What if air quality dictates road pricing? Simulation of an air pollution-based road charging scheme [original]

Journal of Urban Mobility 2 (2022) 100018

Contents lists available at ScienceDirect

Journal of Urban Mobility

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

What if Air Quality Dictates Road Pricing? Simulation of an Air

Pollution-based Road Charging Scheme

☆

Sandro Rodriguez Garzon

∗

, Marcel Reppenhagen, Marcel Müller

Service-centric Networking, Department of Telecommunication Systems, Technische Universität Berlin, Ernst-Reuter-Platz 7, Berlin 10587, Germany

a r t i c l e i n f o

Keywords:

Dynamic road pricing

Road charging scheme

Air pollution

Air quality

Toll

Traﬃc simulation

Intelligent transportation system

Smart city

a b s t r a c t

Road tolls serve various purposes, such as to reﬁnance the road infrastructure or to regulate traﬃc. They are

typically levied for the use of a freeway or the entire road network of a region or country. Toll charges may depend

on the duration of use, the vehicle’s emission class, the time of the day, the distance travelled, or the traﬃc volume.

The primary objective of a few recent toll system deployments is to internalize externalities with respect to

vehicle-caused air pollution. However, the air quality along the toll roads has so far not been considered to directly

inﬂuence road usage prices. This article investigated by simulation the expected monetary expenditure for drivers

and the traﬃc impact of applying a new distance and air pollution-based charging scheme in the metropolitan

region of Berlin. The road usage charges were determined on a per-trip basis by taking the vehicle’s emission

class, the distance travelled, and the air pollution levels along the route into consideration. The simulation results

indicate that it is beneﬁcial for drivers to avoid areas of high air pollution in order to reduce the trip’s total road

usage charges. The average additional detour distance is thereby short in comparison to the route’s length and

the resulting additional emissions do not increase to the same extent as the number of detours, since detours are

partly even shorter in terms of distance. The explorative analysis gives initial insights into the traﬃc eﬀects of a

charging scheme in which air pollution dictates road pricing.

Introduction

Urban air pollution originating from the sector of road transport is

known to signiﬁcantly increase the mortality and morbidity rate of the

aﬀected population ( Künzli et al., 2000 ). For 2018, 18,400 premature

deaths in the U.S. can be attributed to the exposure of traﬃc-related

air pollution ( Dedoussi, Eastham, Monier & Barrett, 2020 ). The situa-

tion is further complicated by the fact that the global traﬃc demand

is expected to increase over the next decade ( International Transport

Forum, 2019 ). Successfully applied regulatory measures to ﬁght traﬃc-

related air pollution include the promotion of environmentally friendly

means of transport ( Xia et al., 2015 ), introduction of traﬃc control

signal systems ( Wood & Baker ), or the development of more eﬃcient

combustion engines and aftertreatment systems. Nevertheless, the air

pollution in many metropolitan regions remains to exceed many times

the limits recommended by the World Health Organization (2016) or

as set by The European Parliament & the Council of the European

Union (2008) .

In Germany, local authorities of cities particularly aﬀected by

urban air pollution are being forced (due to possible penalties) to

☆The authors declare that they have no known competing ﬁnancial interests or personal relationships that could have appeared to inﬂuence the work reported in

this paper.

∗ Corresponding author.

E-mail address: [email protected] (S. Rodriguez Garzon).

introduce driving bans for vehicles of particular emission classes

( Fensterer et al., 2014 ) or to tighten speed limits on particular road

segments ( Vardoulakis et al., 2018 ). However, depending on how driv-

ing restrictions are implemented, they may have little or no eﬀect on

the urban air pollution ( Davis, 2017 ), worsen the air quality situa-

tion ( Zhang, Lin Lawell & Umanskaya, 2017 ), or even encourage ille-

gal behavior among drivers ( Wang, Xu & Qin, 2014 ). Driving restric-

tions that proved to signiﬁcantly reduce air pollution are - among oth-

ers - the vehicle pollution charge Ecopass in Milan ( Rotaris, Danielis,

Marcucci & Massiani, 2010 ), the low emission zones (LEZ) in Germany

( Jiang, Boltze, Groer & Scheuvens, 2017 ; Wolﬀ, 2014 ), and the driving

bans in Quito ( Carrillo, Malik & Yoo, 2016 ). The regulatory measures

are diﬀerent and customized to the local particularities, but the impacts

on the air quality are on a similar order of magnitude. In Milan, the Eu-

ropean vehicle class dictates whether downtown Milan can be entered

for free or by paying a charge that depends upon the vehicle’s emis-

sion class. In Germany, older and high polluting vehicles are generally

prohibited to enter LEZs. In Quito, the last digit of the vehicle plate

determines the day a vehicle is prohibited to enter the restricted area

during peak hours. These and similar driving restrictions are of intrusive

https://doi.org/10.1016/j.urbmob.2022.100018

Received 13 August 2021; Received in revised form 23 February 2022; Accepted 25 February 2022

S. Rodriguez Garzon, M. Reppenhagen and M. Müller Journal of Urban Mobility 2 (2022) 100018

nature and are not meant to change over a long period of time. They are

set by municipalities and apply no matter how the urban air pollution

eventually develops at each day and time of the day. Vehicles may be

banned, and drivers be charged although on some days, for example,

the ﬁne dust concentration within the air might turn out to lie far below

a value that is harmful to health.

To be more ﬂexible, a few municipalities and urban areas started to

introduce temporary LEZs that are only active in case high urban air

pollution is predicted or happened in the past days. In Geneva, only ve-

hicles with a state-issued vignette can drive within the LEZ during its

active period. The last day’s level of air pollution determines thereby

the type of restrictions that apply for the vehicles. Despite operating in

a more ﬂexible manner with respect to changing environmental circum-

stances, an activation still needs to be announced a few hours in advance

and an activation is valid throughout the day or at peak hours. These

regulations prevent the system to react spontaneously to an unforeseen

change of the air quality at the day of activation. A further drawback

of currently deployed systems with temporary LEZs is that the extent of

LEZs corresponds to administrative areas or are limited by well recog-

nizable ring-shaped closed transport routes such as city-rings or circular

railways. The range of a ﬁxed LEZ roughly corresponds to the extent

of the urban area that is aﬀected by urban air pollution, but it does not

necessarily represent it accurately on days with exceptional weather and

traﬃc conditions.

Another principal challenge of applying access-based charging

schemes to reduce air pollution lies in the fact that the costs of using

the road network within an LEZ do not depend upon the real usage. The

more a vehicle drives within an LEZ, the more it pollutes, but the costs

remain constant in Geneva and for all LEZs deployed around the globe.

The Green Activity Zones research project proposed a ﬁrst-best toll,

namely, to charge a driver of heavy-duty vehicles within a LEZ based

on the emissions measured within the vehicle ( Tretvik, Nordtømme,

Bjerkan & Kummeneje, 2013 ). A similar ﬁrst-best toll approach sim-

ulates vehicles in Munich that are charged based on their modeled

emissions ( Kickhöfer & Kern, 2015 ; Kickhöfer & Nagel, 2016 ). Both ap-

proaches propose charging schemes that are closely related to distance-

based charging; a charging scheme that has gained more and more popu-

larity in recent years. At toll plazas in France, a driver pays a charge that

depends upon the distance travelled on the freeway. In Germany, drivers

of heavy goods vehicles are obliged to pay a more ﬁne-grained distance-

based toll in case they use the federal highway network ( Broaddus &

Gertz, 2008 ). Instead of toll plazas, an on-board unit –as part of an elec-

tronic toll system - records the exact route so that a ﬁne-granular charge

based on a price per driven kilometer can be applied. But distance-based

charging is currently not applied to deal with urban air pollution de-

spite the proportional correlation between the distance travelled and

the amount of emissions ( Etyemezian et al., 2003 ).

Yet, the technological progresses of the last decade in the ﬁelds of

mobile computing, mobile communication, outdoor localization, and

electronic road pricing made it possible - today - to determine the route

of a vehicle by satellite, accurate to the meter, at low cost, and to trans-

mit it reliably to a central authority for control and accounting purposes

( Donath et al., 2009 ). Hence, with the proven technical and economic

feasibilities of distance-based charging schemes, the European Union

plans that until 2023, all European toll systems for heavy good vehi-

cles as well as buses must switch to distance-based charging to improve

fairness and environmental protection

1 https://www.europarl.europa.eu/news/en/press-room/

20181018IPR16551/reform-of-road-use-charges-to-spur-cleaner-transport-

and-ensure-fairness

At the same time, sensors for measuring air quality have become

more compact, more reliable, and less expensive ( Jova š evi ć-Stojanovi ć

et al., 2015 ). The unit price has even reached a level that allows private

households to measure air quality with their own sensor stations and

share the results with the community ( Muller et al., 2015 ). Coupled with

communication modules, they can now be an integral part of city-wide

dense sensor networks, enabling the urban air quality to be measured in

real time and across the entire city ( Sivaraman, Carrapetta, Hu & Luxan,

2013 ). It is therefore technically possible to develop an intelligent trans-

port system (ITS) that takes the current air quality distribution within

a metropolitan region into account. What role the urban air quality can

play in an ITS depends on the primary objectives, local conditions, and

political decision-makers. If, for example, air pollution hotspots should

be relieved by additional traﬃc-related emissions, then the spatial and

temporal extent of LEZs could be dynamically determined based on the

time-dependent spatial distribution of the air pollution across an urban

area ( Rodriguez Garzon & Küpper, 2019 ). In other words, the city-wide

air quality dictates the shape, size, and location of temporary LEZs.

In combination with distance-based charging, a driver could then be

charged based on the real distances travelled through dynamically de-

termined LEZs. A temporary LEZ based on high air pollution levels can

thereby be associated with high transit prices per kilometer to make a

transit through an air pollution hotspot less attractive for a polluter than

passing through low polluted area. The charging scheme can also option-

ally be combined with emission class-dependent pricing. The tariﬀ per

kilometer and air pollution level will then - in addition - depend upon

the vehicle’s emission class. A distance and air pollution-based charg-

ing scheme may lead drivers to avoid passing through heavily polluted

LEZs, respectively to stop worsening the situation by an additional pol-

luter, and to switch, instead, to more environmentally friendly means of

transport.

To the best of our knowledge, a highly dynamic road pricing scheme

with spatially and temporally evolving LEZs has not been investigated

yet. To do so, a multitude of research questions arise and need to be

examined with respect to, among many others, the technical feasibility,

usability from the perspective of the road users, the acceptance among

road users, the predictability of returns for the toll operators or collec-

tors, its implications on the mobility behavior, its inﬂuence on the urban

air pollution and its socio-economic eﬀects. This article investigates the

implications of such a distance-based and air pollution-aware charging

scheme on the mobility behavior by exemplarily simulating road net-

work usage within the city of Berlin. LEZs are hereby modeled based

on the spatial distribution of the urban air pollution across the city. The

traﬃc of 10% of the population was simulated for a period of 24 h on

three days with signiﬁcantly diﬀerent air pollution characteristics. An

open data-based transport demand model for Berlin was used in con-

junction with real measurements of the particulate matter concentration

taken by privately-owned sensors that were located within the limits of

Berlin. Although needed to ﬁnally evaluate the charging scheme with

respect to the objective of improving the overall air quality in an urban

area, its inﬂuence on the urban air pollution was not determined because

of a lack of yet to be developed particulate matter dispersion models,

a ﬁne-granular spatiotemporal wind and weather model for the city of

Berlin and a comprehensive list of location-speciﬁc polluters other than

motorized vehicles with their individual contributions to the particulate

matter concentrations. However, the results of simulating road usage -

with a hypothetical charging scheme being in place and the air pollution

being modelled by means of real world measurements - are intended to

give ﬁrst impressions of how parts of the traﬃc might shift to alter-

native routes, what daily volume and compositions of toll trips can be

expected in Berlin for days with low, medium, and high urban air pol-

lution and how the vehicle-caused emissions might change due to road

users intentionally bypassing LEZs with high transit costs.

S. Rodriguez Garzon, M. Reppenhagen and M. Müller Journal of Urban Mobility 2 (2022) 100018

In the following section, recent approaches to set road, parking facil-

ity, and public transport prices based on the predicted or current urban

air pollution are discussed. The core concept of the new air quality and

distance-based charging scheme is then introduced in Section Concept.

In Section Simulation, the simulation setup for Berlin and the simula-

tion results are presented. Assumptions, limitations of the simulations,

and the charging scheme in general are summarized and discussed in

Section Discussion. The article concludes with the major ﬁndings, open

research questions, possible applications, and extensions of the charging

scheme.

Related work

The pros and cons of applying dynamic road pricing schemes to

tackle congestion and air pollution have been discussed extensively over

the past decades ( Saharan, Bawa & Kumar, 2020 ; Yang & Huang, 2005 ).

The main property that distinguishes the concept of dynamic road charg-

ing from its static counterparts is the fact that either the road usage fee

as a whole or only a fraction of it may vary depending on the situa-

tional circumstances. Time of the day ( Chen, Xiong, He, Zhu & Zhang,

2016 ), road network load ( Supernak, Steﬀey & Kaschade, 2003 ), and

car occupancy are the most widespread discussed and considered price-

inﬂuencing situation factors in literature or in real world deployments.

Despite the considerable amount of work examining the impact of dif-

ferent charging schemes on the air quality ( Beevers & Carslaw, 2005 ;

Cavallaro, Giaretta & Nocera, 2018 ; Johansson, Burman & Forsberg,

2009 ; Rotaris et al., 2010 ), only a few research articles suggest that the

urban air quality should be considered, directly or indirectly, through

metrological properties, as an additional variable for dynamic road pric-

ing. Coria, Bonilla, Grundström & Pleijel (2015) propose to adjust road

prices dynamically so that the sum of traﬃc-related emissions do not

exceed the maximum air pollution capacity of an environment (assim-

ilative capacity). Coria et al. (2015) use the wind speed to estimate the

current assimilative capacity and the current traﬃc load to vary the

prices accordingly, instead of measuring the emissions respectively air

quality. The location-speciﬁc correlations between wind speed, air pol-

lution dispersion, and traﬃc network load are thereby predetermined

based on historical hourly wind speed, air quality, and traﬃc ﬂow data

for the city of Stockholm. The road usage prices do not reﬂect the cur-

rent air quality itself but the atmosphere’s ability under the current

wind and traﬃc conditions to cope with traﬃc-related air pollution,

without suﬀering from long-term environmental damage. Further in-

vestigations revealed that pollution peaks indeed can be smoothed if

road charges take the air pollution dispersion into consideration ( Coria

& Zhang, 2017 ). Coria’s approach and the approach taken in this arti-

cle diﬀer in the sense that Coria et al. (2015) examined the impact of

a traﬃc-, weather and environment-aware charging scheme on the air

quality while this article deals with the impact of an air quality-based

charging scheme on the traﬃc itself. Despite not dealing with road usage

charges, Reddy, Yedavalli, Mohanty, & Nakhat, 2017 sketched the idea

to link future public transit prices in general to the predicted urban air

pollution. Poorzahedy, Aghababazadeh and Babazadeh (2016) follow a

similar idea and propose to determine the next day’s cordon access fees

and rates for park & ride facilities at public transport hubs based on the

current carbon oxide concentrations and the next day’s weather fore-

casts. The objective is to make public transport pricewise more and pri-

vate vehicle transport less attractive in case high air pollution levels are

forecasted. Although lesser ﬂexible with respect to price determination,

Costabile and Allegrini (2008) present a prototypical ITS that forecasts

air pollution levels for Beijing and automatically imposes driving restric-

tions in case thresholds will likely be exceeded. Today, similar systems

for air quality-dependent driving restrictions are successfully deployed

in Geneva, Budapest, Madrid, and Oslo. But road prices remain to be de-

termined or driving restrictions be imposed based on air quality estima-

tions or forecasts. The estimated or forecasted air quality does not only

deviate from the actual air quality at the time a car ride is charged but

may even not be experienced everywhere in a ﬁxed LEZ or cordon with

the same intensity. Rodriguez Garzon and Küpper (2019) , in contrary,

describe the concept of an air pollution-aware and distance-based charg-

ing scheme in which the price per kilometer depends upon the urban

air pollution that is measured along the route. It incorporates distance-

based charging to fairly price the environmental damage caused by a

single car ride and it adjusts the fees dynamically to the urban air pollu-

tion to make the worsening of an already critical air pollution situation

more costly than the worsening of a non-critical situation. This article

adopts this new and highly dynamic concept of road pricing and applies

it in a customized way hypothetically to the city of Berlin.

Air pollution-aware charging

The road charging concept under investigation comprises of

distance-based charging and air pollution and emission class-dependent

road usage prices. The principal idea of the concept is that the road us-

age price per kilometer increases as urban air pollution levels rise along

the way, no matter, whether the air pollution originates primarily from

traﬃc or other sources such as wood burning or the industry sector. The

concept’s main objective is to reduce the urban air pollution at pollu-

tion hotspots by encouraging or incentivizing the drivers to reconsider

the transport mode and/or the route decision. By changing their route,

drivers are supposed to spread the emissions more evenly across the city.

The charging happens independently of the traﬃc load or assimilative

capacity of the environment. Even if the environment might be able to

cope with a very high air pollution without suﬀering long term damages,

prices per kilometer will be high if air pollutions levels are high. This

does not mean that the proposed charging scheme is not able to take the

assimilative capacity or traﬃc load into consideration. However, in this

article, it will solely be investigated in an isolated and unbiased manner

to simplify the interpretation of the results. In further studies and pos-

sible extensions, the assimilative capacity could, for example, be used

to support the tariﬀ speciﬁcation process while the variable traﬃc load

can serve as another dynamic variable for the price calculation.

In general, air pollution caused by traﬃc is made up of diﬀerent

components. Gas emissions in form of nitrogen dioxide, carbon monox-

ide, and sulfur dioxide or particulate matter are only some of the well-

known waste products of burning fossil fuel in combustion engines

( Berkowicz, Winther & Ketzel, 2006 ). The urban air pollution is usu-

ally given in the form of an air quality index (AQI) ( Cheng et al., 2007 ).

An AQI groups together diﬀerent pollutants and maps the combination

of pollutant concentrations to a few pollution levels, e.g., low, medium,

and high. Instead of progressively linking the road usage price directly

to the individual concentrations of pollutants in the air, in the proposed

concept, the road usage price at a location depends primarily on the

location-speciﬁc and discretized AQI level. A current AQI level can ei-

ther be determined for a whole toll area as applied in Geneva for the air

pollution-dependent LEZ, or at each position individually. An individ-

ual AQI level per arbitrary position, however, is only possible if the air

quality can be determined with high accuracy, at any road within the

toll area. Hence, the air quality sensor density across the toll area needs

to be suﬃciently high to be able to accurately interpolate air pollution

levels without signiﬁcant deviations from the real exposure. Given these

requirements are fulﬁlled, then, at each point in time, a toll area can be

spatially segmented into coherent zones with the same AQI level. Fig. 1

illustrates exemplarily the particulate matter-driven AQI zones as deter-

mined for the 24th of January 2020 at 9:00 AM in the downtown of

Berlin. The charging scheme interprets each blueish AQI zone as a tem-

porary LEZ with a ﬁxed price per kilometer as applied throughout the

whole LEZ. AQI zones might contain holes in which diﬀerent AQI levels

are measured or interpolated. The resulting temporary LEZs may there-

fore contain smaller LEZs that are associated with diﬀerent road usage

prices. Since temporary LEZs are not statically deﬁned by a human in-

stance but being processed based on the urban air pollution distribution,

S. Rodriguez Garzon, M. Reppenhagen and M. Müller Journal of Urban Mobility 2 (2022) 100018

Fig. 1. Particulate matter-driven AQI zones for downtown Berlin on the 24th of January 2020 at 9:00 AM, given in μg/m

city, district limits, or distinctive infrastructures, such as city highways,

do not serve as delimiters for the temporary LEZs

As the measured air quality at a given position can vary considerably

within a short period of time, e.g., due to a single gust of wind, so can the

extent of the temporary LEZs vary signiﬁcantly or the LEZs even “move ”

or disappear. To deal with the dynamics of urban air pollution and the

operational implications on an air pollution-based charging scheme, two

distinct measures are taken. To cope with strong short-term ﬂuctuations

at a sensor, the air pollution is measured over a certain period and the

sensor readings being averaged using the inverse distance weighting

method. The values measured last are thereby given a higher weight to

take trends into account. The resulting LEZs thus reﬂect the geographical

distribution of air pollution more accurately, but still change over time.

If temporary LEZs were determined in very short time intervals under

adverse weather conditions, it would be diﬃcult, if not impossible, to

predict the road usage charges for a longer car ride. However, according

to the Smeed Report ( Ministry of Transport, Great Britain, 1964 ), price

stability and ascertainability by the road user are crucial factors for the

success or acceptance of a road charging scheme. To make the price

of a journey more predictable, temporary LEZs are active for a certain

amount of time by freezing the measurement results (AQI zones) for an

activation period. For example, the air quality is measured from 1:45 PM

until 2:00 PM and the resulting temporary LEZs are active from 2:00 PM

until 3:00 PM. The longer the activation period is, the more predictable

are road charges for a car ride. On the other hand, towards the end of

an activation period, LEZs may tend to inappropriately represent the air

pollution distribution because of changing weather conditions. Hence,

there is always a trade-oﬀ between the accuracy of the air pollution dis-

tribution representation by means of temporary LEZs and the length of

the activation period. To ﬁnd the right balance, local conditions such

as average driving time, driver acceptance, and the local dynamics of

weather conditions must be considered.

In the proposed concept, the road usage price is not only linked to

the temporary LEZ a vehicle is driving through but also to the vehicle’s

emission class. As it is common in today‘s ITSs and toll system installa-

tions, the higher the emission of a vehicle is, the higher is the base price

per kilometer. The total road usage charge 𝑐

𝑡𝑜𝑡𝑎𝑙

of a car ride results from

the LEZ transits charges for all passed AQI zones 1 to 𝑧 :

𝑐

𝑡𝑜𝑡𝑎𝑙

𝑧

∑

𝑖 =1

𝑑

𝑖

∗ 𝑝

(𝑙

𝑖

, 𝑒

)

An LEZ transit charge is determined by multiplying the distance trav-

elled within the LEZ 𝑑

𝑖

with the price p() per emission class 𝑒 and AQI

level 𝑙

𝑖

of the AQI zone i . Thus, road usage charges are determined in-

dividually per trip, depending on the distance a vehicle traveled across

air polluted zones. In contrast to LEZs in Germany and Switzerland, all

vehicle types are permitted to enter temporary LEZs with arbitrary AQI

levels.

For an operational deployment of the proposed charging scheme

within an ITS, vehicles must be accurately locatable in a continuous

fashion and air pollution must be directly measurable by connected sen-

sors on a ﬁne granular manner in a city-wide scale and accurately in-

terpolatable at positions where no air pollution sensors are located. In

addition, road users should be able to examine expected road usage fees

prior to a trip and be notiﬁed of sudden price changes via a sort of con-

nected mobile device such as a smartphone or on-board unit. But the

investigations conducted in this article assume the dynamic charging

scheme to be hypothetically applied to road network users independent

S. Rodriguez Garzon, M. Reppenhagen and M. Müller Journal of Urban Mobility 2 (2022) 100018

of any concrete technical implementation of it. The reader is referred

to the article “Pay-Per-Pollution: Towards an Air Pollution-Aware Toll

System for Smart Cities ”( Rodriguez Garzon & Küpper, 2019 ) for a more

detailed discussion about the requirements of such a charging scheme

like price predictability, the corresponding challenges like user accep-

tance and its technical feasibility.

Simulations

The traﬃc in Berlin was simulated with the air pollution-aware

charging scheme being applied in order to determine the daily volume

of transit journeys through AQI zones exemplarily on days with varying

levels of urban air pollution. This makes it not only possible to estimate

future revenues for the toll collector and local authorities, but also to

examine the impact of the road charging scheme on the traﬃc and the

traﬃc-induced emissions. The research questions investigated in this ar-

ticle are the following:

■RQ1 : What is the overall toll volume in terms of expected AQI zone

transits per day?

■RQ2 : How often are tariﬀ changes experienced by the drivers during

a trip?

■RQ3 : To what extent are AQI zones bypassed?

■RQ4 : What are the characteristics of the detours?

■RQ5 : To what extent are emissions from motorized vehicles chang-

ing?

■RQ6 : What is the impact of detours on the traﬃc distribution?

Whether the proposed air pollution-aware charging scheme reduces

the overall air pollution or, at least, helps to evenly distribute the emis-

sions across the urban area are not investigated and remain as open

questions for further research.

The greater area of Berlin was chosen as the hypothetical toll area

because Berlin has suﬃcient air quality sensors distributed across the

city area, only a minor amount of deep urban canyons due to mostly

low multi-story or single-family buildings, ﬂat terrain, and city-wide ho-

mogenous weather conditions. In addition, there is an open data-based

transport demand model for Berlin publicly available ( Ziemke, Kad-

doura & Nagel, 2019 ). The synthesized transport demand model was

developed by Ziemke et al. (2019) using diﬀerent open data sources

such as, e.g., a nation-wide census of Germany, commuter statistics,

and local traﬃc counts. A comparison of the model’s results with those

of two independent travel surveys shows that it realistically emulates

transport demand in and around Berlin. The reader is referred to the

work of Ziemke et al. (2019) for more details about the synthetization

and validation of the transport demand model. The investigations of the

proposed charging scheme can then be conducted with urban air pol-

lution and transport demand models that represent or, in case of the

transport demand model, well approximate real-world conditions.

In the following, the experimental setup of the simulations is de-

scribed, including the general assumptions and conﬁgurations. After-

wards, the simulation results are presented in detail.

Setup

Berlin has an urban area of approx. 891.68 km

, a population den-

sity of approx. 4115 inhabitants per km

2 and a road network of ap-

prox. 5437 km length. Berlin is in a moderate climate zone with prevail-

ing continental southwest winds and maritime northwest winds. Due to

lesser traﬃc, ﬂat terrain, weather conditions, and only a small manufac-

turing industry within the city limits, the air pollution is not as severe

as in cities with a similar population of about 3.7 million. Neverthe-

less, on 27 days in 2018, a state-operated measuring station in Berlin

recorded an exceedance of the limits for ﬁne dust pollution with PM

2 https://www.berlin.de/sen/uvk/presse/pressemitteilungen/2019/

pressemitteilung.775788.php

Table 1

Particulate matter-driven air quality index.

(μg/m

) 0 - 20 20.1 - 35 35.1 - 50 50.1 - 100 > 100

AQI level 0 1 2 3 4

Air quality very good good moderate poor very poor

(particulate matter with a diameter of 10 𝜇m or less). However, 6 of

the 16 state-operated and connected air quality measuring stations are

located especially at major roads and 5 out of 16 are located at the

outskirts or forests where a particularly high respectively low air pollu-

tion is expected

. Measurements taken in urban background roads may

signiﬁcantly vary from the air pollution levels measured along major

roads ( Boogaard et al., 2011 ). Hence, measurement stations along ma-

jor roads are only of limited help to properly interpolate air pollution

levels in their surroundings due to their strong bias towards high air

pollution. Only a “well-placed site ” allows to estimate proper air qual-

ity values for the surroundings ( Williamsand et al., 2014 ). Privately-

owned and operated air pollution sensors are installed at windows, on

balconies, walls, or roof-top terraces and are arbitrarily located across

a metropolitan region. Community-driven air quality measurement ini-

tiatives, in general, gained popularity over the last decade, leading to

multiple deployments worldwide. In Berlin and similar metropolitan re-

gions, the amount of privately-owned and operated sensors, as used for

crowdsourcing air quality data, exceeds the number of state-operated

sensor installations many times. During the ﬁrst half of 2020, between

375 and 430 private air pollution sensors within the limits of Berlin

contributed their sensor readings to the citizen science initiatives Luft-

daten.info

4 ( Blon, 2017 ) and OpenSenseMap

. Fig. 2 shows the loca-

tion of Luftdaten.info- and OpenSenseMap-connected sensors across the

metropolitan region of Berlin. Nine most probably faulty sensors were

not considered because they delivered constantly air quality values that

reached far beyond the ones of their closest neighboring sensors. Au-

tomatic outlier detection and ﬁltering was deployed as well. The local

sensor density correlates in Berlin roughly with the population density

( Arandelovic & Bogunovich, 2014 ). The low sensor density in the north-

west, mid-west, south-west, and south-west is attributed to a low pop-

ulation density because lakes and forests predominate the respective

outskirts of Berlin.

Every two to three minutes, the air quality is measured by the

privately-owned and operated sensors and the readings being trans-

mitted to a central unit managed by the initiatives. Based on these

crowdsourced and publicly available measurements in Q1 and Q2 of

2020, three days were selected to adequately represent days with low,

medium, and high PM

concentrations in Berlin. Fig. 3 visualizes the

AQI zones for Berlin at each of the selected days at diﬀerent day times.

The discrete mapping of PM

concentrations to AQI levels is based on

the air quality index provided by the Federal Environment Agency in

Germany

. Table 1 shows the mapping from PM

values to AQI levels

and air qualities as used throughout this article. AQI level 0 is consid-

ered as harmless to health and a transit through an AQI zone of level

0 will therefore not be charged. At the 25.03.2020 (low average PM

concentration), southeast winds with wind speeds of about 7–17 km/h

were measured throughout the day. At the 11.06.2020 (medium av-

erage PM

concentration), northeast winds of about 6–8 km/h were

registered while on the 24.01.2020 (high average PM

concentration),

north and east winds of about 2–14 km/h were measured. In general,

wind speeds were low which in turn increased the accuracy of inter-

polations as described in the next paragraph. In this study, gases such

3 https://www.berlin.de/senuvk/umwelt/luftqualitaet/de/messnetz/blume.shtml

4 https://www.luftdaten.info

5 https://www.opensensemap.org

6 https://www.umweltbundesamt.de/berechnungsgrundlagen-

luftqualitaetsindex

Loading more pages...