urban sensor,
device, person, vehicle, building, and street can be potentially used to probe city dynamics to
enable city-wide computing that serves the population’s travel activities. UTES aims to
enhance urban transportation through an iterative process of sensing, data mining,
understanding and improving urban transport systems. In the following sections, the system
architecture, design, and implementation of UTES are discussed. The design of UTES
focuses on interactive visual analytics to provide the user with the most appropriate
information. The design will eventually consider factors such as urban residential density,
travel styles, alternative options, costs, time restraints, travel experience, etc.
In contrast to traditional urban transportation systems, UTES would leverage the cloud to
profile and optimize data classifiers for mobile devices, depending on the current device
context and sensor data characteristics, to provide interactive visual analytics for supporting
decision making. This project focuses on analysing and mining dynamic information.
Recently, there has been a trend to using a combination of real-time Global Positioning
System (GPS) and Radio Frequency Identification (RFID) sensors deployed in vehicles,
together with static city sensors and cloud service to optimize eco-efficiency. This approach
can benefit from the ‘city portriait’ concept by embracing a wide range information from
static sensors such as street cameras, traffic lights location and cycles, speed of cars, road
lighting, temperature and general weather information and vehicles themselves, including car
speed and GPS location information. This allows optimal management of vehicles and traffic
in real-time, potentially improving system energy efficiency.
The UTES application will be developed and examined through a series simulations based
on real Graphical Information System (GIS) data, potentially with real-time data feeds. As an
example, Figures 5.3 and 5.4 show an indicative travel simulation within a simulated urban
environment for a given trip. In the window at the right of Figure 5.3, the following elements
are presented:
• A GIS map, with the lines presenting roads in the area. The color of the roads changes
with traffic volumes, with green representing lightly-trafficked roads.
• Buildings are currently classified on the map as either public/business (yellow)
residential (grey)
• Users (current travellers and trip planners) are represented by circular dots on the map;
there are 50 users in total in this simulation
• Transport modes are represented by different circle colours This simulation has three
modes: private vehicles (blue), public transport, ie buses, trains or underground rail
(yellow), bicycle/walk (green).
The top left graph in Figure 5.3 gives the transport mode share: blue for car (20%), yellow
for public transport (16%), green for walk/bicycle (64%). The bottom left corner lists the
mode currently used for each traveller. Figure 5.4 gives the energy consumption of the
overall transportation system by those presently travelling. The blue curve shows maximum
energy consumption assuming all travel is by private vehicles, the red curve energy
consumption optimized by UTES method. As the result of implementing UTES method to
optimize transport, the energy consumptions savings are seen to be significant.
Figure 5.3 UTES project, screen display for travel simulation for a given trip.
5.4. Simulation results for energy cost comparisons.
UTES provides users with a variety of interface options, and the underlying
implementation and technology stack is quite diverse as a result. The UTES server back-end
is written in Java and uses a variety of standard open source development libraries and
frameworks for its implementation. The system is composed of a number of service modules,
each providing specific functionality, coupled together.
In order to verify the performance of the proposed algorithm, a web-based navigation
system was developed. The map information used is the open source map from the Baidu
platform.
1. After the user has registered online, a unique user database is created in the Azure
Platform. In the database, one table stores the user’s current energy quota, another stores the
travel logs.
2. The navigation website at present provides for three different transportation types:
private car, public transport, and walk/cycle. Once the origin and destination and transport
mode are selected, the total trip energy use will be provided immediately from the Baidu
platform.
What levels of data would be generated by such a system? A rough calculation gives some
guidance. Annual Beijing road traffic data are estimated to be about 12 terabytes (TB) per
year, consisting of data from both GPS systems and loop detector data feeds. In 2015, the UN
estimate for Beijing’s population was 18 million people. Assume 5% of the population used
the system twice daily on average, then total daily visits will be 1.8 million. Further assume
uniform daytime distribution over about 10 hours; the system would then need to meet about
60 access requests every second, although at peak hours this might rise to perhaps 300-500
requests every second [51].
5.7. Discussion: Future Urban Travel
Predicting what is likely to happen can help planning, not only in urban transport but also in
energy and health. Even though our forecasts for transport and other areas are often wrong,
we really have no other options. Consider the decision to investigate the construction of a
new urban freeway or metro rail system. The investigation, design, and construction alone
could take a decade, and the freeway or metro will have an operation life of many decades.
Hence the planners have to try to predict its likely use many decades in the future if they are
to assess the project’s viability. Of course, in some cases, provision can be made for changing
future use by stage construction, providing space for extra runways for an airport, or extra
lanes for a freeway to be added if required in future.
Schäfer and Victor [42] forecast global vehicular travel out to 2050, on the basis of travel
time and money budgets. Global air travel was expected to grow at a faster rate than car
travel to accommodate growing passenger-km per capita within a daily travel time budget
constraint. By 2050, they predicted that public transport and non-motorised modes would be
marginal, given their slower speeds. Nevertheless, they anticipated global car travel to
continue to grow strongly out to 2050, reaching about 45 trillion pass-km in that year, a
prediction which receives some support from the Organization of the Petroleum Exporting
Countries (OPEC) car ownership projections. For 2014, OPEC [38] gave an estimate of 1022
million cars globally (142 per 1000 population) and projected this to rise to 2167 million by
2040 (240 per 1000 population). Most of the increase was forecast to occur outside the
OECD countries. Since fuel efficiency improvements were forecast to be insufficient to
compensate for the increased vehicle usage, total passenger car energy use would continue to
rise. Although urban car ownership is often lower than the national average in OECD cities,
because of lower than average inner city resident car ownership, the opposite is usually true
in non-OECD cities.
The global car ownership and travel forecasts both assume a business-as-usual world. But
given the challenges faced by transport discussed in Chapter 1, it is difficult to see how
climate mitigation goals for transport could be met unless actual travel and car ownership are
a mere fraction of these forecasts. It then follows that the benefits obtained from each
passenger-km of urban vehicular travel will need to be far higher than is the case today.
Chaoming Song and colleagues [43] studied the predictability of individual travel patterns
by using the data traces left by mobile phone users. Mobile phone companies have data on
the ‘closest mobile tower each time the user uses his or her phone.’ The researchers found ‘a
93% potential predictability in user mobility’. They further claimed that, with the use of data
mining algorithms, actual predictions about mobility could be obtained. Yu Zheng [54] of
Microsoft Asia, has reviewed how trajectory analysis can be used to better understand the
mobility patterns of not only people and vehicles, but even of wildlife, and has discussed how
data collected from diverse sources can be mined to better understand how urban transport
systems actually function.
Such data could be helpful to cities in moving toward sustainable urban transport systems.
The results of such analyses are, of course, only valid for current transport patterns, which in
turn reflect current transport policies. Nevertheless, these and similar approaches would allow
frequent snapshots of travel to be made, and so allow tracking the changes occurring as a
result of policies needed to support urban sustainability, and if found necessary, to alter
policies in the direction of greater sustainability. They can also be used to track travel
patterns and its changes at very local levels.
References
[1] Aguiléra A , Guillot C, Rallet A 2012. Mobile ICTs and physical mobility: Review and research
agenda. Transportation Research Part A 46: 664–672
[2] Alexander LP, González MC 2015. Assessing the impact of real-time ridesharing on urban traffic
using mobile phone data. UrbComp’15, August 10, Sydney, Australia. Available at
(http://humnetlab.mit.edu/wordpress/wp-content/uploads/2014/04/sig-alternate.pdf)
[3] Batty M 2013. Big data, smart cities and city planning. Dialogues in Human Geography 3(3): 274–
279.
[4] Bodhani A 2012. Smart transport. Engineering & Technology July: 70-73.
[5] BP, 2017. BP statistical review of world energy 2017. BP: London.
[6] Burns LD 2013. A vision of our transport future. Nature 497: 181-182.
[7] Bruun E, Givoni M 2015. Six research routes to steer transport policy. Nature 523: 29-31.
[8] Cameron I, Lyons TJ, Kenworthy JR 2004. Trends in vehicle kilometres of travel in world cities,
1960–1990: underlying drivers and policy responses. Transport Policy 11: 287–298.
[9] Calvillo CF, Sánchez-Miralles A, Villar J 2016. Energy management and planning in smart cities.
Renewable and Sustainable Energy Reviews 55: 273–287.
[10] Cohen-Blankshtain G, Rotem-Mindali O. 2013. Key research themes on ICT and sustainable
urban mobility. International Journal of Sustainable Transportation (DOI:
10.1080/15568318.2013.820994).
[11] Coughlan A 2012. The best medicine. New Scientist 25 August: 38–41
[12] Dal Fiore F, Mokhtarian PL, Salomon I, et al 2014. “Nomads at last?” A set of perspectives on
how mobile technology may affect travel. Journal of Transport Geography 41: 97-106
[13] Department for Transport (DtF) 2015. National travel survey: England 2014. DtF: London, UK.
(Also earlier surveys)
[14] Funk JL 2015. IT and sustainability: New strategies for reducing carbon emissions and resource
usage in transportation. Telecommunications Policy 39: 861–874.
[15] Gabrielli S, Maimon R 2013. Digital interventions for sustainable urban mobility: a pilot study.
UbiComp’13, September 8–12, Zurich, Switzerland.
[16] Goel S 2016. Special Issue on connected vehicles. IEEE Intelligent Transportation Systems
Magazine Fall: 5-7.
[17] Gomes L 2016. When will Google’s self-driving car really be ready? IEEE Spectrum May: 13-
14.
[18] Hannes E, Liu F, Vanhulsel M et al 2012. Tracking household routines using scheduling
hypothesis embedded in skeletons. Transportmetrica, 8(3): 225-241 (DOI:
10.1080/18128602.2010.539418).
[19] Henriksson G, Hagman O, and Andréasson H 2011. Environmentally reformed travel habits
during the 2006 congestion charge trial in Stockholm—a qualitative study. International Journal of
Environmental Research and Public Health 8: 3202-3215 (doi:10.3390/ijerph8083202)
[20] Hodson H 2015. A city of numbers. New Scientist 24 January: 22- 23.
[21] Khan SM, Chowdhury M 2014. ITS for one of the most congested cities in the developing
world—Dhaka, Bangladesh: challenges and potentials. IEEE Intelligent transportation systems
magazine. Summer: 80-83.
[22] Lo AY 2016. Challenges to the development of carbon markets in China. Climate Policy 16(1):
109-124.
[23] Luo X, Dong L, Dou Y et al 2017. Factor decomposition analysis and causal mechanism
investigation on urban transport CO2 emissions: comparative study on Shanghai and Tokyo.
Energy Policy (http://dx.doi.org/10.1016/j.enpol.2017.02.049).
[24] Lyons G 2015. Transport’s digital age transition. The Journal of Transport and Land Use 8(2): 1–
19 (http://dx.doi.org/10.5198/jtlu.2014.751).
[25] Mayer-Schӧnberger V, Cukier K 2014. Big Data. Mariner Books: Boston and NY.
[26] Menon A, Sinha R, Ediga D, Iyer S 2013. Implementation of internet of things in bus transport
system of Singapore. Asian Journal of Engineering Research 1(4): 8-17.
[27] Miller HJ 2013. Beyond sharing: cultivating cooperative transportation systems through
geographic information science. Journal of Transport Geography 31: 296–308.
[28] Moriarty P (1980) The potential for non-motorised transport in Melbourne. Proceedings of the
10th ARRB Conference, Sydney 10 (5): 43-50.
[29] Moriarty P 2016. Reducing levels of urban passenger travel. International Journal of Sustainable
Transportation 10(8): 712-719.
[30] Moriarty P, Honnery D 2011. Rise and fall of the carbon civilisation. Springer: London.
[31] Moriarty P, Honnery D 2017. Reducing personal mobility for climate change mitigation. In
Handbook of Climate Change Mitigation and Adaptation, Second Edition, W-Y Chen, T Suzuki
and M. Lackner (Eds.-in-chief) Springer Science+Business Media: New York.
[32] Moriarty P, Honnery D 2017. Sustainable energy resources: Prospects and policy. Chapter 1 in
M.G. Rasul et al. (Eds) Clean Energy For Sustainable Development, Academic Press/Elsevier,
London.
[33] Moriarty P, Kennedy D 2000. Will telework reduce travel? IS 2000 Conference, Univ. of
Woollongong, NSW, December.
[34] Moriarty P, Wang SJ 2014. Low-carbon cities: Lifestyle changes are necessary. Energy Procedia
61: 2289-2292.
[35] Moriarty P, Wang SJ. 2015. Eco-efficiency indicators for urban transport. Journal of Sustainable
Development of Energy, Water, and Environment Systems 3(2): 183-195.
[36] Newman P, Kenworthy J 1999. Sustainability and cities: Overcoming automobile dependence.
Island Press: Washington, DC.
[37] Nilles JM 1976. Telecommunications-transportation tradeoff: options for tomorrow. John Wiley
& Sons Inc: New York.
[38] Organization of the Petroleum Exporting Countries (OPEC) 2016. World oil outlook 2016.
Vienna: OPEC. (Also earlier editions.)
[39] Pan G, Qi G, Zhang W, et al 2013. Trace analysis and mining for smart cities: issues, methods,
and applications. IEEE Communications Magazine 51(6): 120-126.
[40] Pappalardo L, Rinzivillo S, Qu Z et al 2013. Understanding the patterns of car travel. European
Physics Journal Special Topics 215: 61–73.
[41] Peterson B 2014. The vision of automated driving: what is it good for? Answers from society
with economical and organizational perspectives. Proceedings of the 44th European Microwave
Conference (Rome, Italy): 1723-1729.
[42] Schäfer A, Victor D 2000. The future mobility of the world population. Transportation Research
Part A 34(3): 171-205.
[43] Song C, Qu Z, Blumm N, et al 2010. Limits of predictability in human mobility. Science
327(5968): 1018-1021.
[44] Starkey R 2012. Personal carbon trading: A critical survey. Part 1: Equity. Ecological Economics
73: 7–18.
[45] Statistics Bureau Japan 2015. Japan statistical yearbook 2016. Statistics Bureau, Tokyo. (Also
earlier editions). Available at (http://www.stat.go.jp/english/data/nenkan/index.htm).
[46] Taniguchi E, Thompson RG, Yamada T 2016. New opportunities and challenges for city
logistics. Transportation Research Procedia 12: 5–13.
[47] Tao S, Corcoran J, Mateo-Babiano I, et al 2014. Exploring Bus Rapid Transit passenger travel
behaviour using big data. Applied Geography 53: 90-104.
[48] Thomopoulos N, Givoni M 2015. The autonomous car—a blessing or a curse for the future of
low carbon mobility? An exploration of likely vs. desirable outcomes. European Journal of Futures
Research 3: 14pp (DOI 10.1007/s40309-015-0071-z).
[49] United Nations (UN) 2014. World urbanization prospects: the 2014 revision. Accessed on 30
November 2016 at (https://esa.un.org/unpd/wup/CD-ROM/). Also the 2011 revision.
[50] Waldrop MM 2015. No drivers required. Nature 518: 20-22.
[51] Wang SJ, Moriarty P, Ji YM, et al 2015. A new approach for reducing urban transport energy.
Energy Procedia 75: 2910–2915 (Doi:10.1016/j.egypro.2015.07.588).
[52] Wikipedia 2017. Congestion pricing. Accessed on 22 May 2017 at
(https://en.wikipedia.org/wiki/Congestion_pricing).
[53] Wikipedia 2017. Floating car data. Accessed on 14 August 2017 at
(https://en.wikipedia.org/wiki/Floating_car_data).
[54] Zheng Y 2015. Trajectory data mining: an overview. ACM Transactions on Intelligent Systems
and Technology 6(3). Available at (https://www.microsoft.com/en-us/research/wp-
content/uploads/2015/09/TrajectoryDataMining-tist-yuzheng.pdf).
[55] 龙瀛, 张宇, & 崔承印 2012. 利用公交刷卡数据分析北京职住关系和通勤出行. 地理学报
67(10): 1339-1352.
NOTE
*This section is a revised and updated version of Section 2.2 in Reference [51]. Used by permission of
Elsevier.
