Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Sustainable Cities and Society 99 (2023) 104976

Contents lists available at ScienceDirect

Sustainable Cities and Society


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

Spatiotemporal analysis of CO2 emissions and emission reduction potential


of Beijing buses using smart card data
Jiangpeng Zheng a, b, Zhou Huang a, b, *, Xiao Zhou c, Bronte Scheuer a, b, Han Wang d
a
Institute of Remote Sensing and Geographic Information System, School of Earth and Space Sciences, Peking University, Beijing 100871, China
b
Beijing Key Lab of Spatial Information Integration & Its Applications, Peking University, Beijing 100871, China
c
College of Urban and Environmental Sciences, Central China Normal University, Wuhan 430079, China
d
Future Urbanity & Sustainable Environment (FUSE) Lab, Division of Landscape Architecture, Faculty of Architecture, The University of Hong Kong, Hong Kong SAR,
China

A R T I C L E I N F O A B S T R A C T

Keywords: Human activities, primarily carbon dioxide emissions, have undeniably caused global warming. The trans­
CO2 emissions portation sector contributes about a quarter of global CO2 emissions. While replacing traditional buses with
Bus smart card electric ones has reduced emissions, it is crucial to consider the indirect emissions resulting from electricity
Spatiotemporal characteristics
consumption. This study proposes a framework for modeling bus emissions using smart card data, integrating
Per person-kilometer
spatiotemporal distribution characteristics and emission reduction potentials. Our analysis reveals that routes
Emission reduction potential
spanning 10–30 km contribute to 81% of total bus emissions, with an average emission rate of 56.2 gCO2/per-km
for residents traveling by bus. Bus emissions also exhibit cyclical variations during holidays, weekdays, and
weekends, indicating spatial clustering and trends. Although the area within Beijing’s 4th Ring Road constitutes
only 13% of the total area within the 6th Ring Road, it generates almost half of the CO2 emissions. With urban
expansion, total bus emissions increase gradually, but emission intensity decreases. This study emphasizes the
potential for reducing emissions through improved public transportation operations. It recommends fully elec­
trifying the bus fleet and employing low grid emission factors, which could reduce emissions by up to 71%
compared to diesel options. Electrification of buses and optimizing power generation on the grid are essential
priorities for emission reduction.

1. Introduction average growth rate of CO2 emissions in the transportation sector be­
tween 2013 and 2019 remained above 5%, making it the fastest-growing
1.1. Background sector in the country. This underscores the urgent need to address
transportation-related emissions in order to evaluate potential risks. As a
Since around 1750, human activities, primarily through the emission result, transportation industry is considered one of the most challenging
of carbon dioxide (CO2) from the combustion of fossil fuels and indus­ sectors to achieve ‘carbon neutral’.
trial processes, have unequivocally caused global warming (IPCC, To address climate change, a worldwide environmental issue caused
2023). As a major contributor to energy consumption and CO2 emis­ by human activities, governments around the world have been actively
sions, the transportation sector presents great potential to save energy engaged in global ecological governance. Currently, governments
and reduce emissions. According to the International Energy Agency, committed to achieving ‘net zero emissions’ aim to decrease global
the transportation sector accounts for 24% of global carbon emissions in greenhouse gas emissions by roughly 90% (Rogelj et al., 2018). China,
2019, making it the second-largest emitting sector in the world after the the world’s largest carbon emitter, has committed to peak emissions by
energy sector (IEA, 2021). In China, the transport sector accounts for 2030 and achieve net-zero emissions by 2060 (Xi, 2020).
about 11% of the country’s total CO2 emissions, with roads accounting Cities are recognized as a key area for the mitigation of global carbon
for about 90% of transport emissions (Li et al., 2021). Furthermore, the emissions. Exhaust emissions from the burning of fossil fuels in motor

* Corresponding author at: Institute of Remote Sensing and Geographic Information System, School of Earth and Space Sciences, Peking University, Beijing 100871,
China.
E-mail address: huangzhou@pku.edu.cn (Z. Huang).

https://doi.org/10.1016/j.scs.2023.104976
Received 24 April 2023; Received in revised form 14 July 2023; Accepted 28 September 2023
Available online 5 October 2023
2210-6707/© 2023 Elsevier Ltd. All rights reserved.
J. Zheng et al. Sustainable Cities and Society 99 (2023) 104976

vehicles significantly reduces air quality within urban areas. The shift approaches typically employ energy consumption and conversion fac­
from private cars to buses has the potential to reduce energy con­ tors to estimate CO2 emissions from transport, providing a broader
sumption and emissions from road passenger transportation (Wei & regional or national-level assessment of emissions (Li et al., 2019;
Frey, 2022; Yin et al., 2023). Buses can carry a large number of pas­ Morrow et al., 2010). However, for a specific mode of transport, a
sengers per trip, and the number of passengers carried is inversely bottom-up model based on ‘activity-transport-density-fuel consumption’
proportional to emissions on a per capita basis. A 20% increase in is more commonly used (Schipper et al., 2000). This approach relies on
ridership is associated with a 1.1% increase in total emissions and a 13% accurate data such as vehicle type, distance traveled, and unit fuel
reduction in per capita emissions (Waraich et al., 2020). Therefore, consumption, and is better suited for estimating CO2 emissions from
buses are widely regarded as a more environmentally friendly option for small-scale transport systems, such as those in cities (Zhang et al., 2019).
reducing per capita CO2 emissions in the road transport industry. Additionally, it is particularly useful for comparative assessments of
However, due to the low conversion efficiency of their engines, they are low-carbon transport levels between cities (Li et al., 2019).
a serious source of air pollution (Ma et al., 2021). The advantage of new High-resolution vehicle CO2 emission inventories at various scales
energy vehicles, especially pure electric and fuel cell vehicles, is that are crucial for informed decision-making and further research on
they do not produce emissions while in use. Therefore, significantly achieving carbon neutrality in a country or region. By employing a
increasing the share of electrification in transportation is a critical effort bottom-up approach, researchers created a high spatial resolution (1 km
to reduce global fossil energy consumption and CO2 emissions to ensure × 1 km) inventory of China’s vehicle CO2 emissions and found that the
sustainable human development. distribution of motor vehicle emissions in China is unbalanced (Xu et al.,
By the end of 2021, there will be 709,400 urban public buses in 2021), with most of the high-emission networks within city limits
China, of which 419,500 will be purely electric (MOT, 2021). The concentrated in the city center (Sun et al., 2021). Another study
widespread use of electric buses is often seen as an effective measure to analyzed CO2 emissions from road traffic in Shenzhen at a high spatio­
reduce energy consumption and thus CO2 emissions. However, when temporal resolution using traffic trajectory data, which found that the
measuring their environmental efficiency from a whole life cycle city’s main roads are high-emission areas with notable daytime emission
assessment perspective, the emissions from the production process of trends (Zhou et al., 2022). Similarly, emission inventories at different
electric buses are not lower than those of regular fuel vehicles (Wang & scales, such as link-scale, county-scale, and city-scale, can help uncover
Tang, 2022). With the continuous optimization of energy mix and the the spatiotemporal distribution and aggregation of pollutants (Cheng
downward adjustment of the emission factors of the local grid, the et al., 2023). Diesel and natural gas buses are the primary contributors to
emission reduction benefits of electric buses are gradually increasing. daytime NOx emissions in urban areas, while buses and heavy trucks
The diversification and integrated use of public transport modes has play major roles in suburban areas (Liu et al., 2022). However, given the
influenced the travel patterns and structures of urban residents (Wang current widespread electrification of buses, there is still a lack of
et al., 2023; Zhou et al., 2023), which in turn affects the CO2 emissions research on the benefits of electrification in reducing emissions in cities.
patterns of urban transportation. Most previous studies have used a Buses exhibit spatial heterogeneity in their emission characteristics
bottom-up approach to determine bus emissions by using vehicle tra­ across urban areas throughout the day due to the large number of pas­
jectory data. While this is a feasible approach to evaluate CO2 emissions sengers they carry. The use of trajectory and smart card data can reveal
from individual vehicles or over a short period, the spatial and temporal the spatiotemporal distribution of bus CO2 emissions, providing insight
characteristics of overall bus emissions in urban areas remain into the pattern of public transport emissions across the city. For
under-researched, and the emission variability and reduction potential instance, it has been observed that during off-peak hours, the suburbs
of bus routes have not been explored. exhibit lower emission benefits (Sui et al., 2020) due to higher fuel and
Based on bus smart card data in Beijing, this study uses a bottom-up emissions per person-kilometer caused by lower passenger volume
approach to calculate CO2 emissions for each route over a multi-day (Chen et al., 2017). Therefore, it is not always the case that low pas­
range. The Emerging Hot Spot Analysis is used to explore the spatial senger volume buses are inefficient in terms of energy use and CO2
and temporal characteristics of bus emissions under a high-resolution emissions reduction (Wei & Frey, 2022). However, most studies have
holiday-weekday-weekend days. The results of this study can help pol­ only examined vehicle CO2 emissions on a one-day scale, with insuffi­
icy makers and transport planners to implement targeted emission cient information on the spatiotemporal characteristics of buses emis­
reduction measures in key high-emission areas, allocate bus transport sions as a daily urban commuter over multiple days. Furthermore, there
capacity more efficiently and improve the overall sustainability of the is a lack of in-depth research on spatial mining of route emissions
transportation system. reduction using variations in route metrics in conjunction with the
This study makes three key contributions. First, a novel methodo­ proliferation of electric buses and the reduction of vehicle energy
logical framework is proposed to model bus emissions using smart card consumption.
data. The approach integrates spatiotemporal distribution character­
ization and multi-perspective potential mining for emission reduction. 2. Methodology
Second, the research addresses a significant research gap concerning the
spatiotemporal analysis of bus carbon emissions over multi-day periods. 2.1. Framework
Third, this study combines multiple emission factors to assess emission
reduction benefits, while also utilizing passenger flows to explore the Fig. 1 illustrates the methodological framework proposed in this
potential for emission reduction based on per person-kilometer and study. Bus smart card data were used to estimate the number of daily
passenger load factor. departures (NoD) and to calculate the daily passenger flow (PF), average
The rest of the paper is organized as follows: Section 2 describes the passenger ridership (APR), passenger turnover (PT), and average dis­
methodology used in this study, Section 3 presents the study area and tance traveled (ADT). A bottom-up CO2 emission model was used to
data processing, Section 4 presents the results and discussion, and finally calculate the daily emission indicator for each route. A spatiotemporal
Section 5 summarizes the paper and future research directions. cube model was developed for different cycles of CO2 emission data, and
the Emerging Hot Spot is employed to analyze the spatiotemporal
1.2. Literature review characteristics of the routes. Three emission reduction schemes were
established: (1) using of a multi-year integrated emission factor to
According to the Intergovernmental Panel on Climate Change (IPCC) explore emission reduction benefits, (2) comparing the CO2 emission
guidelines, methods for calculating CO2 emissions from mobile sources indicator per person-kilometer (per-km) with private cars to assess the
can be classified as ‘top-down’ and ‘bottom-up’ (IPCC, 2006). Top-down emission reduction potential of low-emission benefit routes, and (3)

2
J. Zheng et al. Sustainable Cities and Society 99 (2023) 104976

Fig. 1. The flowchart of the proposed methodological framework.

comparing the APR of routes with private cars to examine the optimi­ ensure minimal waiting time for passengers. However, in scenarios of
zation of low passenger load on CO2 emissions. traffic congestion, buses may arrive more frequently on the same route.
The methodology presented in this paper allows for the analysis of To accommodate such circumstances, a one-minute swipe interval is
CO2 emission patterns using only smart card data, which includes details employed to record the number of swipes. The estimation of NoD in­
of boarding and alighting stations. In some cities with smart card volves filtering the number of card swipes on board at each station and
mechanisms, passengers are only required to swipe their card when comparing them to determine the maximum station departure number.
boarding, and a fixed fare is charged. In such cases, it is necessary to Fig. 2 shows the detailed flowchart of this process. It is important to note
estimate the drop-off stations (Tang et al., 2020). that some passengers may swipe their cards in advance when
approaching the next station. Hence, the estimation of NoD at alighting
2.2. Bus route indicator stations may not be highly reliable.

2.2.1. Bus route 2.2.3. Passenger turnover


A route r = (r1 , r2, ⋯, rn ) is a series of bus stations. Where n is the Fr,t

number of stations on the r route, ri , ri+1 is two adjacent stations Tr,t =
(
dk ri →rj
)
(1)
(1 ≤ i ≤ n − 1). (ri →rj ) is a section of the route between ri station and rj k=1
station, the station spacing is d(ri →rj ), and the route operating length
(ROL) is Lr . NoD of the bus route r at time t is Nr,t , PF is Fr,t . where Tr,t is PT of bus route r at time t and dk (ri →rj ) is the distance
travelled by passenger k from station ri to station rj .
2.2.2. Number of bus departures
Typically, buses follow a predefined time interval for their arrival to

Fig. 2. Flowchart of bus departure estimation.

3
J. Zheng et al. Sustainable Cities and Society 99 (2023) 104976

2.2.4. Average distance traveled


EFcar,m
ECperkm = (9)
Tr,t f
T r,t = (2)
Fr,t
Where f is the number of people in the private car.
Where Tr,t is ADT by bus route r at time t. Fr,t is PF of the bus route r at In this study, at least one driver is present on board the bus, and the
time t. driver’s turnover is included in the calculation of per-km CO2 emissions.
However, when comparing PT to other indicators such as NoD, only the
2.2.5. Average daily ridership number of passengers is taken into account.
The scope of carbon accounting is typically divided into production-
F r,t =
Fr,t
(3) based and consumption-based (Zhou et al., 2022). The Global Protocol
Nr,t for Community-Scale Greenhouse Gas Emissions Inventories provides
guidelines for system boundaries for production-based GHG emissions
Where Fr,t is ADR of bus route r at time t, NoD of bus route r at time t
(WRI, 2014), which include Scope 1 GHG emissions from sources within
is defined by Nr,t .
city boundaries (including CO2, CH4, and N2O) and Scope 2 emissions
from grid supply, including those energy sources that cross city
2.3. CO2 emission model boundaries. This encompasses emissions from city-wide activities such
as heating, transportation, and waste incineration. However, it does not
2.3.1. CO2 emissions include Scope 3 emissions (e.g., oil and gas production), as these fall
A bottom-up model based on vehicle miles travelled and emission under consumption-based accounting (Harris et al., 2020). The British
factors is used to estimate CO2 emissions. The formula for calculating is Standards Institution provides a clear definition of consumption-based
as follows: accounting, which “captures the direct and life-cycle GHG emissions
of all goods and services consumed by urban residents, i.e. GHG emis­
EBr,t = Nr,t × Lr × EFt (4) sions are attributed to the final consumers of goods and services, rather
Where EBr,t (unit: kgCO2) is the CO2 emissions of the bus route r at than the original producers of those GHG emissions” (BSI, 2013).
time t, and EFt (unit: kgCO2/km) is the composite emission factor at time
t, which represents the amount of CO2 emitted per unit of raw material, 2.4. The space time cube model
fuel and energy consumption or product output. Since the smart card
data does not provide information on the energy source used by bus One of the most powerful human abilities is the use of visual
vehicles, EFt for the bus fleet is utilized in this study to estimate their perception, which helps us to identify complex trends and patterns in
actual emissions. The composite emission factor is calculated as follows: spatiotemporal data (Ahmadi et al., 2022). Creating a space-time cube
∑ allows us to visualize and analyze spatiotemporal data, in the form of
EFt = φt,m × EFbus,t,m (5) time-series analysis, integrated spatial and temporal pattern analysis,
m
and 2D and 3D visualization techniques (Esri, 2022).
where φt,m is the proportion factor of buses consuming m energy at time ArcGIS Pro is used to develop spatiotemporal cubes at various times.
t, which EFbus,t,m is the CO2 emission factor of buses consuming m energy Every bin has a fixed position in space (x, y) and in time (t), which
at time t (unit: kgCO2/km). According to their energy consumption, contains the bus CO2 emission information. Bins covering the same (x, y)
buses are divided into diesel buses (DB), natural gas buses (NGB), and area share the same location ID. Bins encompassing the same duration
new energy buses. In this study, NGB and new energy buses are not share the same time-step ID, and these bins can be combined to form a
subcategorized into more specific energy types, and new energy buses time slice (Fig. 3).
are uniformly assumed to be electric buses (EB).
In Section 4.3.3, the emissions estimate for each year is made by 2.5. Emerging hot spot analysis
using the annual operating mileage combined with a composite emission
factor. The calculation formula is as follows: The Emerging Hotspot Analysis tool calculates the Getis-Order Gi*
/
∑ statistic for each element in the dataset. By calculating the Gi* statistic
EBy = εy × EFy EFt × EBr,t (6) for spatial features, it is possible to reflect the degree of aggregation of
r high value areas (hot spots) and low value areas (cold spots) for spatial
Where εy is the ratio of operating miles in the year y to the base year variables. The calculation is as follows (Ord & Getis, 1995):
t. EBy and EFy are the carbon emissions and the composite emission ∑n ∑n
j=1 ωi,j xj − X j=1 ωi,j
factors for the other year scenarios at t = y, respectively. G∗i = √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅ (10)
√[ (∑ )2 ]

The CO2 emission of private cars are calculated as follows: √ n
√ n j=1 ωi,j −2 n
j=1
ωi,j
S
EC = L × EFcar,m (7) n− 1

∑n
Where EC is the CO2 emissions of private car (unit: kgCO2), L is the j=1 xj
distance traveled of private car, and EFcar,m is the emission factor for a X= (11)
n
private car consuming m energy (unit: kgCO2/km).
√∑
̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
n 2
j=1 xj
2.3.2. Per person-kilometer (per-km) CO2 emissions S= − (X)2 (12)
n
The per-km CO2 emissions method considers both PF and PT. The
carbon emissions per-km of buses (unit: kgCO2/per-km) are calculated Where xj denotes the spatial feature attribute value j, which ωi,j is the
as follows: spatiotemporal weight between the feature i and j, n is the total number
of elements.
EBr,t
EBperkm,r,t = (8) Emerging hotspot analysis tools can identify both data trends, such
Tr,t
as new, intensifying, decreasing and sporadic hot and cold spots, and
The formula for calculating the per-km CO2 emissions of private cars patterns that change over time (Esri, 2022). This study utilizes Getis-Ord
(unit: kgCO2/per-km) is as follows: Gi* statistics and Emerging Hotspot Analysis, which are implemented

4
J. Zheng et al. Sustainable Cities and Society 99 (2023) 104976

Fig. 3. Space time cube structure.

using ArcGIS Pro. In the analysis parameter settings, the time step in­ within the 6th Ring Road accounts for 78% of the total population of the
terval is set to 12 h, the spatiotemporal cube size is 500 m, and the city.
neighborhood distance interval is 3500 m.
3.2. Data and preprocessing
3. Study area and data
3.2.1. Smart card data
3.1. Study area Smart Card data of buses is provided by Amap (https://www.amap.
com/). The Beijing bus system uses a smart card mechanism for both
Beijing is the capital of the People’s Republic of China, which has a boarding and alighting operations. By utilizing an integrated circuit (IC)
total area of 16,410.54 km2, a resident population of 21.54 million at the card, the passenger’s ride-related information such as the bus smart card
end of 2019, a motor vehicle fleet of 6.37 million. The city’s large size, ID, ride route, boarding/alighting time, station details, and location
large population and large number of vehicles lead to a wide range, high coordinates are collected and stored. The raw smart card data covers a
volume, and long distances for residents to travel. It is the largest city in period of one month (May 2019). However, the data for May 11 is absent
China in terms of bus traffic, with 3.13 billion passenger trips in 2019. and has been estimated by calculating the average values based on the
The combination of a large vehicle stock, traditional fossil fuel energy data from other weekends. The total number of card swipe data is about
mix, and poor traffic conditions has led to serious greenhouse gas 184.4 million.
emission problems (Li et al., 2020). In this study, we focus on the urban The smart card data contained many abnormal records, including
area within the 6th Ring Road, which has a more concentrated road travel times of less than 1 min and records of boarding and alighting at
network and population, as shown in Fig. 4. The population living the same station. After cleaning these abnormal records, the total

Fig. 4. Study area. (a) location of Beijing in China, (b) locations of the 6th Ring Road in Beijing, (c) bus routes in the study area.

5
J. Zheng et al. Sustainable Cities and Society 99 (2023) 104976

number of passengers in the 6th Ring Road is about 130.6 million, with Table 2
an average daily passenger volume of about 4.2 million, accounting for CO2 emissions (CEs) factor of different vehicle types.
70.8% of the original data, which basically matches the population share No. Vehicle Fuel Fuel CEs CEs factor Applicable
within the 6th Ring Road. Since the smart card data includes the di­ consumption coefficient time
rection of travel and the station coordinates for bus routes in Beijing 1 NGB – – 0.8965 2017, 2019
differ between positive and negative directions, the operating mileage kgCO2/km
also differs by direction, leading to differences in the calculated amount (Yue et al.,
of carbon emissions. To address this issue, we added driving directions 2016)
2 DB 0.435 L/km ( 2.63 kgCO2 1.1441 2017, 2019
to the exported lines and identified a total of 1200 bus operating lines Ma et al., 2021) /L (IPCC, kgCO2/km
within the 6th Ring Road. 2006)
3 EB 1.42 kWh/km ( 0.8843 1.2557 2017
3.2.2. Bus annual operating mileage and type composition Ma et al., 2021) kgCO2/kWh kgCO2/km
(CCD, 2012)
Beijing is a megacity where the expansion and improvement of rail
4 – – 0.6760 2019
transportation has resulted in a decline in the number of bus trips every kgCO2/km
year. Despite the normalization phase of COVID-19 prevention and (Lv et al.,
control in 2021, and the gradual resumption of normal commuting by 2021)
the public, the proportion of bus trips continued to decrease by 0.2% 5 0.677 kWh/km 0.5810 0.3933 2021
(Wang & Yao, kgCO2/kWh kgCO2/km
year on year. The annual operating mileage of Beijing’s urban buses was 2022) (MEE, 2022)
1354.11 million km in 2015, 1323.57 million km in 2017, 1278.60 6 Gasoline – – 0.2030 2019
million km in 2019, and 1139.00 million km in 2021 (BTI, 2022). The Private kgCO2/km
composition of buses in Beijing by energy type is shown in Table 1. Car (Lv et al.,
2021)

3.2.3. Emission factors


This study uses CO2 emission factors for various types of energy- the autocorrelation gradually increases with NoD, PF, and PT.
based buses that are appropriate for the year of analysis, as shown in Since the differences in Moran’s I index are small and the spatial
Table 2. distributions do not vary much, the clustering/outlier distribution of the
daily averages of NoD, PT, and PF are visualized as shown in Fig. 6.
4. Results and discussion Compared with PF and PT, the clustering of NoD is more concentrated in
the 4th Ring Road, but its distribution of their low/high anomalies is
4.1. Bus route indicator more scattered. The high/high clusters of PF and PT are more concen­
trated in the east-west direction, while the low/low clusters are more
In the following discussion, we analyze the route indicator charac­ concentrated in the north-south direction. Overall, the spatial structure
teristics and the relationship with CO2 emissions. of the three indicators shows an unbalanced distribution, which is
consistent with previous studies (Sui et al., 2020). As will be demon­
4.1.1. Bus route indicator characteristics strated in the next section, these uneven distribution characteristics also
Fig. 5 compares the daily average of NoD, PF, and PT for all bus lead to differences in the spatial distribution of emissions.
routes during holidays, weekdays, and weekends. NoD showed less
variation among the three categories, with weekdays having 16.02% 4.1.2. Route indicators and CO2 emissions
and 16.69% more departures than weekends and holidays, respectively. Fig. 7 shows the scatter statistics of ROL, NoD, PF, PT, and CO2
These additional departures resulted in 43.37% and 44.40% differences emissions of buses with Pearson correlation coefficients of 0.73, 0.74,
in PF and 39.45% and 37.95% higher PT on weekdays compared to 0.89, and 0.90, respectively. Although all correlations are positive, the
weekends and holidays, respectively. There is typically more bus traffic correlations between the indicators differ. The relationship between
on weekdays than on weekends and holidays, as commuting to work is a ROL and emissions is concentrated on short and medium distance routes
primary reason for bus use on weekdays. However, this difference in within 40 km. Moreover, as shown in the scatter plot, the correlation
traffic also results in differences in CO2 emissions from buses. between PF/PT and CO2 emissions is higher than that between ROL/
During the May Day holiday, the NoD and PF on buses were lower NoD and CO2 emissions. This suggests that PF and PT provide a more
than that on weekends; however, the opposite was true for PT. This is accurate reflection of buses actual travel patterns, which is useful when
because the tourist attractions in Beijing are scattered, and buses are comparing carbon emissions per person-kilometer.
mostly used for long-distance trips to these attractions during the peak
tourist season. As a result, there may be less PF but higher PT.
Table 3 shows the Moran indices of the bus route indicators for 4.2. Emissions characteristics
holidays, weekdays and weekends. All Moran indices are positive, cor­
responding to a p-value of less than 0.1 for each index, statistically The CO2 emission indicators for bus routes are calculated by Eqs. (4)
rejecting the null hypothesis. The three indicators exhibit significant and (8).
spatial autocorrelation during different time periods, and the strength of
4.2.1. Temporal characteristics
We analyzed the temporal characteristics of CO2 emissions from two
Table 1 perspectives: (1) total daily emissions, (2) daily emissions for different
The composition of various energy buses in Beijing (BPT, 2019). ROL.
Number and Bus type May 1–4, 2019 was the International Labor Day holiday, and May
proportion 5th was a Sunday with a switched workday schedule implemented.
DB NGB EB Fig. 8 presents a comparison of the daily average CO2 emissions and
Year 2017 10,495, 8397, 7815, 29.26% normalized NoD, PF, and PT during this period. The bus route indicators
39.30% 31.44% and CO2 emissions show a cyclical weekly variation pattern influenced
2019 4402, 19.89% 7547, 10,185, by the travel characteristics of residents. The daily average values of
34.10% 46.02%
holiday, weekday, and weekend emissions are 1,140,614 kgCO2,

6
J. Zheng et al. Sustainable Cities and Society 99 (2023) 104976

Fig. 5. The violin plots of NoD, PF and PT on holidays, weekdays and weekends. (The p-value of less than 0.05 indicates a significant difference between sam­
ple groups.).

Table 3
The Global Moran’s I of Bus Route Indicators.
Day Holiday Weekday Weekend

Indicators Moran’s I Z-value P-value Moran’s I Z-value P-value Moran’s I Z-value P-value

NoD 0.3220 17.7381 0.0000 0.3103 17.0919 0.0000 0.3171 17.4687 0.0000
PF 0.3921 21.6688 0.0000 0.3773 20.8430 0.0000 0.3766 20.8085 0.0000
PT 0.4232 23.5103 0.0000 0.4201 23.3596 0.0000 0.4212 23.4166 0.0000

Fig. 6. Clustering/anomaly analysis of the route’s (a) NoD, (b) PF, (c) PT.

1,304,141 kgCO2, and 1,148,486 kgCO2, respectively. Weekday CO2 4.2.2. Spatial characteristics
emissions are 13.55% higher than weekend and 14.34% higher than Fig. 10a shows the full month CO2 emissions of buses in a 500 m ×
holiday due to the higher NoD. Lower emissions are also found on 500 m grid within the 6th Ring Road. The average emission of non-zero
Sundays than on Saturdays, possibly due to the fact that relatively fewer grids is 4174 kgCO2/500m2, and areas with unit emission intensity
residents choose to travel on Sundays because of Monday is a working higher than 40,000 kgCO2/500m2 are Xiyuan, Zoo, Liuliqiao, Sihui, etc.,
day. mostly with high transit throughput or dense road networks. Liuliqiao (a
Fig. 9 shows a heat map of CO2 emissions for different ROLs, which first-class passenger hub with more than 5000 bus trips per day) is the
indicates that emissions of various routes are generally lower on holi­ area with the highest emissions, up to 83,937 kgCO2/500m2. The grid
days and show alternating changes on weekdays and weekends. Most CO2 emissions show obvious spatial heterogeneity, with higher emis­
(93%) of the emissions come from routes that are between 7 and 37 km sions in main urban areas within the 5th Ring Road network.
in length, with the main routes supporting Beijing’s bus passenger Meanwhile, the 3rd and 4th Ring Roads, which serve as principal
transportation being those between 10 and 30 km in length, accounting transportation arteries, incur significant carbon emissions. Furthermore,
for 81% of the overall emissions. highways connecting Beijing surburbs are also associated with high CO2
emissions. The contrast in CO2 emission intensity between these roads is
evident through the gridded emission pattern. Since most of the long

7
J. Zheng et al. Sustainable Cities and Society 99 (2023) 104976

Fig. 7. Comparison of ROL, NoD, PF, PT and CO2 emissions.

Fig. 8. Daily CO2 emissions and normalized NoD, PF, and PT for May 2019 .

8
J. Zheng et al. Sustainable Cities and Society 99 (2023) 104976

Fig. 9. Heat map of daily CO2 emissions for different ROL (0–74 km, 10–30 km).

Fig. 10. Carbon emissions for 500 m × 500 m gridding


(a) within the 6th Ring Road for the whole month, (b) between different ring roads, (c) Pareto cumulative plot.

operating length buses operating on these roads are diesel-powered, within the 6th Ring Road (2267 km2), but accounts for 48.75% of the
future initiatives could potentially introduce electric buses or battery- total emissions. Whereas, areas within the 5th Ring Road (667 km2),
swap modes to reduce emissions. which accounts for 29.42% of the 6th Ring Road’s total area, accounts
The Beijing road network has a prototypical radial distribution in the for 72.24% of the total emissions. Although the main urban area within
form of concentric rings (Li et al., 2020). Fig. 10b illustrates the com­ the t 4th Ring Road is relatively small, it generates nearly half of the CO2
parison of emissions from different ring roads. The area encompassed by emissions and needs to be prioritized for emission reduction measures.
the 4th Ring Road (302 km2) accounts for only 13.32% of the total area The analysis reveals that bus emissions escalate with increasing area

9
J. Zheng et al. Sustainable Cities and Society 99 (2023) 104976

within the ring road, from the city core to the suburbs. Notably, the exists at the periphery of the intensifying hot spots, suggesting that the
highest emissions per square kilometer occur between the 2nd and 3rd emission intensity outside the 4th Ring Road is not as robust as that
rings, and then gradually decline outward. Overall, while the total CO2 within the 4th Ring Road, and forms an outwardly decay over-zone.
emissions from public transport rise with the built environment, the Additionally, consecutive hot spots are observed in the Xi’erqi area
intensity of emissions decreases. and northeast part of the 2nd Ring Road, probably because the former is
Fig. 10c presents the Pareto cumulative plot of the emission grids, a high-tech science and technology park and the latter is a bustling
revealing the heterogeneous spatial distribution of CO2 emissions. Out hutong commercial area in the city center with a large amount of t bus
of the total 9306 grids, 54.49% (5071) exhibit zero values, and as the capacity on both weekdays and weekends.
number of emission grids increases, the cumulative curve flattens from We2Wd shows clear persistent hot spots and persistent cold spots,
steeply rising to ultimately reaching 100%, indicating uneven spatial with overall distribution consistent with the hot and cold spots, but with
distribution of CO2 emissions. Notably, 1545 grids fall into the orange some variation between weeks. It indicates that high/low emission
region, which comprises only 36% of non-zero emission grids, but leads values at these locations remain clustered from non-working days to
to approximately 80% of CO2 emissions. The yellow region contains working days, with no significant change in clustering intensity. In
64% of grids but contributes only 20% of the emissions. These findings contrast, the diminishing cold spots in Hd2Wd marks a decline in the
highlight the heterogeneity of spatial distribution of CO2 emissions clustering of low values of bus emissions outside the 4th Ring Road,
across urban public transport. along with an overall increase in emission intensity. This indicates that
the evolution process of Hd2Wd and We2Wd emissions is spatially non-
4.2.3. Spatiotemporal characteristics uniform and displays spatial heterogeneity.
(1) Spatiotemporal characteristics of emissions Wd2We exhibits an opposite pattern compared to We2Wd, with
We divided the cyclical CO2 emission change pattern into three cy­ fewer hot spots in the main city within the 4th Ring Road and stronger
cles - holiday to weekday (Hd2Wd), weekend to weekday (We2Wd), and cold spots outside the 4th Ring Road, indicating a decline in bus carbon
weekday to weekend (Wd2We). The spatiotemporal distribution of emissions across the entire region during weekends.
emissions for each cycle is presented using the Emerging Hot Spot (2) Spatiotemporal characteristics of per person-kilometer emissions
Analysis, as shown in Fig. 11. Fig. 12 displays the CO2 emissions per person-kilometer for different
The trend boundary of CO2 emissions lies between the 4th and 5th routes on holidays, weekdays, and weekends. A lower value indicates a
ring areas, forming a concentrated hot spot area inside and loosely higher emissions benefit for that route.
defined cold spot area outside. The low population density and sparse The average daily emissions per-km is 0.0562 kgCO2/per-km. Routes
bus network in this area, which falls within the urban-suburban transi­ with emission values below 0.05 kgCO2/per-km per day are mostly
tion zone, explains this phenomenon. The three cycles are distinctly located on the 3rd and 4th ring main roads and highways s connecting to
demarcated in hot and cold spot regions, with a non-clustered area of Beijing’s suburbs due to their long operating length and high PF values.
emissions in space and time. Most routes have an emission value of 0.05–0.10 kgCO2/per-km and are
Hd2Wd and We2Wd exhibit a similar periodicity, both transitioning scattered throughout the main roads in the urban areas of Beijing, with a
from non-working days to working days. In the hot spot analysis, their similar situation in Qingdao (Sui et al., 2020). Furthermore, these routes
intensifying hot spots displays the largest proportion in time and space, are more widely distributed during weekdays. The high emission routes
primarily concentrated within the 4th Ring Road. The overall trend in­ are clustered in the South 2nd Ring Road area, followed by a scattered
dicates a statistically significant increase in both CO2 emissions and distribution in the Northeast 3rd and 4th Ring Road areas, Changping,
clustering intensity over time. A narrow band of consecutive hot spots Shijingshan, and Tongzhou. These routes have lower PF values, resulting

Fig. 11. The Emerging Hot Spot Analysis for Hd2Wd, We2Wd, and Wd2We.

10
J. Zheng et al. Sustainable Cities and Society 99 (2023) 104976

Fig. 12. Histogram and spatial distribution of CO2 emissions per person-kilometer.

in higher CO2 emissions per-km and should be prioritized for emission according to Eqs. (8) and (9) increase and the emissions benefit de­
reduction optimization. Sporadic high-emission routes on non-working creases accordingly. In this scenario, buses are not considered a
days are more than those on working days. This indicates that regular ‘greener’ way to travel than cars (Sui et al., 2020). If a bus route has
commuter trips on weekdays result in relatively higher emission benefits higher daily CO2 emissions per person-kilometer than a car with only a
for the entire urban area. driver on the same mileage, this indicates that the route has a lower
emission benefit and there is potential to reduce emissions. To improve
this situation, the NoD can be reduced so that the emissions per-km
4.3. Carbon reduction potential
emissions are lower than the value for a private car under equivalent
conditions.
Three schemes are implemented to explore the potential for reducing
Fig. 13a shows the distribution of the types of routes and mileage
carbon emissions.
that exceeded the per-km emissions of private cars throughout the
month. There are a total of 158 routes, of which 20.48% are in the
4.3.1. Emission reduction potential per person-kilometer
general category, 26.21% in the night bus category and 53.31% in the
Buses are designed to accommodate more passengers, to travel to
special bus category. Special types of buses, such as those for shorter
distant destinations without private transportation, and to alleviate
routes and smaller coverage areas, and night buses that operate at night,
traffic congestion caused by numerous small cars. However, in terms of
tend to have lower ridership resulting in lower emissions benefits
environmental benefits, buses have higher emissions than small cars
compared to other modes of transportation.
when traveling the same distance. As the number of bus passengers
Fig. 13b shows the daily emission reductions that can be achieved by
decreases, the carbon emissions per person-kilometer calculated

Fig. 13. (a) Routes exceeding private car per-km emissions, (b) Daily reduction potential .

11
J. Zheng et al. Sustainable Cities and Society 99 (2023) 104976

reducing the NoD on specific bus routes with high per-km emissions. By 4.3.3. Emission factor reduction potential
implementing this strategy, there is a potential reduction of 54,269.39 The combined emission factors for multiple years were used to assess
kg of CO2 emissions for the entire month. However, due to the uneven the reduction benefits. Based on the vehicle CO2 emission factors in
reduction in bus ridership caused by the COVID-19 pandemic, the Table 2, five scenarios were created to estimate the emission reduction
emission benefits of buses compared to cars are already reduced during potential, as shown in Table 4.
off-peak hours. The proposed emission reduction scenario involves Fig. 15 shows the spatial distribution and emission reduction po­
measuring the emission benefits of a route by reducing the NoD for tential of bus emissions under five different scenarios. The main urban
certain high per-km routes to achieve direct emission reductions, or areas in all five scenarios are high value emission areas, with ADB
increasing the waiting time at bus stations to attract more passengers having the most widespread high value emissions. Diesel consumption
and increase the emission benefits. directly emits pollutants into the atmosphere, further worsening the
environment in these areas, including the occurrence of haze (Wang
4.3.2. Passenger load factor reduction potential et al., 2021). While diesel buses have higher CO2 emission factors than
The changing bus network allows buses to stop at the right places to electric buses, the emission factor of electric buses is not significantly
attract more residents who want to travel, and with a large amount of better than diesel buses if they are generated with non-clean energy
capacity, buses can achieve high occupancy rates. However, congestion sources. In addition, electric buses consume more battery power when
during peak hours can also result in poor ridership conditions for pas­ driving at low speeds or with heating and air conditioning on. Based on
sengers. Low passenger loads are often a result of routes not being the current year’s model ratio and operating mileage, the CBD2017
saturated with passengers, and even during off-peak hours when buses scenario simulates high EB emission factors, but emissions are only
may have empty seats or zero passengers, resulting in a waste of public 6.12% lower than the ADB scenario. Furthermore, introducing electric
transport resources. Using Eq. (3), we calculated the ADR of bus routes.
If the ADR of a route is less than 4 passengers, it is lower than the ADR of
a private car. In this case, reducing the NoD during off-peak hours can Table 4
increase the number of passengers per trip and thus reduce the carbon CO2 emission reduction scenarios for buses.
emissions of the bus. Scenario Abbreviation Description Reference
Fig. 14 shows the bus routes and corresponding emission reductions emission
that are lower than the private car passenger load. Similar to Scheme 1, factor number
special types of buses continue to account for a large share of emission All Diesel Buses ADB All buses are diesel 2
reductions. By optimizing the number of trips on 54 routes on different types. Operating
days, the scheme is able to reduce 2821.91 kg of emissions. Additionally, mileage in 2015.
Combinations of CDB2017 Based on the proportion 1,2,3
replacing buses on routes with low ridership with minibuses and
Different energy of the number of buses
extending stops to more densely populated neighborhoods far from bus types of Buses, of each energy type in
stops can increase ridership and improve accessibility for residents to 2017 2017. Operating
complete their ‘last mile’ trips. mileage in 2017.
Schemes 1 and 2 approach emission reduction from different per­ Combinations of CDB2019 Based on the proportion 1,2,4
Different energy of the number of buses
spectives. Scheme 1 has a larger scale of emission reduction, while types of Buses, of each energy type in
almost all routes screened in Scheme 2 are almost included in Scheme 1. 2019 2019. Operating
This is because Scheme 1 uses a more stringent condition (i.e., mileage in 2019.
comparing to private car passenger load). In reality, the authorized Electric Buses under EB2019 Electric buses under the 4
grid emission 2019 grid emission
passenger capacity of a typical bus is 60 passengers, but it can be opti­
factors in 2019 factor. Operating
mized based on a higher average passenger capacity (e.g., one-third of mileage in 2019.
the load capacity). Combining Schemes 1 and 2 can further reduce CO2 Electric Buses under EB2021 Electric buses under the 5
emissions effectively and provide a more refined carbon allocation at the grid emission 2021 grid emission
route level. factors in 2021 factor. Operating
mileage in 2021.

Fig. 14. (a) Routes with lower than private car load factor, (b) Daily reduction potential.

12
J. Zheng et al. Sustainable Cities and Society 99 (2023) 104976

Fig. 15. Spatial distribution of emissions and emission reduction potential under five scenarios.

buses shifts direct CO2 emissions to the upstream power plants. (1) Routes ranging from 10 to 30 km in length are the main routes
The CDB2019 scenario uses the current year’s bus fleet composition supporting passenger transport in Beijing, accounting for 81% of
and corresponding energy consumption coefficients to simulate realistic the total emissions produced by the bus fleet. The daily average
emissions, achieving a 25.78% reduction in CO2 emissions compared to per-km emission is 56.2 gCO2/per-km.
the CDB2017. This reduction is mainly due to the increase in EB pro­ (2) Bus carbon emissions show a cyclical pattern of change that is
portion and their reduced energy consumption. Assuming that all buses consistent with residents’ travel patterns during holidays, week­
are electric in 2019, the EB2019 leads to a 19.93% reduction in CO2 days and weekends. Regular commuting trips on weekdays
emissions compared to the CDB2019, significantly improving the envi­ contribute to relatively high emission benefits for the entire
ronmental conditions of the entire urban area. Furthermore, with the urban area.
latest EB energy consumption and grid emission factors, EB2021 can (3) The CO2 emission intensity varies significantly between roads,
reduce emissions by more than 48.17%, providing a significant with high emissions from the 3rd and 4th Ring Roads, and major
improvement for power plants as a source of CO2 emissions. Over the six roads connecting Beijing’s outer city. The area covered by the 4th
years, the reduction ratio from ABD to EB2021 reaches 71.08%. Ring Road is only 13% of the total area within the 6th Ring Road,
From a whole life cycle perspective, electric buses have lower but it generates almost half of the CO2 emissions. As the built
emissions than diesel buses (Mao et al., 2020), and the complete elec­ environment expands, the total CO2 emissions from buses grad­
trification of the bus fleet in combination with the use of e-fuels can ually increase, but the emission intensity decreases.
achieve carbon peaking in the transport sector (García et al., 2022). (4) The evolution process of CO2 emissions from Hd2Wd and We2Wd
Therefore, the two main tasks to reduce emissions are the electrification is spatially heterogeneous, with a higher overall increase in
of the bus fleet as a consumption endpoint and the optimization of the Hd2Wd emissions. The region from 4th to 5th ring road areas
electricity grid in the power generation sector as a production endpoint. serve as a dividing line between hot and cold spots for bus CO2
emissions from Beijing buses. This area also forms decaying over-
5. Conclusions and policy recommendations zone of the hot spots spreading outward.
(5) Full electrification of buses coupled with low emission factors in
The complex bus road network within cities can contribute signifi­ the power grid leads to a significant reduction in emissions of
cantly to CO2 emissions. This study presents a methodological frame­ around 71% compared to the diesel-based scenario. This
work to investigate the spatiotemporal patterns of daily-scale bus approach taps into the effective emission reduction potential
emissions using bus smart card data. The study focuses on exploring the from both the production side of power generation and the con­
spatiotemporal characteristics of daily-scale bus emissions in Beijing, sumption side of public transportation operations.
and analyzing the potential for bus emission reduction through per-km
emissions, passenger load factor and emission factors. In addition, it Based on the results of this study, we have made several policy rec­
provides a typical case for studying the spatiotemporal characteristics of ommendations. First, electrification of buses and optimization of the
emissions in other cities. The main findings of this research are as power grid are the two key tasks to reduce emissions. Making full use of
follows: electric buses on short and medium routes and replacing diesel buses
with power swap modes on existing long routes can effectively reduce

13
J. Zheng et al. Sustainable Cities and Society 99 (2023) 104976

emissions and thus improve environmental conditions. Second, ratio­ Assessment Report of the Intergovernmental Panel on Climate Change. Retrieved
from https://www.ipcc.ch/report/sixth-assessment-report-cycle/. Accessed 1 July,
nalizing route capacity allocation and reducing the NoD to hotspots with
2023.
high emissions per-km. For example, increasing waiting times at stations Li, F., Cai, B., Ye, Z., Wang, Z., Zhang, W., Zhou, P., & Chen, J. (2019a). Changing
for night buses can aid in improving emission benefits. Third, efforts patterns and determinants of transportation carbon emissions in Chinese cities.
should be made to reduce the vacancy rate of buses, such as replacing Energy, 174, 562–575. https://doi.org/10.1016/j.energy.2019.02.179
Li, W., Bao, L., Wang, L., Li, Y., & Mai, X. (2019b). Comparative evaluation of global low-
regular buses with minibuses and setting up stations in densely popu­ carbon urban transport. Sustainable Cities and Society, 143, 14–26. https://doi.org/
lated areas to increase passenger flow. 10.1016/j.techfore.2019.02.008
This study provides an in-depth analysis of the patterns and potential Li, X., Tan, X., Wu, R., Xu, H., Zhong, Z., Li, Y., & Qiao, Y. (2021). Paths for carbon peak
and carbon neutrality in transport sector in China. Chinese Journal of Engineering
of CO2 emissions from urban transit, but the methodology used has some Science, 23(6), 15–21.
limitations. First, estimated departures are used to estimate CO2 emis­ Li, Y., Lv, C., Yang, N., Liu, H., & Liu, Z. (2020). A study of high temporal-spatial
sions, which may lead to inaccuracies. Secondly, the true fleet compo­ resolution greenhouse gas emissions inventory for on-road vehicles based on traffic
speed-flow model: A case of Beijing. Journal of Cleaner Production, 277, Article
sition of each energy type does not match. Finally, the calculations were 122419. https://doi.org/10.1016/j.jclepro.2020.122419
performed using coarse energy consumption coefficients and emission Liu, Y., Huang, W., Lin, X., Xu, R., Li, L., & Ding, H. (2022). Variation of spatio-temporal
factors, without distinguishing between more refined bus types, such as distribution of on-road vehicle emissions based on real-time RFID data. Journal of
Environmental Sciences, 116, 151–162. https://doi.org/10.1016/j.jes.2021.07.018
compressed natural gas, liquefied natural gas, and hybrid buses. In Lv, C., Zhang, Z., Chen, X., Ma, D., & Cai, B. (2021). Study on CO2 emission factors of
future work, we will add multi-source data (trajectory data, vehicle type road transport in Chinese provinces. China Environmental Science, 41(7), 3122–3130.
data) to address these issues. We will also analyze the carbon emissions Ma, X., Miao, R., Wu, X., & Liu, X. (2021). Examining influential factors on the energy
consumption of electric and diesel buses: A data-driven analysis of large-scale public
of transportation compared to other cities worldwide.
transit network in Beijing. Energy, 216, Article 119196. https://doi.org/10.1016/j.
energy.2020.119196
Declaration of Competing Interest Mao, F., Li, Z., & Zhang, K. (2020). Carbon dioxide emissions estimation of conventional
diesel buses electrification: A well-to-well analysis in Shenzhen, China. Journal of
Cleaner Production, 277, Article 123048. https://doi.org/10.1016/j.
The authors declare that they have no known competing financial jclepro.2020.123048
interests or personal relationships that could have appeared to influence MEE. (2022). Notice on the key work related to enterprise greenhouse gas emission report
management in 2022. Ministry of Ecology and Environment of the People’s Republic
the work reported in this paper.
of China. https://www.mee.gov.cn/xxgk2018/xxgk/xxgk06/202203/t20220315_97
1468.html.
Data availability Morrow, W. R., Gallagher, K. S., Collantes, G., & Lee, H. (2010). Analysis of policies to
reduce oil consumption and greenhouse-gas emissions from the US transportation
sector. Energy Policy, 38(3), 1305–1320.
Data will be made available on request. MOT. (2021). Statistical bulletin on the development of the transport industry of China in
2021. Ministry of Transport of the People’s Republic of China. https://xxgk.mot.gov.
cn/2020/jigou/zhghs/202205/t20220524_3656659.html.
Ord, J. K., & Getis, A. (1995). Local spatial autocorrelation statistics: Distributional issues
Acknowledgments and an application. Geographical Analysis, 27(4), 286–306. https://doi.org/10.1111/
j.1538-4632.1995.tb00912.x
We acknowledge the financial support from the National Natural Rogelj, J., Popp, A., Calvin, K. V., Luderer, G., Emmerling, J., Gernaat, D., &
Marangoni, G. (2018). Scenarios towards limiting global mean temperature increase
Science Foundation of China (42271471, 42201454, 42271426, below 1.5 C. Nature Climate Change, 8(4), 325–332.
41830645), the International Research Center of Big Data for Sustain­ Schipper, L., Marie-Lilliu, C., & Gorham, R.B. (2000). Flexing the link between transport
able Development Goals (CBAS2022GSP06). We also thank the editors and greenhouse gas emissions : A path for the World Bank.
Sui, Y., Zhang, H., Shang, W., Sun, R., Wang, C., Ji, J., & Shao, F. (2020). Mining urban
and anonymous reviewers for their detailed comments. sustainable performance: Spatio-temporal emission potential changes of urban
transit buses in post-COVID-19 future. Applied Energy, 280, Article 115966. https://
References doi.org/10.1016/j.apenergy.2020.115966
Sun, S., Sun, L., Liu, G., Zou, C., Wang, Y., Wu, L., & Mao, H. (2021). Developing a
vehicle emission inventory with high temporal-spatial resolution in Tianjin, China.
Ahmadi, H., Argany, M., Ghanbari, A., & Ahmadi, M. (2022). Visualized spatiotemporal
Science of the Total Environment, 776, Article 145873. https://doi.org/10.1016/j.
data mining in investigation of Urmia Lake drought effects on increasing of PM10 in
scitotenv.2021.145873
Tabriz using Space-Time Cube (2004-2019). Sustainable Cities and Society, 76, Article
Tang, T., Liu, R., & Choudhury, C. (2020). Incorporating weather conditions and travel
103399. https://doi.org/10.1016/j.scs.2021.103399
history in estimating the alighting bus stops from smart card data. Sustainable Cities
BPT. (2019). Social Responsibility Report of Beijing Public Transport Corporation.
and Society, 53, Article 101927. https://doi.org/10.1016/j.scs.2019.101927
BSI. (2013). PAS 2070: 2013 Specification for the assessment of greenhouse gas
Wang, H., Zhou, X., Guo, H., et al. (2023). Quantifying spatially varying impacts of
emissions of a city. Direct plus supply chain and consumption-based methodologies.
public transport on NO2 concentrations with big geo-data. Environmental Monitoring
BTI. (2022). Beijing Transportation Development Annual Report 2022.
and Assessment, 195, 702. https://doi.org/10.1007/s10661-023-11289-4
CCD. (2012). Average Carbon Dioxide Emission Factors of Regional Power Grids in China
Wang, J., Li, X., Ding, S., Xu, X., Liu, L., Dong, L., & Feng, Y. (2021). Uncovering
in 2011 and 2012.
temporal-spatial drivers of vehicular NOx emissions in China. Journal of Cleaner
Chen, X., Shan, X., Ye, J., Yi, F., & Wang, Y. (2017). Evaluating the effects of traffic
Production, 288, Article 125635. https://doi.org/10.1016/j.jclepro.2020.125635
congestion and passenger load on feeder bus fuel and emissions compared with
Wang, N., & Tang, G. (2022). A review on environmental efficiency evaluation of new
passenger car. Transportation Research Procedia, 25, 616–626. https://doi.org/
energy vehicles using life cycle analysis. Sustainability, 14(6), 3371. https://www.
10.1016/j.trpro.2017.05.446
mdpi.com/2071-1050/14/6/3371.
Cheng, S., Zhang, B., Zhao, Y., Peng, P., & Lu, F. (2023). Multiscale spatiotemporal
Wang, Z., & Yao, Z. (2022). Annual Report on the Big Data of New Energy Vehicle in
variations of NOx emissions from heavy duty diesel trucks in the Beijing-Tianjin-
China (2022). 10.1007/978-981-19-5508-2.
Hebei region. Science of the Total Environment, 854, Article 158753. https://doi.org/
Waraich, A. S., Anowar, S., Tenaglia, T., Sider, T., Alam, A., Minaei, N. S., & Eluru, N.
10.1016/j.scitotenv.2022.158753
(2020). Disaggregate level simulation of bus transit emissions in a large urban
Esri. (2022). Space Time Pattern Mining toolbox. Retrieved from https://pro.arcgis.co
region. International Journal of Sustainable Transportation, 14(7), 544–553. https://
m/en/pro-app/latest/tool-reference/space-time-pattern-mining/an-overview-of-th
doi.org/10.1080/15568318.2019.1579009
e-space-time-pattern-mining-toolbox.htm. Accessed 7 Nov, 2022.
Wei, T., & Frey, H. C. (2022). Intermodal comparison of tailpipe emission rates between
García, A., Monsalve-Serrano, J., Lago Sari, R., & Tripathi, S. (2022). Pathways to
transit buses and private vehicles for on-road passenger transport. Atmospheric
achieve future CO2 emission reduction targets for bus transit networks. Energy, 244,
Environment, 281, Article 119141. https://doi.org/10.1016/j.
Article 123177. https://doi.org/10.1016/j.energy.2022.123177
atmosenv.2022.119141
Harris, S., Weinzettel, J., Bigano, A., & Källmén, A. (2020). Low carbon cities in 2050?
WRI. (2014). Global Protocol for Community-Scale Greenhouse Gas Emission
GHG emissions of European cities using production-based and consumption-based
Inventories. An Accounting and Reporting Standard for Cities.
emission accounting methods. Journal of Cleaner Production, 248, Article 119206.
Xi, J. (2020). Statement at the general debate of the 75th session of the united nations
https://doi.org/10.1016/j.jclepro.2019.119206
general assembly. Retrieved from http://english.scio.gov.cn/topnews/2020-09/
IEA. CO2 emissions from fuel combustion. https://www.iea.org/reports/co2-emissions-fr
23/content_76731466.htm. Accessed 7 Nov, 2022.
om-fuel-combustion-2021.
Xu, Y., Liu, Z., Xue, W., Yan, G., Shi, X., Zhao, D., & Wang, J. (2021). Identification of on-
IPCC. (2006). Guidelines for national greenhouse gas inventories. Retrieved from
road vehicle CO2 emission pattern in China: A study based on a high-resolution
https://www.ipcc-nggip.iges.or.jp/public/2006gl/. Accessed 6 Nov, 2022.
emission inventory. Resources, Conservation and Recycling, 175, Article 105891.
IPCC. (2023). Climate Change 2023: Synthesis Report. A Report of the Intergovernmental
https://doi.org/10.1016/j.resconrec.2021.105891
Panel on Climate Change. Contribution of Working Groups I, II and III to the Sixth

14
J. Zheng et al. Sustainable Cities and Society 99 (2023) 104976

Yin, G., Huang, Z., Yang, L., Ben-Elia, E., Xu, L., Scheuer, B., & Liu, Y. (2023). How to Zhou, X., Dong, Q., Huang, Z., Yin, G., Zhou, G., & Liu, Y. (2023). The spatially varying
quantify the travel ratio of urban public transport at a high spatial resolution? A effects of built environment characteristics on the integrated usage of dockless bike-
novel computational framework with geospatial big data. International Journal of sharing and public transport. Sustainable Cities and Society, 89, Article 104348.
Applied Earth Observation and Geoinformation, 118, 103245. https://doi.org/ https://doi.org/10.1016/j.scs.2022.104348
10.1016/j.jag.2023.103245 Zhou, X., Wang, H., Huang, Z., Bao, Y., Zhou, G., & Liu, Y. (2022). Identifying
Yue, T., Chai, F., Hu, J., Jia, M., Bao, X., Li, Z., & Zu, L. (2016). Gaseous emissions from spatiotemporal characteristics and driving factors for road traffic CO2 emissions.
compressed natural gas buses in urban road and highway tests in China. Journal of Science of the Total Environment, 834, Article 155270. https://doi.org/10.1016/j.
Environmental Sciences, 48, 193–199. https://doi.org/10.1016/j.jes.2016.01.028 scitotenv.2022.155270
Zhang, L., Long, R., Chen, H., & Geng, J. (2019). A review of China’s road traffic carbon
emissions. Journal of Cleaner Production, 207, 569–581. https://doi.org/10.1016/j.
jclepro.2018.10.003

15

You might also like