ARTICLE IN PRESS
AE International – Europe
Atmospheric Environment 38 (2004) 1891–1901
Intercontinental transport and the origins of the ozone
observed at surface sites in Europe
R.G. Derwenta,*, D.S. Stevensonc, W.J. Collinsb, C.E. Johnsonb
a
rdscientific, Newbury, Berkshire RG14 6LH, UK
Climate Research, Meteorological Office, Exeter, UK
c
Department of Meteorology, University of Edinburgh, Edinburgh, UK
b
Received 8 September 2003; received in revised form 3 January 2004; accepted 7 January 2004
Abstract
A global three-dimensional Lagrangian chemistry-transport model is used to describe the formation, transport and
destruction of tropospheric ozone using 1990s global emissions and 1998 meteorological archives. Using a labelling
technique, the geographical origins of the ozone formed within the troposphere have been revealed, showing whether
the ozone found at the surface in Europe has had its origins above the continents of North America, Europe or Asia or
elsewhere in the world. In this way, contributions to the ozone found at 21 surface monitoring sites across Europe can
be attributed to production over North America and Asia, demonstrating that intercontinental ozone transport is
an efficient process. Sensitivity tests to the global man-made sources of NOx and carbon monoxide indicate that
global ozone precursor emission controls may contribute towards reaching regional air quality policy goals for ozone
in Europe.
r 2004 Elsevier Ltd. All rights reserved.
Keywords: Intercontinental transport; Ozone; Seasonal cycles; Emission controls
1. Introduction
Over the past few decades there have been conflicting
explanations of the origins of surface ozone over Europe
and of its spring-time maximum (Monks, 2000).
Initially, stratosphere–troposphere exchange, an unmistakably natural process, had been proposed to explain
the spring-time maximum in observed ozone mixing
ratios at a mountain-top site at Arosa, Switzerland
(Junge, 1963). However, more recently man-made
processes are thought to have caused a doubling in
surface ozone mixing ratios in Europe since preindustrial times (Volz and Kley, 1988) based on
observations at the Montsouris Observatory near Paris,
France from 1861 to 1871. European Community
emissions of NOx and VOCs have declined by 25–30%
*Corresponding author.
E-mail address: [email protected]
(R.G. Derwent).
between 1990 and 2000 (UNECE, 2002) leading to a
decline in episodic peak ozone mixing ratios (Derwent
et al., 2003) whilst mean ozone mixing ratios continue
to rise at remote locations such as Mace Head, Ireland
(Simmonds et al., 1997) and Strath Vaich, Scotland
(Fowler et al., 2001). These apparent contradictions
can be resolved if there are a number of sources of
ozone in Europe and if a number of different mechanisms are operating to cause the observed seasonal cycle
in ozone.
In this study, we use a global chemistry-transport
model to study the likely origins of the ozone found at
the surface across Europe. We use a labelling technique
to elucidate the geographical origins (stratosphere,
North America, Asia, Europe) of surface ozone and
we identify intercontinental transport from North
America and Asia as an important source of ozone
within Europe. Finally, we show how global reductions
in man-made emissions of NOx and carbon monoxide
could reduce surface ozone levels over Europe.
1352-2310/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.atmosenv.2004.01.008
ARTICLE IN PRESS
R.G. Derwent et al. / Atmospheric Environment 38 (2004) 1891–1901
1892
2. The global 3-D chemistry-transport model
STOCHEM
The model used in this study is the UK Meteorological Office global Lagrangian tropospheric 3-D chemistry transport model (STOCHEM). In the version
employed here, the troposphere has been divided into
50,000 air parcels which are advected with a 3 h time
step using archived winds from the Meteorological
Office Unified Model. These winds are 6-hourly
instantaneous values, taken from operational meteorological analyses at 0.83 1.25 resolution for 1997
and 1998. The model calculations began at 1 July 1997
and run through to 31 December 1998. Results
were taken from the last full year of the calculations
with a 6-month period for initialisation. This model
version is not the most recent (Collins et al., 2003)
but is an earlier offline version which is faster and is
run from operational meteorological analyses, full
details of which are given elsewhere (Collins et al.,
1997, 1999).
Each parcel holds the concentrations of 73 chemical
species including NOx, O3, OH, HO2, PAN, methane,
ethane, ethylene, propane, propylene, butane, toluene,
o-xylene, isoprene, formaldehyde, acetaldehyde, acetone, methyl glyoxal and methyl vinyl ketone. The
chemical species react according to 174 photolytic and
thermal reactions, with a 5-min chemistry time step.
Surface uptake, dry deposition and wet removal are
unchanged from Collins et al. (1997). Emissions into
the model are implemented as an additional term in the
production flux for each species during each integration
time-step. The emissions compiled to represent the
situation of the 1990s are listed in Table 1. Biomass
burning, vegetation, soil, oceans and ‘other’ are all
surface sources based on two-dimensional (latitude,
longitude) distributions as in Collins et al. (1997). The
aircraft and lightning NOx sources are three-dimensional. Methane sources from paddies, tundra, wetlands
and animal sources (see Table 1) are held constant
throughout the year at their yearly average values.
Natural biogenic sources from soils, vegetation and the
oceans, vary by calendar month. No seasonal nor
diurnal dependences are assumed for any of the
industrial emissions. The 6-hourly vertical wind fields
were used to calculate a stratosphere–troposphere
exchange ozone flux across 100 mb surface on a
5 5 grid using monthly ozone fields from Li and
Shine (1995). The total annual injection of ozone
through the 100 mb surface amounted to 1216 Tg yr1,
with a NOy flux of one thousandth of the ozone flux by
mass (as N), based on the study of Murphy and Fahey
(1994). Inter-parcel mixing was treated by relaxing the
mixing ratios in each parcel towards the average
Table 1
Model emissions for the 1990s in Tg yr1 from the major source categoriesa,b
Species
Industrial
Biomass burning
NOx
SO2
H2
CO
CH4
C2H4
C2H6
C3H6
C3H8
C4H10
C5H8
C8H10
C7H8
H2CO
CH3CHO
CH3OH
Acetonee
DMS
NH3
31.7
71.2
20.0
650.0
155.0
12.3
7.0
19.9
13.1
91.4
7.1
1.4
20.0
700.0
40.0
7.7
3.6
10.9
1.1
1.7
17.6
18.7
1.2
2.5
7.6
4.5
1.2
7.2
0.6
3.2
4.3
0.4
39.3
3.5
Vegetation
Soil
Oceans
5.6
5.0
150.0
Aircraftc
Lightning
0.5
5.0
5.0
50.0
260.0
10.0
3.5
10.0
5.0
3.0
249.3
1.9
1.0
1.6
1.3
18.8
2.4
15.0
8.2
Emissions are in Tg yr1 except for NOx which are in Tg N yr1.
Emission totals and their spatial distributions were based on entries in the EDGAR database (Olivier et al., 1996).
c
Aircraft NOx emissions for 1992 were taken from IPCC (1999).
d
Includes paddies, tundra, wetlands, termites and animals.
e
Acetone source from vegetation represents production from terpene oxidation.
a
b
Otherd
ARTICLE IN PRESS
R.G. Derwent et al. / Atmospheric Environment 38 (2004) 1891–1901
background mixing ratio as described by Collins
et al. (1997).
We have participated in a number of global 3-D
chemistry-transport model intercomparisons to establish
the level of STOCHEM model performance. These
include intercomparisons of fast photochemistry (Olson
et al., 1997), tracer transport (Stevenson et al., 1998),
ozone (Kanakidou et al., 1998), carbon monoxide
(Kanikadou et al., 1999) and of past, present and future
tropospheric ozone distributions (Gauss et al., 2003). In
all of these intercomparisons, STOCHEM performs
within the range of current models. However, in
common with all other complex models, it is sometimes
difficult to reconcile good model performance with one
species in one location with poor model performance
elsewhere or with other species. In previous work,
Collins et al. (1997) have described comparisons of
model results with observations for global mean OH,
surface ozone for the Atlantic Ocean, for remote
background sites, with ozonesondes and NOx measurements throughout the troposphere. From these comparisons it was concluded that the model was well able to
provide a good approximation to the main controlling
processes for OH, NOx and ozone in both the winter
and the summer and that the model ozone distribution
contains all the most important features of the observed
distribution. In addition, Collins et al. (1999) presented
comparisons between the model results and observations
for hydrogen peroxide, formaldehyde, methyl hydroperoxide and acetone, all potentially important free radical
sources in the troposphere. Furthermore, the methane
lifetime due to OH removal was found to be 9.0 years in
these model experiments, in agreement with Prinn et al.
(1995) who reported a lifetime of 8.970.6 years. The
global ozone budgets, burdens and turnover times
calculated here are well within the ranges in the
literature as reviewed by Collins et al. (2000).
The mixing ratios of each of 73 model species, i, were
calculated from:
dxi =dt ¼ Pi Li xi þ Mi :
ð1Þ
The production rate Pi for each species was calculated
at each time step from the local, instantaneous rates of
emission and chemical production from the set of 174
reactions. The loss rate Li xi for each species was
calculated from the local, instantaneous destruction rate
due to chemical reaction, dry and wet deposition and
aerosol scavenging. Mi represents the net production
from interparcel mixing. In addition to ozone, the air
B
C
parcels carried other forms of ozone, OA
3 , O3 , O3 , y, in
which ozone is labelled by the geographical region in
which it was formed. Our approach adapts and extends
that pioneered by Roelofs and Lelieveld (1997) for
ozone of stratospheric origins. Each of the additional
forms of ozone has the same loss coefficient, Li ; hence
the sum of all of the mixing ratios of the different forms
1893
of ozone, xA
O3 y, in each air parcel adds up to the
mixing ratio of ozone, henceforward, ‘total ozone’, xO3 ;
to distinguish it from the labelled forms.
The ‘total ozone’ production rate is given by:
dxO3 =dt ¼ ESTE þ k1 ½HO2 ½NO þ k2 ½RO2 ½NO
k3 ½O1 D½H2 O
other O3 loss terms other NO2 loss terms;
ð2Þ
where ESTE is the local, instantaneous ozone production
rate from stratosphere–troposphere exchange, k1 represents the rate coefficient for the HO2+NO=OH+NO2
reaction, k2 for RO2+NO=NO2+RO, k3 for O1D+
H2O=OH+OH, the other loss terms for ozone involve
loss through its reactions with OH, HO2, NO2 and its
ozonolysis reactions with ethylene, propylene and
isoprene and the other NO2 loss terms include its
reactions with OH, peroxy acyl radicals, NO3 radicals
and ozone. The two terms k1[HO2][NO] and
k2[RO2][NO] were then used to drive the rate of ozone
production of the ozone tracer labels, depending where
the air parcel was located with respect to the Northern
Hemisphere continental land masses, North America
(defined arbitrarily as 140 W–20 W), Asia (60oE–
140 W) and Europe (20 W–60 E), or elsewhere in the
world southwards of 23 N. Since the selected regions fill
the entire model domain, both horizontally and vertically up to the model top at 100 mb, then the sum of the
tracer mixing ratios satisfies Eq. (1) and the mixing
ratios of the ozone tracers could be determined along
with the other 73 model species. Tracer conservation is
assured in the formulation adopted for the Lagrangian
transport scheme in STOCHEM.
3. Intercontinental sources of ozone in Europe
3.1. Comparison of model with observed seasonal cycles
in Europe
Because this study focuses on the origins of ozone
observed at the surface in Europe, attention is first given
to how well the STOCHEM model describes the
seasonal cycle of ozone across Europe. Here, the model
results for ozone of all origins are used, the results for
the so-called ‘total ozone’. A record is kept of the air
parcels which are in the vicinity (within 1 ) of particular
locations and their ozone mixing ratios are noted at each
time step. For the chosen locations, model points were
found for 87% of the days in 1998. The model mixing
ratios are plotted out as individual points in Fig. 1 and
as monthly ranges, shown as monthly mean71 standard
deviation, as broken lines. Fig. 1 also presents the
observed monthly mean maximum daily ozone mixing
ratios in ppb (1 part in 109 parts by volume) during 1998
ARTICLE IN PRESS
1894
R.G. Derwent et al. / Atmospheric Environment 38 (2004) 1891–1901
Fig. 1. Observed mean monthly daily maximum ozone mixing ratios in ppb (solid lines), model calculated ozone mixing ratios (dots)
and monthly model ranges as mean71 standard deviation (broken lines) for 1998 at selected EMEP ozone monitoring network
locations in Europe, country locations are given in Table 3.
ARTICLE IN PRESS
R.G. Derwent et al. / Atmospheric Environment 38 (2004) 1891–1901
at each of 10 selected remote, rural site locations in
the EMEP ozone monitoring network (Hjellbrekke,
2000), as solid stepped lines. The observed monthly
means of the maximum daily (i.e. mid-afternoon) ozone
mixing ratios were plotted instead of the monthly
mean values because of the marked diurnal cycles
found at these continental rural sites due to surface
uptake of ozone under shallow nocturnal stable layers
which are not simulated in the global model. During
the mid-afternoon, surface observations reflect ozone
levels in a deeper atmospheric layer and provide a
better comparison with model results that are for
the lowest model layer at 950 mb. Details of the
latitudes, longitudes, siting criteria and measurement
techniques for the EMEP network are given in
Hjellbrekke (2000) and the country locations are given
in Table 3.
The model points scatter around the stepped solid
lines at most locations during much of the year and
the observed monthly mean daily maximum ozone
values (solid lines) fall within the ranges (broken lines)
of the model results. Exceptions were found during some
winter months at central European locations, where the
model results overestimate the observations. This is
caused by the coarse scale of the NOx emission
inventories used, leading to an underestimation of
wintertime ozone depletion. At the remote Scandinavian
locations, the model tends to overestimate the observations during some summer months because of its coarse
resolution which means that too many regional pollution episodes have reached these remote locations. The
issues of too coarse model resolution have been
minimised by using ozone observations at regionally
representative rural locations during the mid-afternoon
when boundary layer depths are at their maximum,
minimising the effects of local NOx emissions and
surface uptake on ozone levels. No attempt has been
made to assess the impact on model ozone of individual
power station or urban plumes.
Seasonal cycles are clearly evident in both observations and model results in Fig. 1. At most sites,
summertime ozone peaks dominate the seasonal patterns. However at some of the more remote sites,
secondary maxima are apparent during the springtime in
both observations and model. Such maxima are evident
at Mace Head, Tortosa, Kollumerwaard, Birkenes and
Aspreveten. At Waldhof, Montilibretti, and Iskrba, the
observations reveal a summertime maximum that is also
reproduced faithfully by the model. On this basis, it is
concluded that the model exhibits some skill in
reproducing qualitatively the observed seasonal cycles
at a range of selected site locations in Europe. A clear
geographical pattern has emerged from both the
observations and the model calculations, with summertime maxima dominating in the Mediterranean region
and central Europe and with some evidence of
1895
secondary springtime maxima in Scandinavia, and in
the Atlantic Ocean fringes of north and west Europe.
In Table 2, the seasonal cycles in the observations and
in the model calculations are compared in a more
rigorous manner for 19 locations across Europe. The
biases, fractional biases, root mean square errors
(RMSE) and the fraction of the monthly model results
within 720% of the observations (i.e. about 77 ppb)
were calculated as described in the footnotes to Table 2.
Taking the locations with springtime maxima, mean
biases fell in the range 2.06 to 3.34 ppb, 5% to 8.5%,
with an average of 1.3 ppb. They showed a wide range in
performance as indicated by the fraction of monthly
means within 720%, from 0.11 to 0.92 and the RMS
errors from 5.2 to 13.8 ppb. For the locations with
summertime maxima, mean biases fell in the range 3.5
to 2.2 ppb, 7.5% to 5.9%, with an average of
1.1 ppb. They showed a much narrower range in
performance compared to the locations with spring-time
maxima, as indicated by the fraction of monthly
means within 720% of 0.42–0.67, and RMSE errors
of 7.1–10.8 ppb.
Taking all 19 locations together, it appears that the
model is capable of reproducing observed monthly mean
daily maximum surface ozone mixing ratios with an
accuracy of 78.7 ppb, based on RMSE errors, during
1998 and with about two-thirds of individual monthly
mean values within 720% of the observations. This is a
workable level of model performance that gives some
confidence in the model-derived continental attributions
which follow.
3.2. Intercontinental attribution of the origins of surface
ozone in Europe
By using the tracer-labelling technique, the model is
able to give an indication of the geographical origins
where the ozone was produced in each air parcel arriving
at each location. By averaging overall the air parcels
arriving during 1998, a picture has been built up of the
geographical origins of the ozone calculated for selected
locations and this is presented in Table 3. Broadly
speaking, adding up the columns across Table 3 gives
the annual mean ‘total’ model ozone mixing ratio and
these fall in the range 30–45 ppb. As the previous section
shows, these ‘total’ values were directly comparable with
the surface observations. The columns in Table 3,
present estimates of the ozone mixing ratios that could
be attributed to the particular continental source
regions.
The annual mean ozone mixing ratios across Europe
due to stratosphere–troposphere exchange fell in the
range from 5.7 to 10.6 ppb, with Iskrba (Slovenia) at
the lowest end of the range and Mace Head (Ireland) at
the highest. The mean value and seasonal cycle found
here for Mace Head for ozone of stratospheric origins is
ARTICLE IN PRESS
R.G. Derwent et al. / Atmospheric Environment 38 (2004) 1891–1901
1896
Table 2
Summary of the comparison of observed monthly mean daily maximum ozone mixing ratios and model calculated monthly mean
mixing ratios at 19 selected locations across Europe in 1998
Locationa
Model mean
(ppb)b
Observed mean
(ppb)c
Mean bias
(ppb)d
Mean fractional
biase
Root mean
square error
(ppb)f
Fraction within
720%g
Mace head
Illmitz
Taenikon
Kosetice
Waldhof
Frederiksborg
Laheema
Tortosa
Virolahti
Revin
Harwell
Montelibretti
Preila
Kollumerwaard
Birkenes
Jarcsew
Aspreveten
Iskrba
Starina
40.7
45.6
41.4
43.2
39.2
37.7
33.8
44.3
33.3
39.3
37.8
49.0
39.2
36.3
38.6
42.8
36.7
44.5
43.1
42.8
44.3
44.6
44.1
37.0
34.6
34.2
41.0
38.5
37.9
37.1
51.4
38.7
34.4
38.1
40.9
36.6
48.0
36.3
2.06
1.31
3.20
0.88
2.23
3.06
0.43
3.34
5.24
1.43
0.77
2.39
0.46
1.90
0.48
1.83
0.09
3.48
6.77
0.049
0.029
0.074
0.020
0.059
0.085
0.013
0.078
0.146
0.037
0.020
0.047
0.012
0.054
0.012
0.044
0.003
0.075
0.170
5.2
9.9
9.0
8.3
7.1
7.7
7.0
7.6
12.0
7.9
4.3
10.7
9.1
7.2
8.9
13.8
8.9
10.4
10.8
0.92
0.42
0.58
0.67
0.67
0.33
0.33
0.67
0.58
0.67
0.92
0.58
0.33
0.58
0.58
0.25
0.42
0.67
0.11
a
Country locations are given in Table 3.
Model mean is the average of the 12 monthly mean model ozone values.
c
Observed mean is the average of the 12 monthly mean daily maximum values.
d
Mean bias is the average of the 12 (model-observed) values.
e
Mean fractional bias is the average of the 12 (model-observed)/observed values.
f
Root mean square error is the square root of the sum of the 12 (model-observed)2 values divided by 12.
g
Fraction within 720%, is the number of model monthly values that lie within the range of the observed value 20% and the
observed value+20%.
b
in excellent agreement with that already reported by
Collins et al. (2003). Mace Head (Ireland) being the
westerly most location is most exposed to the weather
systems travelling across the North Atlantic Ocean.
These weather systems are thought to account for much
of the stratosphere–troposphere exchange found in
northerly midlatitudes (Stohl, 2001). Iskrba (Slovenia),
on the other hand, is one of the easterly most locations
in Table 3 and is thus farthest removed from the
influence of the weather systems travelling across the
North Atlantic Ocean. Averaged over the 21 locations in
1998, stratosphere–troposphere exchange contributed
about 7.3 ppb (18%) to surface ozone mixing ratios over
Europe.
Table 3 indicates that the bulk of the ozone calculated
in air parcels arriving at the 21 selected locations carried
the ‘European’ label. European contributions covered
the range from 9.5 to 31.5 ppb, with Mace Head
(Ireland) at the lowest end of the range and Montelibretti (Italy) at the highest. Mace Head (Ireland) is the
location in Table 4 with the least exposure to European
regionally polluted air masses, receiving European air on
only 32% of occasions during the 1990–1994 period
(Simmonds et al., 1997). Montelibretti (Italy), on the
other hand, is surrounded by centres of European
population and industry in all directions and so is
greatly influenced by European ozone production.
Averaged over the 21 locations in 1998, European ozone
production contributed about 20 ppb to surface ozonemixing ratios over Europe and provided the largest
continental contribution of about 50%.
The North American region contributed surface
ozone mixing ratios over Europe in the range
5.8–12.2 ppb. Virolahti (Finland) and Laheema (Estonia), both Baltic Sea locations, received the least North
American contribution and Mace Head (Ireland) the
highest. As we have remarked above, Mace Head is the
location most exposed to the weather systems moving
across the North Atlantic Ocean and these systems
appear to provide a mechanism for the intercontinental
transport of ozone as well as stratosphere–troposphere
exchange. The behaviour seen for stratosphere–troposphere exchange and North American intercontinental
transport appear somewhat different pointing perhaps
ARTICLE IN PRESS
R.G. Derwent et al. / Atmospheric Environment 38 (2004) 1891–1901
1897
Table 3
Contributions to the annual mean daily maximum ozone mixing ratios in ppb from the different source areas calculated for 21
locations across Europe
Location
Stratosphere–troposphere
Europe
North America
Asia
Contributions in ppb
Mace Head, Ireland
Illmitz, Austria
Taenikon, Switzerland
Kosetice, Czech Republic
Waldhof, Germany
Frederiksborg, Denmark
Laheema, Estonia
Tortosa, Spain
Virolahti, Finland
Revin, France
Harwell, United Kingdom
K-puszta, Hungary
Montelibretti, Italy
Preila, Lithuania
Kollumerwaard, Netherlands
Birkenes, Norway
Jarcsew, Poland
Monte Vehlo, Portugal
Aspreveten, Sweden
Iskrba, Slovenia
Starina, Slovakia
Average
10.6
6.9
7.0
7.5
7.7
7.8
6.0
6.9
5.8
8.0
8.8
6.1
6.2
7.3
8.7
8.2
6.3
7.5
7.5
5.7
6.2
7.3
9.5
27.1
21.6
23.0
18.3
16.8
16.9
21.8
15.9
16.9
12.9
27.8
31.5
19.7
12.6
15.2
26.6
16.6
16.5
28.5
25.2
20.0
12.2
7.7
8.4
8.3
8.5
8.1
5.9
9.2
5.8
9.5
10.8
7.0
6.9
7.1
9.5
8.8
6.9
9.7
7.4
6.7
7.2
8.2
6.6
3.9
4.4
4.4
4.7
5.0
4.2
4.0
4.2
4.9
5.3
3.7
3.5
4.7
5.5
5.6
3.8
4.2
4.9
3.6
3.8
4.5
to different geographical patterns of influence from
these two mechanisms across Europe as shown in Fig. 2.
Averaged over the 21 locations in 1998, North American
continental ozone production contributed about 8.2 ppb
(20%) to surface ozone mixing ratios over Europe.
The Asian region contributed surface ozone mixing
ratios in the range 3.5–6.6 ppb. Montelibretti (Italy) and
Iskrba (Slovenia) received the least Asian contribution
and Mace Head (Ireland) the highest. The spatial
pattern of the contributions from Asian sources show
similarities with that from stratosphere–troposphere
exchange (see Fig. 2), but differences from North
American sources. Averaged over the 21 locations in
1998, Asian continental ozone formation contributed
about 4.5 ppb (11%) to surface ozone mixing ratios over
Europe.
The mixing ratio of ozone with the stratospheric label
reached a maximum of 11.2 ppb during March 1998,
with a secondary maximum of 10.6 ppb during that
December. The European labelled ozone reached its
maximum of over 30 ppb during May–August 1998,
with its peak of 38 ppb during July. North American
ozone reached a peak of 12.7 ppb during August 1998
and exceeded 10 ppb for the period July–September.
Asian ozone reached its maximum of over 4 ppb during
the period August–December, with a peak of 6.2 ppb
during November. On this basis, intercontinental trans-
port from North America and Asia combined, was a
significantly more important source of surface ozone in
Europe compared with the stratosphere in all months,
except December and February. Broadly speaking, the
stratospheric contribution peaked in the spring, the local
European production peaked during the middle of
summer and the intercontinental transport contributions
from North America and Asia, peaked during the late
summer and autumn, respectively.
3.3. Sensitivity of the intercontinental attribution of
ozone in Europe to global emission source strengths
In this section, the sensitivities of the model surface
ozone mixing ratios in Europe to the global man-made
emissions of NOx and carbon monoxide are examined.
They were studied by reducing by one half the global
man-made source strengths in two separate model
experiments. The 50% reductions were applied acrossthe-board, with no geographical or seasonal dependence. Table 4 presents the impact of a 50% reduction in
NOx emissions on the ozone mixing ratios attributed to
each continental source region at each of the 21 selected
locations. No entries are provided for stratosphere–
troposphere exchange because the ozone mixing ratios
from this source were unaffected by the reductions
emissions. The reductions in annual mean ozone mixing
ARTICLE IN PRESS
R.G. Derwent et al. / Atmospheric Environment 38 (2004) 1891–1901
1898
Table 4
The impact on the contribution to the annual mean ozone mixing ratios from each source area following a 50% reduction in global
man-made NOx and CO emissions
Location
50% percent reduction in man-made NOx emissions
50% percent reduction in man-made CO emissions
Europe
North America
Asia
Europe
North America
Asia
Changes in contributions in ppb
Mace Head
1.2
Illmitz
4.2
Taenikon
3.1
Kosetice
3.3
Waldhof
1.7
Frederiksborg
2.1
Laheema
2.6
Tortosa
4.0
Virolahti
2.4
Revin
2.0
Harwell
1.0
K-puszta
4.3
Montelibretti
5.4
Preila
2.6
Kollumerwaard 0.9
Birkenes
2.3
Jarcsew
3.6
Monte Vehlo
3.0
Aspreveten
2.3
Iskrba
4.0
Starina
3.8
1.9
1.2
1.3
1.3
1.3
1.2
0.8
1.5
0.9
1.4
1.7
1.1
1.1
1.1
1.5
1.3
1.0
1.5
1.1
1.0
1.1
0.9
0.5
0.6
0.6
0.6
0.7
0.5
0.6
0.5
0.7
0.7
0.5
0.5
0.6
0.7
0.7
0.5
0.6
0.7
0.5
0.5
0.3
0.7
0.6
0.7
0.6
0.5
0.4
0.5
0.4
0.5
0.4
0.7
0.8
0.5
0.4
0.4
0.7
0.4
0.4
0.7
0.6
0.3
0.2
0.2
0.2
0.2
0.2
0.1
0.2
0.1
0.2
0.3
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
0.1
Average
1.2
0.6
0.5
0.2
0.1
2.8
ratios due to the 50% reduction in global NOx emissions
were largest for the European source region, 0.9–
4.3 ppb, least for the Asian source region, 0.5–0.7 ppb,
with North America 0.8–1.9 ppb. On a percentage
change basis, the reductions in surface ozone attributed
to the European source region were broadly similar.
The reductions in surface ozone attributed to the
European source region, when averaged over the 21
locations, exceeded 5 ppb during the summer months of
May to August and were close to zero during the winter
months of December to March. In contrast, the
reductions due to the North American source region
showed much reduced seasonal variation, lying between
1.0 and 1.5 ppb, with a summer minimum, an autumn
maximum and a subsidiary maximum during spring.
The reductions due to the Asian source region followed
an exactly similar seasonal pattern as those due to the
North American region but covered a significantly wider
range from 0.3 to 0.8 ppb.
On a percentage change basis, when averaged overall
21 locations, the reduction in surface ozone due to a
50% reduction in man-made NOx emissions attributed
to the North American source region, varied from 10%
to 25% depending on the time-of-year, and from 6% to
20% for the Asian source region, showing their
minimum values during summer and their maximum
values during winter. The seasonal variations in these
percentage changes are characteristically different from
that shown in the base case attributions to each source
region shown in Table 3. Ozone attributed to the
European source region showed little or no sensitivity to
the reduction of man-made NOx emissions in wintertime
and showed its greatest sensitivity of about 15% during
summertime.
The impact of an across-the-board 50% reduction in
global man-made carbon monoxide emissions on the
surface ozone attributed to each continental source
region are presented in Table 4. The reductions in ozone
mixing ratios were again largest for the ozone attributed
to the European source region, 0.4–0.8 ppb, least for the
Asian source region, 0.1–0.2 ppb, with North American
ozone reduced by 0.1–0.3 ppb. On a percentage basis, a
50% reduction in carbon monoxide emissions reduced
surface ozone from Europe by 2.3–3.2%, from North
America by 1.7–3.0% and from Asia by 1.8–3.0%.
These responses to a 50% reduction in CO emissions
appear to be largely independent of ozone production
region, on a percentage basis. However, they are close to
ARTICLE IN PRESS
R.G. Derwent et al. / Atmospheric Environment 38 (2004) 1891–1901
1899
Fig. 2. Spatial patterns of the annual average contributions in ppb to surface ozone from internal production within Europe,
intercontinental transport from Asia and North America and from stratosphere–troposphere exchange.
the levels of noise in our 3-D chemistry-transport model
and so it is difficult to be sure of their significance for air
quality policy.
It is instructive to compare the results from the
present study for a 50% reduction in global man-made
NOx emissions with those of Wild and Akimoto (2001)
for a 10% increase in surface fossil fuel NOx, CO and
VOC emissions. Scaling the results from Wild and
Akimoto (2001) gives a 0.5 ppb decrease in surface
ozone found in Europe but formed over Asia, compared
with 0.5–0.7 reported here, 1.0 ppb decrease in ozone
formed over North America, compared with 0.8–1.9 ppb
reported here and a 2.0 ppb decrease in ozone formed
over Europe, compared with 0.9–4.3 ppb in this
study. The estimates from Wild and Akimoto (2001),
though for NOx, CO and VOC emission changes, are
similar to those reported in this study and provide
confirmation of our tracer modelling technique, bearing
in mind that our calculations are for a larger NOx
reduction and probably take the model into a region of
greater DO3/DNOx slope compared with Wild and
Akimoto (2001).
4. Discussion and conclusions
Evidence for the intercontinental transport of ozone
and particles into Europe from North America and Asia
based on ground-level, aircraft and satellite observations
has been reviewed by Volz-Thomas (2003). The impact
of man-made emissions from North America and Asia
on surface ozone in Europe has been investigated using
the global 3-D GEOS-CHEM model by performing
sensitivity studies in which regional emissions were
turned off (Li et al., 2002). According to these sensitivity
studies, surface ozone in Europe decreased by 2–4 ppb
when man-made emissions in North America were shut
off and by slightly less for Asian emissions.
Shutting off North American man-made emissions in
the GEOS-CHEM model (Li et al., 2002) decreased
mean surface ozone mixing ratios in Europe by 2–4 ppb.
The STOCHEM ozone tracer labelling technique
indicated a somewhat larger contribution of 8.2 ppb
from North America in Table 3. However, using the
emission sensitivity approach in Table 4, a much smaller
North American contribution of 1.2 ppb was identified
ARTICLE IN PRESS
1900
R.G. Derwent et al. / Atmospheric Environment 38 (2004) 1891–1901
by using a 50% reduction in man-made NOx emissions.
Tracer labelling and emission sensitivity techniques
appear to give different estimates of the magnitudes
of the intercontinental transport of ozone because of
important non-linearities in the fast photochemistry of
ozone and its precursors. It may well be that the tracer
labelling technique appears to give higher estimates and
a more accurate quantification of intercontinental
transport whereas the sensitivity approach provides a
somewhat lower estimate but a more policy-relevant
quantification by describing responses to emission
controls.
In European air quality policy terms, the focus of
concern is not so much with annual mean ozone
concentrations (see Tables 3 and 4), as with ozone levels
during the spring-time growing season for the protection
of crops and vegetation as quantified by the accumulated exposure over a threshold of 40 ppb, the AOT40
concept (Fuhrer et al., 1997). An empirical relationship
has been constructed between mean monthly daily
maximum ozone mixing ratios and the AOT40 values
within the EMEP database for 1998 (Hjellbrekke, 2000)
for the 19 selected ozone monitoring sites in Table 2,
as follows:
AOT40 ðin ppb hÞ ¼ 804 ammdm 29; 000;
where ammdm is the average monthly mean daily
maximum ozone mixing ratio in ppb between April
and July. Using this relationship, it has been possible to
estimate the changes in AOT40 exposure levels across
Europe resulting from a 50% reduction in man-made
NOx emissions. Reductions in North American NOx
emissions reduced AOT40 exposure levels by 620–
1480 ppb h, 3.6–36%, with an average of 970 ppb h,
13%, across Europe. Reductions in Asian NOx emissions reduced average AOT40 exposure levels by
410 ppb h, 5% and in European NOx emissions by
4300 ppb h, 43%. Emission sensitivity studies on the
hemispheric scale show that NOx emission controls in
North America and Asia may produce significant
benefits within Europe in terms of reducing future
ozone exposures and protecting crops and natural
vegetation.
On this basis, we conclude that the ozone tracer
labelling technique can provide an account of the
influence of intercontinental transport on surface ozone
levels within Europe. On a year-round basis, the
stratosphere appears to account for about 7 ppb,
whereas intercontinental transport from North America,
8 ppb, and from Asia, 5 ppb. These intercontinental
sources influence the different parts of Europe to
different extents both spatially and seasonally and these
variations help to explain the different seasonal cycles
observed across Europe. Furthermore, because NOx
emissions from North America, Asia and Europe have
exhibited different trends over the 1970s–1990s (Naja
et al., 2003), it may be possible to explain why there have
been different observed trends in surface ozone in
different parts of Europe during the 1970s–1990s.
Background ozone levels may well be rising along the
Atlantic fringes of Europe, influenced strongly by the
steadily increasing North American NOx emissions and
the rapidly rising Asian emissions. Ozone levels in
central Europe may well remain roughly level because
the increasing influence of North American and Asian
emissions is counterbalanced by the influence of
decreasing NOx emissions from central Europe itself.
What could happen to ozone levels in Europe in the
future could depend on future the NOx emissions from
the different continents and this will be the subject of
further studies with the ozone labelling technique.
Acknowledgements
This work was sponsored by the Air and Environmental Quality Division of the Department for Environment, Food and Rural Affairs through contract no.
EPG 1/3/164, by the GMR programme of The Met
Office and by a NERC/Environment Agency Advanced
Fellowship (P4-F02) to one of us (DSS). Our thanks go
to Anne-Gunn Hjellbrekke of the Norwegian Institute
for Air Research for providing the observations of the
monthly mean daily maximum ozone mixing ratios for
the EMEP surface ozone monitoring sites.
References
Collins, W.J., Stevenson, D.S., Johnson, C.E., Derwent, R.G.,
1997. Tropospheric ozone in a global-scale three-dimensional Lagrangian model and its response to NOx emissions
controls. Journal of Atmospheric Chemistry 26, 223–274.
Collins, W.J., Stevenson, D.S., Johnson, C.E., Derwent, R.G.,
1999. The role of convection in determining the budget of
odd hydrogen in the upper troposphere. Journal of
Geophysical Research 104, 26927–26941.
Collins, W.J., Derwent, R.G., Johnson, C.E., Stevenson, D.S.,
2000. The impact of human activities on the photochemical
production and destruction of tropospheric ozone. Quarterly Journal of the Royal Meteorological Society 126,
1925–1951.
Collins, W.J., Derwent, R.G., Garnier, B., Johnson, C.E.,
Sanderson, M.G., Stevenson, D.S., 2003. Effect of stratosphere–troposphere exchange on the future tropospheric
ozone trend. Journal of Geophysical Research 108, 8528
doi:10.1029/2002JD002617.
Derwent, R.G., Jenkin, M.E., Saunders, S.M., Pilling, M.J.,
Simmonds, P.G., Passant, N.R., Dollard, G.J., Dumitrean,
P., Kent, A., 2003. Photochemical ozone formation in
northwest Europe and its control. Atmospheric Environment 37, 1983–1991.
Fowler, D., Coyle, M., ApSimon, H.M., Ashmore, M.R.,
Bareham, S.A., Battarbee, R.W., Derwent, R.G., Erisman,
ARTICLE IN PRESS
R.G. Derwent et al. / Atmospheric Environment 38 (2004) 1891–1901
J.-W., Goodwin, J., Grennfelt, P., Hornung, M., Irwin, J.,
Jenkins, A., Metcalfe, S.E., Ormerod, S.J., Reynolds, B.,
Woodin, S., 2001. Transboundary air pollution. Acidification, eutrophication and ground-level ozone in the UK.
National Expert Group on Transboundary Air Pollution,
Department for the Environment, Food and Rural Affairs,
London.
Fuhrer, J., Skarby, L., Ashmore, M., 1997. Critical levels of
ozone effects in Europe. Environmental Pollution 97, 91–106.
Gauss, M., Myhre, G., Pitari, G., Prather, M.J., Isaksen, I.S.A.,
Berntsen, T.K., Brasseur, G.P., Dentener, F.J., Derwent,
R.G., Hauglustaine, D.A., Horowitz, L.W., Jacob, D.J.,
Johnson, M., Law, K.S., Mickley, L.J., Muller, J.-F.,
Plantevin, P.-H., Pyle, J.A., Rogers, H.L., Stevenson,
D.S., Sundet, J.K., van Weele, M., Wild, O., 2003.
Radiative forcing in the 21st century due to ozone
changes in the troposphere and lower stratosphere.
Journal of Geophysical Research 108, 4292 (doi:10.1029/
2002JD002624).
Hjellbrekke, A.-G., 2000. Ozone measurements, 1998. EMEP/
CCC-Report 5/2000, Norwegian Institute for Air Research,
Kjeller, Norway.
IPCC, 1999. Aviation and the Global Atmosphere. Cambridge
University Press, Cambridge.
Junge, C.E., 1963. Air Chemistry and Radioactivity. Academic
Press, New York.
Kanakidou, M., Dentener, F.J., Brasseur, G.P., Collins, W.J.,
Berntsen, T.K., Hauglustaine, D.A., Houweling, S., Isaksen,
I.S.A., Krol, M., Law, K.S., Lawrence, M.G., Muller, J.F.,
Plantevin, P.H., Poisson, N., Roelofs, G.J., Wang, Y.,Wauben, W.M.F., 1998. 3-D global simulations of tropospheric
chemistry with focus on ozone distributions, EUR 18842
Report. European Commission, Office for Official Publications of the European Communities, Luxembourg.
Kanikadou, M., Dentener, F.J., Brasseur, G.P., Berntsen, T.K.,
Collins, W.J., Hauglustaine, D.A., Houweling, S., Isaksen,
I.S.A., Krol, M., Lawrence, M.G., Muller, J.F., Poisson, N.,
Roelofs, G.J., Wang, Y., Wauben, W.M.F., 1999. 3-D
global simulations of tropospheric CO distributions—
results of the GIM/IGAC intercomparison 1997 exercise.
Chemosphere Global Change Science 1, 263–282.
Li, D., Shine, K.P., 1995. A 4-D ozone climatology for
UGAMP models. UGAMP Internal Report, University of
Reading, UK.
Li, Q., Jacob, D.J., Bey, I., Palmer, P.I., Duncan, B.N., Field,
B., Martin, R.V., Fiore, A., Yantosca, R.M., Parrish, D.D.,
Simmonds, P.G., Oltmans, S.J., 2002. Transatlantic transport of pollution and its effects on surface ozone in Europe
and North America. Journal of Geophysical Research 107,
4166 (doi:10.129/2001JD001422).
Monks, P.S., 2000. A review of the observations and origins of
the spring ozone maximum. Atmospheric Environment 34,
3545–3561.
1901
Murphy, D.M., Fahey, D.W., 1994. An estimate of the
flux of stratospheric reactive nitrogen and ozone into
the troposphere. Journal of Geophysical Research 99,
5325–5332.
Naja, M., Akimoto, H., Staehelin, J., 2003. Ozone in background and photochemically aged air over central Europe:
analysis of long-term ozonesonde data from Hohenpeissenberg and Payerne. Journal of Geophysical Research 108,
4063 (doi:10.1029/2002JD002477).
Olivier, J.G.J., Bouwman, A.F., van der Maas, C.W.M.,
Berdowski, J.J.M., Veldt, C., Bloos, J.P.J., Visschedijk,
A.J.H., Zandveld, P.Y.J., Haverlag, J.L., 1996. Description
of EDGAR version 2.0. RIVM Report no. 771060 002,
Bilthoven, Netherlands.
Olson, J., Prather, M., Berntsen, T., Carmichael, G., Chatfield,
R., Connell, P., Derwent, R., Horowitz, L., Jin, S.,
Kanakidou, M., Kasibhatla, P., Kotamarthi, R., Kuhn,
M., Law, K., Penner, J., Perliski, L., Sillman, S., Stordal, F.,
Thompson, A., Wild, O., 1997. Results from the intergovernental panel on climatic change photochemical model
intercomparison (Photocomp). Journal of Geophysical
Research 102, 5979–5991.
Prinn, R.G., Weiss, R.F., Miller, B.R., Huang, J., Alyea, F.N.,
Cunnold, D.M., Fraser, P.J., Hartley, D.E., Simmonds,
P.G., 1995. Science 269, 187–192.
Roelofs, G.-J., Lelieveld, J., 1997. Model study of the influence
of cross-tropopause ozone transports on tropospheric O3
levels. Tellus B 49, 38–55.
Simmonds, P.G., Seuring, S., Nickless, G., Derwent, R.G.,
1997. Segregation and interpretation of ozone and carbon
monoxide measurements by air mass origin at the TOR
station Mace Head, Ireland from 1987–1995. Journal of
Atmospheric Chemistry 28, 45–59.
Stevenson, D.S., Collins, W.J., Johnson, C.E., Derwent, R.G.,
1998. Intercomparison and evaluation of atmospheric
transport in a Lagrangian model (STOCHEM), and an
Eulerian model (UM), using 222Rn as a short-lived tracer.
Quarterly Journal of the Royal Meteorological Society 124,
2477–2491.
UNECE, 2002. Present state of emission data. United Nations
Economic Commission for Europe, Economic and Social
Council, EB.AIR/GE.1/2002/8, Geneva, Switzerland.
Volz, A., Kley, D., 1988. Evaluation of the Montsouris series of
ozone measurements made in the nineteeth century. Nature
332, 240–242.
Volz-Thomas, A., 2003. Tropospheric ozone and its control. In:
Midgley, P., Reuther, M. (Eds.), Towards Cleaner Air for
Europe—Science, Tools and Applications. Margraf Publishers, Weikersheim, Germany (Chapter 3).
Wild, O., Akimoto, H., 2001. Intercontinental transport
of ozone and its precursors in a three-dimensional
global CTM. Journal of Geophysical Research 106,
27729–27744.