Journal of Environmental Management 177 (2016) 253e263

Contents lists available at ScienceDirect

Journal of Environmental Management

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

Research article

Integrated modelling approach for the evaluation of low emission

zones
Daniela Dias*, Oxana Tchepel, Anto

nio Pais Antunes
CITTA, Department of Civil Engineering, University of Coimbra, Polo II, 3030-788 Coimbra, Portugal

a r t i c l e i n f o a b s t r a c t

Article history: Low emission zones (LEZ) are areas where the most polluting vehicles are restricted or deterred from
Received 28 July 2015 entering. In recent years, LEZ became a popular option to reduce traffic-related air pollution and have
Received in revised form been implemented in many cities worldwide, notably in Europe. However, the evidence about their
8 April 2016
effectiveness is inconsistent. This calls for the development of tools to evaluate ex-ante the air quality
Accepted 10 April 2016
Available online 22 April 2016
impacts of a LEZ. The integrated modelling approach we propose in this paper aims to respond to this
call. It links a transportation model with an emissions model and an air quality model operating over a
GIS-based platform. Through the application of the approach, it is possible to estimate the changes
Keywords:
Low emission zones
induced by the creation of a LEZ applied to private cars with respect to air pollution levels not only inside
Traffic-related air pollution the LEZ, but also, more generally, in the city where it is located. The usefulness of the proposed approach
Transportation modelling was demonstrated for a case study involving the city of Coimbra (Portugal), where the creation of a LEZ is
Emissions modelling being sought to mitigate the air quality problems that its historic centre currently faces. The main result
Air quality modelling of this study was that PM10 and NO2 emissions from private cars would decrease significantly inside the
Impact evaluation LEZ (63% and 52%, respectively) but the improvement in air quality would be small and exceedances to
the air pollution limits adopted in the European Union would not be fully avoided. In contrast, at city
level, total emissions increase and a deterioration of air quality is expected to occur.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction Under European Directive 2008/50/EC, Member States must

1.1. Research context
Increased public concerns have recently elevated the role of air
quality policies. According to the World Health Organization, air
pollution is the largest single environmental health risk, contrib-
uting to around 7 million premature deaths worldwide in 2012, and
traffic-related air pollution is the major factor in this respect (WHO,
2013, 2014).
In order to protect public health, the European Union (EU) has
set forth directives to regulate ambient air quality by setting limit
values for several pollutants, including particulate matter (PM10)
and nitrogen dioxide (NO2) (Directive, 2008/50/EC). However,
despite these directives, air pollution limits are currently being
infringed in many cities throughout Europe, and road transport is
one of the main reasons for this problem to occur (Colvile et al.,
2001; Giannouli et al., 2011; EEA, 2012; WHO, 2013).
* Corresponding author.
E-mail address: daniela.dias@uc.pt (D. Dias).
provide Action Plans for those areas that do not comply with air
pollution limits. In the context of such plans, numerous European
cities have established and implemented ambitious traffic man-
agement measures to reduce air pollution levels, focusing espe-
cially on road transport emissions. Examples of such measures
include congestion charges, old vehicle scrapping programs, and
biking and transit network improvements. However, one of the
most frequent responses to the directive has been the creation of
low emission zones (LEZ).
The concept of LEZ, which first appeared in Sweden in the late
1990s under the name of miljo € zon (environmental zone in English),
designates an area where the most polluting vehicles are restricted
or deterred from entering (EC, 2014). For this purpose, criteria
based on Euro emission classes are used to select the vehicles to be
banned from the LEZ or charged if they enter them. Such emission
reduction scheme can take several forms depending on the
geographical area they cover, the time period during which the LEZ
is in force, and the type and emission class of the vehicles to which
they apply. In the EU, over 150 cities in nine Member States have
meanwhile implemented LEZ schemes, in most cases focusing on

http://dx.doi.org/10.1016/j.jenvman.2016.04.031
0301-4797/© 2016 Elsevier Ltd. All rights reserved.
254 D. Dias et al. / Journal of Environmental Management 177 (2016) 253e263
heavy vehicles and valid for 365 days and 24 h per day (EC, 2014).
Despite LEZ are generally considered to be an appropriate
measure towards achieving the air pollution reduction targets of
the EU, the evidence about their effectiveness is somewhat incon-
sistent, which makes their application debatable. Indeed, this type
of measure has not always been successful in meeting European air
pollution limits, notably with regard to PM10 and NO2 levels
(Boogaard et al., 2012; Ellison et al., 2013). In our view, there is not
enough understanding on how the adoption of LEZ will help
reducing air pollution in a given city, and there is consequently an
urgent need for tools that can assist local authorities in the making
of well-informed decisions on this important matter.
The research described in this paper aims to respond to this
urgent need. Specifically, it consists in the development of an in-
tegrated modelling approach to evaluate (ex-ante) the air quality
impacts of LEZ in urban areas. This approach is intended to be a key
component of a decision-support tool for the complex decision-
making processes concerning the impacts of LEZ. It encompasses
three interconnected models: the first is a macroscopic trans-
portation model that describes road traffic in the urban area where
the approach is to be applied; the second is an emissions model
that quantifies the amount of pollutants produced by line sources,
thus by road traffic; and the third is an air quality model that de-
scribes the dispersion of pollutants in the air. The usefulness of the
approach is demonstrated through a study involving the city of
Coimbra (Portugal), where two large areas of the historic centre
were recently classified as UNESCO World Heritage.
1.2. Literature review
The concept of LEZ is quite recent, and research about it is still
rather scarce. In particular, there are only a limited number of
studies dealing with the evaluation of LEZ as a tool to improve
urban air quality. This literature can be divided in two streams: ex-
post evaluation and ex-ante evaluation. On ex-post evaluation
studies, the effectiveness of a LEZ is analysed after a certain period
has passed since its implementation. Ex-ante evaluation studies are
conducted prior to the implementation of a LEZ, to anticipate its
outcomes and confirm whether this measure is effective. The in-
sights gained from the review of this literature and its major gaps
are described in this section. In Table 1, we enumerate the ex-post
and ex-ante LEZ evaluation studies published in journals included
in the ISI and/or SCOPUS bibliographic databases. Studies aimed
solely to quantify the impacts of LEZ on pollutant emissions were
not considered in the literature review.
1.2.1. Ex-post evaluation
To the best of our knowledge, Boogaard et al. (2012) contains the
first ex-post evaluation of LEZ from the viewpoint of air quality.
Based on a measurement campaign in five Dutch cities including
Amsterdam, the authors concluded that the introduction of LEZ for
heavy-duty vehicles has produced non-significant reductions in
PM10 and NOx concentrations. In contrast, focusing specifically on
the effects of the LEZ in Amsterdam, Panteliadis et al. (2014) re-
ported a significant decrease in PM10 and NOx concentrations (7.7%
and 5.9%, respectively) in the vicinity of roadside environmental
monitoring stations. Instead of measured air pollution concentra-
tions, Keuken et al. (2012) used modelled concentrations to esti-
mate the effects of this same LEZ on elemental carbon (EC) levels.
Actual traffic data, meteorological parameters and emission factors
were used as input for a street-canyon model (URBIS model) to
characterize the spatial distribution of EC concentrations. However,
the model was only applied to roads inside the LEZ (with a traffic
volume exceeding 7500 vehicles in 24 h), and therefore the city-
wide impact of the LEZ was not addressed. Specific EC emission
factors were determined for heavy-duty vehicles in this study, but
no distinction was made regarding the EC emissions relating to the
various Euro classes. The conclusion stated by the authors was that
the LEZ led to a negligible reduction of EC concentrations due to the
low proportion of highly polluting heavy-duty vehicles (1% of traffic
volume) circulating in the urban area.
In the United Kingdom, Ellison et al. (2013) carried out an ex-
post evaluation study on the impacts of a LEZ created in London
for heavy diesel vehicles, buses and coaches. This study concluded
that London's air quality improved marginally with respect to PM10
concentrations. Indeed, ambient air quality measurements showed
that concentrations of PM10 within the LEZ have dropped by
2.5e3.1% compared to just over 1% for areas in its vicinity. In
contrast, no discernible differences were found for NOx
concentrations.
A first ex-post evaluation study of a LEZ in Munich has been
conducted by Qadir et al. (2013) based on PM2.5 measurements
made before and after its creation. These measurements were
analysed for elemental carbon (EC) and particulate organic com-
pounds (POC). The authors evidenced that the average daily con-
centration of EC from traffic decreased 60%, from 1.1 mg/m3 before
to 0.5 mg/m3 after the implementation of the LEZ, and no significant
differences were noticed in POC concentrations. More recently, also
evaluating the effects of the LEZ in Munich, Fensterer et al. (2014)
reported a substantial decrease of PM10 levels at the traffic-
monitoring site analysed, especially in summer (19.6%) but also in
winter (6.8%). However, it should be noted that Fensterer et al.
(2014) considered not only the impact of the LEZ but also the
impact of a transit ban for all trucks on Sundays. Therefore, the
authors did not assess the effects that can be specifically attributed
to the implementation of the LEZ.
The most recent ex-post evaluation study we encountered in the
literature was performed by Da Silva et al. (2014) to assess the
effectiveness of the LEZ established in 2012 in the most polluted
area of Lisbon (Baixa and Avenida da Liberdade). The conclusion
was that its implementation led to reductions of 16% and 6% in
annual average PM10 and NO2 concentrations, respectively. How-
ever, the air pollution levels achieved outside the LEZ were not
addressed by the authors, therefore the citywide impact of the LEZ
was not fully appraised.
Globally, as evidenced by the studies reviewed, it remains un-
clear how successful LEZ have been with respect to the fulfilment of
EU air pollutant limit values and what improvements in air quality
levels can be attributed to such traffic management measures. To
clarify the situation, ex-post evaluations can certainly be very
useful, but need to be made considering also what is happening
outside the LEZ, as well as separating their impacts from those of
other measures that may likewise be affecting air pollution levels in
the city where they are being applied.
1.2.2. Ex-ante evaluation
Despite the important information that can be obtained through
the ex-ante evaluation of the air quality impacts of a LEZ, only two
studies were found in the literature where such impacts have been
estimated.
The first of these studies was performed for London by Carslaw
and Beevers (2002). Using NO2 as air pollution indicator, two main
types of potential LEZ schemes were analysed: the reduction of
vehicle flows and the restriction of higher polluting vehicles. For
this purpose, the COPERT II model (HA, 1999) and the CAR Inter-
national model (Eerens et al., 1993) were used to calculate the
emissions and near-road concentrations of NO2, respectively.
Considering the most effective scenario, the authors predicted an
overall reduction between 3.6% and 11.1% in NO2 concentrations.
However, it should be noted that traffic data for all major roads in
 D. Dias et al. / Journal of Environmental Management 177 (2016) 253e263 255

Table 1
Summary of ex-ante and ex-post evaluations of LEZ impacts on air pollution.

Reference City Measuring Modelling Passenger Heavy duty Buses and Highlights
approach approach vehicles vehicles coaches

Ex-post Boogaard et al., Amsterdam ☑ < EURO 4 No measurable

evaluation 2012 The Hague effect of the LEZ
Den Bosch in traffic-
Tilburg related air
Utrecht pollutants
Keuken et al., Amsterdam ☑ < EURO 3 and Negligible
2012 EURO 3 (old reductions on
than 8 years or modelled EC
without diesel concentrations
PM filter)
Ellison et al., London ☑ < EURO 3 < EURO 3 Small
2013 improvement
in PM10 and
NOx compared
to other areas
since LEZ
introduced.
Fensterer et al., Munich ☑ < EURO 2 < EURO 2 < EURO 2 Significant
2014 reduction of
PM10
concentrations
but not only
attributed to
LEZ scheme.
Panteliadis Amsterdam ☑ < EURO 3 Significant
et al., 2014 decreases in
PM10 and NO2
concentrations
in the vicinity
of a roadside
monitoring
station
Da Silva et al., Lisbon ☑ < EURO 2 (zone < EURO 2 (zone < EURO 2 (zone Significant
2014 1) 1) 1) decreases in
< EURO 1 (zone < EURO 1 (zone < EURO 1 (zone PM10 and NO2
2) 2) 2) concentrations
within LEZ area
Ex-ante Carslaw and London ☑ < EURO 2 < EURO 2 < EURO 2 Even ambitious
evaluation Beevers, 2002 (scenario 3) (scenario 3) (scenario 3) LEZ scenarios in
< EURO 2 < EURO 4 < EURO 4 central London
(scenario 4) (scenario 4) (scenario 4) produce
concentrations
of NO2 that are
achieved
through natural
fleet renewal
only five years
later
Cesaroni et al., Rome ☑ < EURO 1 Significant
2012 (optimistic decreases in the
scenario) intervention
Replacement of zones and
90% of < Euro 2 reductions in
(pessimistic NO2
scenario) concentrations
are larger than
PM10.

London were provided by manual counts and that LEZ scenarios
were evaluated considering that road traffic would not be affected
in any other locations, thus disregarding the displacement of traffic
to outside the LEZ.
The other study is due to Cesaroni et al. (2012). In addition to air
quality, these authors evaluated the effectiveness of a LEZ to be
created in Rome in terms of years of life gained. For this purpose,
two alternative scenarios, respectively classified as “optimistic” and
“pessimistic”, were simulated. The COPERT II model was used also
in this case to determine average PM10 and NO2 emission values for
each year considered in the study. The effect of the LEZ on the
number and type of vehicles was analysed using traffic data ob-
tained from the national automobile statistical database. Road ge-
ometry, traffic volumes and emission factors were used as input for
the CAR II model (Jonkers, 2007) used to estimate the dispersion of
those two pollutants at the street level. However, concentrations
were calculated only for roads with traffic exceeding 10,000 vehi-
cles per day and the dispersion model did not take nearby streets
into account. Therefore, air pollution levels inside the LEZ were not
properly addressed, leading to underestimation of modelled
256 D. Dias et al. / Journal of Environmental Management 177 (2016) 253e263
Fig. 1. Modelling approach.
benefits. As concluded by the authors, the LEZ in Rome would have
a large effect on years of life gained in the intervention area, but its
global urban impact would be small due to modest changes in air
pollutant concentrations for the city as a whole.
The review of this (meagre) literature makes clear that an in-
tegrated modelling approach for the ex-ante evaluation of the air
quality impacts of a LEZ is lacking. The approach we propose and
test in the next sections is a step in that direction.
2. Materials and methods
2.1. Modelling approach
The approach we have developed to evaluate the air pollution
impacts of LEZ consists of three interconnected models (Fig. 1): a
transportation model that describes road traffic in the city under
study; an emissions model that quantifies the emissions generated
by road traffic; and an air quality model that describes the
dispersion of air pollutants across the city. This approach is
implemented on a GIS-based information system that manages the
inputs and outputs associated with each model. Detailed informa-
tion about the models is provided below in separate subsections.
2.1.1. Transportation model
The main classification of transportation models distinguishes
between macro and micro models. Macro models describe traffic in
an aggregate manner, without taken into account the behaviour of
individual travellers. Such models date back to the 1950s, being
generally considered to be among the first large planning models
with spatial content. Micro models appeared much later, and after
remarkable progresses in recent years they are becoming a popular
choice in traffic studies. However, they are being used in studies for
parts of an urban centre, and are still not an option for citywide
analyses.
Macro transportation models have evolved considerably over
time in several respects, but the most used in practice continue to
rely on the same trip-based approach and four steps (sub-models)
that characterized the initial models (McNally, 2008). The four
steps are: trip generation; trip distribution; modal split; and traffic
assignment. They determine in succession the total incoming and
outgoing trips generated in each zone of a city (or other study
domain), distribute them to the other zones, allocate them to the
different transport modes available, and finally assign the vehicles
to the road network. Traffic assignment is carried out according to
Wardrop's principle (at equilibrium drivers cannot improve their
travel times by changing routes), to take into account congestion
effects (Ortúzar and Willumsen, 2011; Chaps. 10 and 11).
In light of these circumstances, we decided to base our approach
in the four-step model, as implemented in the VISUM software
(PTV, 2013). Through this model, it is possible to estimate how
traffic volume and vehicle speed will change in each road segment
after the creation of a LEZ, which is the key information necessary
for the emissions model. However, our plan it to complement the
four-step model with a micro model as soon as the combined
evolution of computing software and hardware makes the latter
type of model apt for application to mid-size cities (which should
not take long).
2.1.2. Emissions model
Several traffic-related emissions models are available to predict
the pollutant amount emitted to the atmosphere at microscopic,
mesoscopic, and macroscopic scale (Agostini et al., 2005;
 D. Dias et al. / Journal of Environmental Management 177 (2016) 253e263
Fig. 2. Coimbra's urban centre (left) and central area (right).
Fig. 3. Historic centre and UNESCO World Heritage sites of Coimbra (1 - Rua da Sofia, 2
- Alta Universita
ria) considered for the LEZ.
Gkatzoflias et al., 2007). Macroscopic emission models are usually
used when emission inventories are required at regional, national
or global scales. Microscopic models consider instantaneous
vehicle speeds thus accounting for second-by-second speed pro-
files including acceleration, deceleration, idling and cruising.
However, such increase in resolution compromises their applica-
tion at urban scale, where larger networks are required to be
257
Fig. 4. Composition of Coimbra's car fleet according to Euro emission classes.
characterized. Mesoscopic models do not account for such details,
but are sensitive to spatial variability across a transportation
network (Boulter et al., 2007).
In the proposed modelling approach, we opted to use the TREM
mesoscopic emissions model due to its ability to provide the
detailed traffic emission information at road segment level
required by air quality modelling at urban scale (Tchepel et al.,
2012). The selected model is particularly designed for line sour-
ces, estimating emissions individually for each road segment based
on emission factors determined according to average speed and
vehicle type (classified according to engine age, type and capacity,
vehicle weight, fuel type, and emission reduction technology).
Overall, the input data required by the emissions model have
been compiled on the basis of statistical information on fleet
composition (vehicle age, technology and fuel type), spatial data on
road network, and traffic volume and vehicle speed for each road
segment provided by the transportation model. The model output
data consists of the emission rate for each road segment with a
258 D. Dias et al. / Journal of Environmental Management 177 (2016) 253e263
Fig. 5. Concentrations of daily average PM10 (left) and hourly NO2 (right) at Avenida Fern~ao de Magalh~aes monitoring station for the selected days in 2011.
Table 2
Estimated daily vehicle-kilometres travelled by car in Coimbra before and after the
implementation of the LEZ.
Geographical area Vehicles-kilometres travelled
Value (km/day) Variation (%)
Baseline scenario LEZ scenario
LEZ 92,386 67,253
27.2
Other urban areas 1,336,021 1,392,595 4.2
Coimbra 1,428,407 1,459,848 2.2
temporal resolution identical to the input information on traffic
volume. In this way, emissions can be estimated on an hourly, daily
or even yearly basis.
2.1.3. Air quality model
In general, air quality models can be divided into statistical
models and process oriented models (Moussiopoulos et al., 1996).
Statistical models are valuable tools to screen air quality by means
of interpolation and extrapolation of measurement data (e.g. the
concentrations measured show a statistically significant depen-
dence on the volume of traffic). However, this type of models
should be used with caution. They may be considered valid only
within the range of the data from which they were derived. In
contrast, process oriented models are based on the description of
physical and chemical processes that occur in the atmosphere.
Compared to statistical models, process oriented models have the
main advantage of incorporating both spatial and temporal varia-
tions of pollutant concentrations even in locations without dense
monitoring networks, and of establishing the cause-effect re-
lationships required to support decision-making.
Given the complex spatial and temporal variability of urban air
pollution, we selected the Lagrangian dispersion model AUS-
TAL2000 that allows estimating the impacts of emission changes on
pollutant concentrations over a city. AUSTAL2000 is the official
reference air dispersion model of the German Regulation on Air
Quality Control for short range applications (Janicke and Janicke,
2002). Despite the availability of more complex regional air qual-
ity models already applied for Portugal in previous studies
(Monteiro et al., 2007; Ferreira et al., 2012), the ability of AUS-
TAL2000 to simulate air pollution dispersion at local scale with high
spatial resolution (10 m grid resolution in this study), considering
traffic emissions explicitly as line sources for each road segment and
the effects of buildings on the wind flow, was decisive for the choice
of this model. Moreover, the model can handle complex terrain
profiles and account for atmospheric advection and turbulent
mixing and deposition. With respect to chemistry, AUSTAL2000
adopts a simple solution for NO to NO2 conversion. However, this
solution is widely used at the spatial and temporal resolution
considered by the local-scale model (Vardoulakis et al., 2003).
Additionally to input data on emissions, a continuous time se-
ries of meteorological parameters, including wind direction, wind
speed and atmospheric stability obtained at one monitoring station
located in the city are required by the model. Because topography
and building characteristics have an important influence on the
dispersion of atmospheric pollutants in the urban environment, the
air quality model requires the characterization of the spatial vari-
ation of terrain surface elevation, building 3D coordinates and
emission sources location and dimensions. As output, the air
quality model provides the spatial distribution of pollutants con-
centration with 1-h resolution. In this study the model was applied
for PM10 and NO2 because of the exceedances of the limit values
observed for these pollutants. Despite the relevance of PM2.5
fraction to human health, it is not addressed in this study due to the
unavailability of measured data by the air quality monitoring sta-
tion located in the study area.
2.2. Case study
The usefulness of the integrated modelling approach developed
in this research is demonstrated through an application to the city
of Coimbra (Fig. 2), the largest urban centre in central Portugal. It is
a city laid out on hilly terrain with a population of approximately
150,000 (including suburbs).
Despite some decline in the last two decades, the central area of
Coimbra is still by far the city's major traffic generator, being origin
or destination for a daily total of about 94,000 motorized trips ac-
cording to the latest mobility survey. Automobile accounts for 69%
of these trips, bus (and trolley car) for 27%, train for 3%, and
motorcycle for the remaining 1%. The road network of this area
consists of three main axes e Avenida Fern~ ao de Magalha ~es, Rua da
Sofia, and Avenida S
a da Bandeira e laid out upon a labyrinth of
narrow streets with origins in the 19th century or before. The
configuration of this road network is responsible for the major
traffic-related problems faced by Coimbra. On the one hand, the
narrow streets do not offer suitable conditions for motorized traffic,
and indeed several of those streets have been pedestrianized since
the 1980s. On the other hand, and more importantly, those main
axes are often congested during weekdays (and not only during
rush hours), leading to excessive air pollution levels. This is attested
by measurements made at the air quality monitoring station
located near Avenida Fern~ ao de Magalh~ aes, where the limits
established in Directive 2008/50/EC with respect to PM10 (daily
average of 50 mg/m3) and NO2 (hourly average of 200 mg/m3) have
not been observed in several recent occasions.
The fact that a large part of Coimbra's central area was recently
classified as UNESCO's World Heritage is making these traffic-related
problems more visible and urgent, and is pushing the city council to
look for solutions to mitigate them. One the main solutions that is
 D. Dias et al. / Journal of Environmental Management 177 (2016) 253e263
Fig. 6. Frequency distribution of average car speed in Coimbra before and after the implementation of the LEZ.
Fig. 7. Estimated daily emissions of PM10 (left) and NOx (right) in Coimbra before and after the implementation of the LEZ.
currently being envisaged is the implementation of a LEZ.
The case study we present in this paper arises in this context.
Specifically, it consists in analysing the impacts of creating a LEZ in
Coimbra with the same geographical delimitation as the city's
historic centre (Fig. 3) and the following characteristics:
(1) Enforcement 24 h per day.
(2) Entry permission for public transport vehicles, cars owned by
residents in the area, and cars owned by non-residents of
emission class Euro 3 or higher.
(3) Entry restriction for Pre-Euro, Euro 1 and Euro 2 cars owned
by non-residents (these vehicles, which constitute 47% of
Coimbra's car fleet as shown in Fig. 4, are in particular
responsible for 74% and 68% of the PM10 and NO2 emissions
from private cars at Avenida Fern~ ao de Magalha ~es,
respectively).
The transportation network model considered in the study
consisted of 283 traffic generation zones (12 of which located inside
the LEZ, 107 in other parts of the urban centre, and 164 outside the
urban centre) and 17,898 road segments. The number and modal
share for the trips made between these zones were assumed to be
the same as in the latest mobility survey, including for trips with
destination in the LEZ. That is, we considered that non-residents
who drive a restricted car (Euro 2 or lower) would not change to
bus because of being unable to reach a destination in the LEZ.
Instead, it was assumed that they would continue to use their car,
parking it as close as possible to their destinations (and walking
from there). These trips and any other trips made with restricted
cars were assumed to be made through the fastest routes that do
259
not comprise any segments included in the LEZ, taking congestion
effects into account (that is, traffic was assigned to the road
network according to the Wardrop equilibrium principle). Finally, it
was assumed that the composition of the fleet would remain un-
changed, thus disregarding possible decisions to change car that
could occur in reaction to the implementation of the LEZ. All these
assumptions are suitable for a short-term evaluation of its air
quality impacts. However, we may be underestimating its long-
term effects.
The analysis of air quality carried out in the study involved an
area of 8.9  9.9 km2 with a 10 m grid resolution covering the
whole city of Coimbra and containing about 50,000 buildings. It
focused on six days with meteorological characteristics similar to
those of 25th February, 1st of July, 6th October, 7th October, 13th
October and 30th November 2011, for which exceedances to PM10
and, in one case, to NO2 limit values were registered in the air
quality monitoring station of Avenida Ferna ~o de Magalh~aes (Fig. 5).
In this analysis, additionally to traffic-induced local pollution we
consider background air pollution (i.e., pollution caused by non-
traffic sources and pollution transported from surrounding areas)
using the method described in Tchepel and Borrego (2010) and
Tchepel et al. (2010). The background pollution levels were
assumed to remain unchanged after the implementation of the LEZ.
3. Results and discussion
3.1. Transportation results
The aggregate traffic flows estimated through the transportation
model for the central area of Coimbra before the implementation of
260 D. Dias et al. / Journal of Environmental Management 177 (2016) 253e263
Fig. 8. Expected spatial variation of daily PM10 emissions in Coimbra after the implementation of the LEZ.
the LEZ (baseline scenario) and after it (LEZ scenario) are presented
in Table 2. As shown in this table, the implementation of the LEZ
leads to a reduction of 27.2% in the vehicle-kilometres travelled
(VKT) inside the historic centre of Coimbra (from approximately
92,400 to 67,300 on average per day). The most striking traffic
volume decrease takes place at Avenida Ferna ~o de Magalha ~es,
where the daily flow of cars drops from 14,400 to 8600 (40.3%).
However, since restricted cars are numerous (47% of the total fleet)
and will take longer routes to avoid the LEZ, the VKT in Coimbra
globally increases by 2.2%.
The impact of the implementation of the LEZ is globally negative
not only in relation to traffic flows but also with respect to travel
speed and fuel consumption. Indeed, the average car speed for the
baseline scenario is 37.2 km/h and for the LEZ scenario is 35.1 km/h.
The reason for this decrease in speed is because the capacity of the
roads that restricted cars will use for a large number of trips is
generally much smaller than that of Avenida Ferna ~o de Magalha ~es,
Rua da Sofia, and Avenida Sa
da Bandeira, which are included in the
LEZ and can no longer be crossed by those cars. The frequency
distribution of the average car speed in both scenarios under
analysis is presented in Fig. 6. Regarding fuel consumption, an in-
crease of 4.9% is expected as a combined outcome of the increase in
VKT and the decrease in average car speed.
3.2. Emissions results
According to the outcomes of the emissions model, the imple-
mentation of the LEZ leads to a global increase of about 1.2% and
1.5% respectively in the PM10 and NOx emissions for Coimbra
(Fig. 7), which is consistent with the increase in traffic volumes and
fuel consumption and with the decrease in travel speeds that occur
in the city. In contrast, the same emissions in the historic centre
drop sharply, by 63% and 52%. The drop near the air quality
monitoring station of Avenida Ferna ~o de Magalha~es is even sharper,
reaching 71% and 66%, respectively.
The spatial variation in car emissions across Coimbra after the
implementation of the LEZ is illustrated in Fig. 8 for PM10. As
shown in this figure and as could be anticipated, strong changes in
PM10 emissions are expected to occur in road segments next to the
borders of the LEZ, both in the negative effect (outside the LEZ) and
in the positive effect (inside the LEZ). This is where the segments
with the largest emission reductions (234 g/km per day) are
located. In the rest of the city, the pattern of change is complex, but
it should be emphasised that strong changes will happen also
relatively far from the LEZ, as is notably the case of the north section
of the ring road, where PM10 emission increases are expected to
reach up to 290 g/km per day.
3.3. Air quality results
The outputs of the air quality model indicate that, similarly to
the pollutant emissions, the most positive effects of the imple-
mentation of the LEZ occur inside the Coimbra's historic centre and
the more negative along the north section of the ring road. How-
ever, in contrast to the pollutant emissions, there are no strong
negative air quality impacts in the areas immediately outside the
LEZ. On the contrary, pollution in these areas does not increase
significantly and even decreases as a consequence of the dispersion
of air of a better quality coming from the LEZ.
The pattern of spatial variation in daily PM10 concentrations
expected to occur after the implementation of the LEZ is illustrated
in Fig. 9 for one of the selected days with PM10 exceedances. Inside
 D. Dias et al. / Journal of Environmental Management 177 (2016) 253e263 261

Fig. 9. Expected spatial variation of PM10 concentrations in Coimbra after the implementation of the LEZ for one of the selected days.

the LEZ, the reduction of PM10 concentrations is expected to reach
7.4 mg/m3, but will only be of 1.03 mg/m3 in the vicinity of the air
quality monitoring station of Avenida Ferna ~o de Magalha ~es. In
contrast, near the north section of the ring road PM10 concentra-
tions can increase up to 6.7 mg/m3. A similar behaviour was
observed for the expected spatial variation of NO2 concentrations.
In Table 3 the expected variations in PM10 and NO2 concen-
~o
trations near the air quality monitoring station of Avenida Ferna
de Magalha ~es are presented for all days selected. As evidenced
there, the reductions of daily PM10 concentrations and hourly NO2

Table 3
~o de Magalha
Expected variation of the PM10 and NO2 at Avenida Ferna ~es monitoring station for all days selected.
262 D. Dias et al. / Journal of Environmental Management 177 (2016) 253e263
Fig. 10. Modelled vs. observed daily PM10 (left) and NO2 (right) concentrations in Avenida Fern~ao de Magalh~aes monitoring station.
Table 4
Statistics of modelling performance for PM10 and NO2.
Parameter
Fractional BIAS
Normalized mean square error (NMSE)
Fraction of predictions within a factor of 2 of observations
M: modelled values, O: observed values.
concentrations after the implementation of the LEZ range from 0.64
to 1.34 mg/m3 and from 3.11 to 10.64 mg/m3, respectively. Therefore,
for the days analysed, despite the substantial decrease in the
emissions of PM10 and NOx (71% and 66%, respectively), the
maximum reduction achieved by the creation of the LEZ amounts
only to 1.34 mg/m3 for daily PM10 levels (day similar to the 1st July
2011) and 10.64 mg/m3 for hourly NO2 levels (6th October 2011).
Despite the changes in air pollution levels inside the LEZ are
expected to occur in the positive direction, they are small and not
enough to ensure that the air pollution limits established in
Directive 2008/EC/50 are accomplished. For instance, in a day
similar to the 7th October 2011, the one corresponding to the
lowest PM10 exceedances, the daily concentration of this pollutant
near Avenida Ferna ~o de Magalh~ aes monitoring station will still
reach 52 mg/m3, thus being above the maximum of 50 mg/m3
considered in the directive. Likewise, in a day similar to the 13th
October 2011, the hourly concentration of NO2, 230 mg/m3, will
clearly exceed the limit of 200 mg/m3. Hence, the creation of the LEZ
attenuates the air pollution problems of Coimbra's historic centre,
but does not completely eliminate them. At the same time, such
problems are aggravated in other parts of the city, and particularly
near the north section of the ring road, where daily PM10 con-
centrations as high as 69.7 mg/m3, almost 40% above the limit, are
expected to occur.
3.4. Modelling performance
In order to assess the performance of the proposed modelling
approach, we compared the PM10 and NO2 concentrations esti-
mated by the modelling approach (including background concen-
trations) and measured by the air quality monitoring station of
Avenida Ferna ~o de Magalha ~es (before the implementation of the
LEZ).
As shown in Fig. 10 and Table 4, a good agreement between the
modelled and observed PM10 concentrations is obtained. The cor-
relation coefficient (R) of 0.77 between the two datasets confirms
Formula Value
PM10 NO2
FB ¼ M
O 0.04 0.2
0:5ðMþOÞ
2 0.004 0.06
NMSE ¼ ðM
OÞ
MO
M 0.99 0.79
O
the model capability to describe data variability, and the normal-
ized mean square error (NMSE) is practically zero. Moreover, the
positive fractional BIAS is 0.04, therefore very close to the ideal
value of zero, and 99% of the modelled values are within a factor of
two from the observed values. The equivalent figures for NO2
concentrations are less satisfactory, which could be related with the
assumed chemical conversion of NO to NO2. In particular, the cor-
relation coefficient is 0.55, lower than the one we have obtained for
PM10 but still reasonable, and the positive fractional BIAS of 0.2,
though indicating that NO2 concentrations are overestimated, re-
mains clearly within the range of acceptable values (
2 to 2).
Overall, this signifies that the performance of the proposed
modelling approach is quite satisfactory, providing us with mean-
ingful results regarding the expected impacts of the creation of a
LEZ in Coimbra.
4. Conclusion
The research we presented in this paper was aimed to develop
an integrated modelling approach for the evaluation of low emis-
sion zones (LEZ). The approach links a transportation model with
an emissions model and an air quality model, and its application
allows to estimate the changes induced by the creation of a LEZ
with respect to air pollution levels not only inside the LEZ, but also,
more generally, in the city where it is located.
The usefulness of the proposed approach was demonstrated for
a case study involving the city of Coimbra, where the creation of a
LEZ is being sought to mitigate the air quality problems that its
historic centre currently faces. Specifically, we focused on two
pollutants, PM10 and NO2, with results that were more accurate
with respect to the former than to the latter., Overall, the modelling
performance was rather satisfactory. We concluded that a LEZ
restricting entry in the historic centre to cars of non-residents of
emission class Euro 2 or lower would have a beneficial local effect,
though not strong enough to fully eliminate exceedances to the air
pollution limits established in EU Directive 2008/50/EC. We also
 D. Dias et al. / Journal of Environmental Management 177 (2016) 253e263
concluded that, at the same time, air pollution levels would in-
crease outside the historic centre as a consequence of the changes
in traffic patterns induced by the creation of the LEZ.
Acknowledgments
This research was carried out within the framework of projects
EMSURE (CENTRO-07-0224-FEDER-002004) and TRAPHIC (PTDC/
ECM-URB/3329/2014). The meteorological data were provided by

rio Geofísico e Astrono
Observato
mico da Universidade de Coimbra.
References
Agostini, A., Lelli, M., Negrenti, E., Parenti, A., 2005. An advanced software tool for
vehicular emissions disaggregate estimation. In: Brebbia, A. (Ed.), Air Pollution
XIII. Wessex Institute of Technology, Southampton, UK, pp. 357e367.
Boogaard, H., Janssen, N.A., Fischer, P.H., Kos, G.P., Weijers, E.P., et al., 2012. Impact of
low emission zones and local traffic policies on ambient air pollution concen-
trations. Sci. Total Environ. 435, 132e140.
Boulter, P.G., McCrae, I.S., Barlow, T.J., 2007. A Review of Instantaneous Emission
Models for Road Vehicles. Project Report PPR267. Transport Research Labora-
tory, Wokingham, UK.
Carslaw, D.C., Beevers, S.D., 2002. The efficacy of low emission zones in Central
London as a means of reducing nitrogen dioxide concentrations. Transp. Res.
Part D Transp. Environ. 7 (1), 49e64.
Cesaroni, G., Boogaard, H., Jonkers, S., Porta, D., Badaloni, C., et al., 2012. Health
benefits of traffic-related air pollution reduction in different socioeconomic
groups: the effect of low-emission zoning in Rome. Occup. Environ. Med. 69 (2),
133e139.
Colvile, R., Hutchinson, E., Mindell, J., Warren, R., 2001. The transport sector as a
source of air pollution. Atmos. Environ. 35 (9), 1537e1565.
Da Silva, F.N., Custo
dio, R.A.L., Martins, H., 2014. Low emission zone: Lisbon's
experience. J. Traffic Logist. Eng. 2 (2), 133e139.
EC, 2014. Low Emission Zones in Europe. European Comission, Brussels, Belgium.
http://www.lowemissionzones.eu/ (accessed 29.06.15).
EEA, 2012. Indicator: Exceedance of Air Quality Limit Values in Urban Areas. Eu-
ropean Environment Agency, Copenhagen, Denmark.
Eerens, H.C., Sliggers, C.J., Van den Hout, K.D., 1993. The CAR model: the Dutch
method to determine city street air quality. Atmos. Environ. Part B Urban
Atmos. 27 (4), 389e399.
Ellison, R.B., Greaves, S.P., Hensher, D.A., 2013. Five years of London's low emission
zone: effects on vehicle fleet composition and air quality. Transp. Res. Part D
Transp. Environ. 23, 25e33.
Fensterer, V., Küchenhoff, H., Maier, V., Wichmann, H.E., Breitner, S., et al., 2014.
Evaluation of the impact of low emission zone and heavy traffic ban in Munich
(Germany) on the reduction of PM10 in ambient air. Int. J. Environ. Res. Public
Health 11 (5), 5094e5112.
Ferreira, J., Rodriguez, A., Monteiro, A., Miranda, A.I., Dios, M., Souto, J.A.,
Yarwood, G., Nopmongcol, U., Borrego, C., 2012. Air quality simulations for
North America - MM5-CAMx modelling performance for main gaseous
263
pollutants. Atmos. Environ. 53, 212e224.
Giannouli, M., Kalognomou, E.A., Mellios, G., Moussiopoulos, N., Samaras, Z., Fiala, J.,
2011. Impact of European emission control strategies on urban and local air
quality. Atmos. Environ. 45 (27), 4753e4762.
Gkatzoflias, D., Kouridis, C., Ntziachristos, L., Samaras, Z., 2007. Passenger Cars,
Light-duty Trucks, Heavy-duty Vehicles Including Buses and Motor Cycles.
European Topic Centre on Air and Climate Change. Bilthoven, The Netherlands.
HA, 1999. Design Manual for Roads and Bridges, vol. 11. Air Quality. Highways
Agency, London, UK. Section 3, Part 1.
Janicke, L., Janicke, U., 2002. Entwicklung Einesmodellgestutzten Beurteilungssys-
tems Furden Anlagenbezogenen Immissionsschutz. UFOPLAN200 43 256,
Umweltbundesamt. German Federal Environment Agency, Dessau-Roßlau,
Germany.
Jonkers, S., 2007. Handleiding CARII, Versie 6.1. TNO-rapport 2007-A-R0788/B. TNO
Bouw en Ondergrond. Apeldoorn, The Netherlands.
Keuken, M.P., Jonkers, S., Zandveld, P., Voogt, M., 2012. Elemental carbon as an in-
dicator for evaluating the impact of traffic measures on air quality and health.
Atmos. Environ. 61, 1e8.
McNally, M.G., 2008. The four-step model. In: Hensher, D.A., Button, K.J. (Eds.),
Handbook of Transport Modelling, second ed. Emerald Books, Bingley, UK,
pp. 35e52.
Monteiro, A., Miranda, A.I., Borrego, C., Vautard, R., Ferreira, J., Perez, A.T., 2007.
Long-term assessment of particulate matter using CHIMERE model. Atmos.
Environ. 41, 7726e7738.
Moussiopoulos, N., Berge, E., Bøhler, T., de Leeuw, F., Grønskei, K., et al., 1996.
Ambient Air Quality, Pollutant Dispersion and Transport Models. Topic Report
19. European Topic Centre on Air Quality. European environment agency,
Copenhagen, Denmark.
Ortúzar, J.D., Willumsen, L.G., 2011. Modelling Transport, fourth ed. Wiley, Chi-
chester, UK.
Panteliadis, P., Strak, M., Hoek, G., Weijers, E., Van der Zee, S., Dijkema, M., 2014.
Implementation of a low emission zone and evaluation of effects on air quality
by long-term monitoring. Atmos. Environ. 86, 113e119.
PTV, 2013. VISUM 13 User Guide. PTV-AG, Karlsruhe, Germany.
Qadir, R.M., Abbaszade, G., Schnelle-Kreis, J., Chow, J.C., Zimmermann, R., 2013.
Concentrations and source contributions of particulate organic matter before
and after implementation of a low emission zone in Munich, Germany. Environ.
Pollut. 175, 158e167.
Tchepel, O., Borrego, C., 2010. Frequency analysis of air quality time series for traffic
related pollutants. J. Environ. Monit. 12 (2), 544e550.
Tchepel, O., Costa, A.M., Martins, H., Ferreira, J., Monteiro, A., et al., 2010. Deter-
mination of background concentrations for air quality models using spectral
analysis of monitoring data. Atmos. Environ. 44 (1), 106e114.
Tchepel, O., Dias, D., Ferreira, J., Tavares, R., Miranda, A.I., 2012. Emission modelling
of hazardous air pollutants from road transport at urban scale. Transport 27 (3),
299e306.
Vardoulakis, S., Fisher, B.E., Pericleous, K., Gonzalez-Flesca, N., 2003. Modelling air
quality in street canyons: a review. Atmos. Environ. 37 (2), 155e182.
WHO, 2013. Health Risks of Air Pollution in Europe e HRAPIE Project. World Health
Organization, Regional Office for Europe, Copenhagen, Denmark.
WHO, 2014. Ambient (Outdoor) Air Quality and Health. Fact sheet N 313. World
Health Organization, Geneva, Switzerland. http://www.who.int/mediacentre/
factsheets/fs313/en/ (accessed 29.06.15).