Environmental Modelling & Software 16 (2001) 399–414

www.elsevier.com/locate/envsoft

Urban air quality in larger conurbations in the European Union

a,*
Frank A.A.M. de Leeuw , Nicolas Moussiopoulos b, Peter Sahm b, Alena Bartonova c

a
RIVM, P.O. Box 1, 3720 BA Bilthoven, The Netherlands
b
Aristotle University Thessaloniki, Box 483, 54006, Thessaloniki, Greece
c
NILU, P.O. Box 100, N-2027, Kjeller, Norway

Received 23 March 2000; received in revised form 5 July 2000; accepted 22 January 2001

Abstract

The air quality in about 200 urban agglomerations within the European Union (EU) is calculated for a reference year (1995)
and for the year 2010. Relatively simple, robust tools have been applied which allow for a generalisation of the results on the scale
of the whole EU. The projected air quality gives information on the frequency and severity of exceedance of air quality objectives
and on the fraction of EU urban population potentially exposed. The parameter calculated is the urban background concentration,
which is representative for the concentration in most of the urban area, with the exception of places under direct influence of
sources, such as streets. Pollutants considered are SO2, NO2, PM10, Pb, O3, CO, and benzene. In 2010 the urban background
concentrations will decrease strongly in the set of 200 modelled cities. It is projected, however, that the agreed or proposed air
quality standards will still be exceeded in the future. The most serious problems are exceedances of the short- and long-term
objectives for PM10 and exceedance of the long-term objective for NO2.  2001 Elsevier Science Ltd. All rights reserved.

Keywords: Urban air quality; Traffic emissions; Abatement strategy

Software availability Software requirements: FORTRAN 77 compiler

(1) Program title: cQ Program language: FORTRAN
Developers: A. Bartonova, S Larssen Availability: currently internal use only.
Contact address: NILU, P.O. Box 100, N-2007, Kjeller, (3) Program title: UAQAM
Norway. Tel.: +47 63 89 8152; fax: +47 63 89 Developers: A. van Pul, F. de Leeuw
80 50; e-mail: alena.bartonova@nilu.no Contact address: RIVM, P.O. Box 1, 3720 BA Bil-
First year available: 1996 thoven, the Netherlands. Tel.: +31 30 274 2806;
Hardware requirements: PC fax: +31 30 228 7531; e-mail: tkk.de.
Software requirements: none leeuw@rivm.nl
Program language: Excel spreadsheet First year available: 1996
Availability: currently internal use only. Hardware requirements: PC or UNIX
(2) Program title: OFIS Software requirements: FORTRAN 77 compiler
Developers: N. Moussiopoulos, P. Sahm Program language: FORTRAN
Contact address: Aristotle University Thessaloniki, Box Availability: currently internal use only
483, 54006 Thessaloniki, Greece. Tel.: +30 31
996011; fax: +30 31 996012; e-mail:
moussio@eng.auth.gr
First year available: 1996
Hardware requirements: UNIX 1. Introduction

Europe is a highly urbanised continent. In 1990,

* Corresponding author. Tel.: +31-30-274-2806; fax: +31-30-228- ⬎70% of the total population lived in urban agglomer-
7531. ations. The concentration of human activities in a rela-
E-mail address: frank.de.leeuw@rivm.nl (F.A.A.M. de Leeuw). tively small area puts enormous pressure on the urban

1364-8152/01/$ - see front matter  2001 Elsevier Science Ltd. All rights reserved.
PII: S 1 3 6 4 - 8 1 5 2 ( 0 1 ) 0 0 0 0 7 - X
400 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414

system and has led to numerous environmental problems full coverage of urban areas. Results of these comp-
(e.g. noise, waste, air pollution). On the basis of pan- lementary procedures are presented together and dis-
European surveys of the state of the environment (EEA, cussed.
1995, 1997, 1998) urban air pollution was regarded by The air quality objectives used in this study are all
the European Environment Agency (EEA) as one of the related to the protection of human health and are based
most important environmental problems in Europe. on adopted or proposed daughter directives (European
Whereas in the past industry and energy production have Commission, 1999a,b,c, see Table 1). Ambient concen-
been the dominant sources of urban air pollution, now- trations must be in compliance to these objectives by the
adays transport, in particular road traffic, is the most year 2010. In addition, assessments have been made for
important source. benzo(a)pyrene [B(a)P], but they are not reported here.
In order to assess the urban air quality on European Emissions estimates and dispersion calculation for B(a)P
scale, a combination of atmospheric dispersion models are too uncertain to derive any firm conclusion.
and air quality measurements should be used. The extent Assessments for lead are not reported here in detail.
of observational data available on urban air pollution is The calculations for lead clearly show that under base
not large enough to fully characterise urban air quality case scenario conditions, in 2010 the urban background
on an European scale (Larssen and Hagen, 1996). concentration will be well below the objective of
Atmospheric dispersion models on urban scale have 0.5 µg/m3 (European Commission, 1999a) as annual
been used, for example, in the Auto Oil programmes I mean. In 1990 (the reference situation in the Pb-
and II, launched by the European Commission in 1992 calculations) there are 14 cities (23% of the total urban
and 1997. These programmes used detailed modelling to population) out of compliance.
study cost-effective measures to reduce road transport
emissions, and to provide a foundation for a transition 2.1. Selection of cities
towards longer term air quality studies covering all emis-
sion sources. In the Auto Oil programmes, compu- Violation of air quality objectives is primarily
tationally demanding dispersion models were used in up expected in larger conurbations with a high emission
to ten European cities. density. All cities in the 15 Member States of the Euro-
A simpler methodology, based on less demanding pean Union (EU15) with ⬎250 000 inhabitants were
atmospheric dispersion and empirical modelling, has included in this modelling exercise. In order to fully util-
been proposed by the EEA’s European Topic Centres on ise the available measurement data, this list has been
Air Quality (ETC/AQ) and on Air Emissions (ETC/AE)
to complement the Auto Oil II programme (De Leeuw Table 1
et al., 2000). In contrast to the detailed modelling of ten Environmental objectives used in this study
cities, this approach aimed at a more general assessment
Pollutant Averaging period AQ standard/objective
of urban air quality in the European Union (EU). More
precisely, the goal of this “Generalised Exposure Assess- SO2a 1h 350 µg/m3 not to be exceeded
ment (GEA)” is to estimate the fraction of the urban more than 24 times a calendar
population that is living in European cities that are not year
in compliance with air quality objectives in future years SO2a 24 h 125 µg/m3 not to be exceeded
more than three times a calendar
and to estimate additional emission reductions needed to year
reach compliance. In the GEA-approach, simple robust NO2a 1h 200 µg/m3 not to be exceeded
tools are used to calculate, in a consistent way, air qual- more than 18 times a calendar
ity in a relatively large number of cities. The consistency year
allows for a generalisation of the results on the scale of NO2a Calendar year 40 µg/m3
PM10a 24 h 50 µg/m3 not to be exceeded
the whole EU. Pollutants considered are SO2, NO2, more than seven times a calendar
PM10, O3, CO, and benzene. year
PM10a Calendar year 20 mg/m3
COb 8h 10 mg/m3
2. Methodology O3c Daily 8-h max 120 µg/m3 not to be exceeded
⬎20 days per calendar year
Benzeneb Calendar year 5 µg/m3
The main intention of this study is to provide infor- Calendar year 0.5 µg/m3
mation on current and future non-attainment of air qual-
a
ity objectives related to the protection of human health Limit values for the protection of human health from the Directive
for the urban population in the EU. A combination of on SO2, NO2, PM10 and Pb (European Commission, 1999a).
b
Limit values for the protection of human health from proposed
measuring and modelling tools is applied. Measurement Directive on CO and Benzene (European Commission, 1999b).
data is used wherever possible; it is complemented using c
Target value for the protection of human health from proposed
the results of atmospheric dispersion models to obtain a daughter Directive on Ozone (European Commission, 1999c).
 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414 401

extended with about 50 smaller cities with reliable moni-

toring data. The full list contains 192 conurbations with
a total population of 113 million, see Fig. 1. The selected
cities cover about 30% of the total population in the
EU15. The calculations for ozone concentrations have
been carried out for a sub-set of 57 cities, including
almost all cities with ⬎0.5 million inhabitants (Fig. 2).
It is difficult to relate the selected population quanti-
tatively to the total urban population within the EU15 as
different definitions of urban areas are in use (e.g.
according to population density, on the basis of existing
infrastructure and services, or for administrative
reasons). The coverage varies from 40% — using popu-
lation statistics based on UNEP (1993) — to at least
70% — using data from urban agglomerations with
100 000 inhabitants or more (UN, 1997).
The built-up area of the selected cities has been esti-
mated by a procedure developed by the ETC on Land
Cover (ETC/LC, 1997) using detailed land-use and land-
cover information.
Measurement data for urban background stations for
SO2, NO2, particulate matter (PM10) and Pb have been
taken from a recent update of AIRBASE [an European
air quality database and information system developed

Fig. 2. Selected cities for OFIS model applications.

by ETC/AQ, see EEA (1999)] for the period 1992–1996.

Only cities with more than one site-year of data have
been selected. Depending on the component, monitoring
data is available for cities covering 30–80% of the selec-
ted population.

2.2. Urban emissions

Urban emissions were estimated by downscaling

national emission totals (EEA, 1996b). While this simple
procedure is clearly approximate, it offers the advantage
of providing comparable emissions for all cities.
Estimates were made on the basis of available data:

앫 national emissions per sector as given in the Auto Oil

base case scenario (SENCO, 1999);
앫 detailed information on emissions from CORINAIR-
90 at a NUTS-3 administrative level and a SNAP1
level of sector detail, available for SO2, NOx, CO and
VOC from the CORINAIR emission inventory
(EEA, 1996a).

CORINAIR provides an inventory based on well-

Fig. 1. Selected cities for cQ and UAQAM model applications. described and accepted procedures (UN-ECE, 2000). To
402 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414
guarantee the consistency in the urban emission, no
modifications have been made in the emission data set.
In a few cases, the model results suggested possible
errors in the data sets; in those cases sensitivity calcu-
lation have been made to test this hypothesis (see
below).
CORINAIR defines source categories of air emissions
in the Selected Nomenclature for sources of Air Pol-
lution (SNAP) at various level of detail. The SNAP1
level defines 11 broad economic sectors such as traffic
(sector 7) or agriculture (sector 10). The geographical
distribution of emissions is based on the NUTS-classi-
fication (Nomenclature of Territorial Units for Statistics,
a classification of administrative regions defined by
EUROSTAT, the Statistical Office of the EU).
The top–down approach is different for large point
sources and for low-level area sources. Large point
sources with known co-ordinates were allocated to a city
when their distance to the city centre is less than the
radius of the city. For the area sources, the top–down
approach involved scaling of NUTS-3 emission esti-
mates to a local level through the use of indicators of
the proportion of a particular activity occurring in the
specified local area. In the current application the popu-
lation was used as a proxy for the statistical indicator for
all sectors. Emissions from agriculture and nature were
assumed to occur in rural areas only and were excluded
from the urban emissions.
Emissions for years other than 1990 were not avail-
able for any of the pollutants at a NUTS-3 level of detail.
Urban emissions for other years were estimated by sca-
ling the 1990 emission data according to the ratio of
national emissions (at a SNAP1 level) between 1990 and
the selected year.
The allocation of point sources critically depends on
the assumed size of the urban area. This effect was
investigated in two sensitivity runs for urban NOx emis-
sions recalculated:
1. with an urban radius increased by 5 km; and
2. with a uniform increase in radius of 10%.
The results (see De Leeuw et al., 2000) indicate that for
small cities, the 5 km increase may have a large effect
since it may easily lead to a doubling of the urban area.
However, for the vast majority of cities, the sensitivity
calculation results in emission differences not
exceeding 10%.
For major French cities, urban emissions were also
calculated by CITEPA (Fontelle et al., 1997). In the UK
the London Research Centre has collected emission data
for eight of the cities considered here (Hutchinson and
Clewley, 1996; Buckingham et al., 1997a,b,c, 1998). A
comparison shows a fair agreement between the GEA
approach and the independent French and British esti-
mates. The largest differences are found in the SO2 emis-
sion estimates, most probably caused by different treat-
ment of large point sources. For NOx, GEA estimates
are higher than the CITEPA results.
2.3. Air pollution models
In the GEA study, three air pollution models have
selected (EEA, 2000) for the calculation of air quality
parameters from urban emissions:
앫 The cQ-model for ‘inert’ species where sufficient
monitoring data were available, that is NOx/NO2, SO2
and PM10.
앫 The UAQAM model for ‘inert’ species in all cities,
that is NOx/NO2, SO2 and PM10, Pb, CO and benzene.
앫 The OFIS model which was applied to calculate
ozone concentrations for a limited number of cities.
Strictly speaking none of the considered pollutants is
‘inert’ as all are involved in atmospheric chemistry. In
this study, however, pollutants are indicated as ‘inert’
when the photochemical decay rate is small compared
to the residence time over the urban area. With a rep-
resentative wind velocity of 3 m/s the residence time will
not exceed 2 h for cities with a diameter of 20 km or
less. The pollutants NOx, SO2, PM10, Pb, CO and ben-
zene are approximately inert at these time scales and
have been treated here as ‘inert’.
It is recognised that the methodology of the selected
models may not be appropriate for locations with
extreme orographic influence resulting in inhomo-
geneous flow patterns; such conditions cannot be
resolved by the OFIS and UAQAM model.
All models calculate the contribution to the urban
background concentrations resulting from the local,
urban emissions. The regional background contributions
from emissions outside the considered urban area were
derived from the EMEP acidifying (Tsyro, 1998) and
photochemical model (Simpson et al., 1997) or from the
TREND model (van Jaarsveld, 1995) using emission
data for 1995 and 2010.
For estimation of CO background concentrations a
different approach has to be taken. Although CO is an
important precursor for ozone and is therefore included
in photochemical models, regional background concen-
trations of CO are commonly not included in the output
of these long-range transport models. Based on measure-
ments at rural stations in the United Kingdom and the
Netherlands an annual averaged level of 0.45 mg/m3 is
estimated. The temporal variation in background levels
is described as a log-normal distribution with a 50- per-
centile value of 0.45 mg/m3 and a sigma value of 0.6.
In UAQAM calculations it is further assumed that the
temporal variations in background concentrations corre-
late perfectly with the variations in urban levels, that is,
 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414
the 98-percentile values of the urban and regional contri-
butions are reached at the same moment.
Between 1990 and 2010 the CO-emissions in EU 15
decrease with ⬎50%; as there are contributions from
other European countries and as there is an in situ contri-
bution to CO from photochemical oxidation, it is
assumed that the decrease in regional background con-
centrations is less than the decrease in emissions. The
50 percentile value has been reduced to 0.35 mg/m3, that
is, a reduction of about 20–25%.
2.3.1. cQ-model
The cQ-model is an empirical model that relates urban
emissions to observed concentrations corrected for
actual regional background, assuming a proportionality
that is determined by local dispersion meteorology and
emission conditions as well as spatial distribution of
sources. A full description of the model and some of its
applications can be found in CEC (1998) and Olsthoorn
et al. (1999). This model has been applied for ‘inert’
species and for cities with sufficient air quality monitor-
ing data to estimate future air quality. The requirements
on monitoring data limit the use of the cQ-model to SO2,
NO2 and PM10. The model uses current or historical
emissions [based on the CORINAIR inventory, see, for
example, EEA (1996a)] and air concentrations data
[based on AIRBASE (EEA, 1999) or Larssen and Hagen
(1996)] to derive an empirical ratio between urban emis-
sions Q and the air concentration c corrected for regional
contributions. Using this ratio (representing an average
for several years) and predicted emission data, the model
then estimates future air concentrations. Different types
of sources are assumed to contribute differently to the
calculated concentrations by using effectivity factors
related to the assumed emission height. The relationships
for NO2 are calculated after a conversion of NOx to NO2
concentrations using an empirically derived relation. For
consistency this relation is identical with the relationship
obtained from the Auto Oil I study (DGXI, 1996).The
cQ-model was previously used in a study for DG
Environment (CEC, 1998; Olsthoorn et al., 1999). For
many of the cities, the cQ-relationships for SO2,
NOx/NO2 and (for some cities) also PM10 gave consistent
results for all compounds. This supports the validity of
the method.
One strength of the cQ-model is its basis in actually
measured air quality. Uncertainties include:
앫 uncertainties in reference emission estimates. Results
are affected by the accuracy of relative sector source
strengths and relative reductions, rather than absolute
emission figures;
앫 uncertainty related to the effectivity factor for traffic
emissions vs. other emission categories. This plays a
minor role in the current study, where only traffic
related measures are considered;
403
앫 representativity of monitoring stations for the city
background concentrations. Data for PM10 is con-
siderably less complete than for NO2 or SO2. Any EU-
wide PM10 analysis will be hampered from this lack
of data;
앫 representativity of monitoring period for all relevant
meteorological conditions.
2.3.2. UAQAM model
The Urban Air Quality Assessment Model (UAQAM)
is a dispersion model calculating (annual) average city
background concentrations and exceedances of air qual-
ity thresholds on an hourly or daily basis from actual
meteorology and urban emissions.
Input data are limited to urban emissions, city area,
regional background concentrations and meteorological
observations only. The structure of the model is trans-
parent; simple parametrisations simulate the most
important phenomena. A full description of the model
has been given by van Pul et al. (1996).
UAQAM contains three modules in which emissions,
meteorological (boundary-layer) parameters and disper-
sion are modelled. Statistical information on activity pat-
terns (traffic intensity, industrial activity, relation
between heating demand and ambient temperature) is
used to obtain seasonal and diurnal variations in emis-
sions. The UAQUAM model does not include atmos-
pheric chemistry and it is therefore only applied for inert
pollutants. To account for the NO2/NOx conversion, the
NO2 concentration is estimated from the calculated NOx
concentration using an empirical relation (BUWAL,
1997) which is based on monitoring data in Germany,
Austria and Switzerland. Compared to the NO2/NOx
relation used in Auto Oil I, this relation is nearly equiv-
alent at lower NOx concentrations. At high NOx-levels
the BUWAL-equation estimates lower NO2 concen-
trations which are in better agreement with measure-
ments.
2.3.3. OFIS model
The OFIS model (Ozone Fine Structure Model) is a
photochemical dispersion model for calculating ground
level ozone concentrations in and around urban areas
(Sahm and Moussiopoulos, 1999). It belongs to the Eur-
opean Zooming Model system (EZM), a comprehensive
model system for simulations of wind flow and pollutant
transport and transformation (Moussiopoulos, 1995).
The EZM may be used either in conjunction with a
regional scale model, for example, the EMEP MSC-W
model (Simpson 1993, 1995), or as a stand-alone model
system driven directly with measured data.
Having been derived from the well-tested core models
of the EZM, the OFIS model retains all elements neces-
sary for a realistic assessment of urban scale ozone lev-
els. The Eulerian photochemical dispersion model OFIS
404 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414
is capable of simulating transport and photochemical
transformation processes in an urban plume. This allows
an adequate description of urban photochemistry and
atmospheric dynamics with a very low computational
effort, which extends its applicability to longer time per-
iods. Thus it may be used for calculating urban scale
ozone concentration (e.g. for exposure analyses or for
assessments of control strategies) based on large-scale
meteorological information and long range transport
information over a longer time period.
The conceptual basis of OFIS is a coupled 1D/2D
approach. Background boundary layer concentrations are
computed with the multi-layer box model embedded in
OFIS for a domain of typically 150×150 km2 with the
city in the centre and rural area all around. For each day
in the period considered, pollutant transport and trans-
formation downwind of each city is calculated with a
multi-layer multi-box model assuming the prevailing
wind direction valid for the respective day, an initial
plume width according to the city diameter and a plume
widening angle of 30°. As an option, OFIS allows taking
into account the influence of large emission sources in
the vicinity of the urban area depending on the wind
direction prevailing on the respective day. Local circu-
lation systems (such as the sea-breeze in coastal areas)
can be taken into account by inverting the wind direction
of the urban plume in the lower layers (i.e. up to the
height of the mixed layer) in the afternoon hours of days
with weak synoptic forcing and off-shore wind direction.
Dry deposition is accounted for by using a three-resist-
ance model approach. Biogenic emissions are taken into
account for rural areas.
OFIS was applied to the Athens and Stuttgart area in
the context of the European Commission’s Communi-
cation on an Ozone Strategy (Moussiopoulos et al.,
2000). Regional background concentrations were
derived from results of the EMEP model (Simpson et
al., 1997).
The EMEP/OFIS model cascade describes satisfac-
torily the levels of ozone exposure, resolving both down-
town ozone titration and ozone formation in the urban
plumes of Athens and Stuttgart.
2.4. Population exposure
Population exposure can be expressed in several ways,
giving emphasis to different features of exceedances.
The simplest way is to calculate a number of inhabitants
exposed to values above given environmental objectives.
For evaluating the extent of exceedance of environmen-
tal objectives, the population exposure above a threshold
(PET) is defined as:
冘冘
Ncity En
PET⫽ (Ci,n⫺T)popn
n⫽1 i⫽1
where Ci,n is the concentration in excess of the threshold
value T in city n during exceedance i; Ncity is the number
of cities where an exceedance is calculated; En is the
number of exceedances; and popn is the population of
city n. PET is expressed in persons µg/m3.
Note that the urban background concentration is used
here to estimate the potential exposure of the population.
In estimating the actual exposure improvements can be
made by considering other micro environments, such as
indoor or street, in combination with information on the
time spent in the various micro environments. However,
in epidemiological studies to health effects of air pol-
lution the urban background concentration is frequently
used as proxy for a personal exposure. Urban back-
ground concentrations may well reflect the average con-
centration experienced by the resident when moving
within the town and may therefore be used in health
impact assessments (Krzyzanowski, 1997; WHO, 1995).
3. Results for the reference case 1995
3.1. Ozone
Simulations with the OFIS model were performed for
summer 1995. Scatter plots of observations and mod-
elled results show in general a satisfactory agreement
although notable mismatches are found. Fig. 3 shows the
cumulative distribution of the urban area maximum and
urban area average number of days with exceedance of
the running 8 h-average of 120 µg/m3 ozone versus the
percentage of population subject to these exceedances.
The distribution is shown separately for cities in North-
ern Europe (north of 46°N) and cities in Southern Europe
(south of 46°N).
Highest maximum values exceedances during 100
days or more) are calculated for cities in Southern Eur-
ope. About 75% of the population of cities in Southern
Europe and about 40% of the population of cities in
Northern Europe is exposed to maximum urban ozone
levels exceeding the running 8 h-average of 120 µg/m3
ozone for ⬎20 days. Evaluating the number of
exceedance day for urban average concentrations, 48%
of the urban population in EU15 lives in non-attainment
cities. More than 25% of the urban population in the
Southern Europe and about 5% of the urban population
in Northern Europe is exposed to exceedances of the
running 8 h-average of 120 µg/m3 ozone on ⬎20 days.
3.2. NO2
Fig. 4 presents a comparison between observed
(period 1992–1996) and modelled annual mean concen-
trations using the UAQAM model. The agreement, with
a root-mean-square-error (RMSE) of 12 µg/m3 and 94%
of the deviations within a factor of 2, is reasonable, tak-
 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414
Fig. 3. Cumulative distribution of days with exceedance of the running 8 h-average of 120 µg/m3 ozone (1995) over the population in the modelled
cities in Northern Europe (䊏) and cities in Southern Europe (왖); left, maximum in the urban area; right, average over the urban area.
Fig. 4. Comparison between observed and modelled urban NO2 back-
ground concentrations (annual mean, µg/m3); reference year 1995. The
2:1 overprediction and 1:2 underprediction lines are shown.
ing into account limitations in representativity of the
monitoring stations for urban background conditions.
Observations indicate that about 70% of the urban popu-
lation is exposed to an annual mean concentration in
excess of the objective; the model estimates this fraction
on 65%.
3.3. PM10
Emission estimates for particulate matter are less
developed than for other compounds. To investigate the
impact of some of the uncertainties, two versions of
PM10 emission estimates (identified as Version 4 and
405
Version 5) have been used that differ in the assumed
emission factors for tail pipe and non-exhaust transport
emissions and in the estimates for waste incinerators.
Version 5 results in the lowest emissions (SENCO,
1999).
Fig. 5 presents a comparison between observed and
UAQAM-modelled concentrations using both sets of
emission estimates. The modelled urban concentration
includes the contributions from local primary PM10
emissions as well as the regional background contri-
butions of primary and secondary PM10. Observations
cover the period 1992–1996. In addition to the modelled
urban background concentrations, the contribution of sea
salt (Eerens et al., 1998) is given. The scatter in the two
plots is too large to draw conclusions on the reliability
of the two emission sets. About 90–95% of the model
results is within a factor of 2 of the observations. Fig.
5 shows that for Lisbon (observed 75 µg/m3, modelled
24–32 µg/m3) and Setubal (observed 64 µg/m3, modelled
19–24 µg/m3) the modelled concentration strongly devi-
ates. Here, as well as for many other cities, the (semi)-
natural contribution of resuspended soil might be of
importance.
3.4. CO
A comparison between observed and UAQAM-mod-
elled concentrations is given in Fig. 6. The observed data
is averaged for 1994 and 1995 and obtained for all sta-
tions labelled as ‘urban background’ in AIRBASE. In
view of the uncertainties in the estimates of the regional
background and the uncertainties in the representativity
of the urban stations, there is a fair agreement (95% of
the model results fall within a factor 2 of the
measurements).
Exceedances of the environmental objective (Table 1)
have been calculated for 11 cities (14% of the total popu-
lation in all modelled cities). All of these cities are
located in the Southern part of Europe.
406 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414
Fig. 5. Comparison between observed and modelled urban PM10 background concentrations (annual mean, µg/m3); reference year 1995. Model
calculations are based on two sets of emissions (Versions 4 and 5, see text for further explanation); modelled concentrations are given with (䊏)
and without (䉬) a sea salt contribution.
Fig. 6. Comparison between observed and modelled urban CO back-
ground concentrations (annual mean, mg/m3); reference year 1995. The
2:1 overprediction and 1:2 underprediction lines are shown.
3.5. SO2
Fig. 7 gives a comparison between observed and
UAQAM-modelled annual mean concentrations. All
available annual observations in the period 1990–1994
are included. The modelled concentrations refer to the
year 1995. The agreement is reasonable: the RMSE is
16 µg/m3 and 63% fall within a factor of 2.
For the reference situation a frequent violation of both
the daily and hourly objective is noted. A total of 25%
of the urban population is exposed more than three times
per year to a daily averaged concentration in excess of
125 µg/m3; for the hourly objective, this number is
slightly less (23%).
The dependence on meteorological conditions is
shown in Fig. 8. Using the emissions for the reference
year 1995, UAQAM calculations are made using the
actual meteorological data for 5 years in the period
1990–1995; for technical reasons 1992 could not be
included. For each city the ratio between the actual
annual mean and the multi-year average has been calcu-
lated, similar ratios have been calculated for the 98-per-
centile values for daily means. Fig. 8 shows that, on the
European scale, a ‘worst case’ situation cannot be
defined. Selecting one meteorological period will result,
depending on the region, both in under- and overpredic-
tion of the multi-annual average. The graph for the
 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414
Fig. 7. Comparison between observed and modelled urban SO2 back-
ground concentrations (annual mean, µg/m3); reference year 1995. The
2:1 overprediction and 1:2 underprediction lines are shown.
annual means suggests that for a large number but not
all of cities the 1991-meteorology results in the highest
annual mean; for the 98 percentile values this is less
outspoken. The interannual variability is 6% for annual
means and 10% for the 98 percentiles.
The fraction of the population living in non-attainment
cities is relatively insensitive to the meteorological con-
ditions: for the hourly objective the fraction varies from
19.7 to 22.8%; for the daily objective the range is 24.2–
26%. The total exposure shows larger variations: PET
varies from 3.0 to 5.9 1011 persons µg/m3 and from 0.91
to 1.1 1011 persons µg/m3 for the hourly objective and
daily objective, respectively. Judging the population
fraction and PET-values a worst-case situation cannot be
defined. The 1991-meteorology results in the lowest
PET-values but in the largest population fraction living
in cities where the daily objective is exceeded.
3.6. Benzene
The comparison with observations is strongly ham-
pered both by the scarcity of data and by the fact that
measurements are frequently made at stations in a traffic
environment. In view of these complications, Fig. 9
shows a reasonable agreement with measurements (88%
fall within a factor of 2).
About 50% of the urban population in the modelled
cities is exposed to city background levels exceeding the
threshold value of 5 µg/m3. High concentrations are cal-
culated for Paris. This is largely due to an extremely
407
high regional background value which is calculated from
total emissions, including those from Paris. Incidental
conditions relating to the grid cell orientation in the
background model cause this double counting of the
urban emissions to lead to overestimating the concen-
trations in Paris (De Leeuw et al., 2000).
4. Results for base case 2010
The base case scenario consists, broadly speaking, of
a projection over the period 1990–2020 of emissions of
all sources at 5-year intervals. It incorporates the
expected effects of the Auto-Oil-I and other adopted or
anticipated legislation. For the road transport component
detailed information collected at national levels has been
used. The base case is essentially a ‘business-as-usual’
scenario. This does not mean that it is a prediction of
what will actually happen but rather of what could hap-
pen if no further interventions are made beyond what is
now agreed or firmly in the pipeline. It does not neces-
sarily follow, however, that the base case gives a worst
case view of trends in emissions since it is possible that
economic activities increase faster or that implemen-
tation of legislation and other measures is less successful
than originally hoped.
Air quality calculations are only made for the year
2010. The urban emissions for 2010 are obtained by sca-
ling the 1995 data according to the emission trend at a
national level. Note that in the current procedure changes
in urban population, built-up area and in local conditions
(e.g. major investments in public transport, changes in
traffic congestion) are not accounted for in the emission
scenarios. If this type of information becomes available
for individual cities, emissions can be updated by introd-
ucing appropriate correction factors for these cities.
Regional background concentrations have been updated
for the year 2010. An overview of the emission changes
is presented in Table 2 and Fig. 10.
4.1. Ozone
Fig. 11 shows the cumulative distribution of the
maximum and urban area average number of days with
exceedance of the running 8 h-average of 120 µg/m3
ozone over the population given again separately for
Northern European cities and Southern European cities.
About 15% of the urban population (two cities) in South-
ern Europe is exposed to urban ozone levels exceeding
the 120 µg/m3 level on ⬎20 days, whereas the number
of days with exposure in all cities in Northern Europe
stays below 20. For the whole EU15, 6% of the urban
population lives in non-attainment cities. With regard to
the number of days with exceedance averaged over the
urban area, the impact of the emission reductions
appears to be significant for all European cities under
408 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414

Fig. 8. Variability of SO2 annual mean and 98 percentile values (daily values) on meteorological conditions. Using constant 1995 emissions and
meteorological input for five different years the ratio of annual mean (top) and 98 percentile (bottom) over the multi-year averaged value is plotted.

consideration: no exceedance of the target value of 20
days is modelled.
The calculations show that for the majority of the cit-
ies the days of exceedances as well as the maximum
ozone concentrations are reduced considerably. Not sur-
prisingly, the emission reductions appear to lead to
higher summer average ozone concentrations in most cit-
ies; under condition with lower NOx emissions there will
be a shift in the photostationary equilibrium in favour
of ozone.
4.2. NO2
Future urban annual averages of NO2 were calculated
by the cQ-model for cities with sufficient reported moni-
toring data. Nitrogen dioxide is related to emissions of
nitrogen oxides directly using an empirical conversion
to nitrogen oxides (DGXI, 1996). The cQ-model is based
on annual averages. Using statistical relations the
measurements indicate that the objective for hourly con-
centrations corresponds to an annual mean of 37 µg/m3.
Both for the hourly and the annual average objective
attainment in all cities except two, representing 5% of
the population has been estimated.
Urban concentrations for 2010 have been calculated
by UAQAM for all selected cities. There is a strong
reduction in urban NO2 concentration as is seen in Fig.
12. Here all 192 cities are sorted according to their NO2
concentrations calculated for 2010. The reduction in
concentration since 1995 is on the average 13 µg/m3.
Differences in emission reductions, in particular traffic
emissions, introduce differences in concentration
reduction. From the 97 cities that had in 1995 a back-
ground concentration in excess of 40 µg/m3, in 2010 still
30 remain. For about 19% of the urban population the
mean value is in excess of the objective (in 1995: 65%).
Here a discrepancy with the cQ-model is found and
further work is needed. A better agreement between the
two models is obtained for exceedances of the hourly
objective. According to UAQAM, exceedances of an
 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414 409

short-term averaged concentrations are equivalent. In

2010 the annual averaged concentrations are calculated
to be about 20% lower than current levels. The environ-
mental objectives are estimated to be exceeded for 62%
of the population covered by the database.
According to UAQAM-calculations and including a
sea salt contribution about half of the urban population
is in 2010 exposed to an annual mean concentration
above the objective. If a correction for sea salt is not
applied, the fraction of exposed population drops to
39%. Breaching of the objective for daily values is
widely found: 71% of the population. There is a reason-
able agreement with the results of the cQ-model.

4.4. CO

Under the assumption that background concentrations

Fig. 9. Comparison between observed and modelled urban benzene
background concentrations (annual mean, µg/m3): reference year 1995.
The 2:1 overprediction and 1:2 underprediction lines are shown.
hourly NO2 concentration of 200 µg/m3 is expected in
one or two cities. The expected number of exceedances
is, however, ⬍18; the objective will therefore be attain-
ed.
4.3. PM10
Future urban annual averages of PM10 were calculated
by the cQ-model for cities with sufficient reported moni-
toring data. Statistical relations were used to relate the
objective for a daily mean to an annual mean. The obser-
vations indicate that the objectives for long-term and
are reduced by 20%, the UAQAM-calculations indicate
that still some minor exceedances of the 10 mg/m3 level
are to be expected. Depending on the meteorological
conditions, the number of cities with exceedances ranges
from 2 (0.5% of the total modelled population) to 6
(1.5% of the population). Using the 2010 emission con-
ditions and meteorological conditions for the period
1990–1995, the annual PET-values varies between
1.1×106 and 2.3×107 person mg/m3. With respect to 1995
the exposure to CO concentrations in excess to the limit
value will in 2010 be reduced by a factor of 10 or more.
Again, exceedances are only modelled for Southern Eur-
opean cities.
4.5. SO2
Future urban annual averages of SO2 were calculated
by the cQ-model for cities with sufficient data. The
annual average concentration is projected to be reduced
in all cities between 1995 and 2010. Based on regression

Table 2
Emissions of NOx VOC, PM10 and benzene (in ktonne/year) per SNAP1 category for 1995 and 2010 in EU15

Snap sector NOx NOx VOC VOC PM10 PM10 Benzene Benzene
1995 2010 1995 2010 1995 2010 1995 2010

1. Public power, cogeneration and district 2396 1312 244 220 437 366 9 8
heating
2. Commercial,institutional and residential 584 620 654 649 436 301 24 24
combustion
3. Industrial combustion 795 748 54 64 373 252 2 2
4. Production processes 692 645 824 968 454 415 18 21
5. Extraction and distribution of fossil fuels 0 0 910 588 0 0 2 1
6. Solvent use 0 0 3861 2973 0 0 0 0
7. Road transport 5131 1631 4364 797 354 160 163 23
8. Other mobile sources and machinery 1695 1468 710 584 24 23 24 19
9. Waste treatment and disposal 0 0 85 88 77 6 5 5
10. Agriculture 315 237 101 101 23 25 3 3
Total EU15 11608 6661 11807 7033 2179 1548 249 108
410 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414

Fig. 10. Relative changes in emissions in each of the member states (indicated by their ISO 3166-1:1977 Alpha-2 code) in 2010 compared to
the situation in 1995.

Fig. 11. Cumulative distribution of days with exceedance of the running 8 h-average of 120 µg/m3 ozone (base case 2010) over the population
in the modelled cities in Northern Europe (䊏) and cities in Southern Europe (왖); left, maximum in the urban area; right, average over the urban area.

analysis of measured data, the objective for hourly

Fig. 12. Calculated NO2 concentrations (annual mean, µg/m3) for
1995 (upper line) and 2010 (lower line) for all 192 cities. Cities are
ranked according to their concentrations in 2010.
values corresponds to annual average of 38 µg/m3; this
level is exceeded only in one city representing 2% of the
population. The environmental objective for daily values
corresponds to 26 µg/m3 and is exceeded for 7% of
population. Industrial hot-spots are not taken account of
in the calculations, so it cannot be ruled out that there
are areas of higher SO2 concentrations within the cities
of the database.
There is a strong reduction in urban SO2 concen-
trations calculated by UAQAM. The objective for the
daily averages is exceeded in nine cities (11% of the
population). For a meteorological more favourable year,
1990-meteorology, 9% of the population is exposed. A
frequent violation during ⬎100 days is modelled for the
Greek cities. There is, however, a questionable attri-
bution of emissions from marine bunker fuel to the urban
areas. The exceedance numbers are most likely in error.
 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414 411

The objective for hourly concentrations is exceeded
in four cities (6% of the population); again the two
Greek cities show a high number of exceedances. In a
sensitivity calculation the emissions in both Greek cities
for ‘other transport’ (snap sector 8, including emissions
from bunker fuel) were constraint to the 1995 values. In
this case, both objectives are still exceeded but the sever-
ity of the exceedances is strongly reduced. Exceedances
of this objective strongly depend on the meteorological
conditions: using meteorological data of 1990 results in
exceedance in two cities (3% of the population).
There is a fair agreement between the forecasts of the
cQ-model and UAQAM: 7% vs 9–11% of the population
for the daily objective and 2% vs 3–6% of the population
for the hourly objective.
4.6. Benzene
The scenario calculations for 2010 indicate that in four
cities (13% of population in the selected cities) the urban
background concentrations will still be in excess of the
threshold value. The number of people exposed and also
the exposure level are strongly reduced. In 1995 50%
of the population was exposed; the population weighted
averaged concentration in cities in excess of the thres-
hold is 10.5 µg/m3; for 2010 a value of 6.3 µg/m3 is cal-
culated.
5. Discussion
An overview of the results obtained by the various
models for the reference year 1995 and 2010 is presented
in Table 3.
The results for the reference year were compared to
measured air quality data. The comparison of modelled
and measured air quality is hampered by a lack of
reliable data. To improve the coverage of the monitoring
data, data over the period 1992–1996 has been used. The
air quality database AIRBASE has proved to be valuable
tool but needs further input of air concentration dam
from member states. Although the monitoring stations
have been selected on their representativity for the urban
background, it cannot be excluded that measurements are
frequently made at stations in a traffic environment,
which is higher than the modelled averaged urban back-
ground concentrations. Additional uncertainties originate
in the downscaling of urban emissions from national tot-
als. Despite all uncertainties the modelled air quality is
in acceptable agreement with the observations; generally
⬎90% of the calculated concentrations is well within a
factor of 2 from the observations. In the case of SO2 this
ratio is about 65%.
A summary of the fraction of the urban population
potentially exposed to concentrations above their quality
objectives is given in Table 3. About 4% of the urban
population is living in 15 smaller cities where air quality
is already in compliance with the 2010 objectives. Non-
attainment of objectives for PM10, NO2, ozone and ben-
zene is modelled for more than half of the urban popu-
lation. The models indicate that breaching of objectives
more frequently occurs in southern European cities. The
agreement between model results and measurements is
reasonable; a large discrepancy is found, however, for
the exposure to the objective of hourly NO2 concen-
trations (Table 3). Partly this can be explained by the
differences in population size; NO2 monitoring data is
only available for 60% of the population included in the
model calculations.
The baseline projected emissions for 2010 result in a

Table 3
Fraction (in %) of total urban population living in non-attainment cities: environmental objectives are defined in Table 1; in all calculations 1995
meteorological data is used except when indicated

Pollutant Averaging period 1995a obs (%) 1995b UAQAM/OFIS (%) 2010c cQ (%) 2010b UAQAM/OFIS (%)

SO2 1h 14 20-23d 2 3-6d

SO2 24 h 38 24-26d 7 9-11d
NO2 1h 77 5 5 0
NO2 Calendar year 73 64 5 19
PM10 24 h 97 89 62 71
PM10 Calendar year 97 87e 62 52e
CO 8h - 11 - 0.5-1.5
O3 Daily 8-h max - 48 - 6
Benzene Calendar year - 50 - 12
Pb Calendar year - 23f - 0

a
Fraction estimated from monitoring data. Note that monitoring data is not available for the full set of 192 conurbations.
b
Fraction estimated from UAQAM and OFIS (ozone only) model calculations.
c
Fraction estimated from cQ-model calculations.
d
Range refers to the modelled range when using meteorological conditions for five individual years.
e
UAQAM model estimates for PM10 include sea salt contribution.
f
Reference year for Pb model calculations is 1990.
412 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414
large improvement in urban air quality, however,
exceedances of several environmental objectives are still
to be expected. The objectives for PM10, both short-term
and long-term, as well as the NO2 long-term objectives
are projected to be most frequently exceeded in 2010.
None or only minor exceedances are to be expected for
lead and CO. Similar to the 1995-calculations the models
indicate that in southern European cities the objectives
are generally less well attained than in the more northern
cities. Partly this will be caused by better dispersion con-
ditions in the northern cities. The number of exceedances
may fluctuate from year-to-year depending on the
meteorological conditions; sensitivity calculations per-
formed for CO and SO2 (see Table 3) demonstrate a
range of a few percent.
Fig. 13 shows that the population is simultaneously
exposed to above-threshold concentrations of different
pollutants. For instance, in 1995 about 25 million inhabi-
tants lived in cities where the objectives for four pol-
lutants are exceeded simultaneously; for ⬎40 million
inhabitants concentrations of 4 or more pollutants are
above the objectives. In 2010 this number is reduced
to ⬍4.5 million according to the projections made in
this study.
Urban air quality not only improves in terms of popu-
lation exposed but also in terms of severity of the
exposure. The changes in population exposure expressed
PET values, are presented for various scenario and sensi-
tivity calculations in Fig. 14. To facilitate the compari-
son between pollutants and averaging periods, the PET-
values calculated for the reference year 1995 have been
scaled to unity and the other PET are expressed as ratio
to the reference situation. When comparing the results
for 1995 and the base case situation in 2010 a reduction
of 45–100% is seen depending on pollutant and AQ
objective. Whereas Table 3 indicates that for PM10 there
is a limited reduction in the total population living in
non-attainment areas between 1995 and 2010, Fig. 14
shows, however, that for the population at risk there is
a large reduction in exposure.
Fig. 13. Urban population simultaneously exposed to concentrations
of the pollutants CO, NO2. SO2, PM10, benzene or Pb in excess of the
air quality objectives. Zero implies that for none of the six pollutant
air quality objectives are exceeded.
Fig. 14 also shows the results for a 2010 scenario
under the assumption that emission reduction measures
as defined in the Auto-Oil I Directives had not been
implemented (indicated as ‘no AOPI’). In this scenario,
emissions for all other categories were taken equal to
the 2010 base case scenario. Comparison of this scenario
with the base case situation for 2010 shows the benefits
of the Auto-Oil I Directives. Large effects are seen for
NO2, CO and benzene. In the 2010 base case scenario
urban air quality is in compliance with the 1 h objective
for NO2 but for the ‘no AOP I’-scenario exceedances
are calculated again. For CO and benzene PET-values
are expected to be twice as high when the Auto Oil I
measures should not have been implemented. For SO2
and PM10 the impact of the AOP-I directives is much
smaller; for these pollutants other source categories
and/or the long-range contributions are more important.
The impact of traffic emissions is illustrated in two
sensitivity calculations in which all emissions for road
transport have been set to zero (indicated as ‘no traffic’).
In 1995 the impact of traffic SO2 emissions on PET is
about 10–15%; in 2010 this is reduced to ⬍5% in agree-
ment with the decrease in relative contribution of traffic
emissions to the total SO2 emission. For PM10 the rela-
tive impact of traffic emissions is more or less constant
over time; excluding traffic emissions reduces PET with
35% (for the annual objective) and 10–15% (for the daily
objective). For NO2, CO and benzene the relative share
of traffic emissions becomes smaller in 2010 compared
to 1995 but traffic is still the dominant contributor to
PET-values. Note that even in this hypothetical no-traffic
situation there are still exceedances of the air quality
objectives. Reaching all objectives will only be possible
when further abatement measures for other source categ-
ories are introduced.
6. Conclusions
In 1995 a large fraction of the urban population is
exposed to concentrations of one or more pollutants
which are in excess of the air quality objectives set for
the year 2010. The situation will be much improved in
2010 under the AOPII base case scenario but a till com-
pliance with all objectives is not expected. Full com-
pliance can only be realised by means of additional
reductions for source categories other than road trans-
port. However, especially for NO2, CO, and benzene
there are clear benefits from the Auto-Oil I directives.
The GEA-approach is a simple modelling method-
ology by which the urban air quality in a large number
of cities can be evaluated for a range of pollutants.
The top–down approach to estimate urban emissions
from national totals may not accurately reflect the actual
situation but comparison with otherwise obtained emis-
sion data looks promising. Further work on estimation
of urban emissions is needed.
 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414
Fig. 14. Relative changes in population exposure (expressed as PET) for various emission scenarios. The situation in the reference year 1995 is
set to unity. Air quality objectives are given in Table 1.
Despite all uncertainties the modelled air quality is in
acceptable agreement with the observations. The com-
parison of modelled and measured air quality is ham-
pered by a lack of (reliable) data. The air quality datab-
ase AIRBASE has proved to be valuable tool but needs
further input of air quality data from member states.
Acknowledgements
This study is developed from the contribution of the
European Topic Centre on Air Quality to the Working
Group I of the Auto Oil II programme under contract to
the European Environment Agency. Valuable contri-
butions have been given by Robert Enesund, Tinus
Pulles, Antoon Visschedijk and various other colleagues
from the European Topic Centres on Air Emissions and
on Land Cover.
References
Buckingham, C., Forrest, G., Sadler, L., Shah, S., Wickham, L., 1997a.
Atmospheric Emissions Inventory for Greater Manchester. London
Research Centre, London, UK.
Buckingham, C., Clewley, L., Hutchinson, D., Sadler, L., Shah, S.,
Banham, S., Forest, G., Spence, E., Wickham, L., 1997b. Atmos-
pheric Emissions Inventories for Four Urban Areas: Merseyside,
Bristol, Southampton & Portsmouth and Swansea & Port Talbot.
London Research Centre, London, UK.
Buckingham, C., Clewley, L., Hutchinson, D., Sadler, L., Shah, S.,
1997c. London Atmospheric Emissions Inventory. London
Research Centre, London, UK.
Buckingham, C., Clewley, L., Hutchinson, D., Sadler, L., Shah, S.,
Forest, G., Jenkins, N., Spence, E., 1998. Atmospheric Emissions
Inventories. Glasgow, Middlesbrough and West Yorkshire. London
Research Centre, London, UK.
BUWAL, 1997. NO2 Immissione in der Schweiz 1990–2010. Schriftre-
413
ihe Umwelt nr 289. Bundesamt für Umwelt, Wald und Landschaft,
Bern, Switzerland.
CEC, 1998. Economic Evaluation of Air Quality Targets for Sulphur
Dioxide, Nitrogen Dioxide, Fine and Suspended Particulate Matter
and Lead. Office for Official Publications of the European Com-
munities, Luxembourg.
De Leeuw, F.A.A.M., Moussiopoulos, N., Sahm, P., Bartonova, A.,
2000. Air Quality in Larger Conurbations in the European Union.
European Topic Centre on Air Quality, European Centre on Air
Emissions (in preparation).
DGXI, 1996. Air Quality Report of the Auto Oil Programme. EC
DGXI, Brussels.
EEA, 1995. Europe’s Environment. The Dobřiš Assessment. European
Environment Agency, Copenhagen.
EEA, 1996a. CORINAIR 1990 Summary Report 1. Topic Report 7-
1996 Air Emissions. European Environment Agency, Copenhagen.
EEA, 1996b. Review Study on European Urban Emission Inventories.
Topic Report 30-1996, Air Emissions. European Environment
Agency, Copenhagen.
EEA, 1997. Air Pollution in Europe 1997. EEA Environmental Mono-
graph No. 4. European Environment Agency, Copenhagen.
EEA, 1998. Europe’s Environment. The Second Assessment. European
Environment Agency, Copenhagen.
EEA, 1999. European Topic Centre on Air Quality. European Environ-
ment Agency, Copenhagen (AIRBASE, available from internet
http://www.etcaq.rivm.nl).
EEA, 2000. Criteria for Selecting Air Pollution Models in ETC/AQ
Applications. Topic Report Air Quality. European Environment
Agency, Copenhagen (in preparation).
Eerens, H.C., van Jaarsveld, J.A., Peters, J., 1998. European status —
air quality. Trends, monitoring, background modelling. In: Schne-
ider, T. (Ed.), Air Pollution in the 21st Century; Priority Issues and
Policy, Studies in Environmental Science 72. Elsevier, Amsterdam.
ETC/LC, 1997. Workshop on land cover applications — needs and
use. 12–13 May 1997, Copenhagen.
European Commission, 1999a. Council Directive 1999/30/EC of 22
April 1999 relating to limit values for sulphur dioxide, nitrogen
dioxide and oxides of nitrogen, particulate matter and lead in ambi-
ent air. Official Journal of the European Communities OJL163, 41.
European Commission, 1999b. Proposal for a Council Directive relat-
ing to limit values for benzene and carbon monoxide in ambient
air. COM/1998/591/FINAL. Official Journal of the European Com-
munities OJC53, 8.
414 F.A.A.M. de Leeuw et al. / Environmental Modelling & Software 16 (2001) 399–414
European Commission, 1999c. Proposal for a Directive of the Euro-
pean Parliament and of the Council Relating to Ozone in Ambient
Air. COM/1999/125/FINAL. European Commission, Brussels.
Fontelle, J.P., Chang, J.P., Audoux, N., Levy, C., Riviere, E., 1997.
Inventaires d’emissions dans l’atmosphere dans le cadre des plan
regionaux pour la qualite de l’air. CITEPA, Paris.
Hutchinson, D., Clewley, L., 1996. West Midlands Atmospheric Emis-
sions Inventory. London Research Centre, London, UK.
van Jaarsveld, J.A., 1995. Modelling the long-term atmospheric behav-
iour of pollutants on various spatial scales. Thesis, University of
Utrecht, the Netherlands.
Krzyzanowski, M., 1997. Methods for assessing the extent of exposure
and effects of air pollution. Occupational and Environmental Medi-
cine 54, 145–151.
Larssen, S., Hagen, L.O., 1996. Air Quality in Europe, 1993 — A Pilot
Report. Topic Report 25-1996, Air Quality. European Environment
Agency, Copenhagen.
Moussiopoulos, N., 1995. The EUMAC zooming model, a tool for
local-to-regional air quality studies. Meteor. Atmos. Phys. 57,
115–133.
Moussiopoulos, N., Sahm, P., Tourlou, P.M., Friedrich, R., Simpson,
D., Lutz, M., 2000. Assessing ozone abatement strategies in terms
of their effectiveness on the regional and urban scales. Atmospheric
Environment 34, 4691–4699.
Olsthoorn, X., Amann, M., Bartonova, A., Clench-Aas, J., Cofala, J.,
Dorland, K., Guerreiro, C., Henriksen, J.F., Jansen, H., Larssen, S.,
1999. Cost benefit analysis of European Air Quality targets for
sulphur dioxide, nitrogen dioxide and fine and suspended particu-
late matter in cities. Environmental and Resource Economics 14,
333–351.
van Pul, W.A.J., van Zantvoort, E.D.G., de Leeuw, F.A.A.M., Sluyter,
R.J.C.F., 1996. Urban Air Quality Assessment Model UAQAM.
Report 722401010. RIVM, Bilthoven, the Netherlands.
Sahm, P., Moussiopoulos, N., 1999. The OFIS model: a new approach
in urban scale photochemical modelling, EUROTRAC Newsletter
21/99, EUROTRAC-ISS, Garmisch-Partenkirchen, 22–28.
SENCO, 1999. AOPII Base case emission
scenarios.http://www.btintemet.com/-SENCO.
Simpson, D., 1993. Photochemical model calculations over Europe for
two extended summer periods: 1985 and 1989. Atmospheric
Environment 27A, 921–943.
Simpson, D., 1995. Biogenic emissions in Europe 2: implications for
ozone control strategies. Journal Geophysical Research 100,
22891–22906.
Simpson, D., Olendrzynski, K., Semb, A., Støren, E., Unger, S., 1997.
Photochemical Oxidant Modelling in Europe: Multi Annual Model-
ling and Source–Receptor Relationships. EMEPNSC-W report
3/97. DNMI, Oslo, Norway.
Tsyro, S., 1998. Description of the Lagrangian Acid Deposition Model.
In: Transboundary Acidifying Air Pollution in Europe. Part 2:
Numerical Addendum. EMEP MSC-W Status Report 1998.
UN, 1997. United Nations Statistical Division Demographic Yearbook
1995. No. E/F.97.XIII.1. United Nations.
UN-ECE, 2000. EMEP/CORINAIR Atmospheric Emission Inventory
Guidebook, 2nd ed. http://themes.eea.eu.int/theme.php/state/air.
UNEP, 1993. United Nations Environmental Programme Environmen-
tal Data Report, 1993–94. Blackwell, Oxford.
WHO, 1995. Concern for Europe’s Tomorrow: Health and the
Environment in the WHO European Region. WHO Regional Office
for Europe/Wiss.Verl.-Ges, Stuttgart.