copernicus atmosphere monitoring service
TRANSCRIPT
ECMWF COPERNICUS REPORT
Copernicus Atmosphere Monitoring Service
Annual air quality assessment report 2015
Issued by: INERIS/ Laurence ROUÏL
Date: 22/01/2018
Ref: CAMS71_2016SC2_D71.2.7_201802_2015AAR_v1
This document has been produced in the context of the Copernicus Atmosphere Monitoring Service (CAMS).
The activities leading to these results have been contracted by the European Centre for Medium-Range Weather Forecasts,
operator of CAMS on behalf of the European Union (Delegation Agreement signed on 11/11/2014). All information in this
document is provided "as is" and no guarantee or warranty is given that the information is fit for any particular purpose.
The user thereof uses the information at its sole risk and liability. For the avoidance of all doubts, the European Commission
and the European Centre for Medium-Range Weather Forecasts has no liability in respect of this document, which is merely
representing the authors view.
Copernicus Atmosphere Monitoring Service
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 3 of 42
Annual air quality assessment report -2015
INERIS Laurence ROUÏL
Frédérik MELEUX
Date: 10/02/2018
Ref: CAMS71_2016SC2_D71.2.7_201802_2015AAR_v1
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 4 of 42
Table of Contents
1. Rationale 9
2. Ozone 10
2.1 Annual and seasonal averages 10
2.2 Exposure indicators 14
2.3 Peaks indicators 19
2.4 Conclusions for ozone 20
3. Nitrogen dioxide 21
3.1 Annual and seasonal averages 21
3.2 Conclusions for nitrogen dioxide 24
4. Particulate Matter (PM10) 25
4.1 Annual and seasonal averages 25
4.2 Daily exceedances 28
4.3 Conclusions for PM10 31
5. Particulate Matter (PM2.5) 32
5.1 Annual and seasonal averages 32
5.2 Conclusions for PM2.5 36
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 5 of 42
Table of figures Annual average of ozone concentrations in 2015 11
Annual average of ozone concentrations in 2014 (source : CAMS – 2014 air quality assessment report) 11
Seasonal averages of ozone concentrations in 2015: Spring (a), Summer (b), autumn (c), winter (d) 12
Surface air temperature in 2015 relative to its 1981-2010 average (a) whole year, (b) May, (c) July, (d)
August (source: ECMWF, Copernicus Climate Change Service -C3S temperature re-analyses) 13
AOT40 indicator in 2015 15
AOT40 indicator in 2014 (source : CAMS – 2014 air quality assessment report) 15
SOMO35 indicator in 2015 17
SOMO35 indicator in 2014 (source : CAMS – 2014 air quality assessment report) 17
Number of days when 120 µg/m3 (maximum daily 8-hours average) was exceeded in 2015 18
Number of days when 120 µg/m3 (maximum daily 8-hours average) was exceeded in Spring (left) and
summer (right) 2015 18
Number of days when 120 µg/m3 (maximum daily 8-hours average) was exceeded in 2015 at the
monitoring stations reported to the EEA (source: EEA data viewer) 19
Number of hours when the information ozone threshold (180 µg/m3) was exceeded in 2015 20
Annual average concentrations of NO2 in 2015 22
Annual average concentrations of NO2 in 2014 (source : CAMS – 2014 air quality assessment report) 22
NO2 annual average in 2015 at the monitoring background stations reported to the EEA (source: EEA
data viewer) 23
Seasonal averages of NO2 concentrations in 2015; Spring (a), Summer (b), autumn (c), winter (d) 24
Annual average of PM10 concentrations in 2015 26
Annual average of PM10 concentrations in 2014 (source : CAMS – 2014 air quality assessment report) 26
PM10 annual average in 2015 at the monitoring background stations reported to the EEA (source: EEA
data viewer) 27
Seasonal averages of PM10 concentrations in 2015; Spring (a), Summer (b), autumn (c), winter (d) 27
Number of days when the daily limit PM10 value was exceeded in 2015 29
Number of days when PM10 daily average exceeded the daily limit value in 2015 at the monitoring
background stations reported to the EEA (source: EEA data viewer) 30
Number of days when the daily limit PM10 value was exceeded by season in 2015; spring (a), summer (b),
autumn (c), winter (d) 30
Surface air temperature in December 2015 relative to its 1981-2010 average (source: ECMWF,
Copernicus Climate Change Service -C3S temperature re-analyses) 31
Annual average of PM2.5 concentrations in 2015 33
Annual average of PM2.5 concentrations in 2014 (source: CAMS – 2014 air quality assessment report) 33
PM2.5 annual average in 2015 at the monitoring background stations reported to the EEA (source: EEA
data viewer) 34
Seasonal averages of PM2.5 concentrations in 2015; Spring (a), Summer (b), autumn (c), winter (d) 34
Number of days when daily average exceeded 25 µg/m3 in 2015 35
IRA 2015: Taylor diagram showing the CAMS regional air quality models performances for the simulation
of ozone daily maximum – Interim production (2016) 39
VRA 2015: Taylor diagram showing the CAMS regional air quality models performances for the
simulation of ozone daily maximum – validated production (2016) 39
Ozone summer average simulated by CAMS Ensemble model : interim simulation (left) and validated
simulation (right) 40
IRA 2015: Taylor diagram showing the CAMS regional air quality models performances for the simulation
of PM10 daily average – Interim production (2016) 40
VRA 2015: Taylor diagram showing the CAMS regional air quality models performances for the
simulation of PM10 daily average – validated production (2016) 41
Number of days when daily average exceeded 35 mg/m3 simulated by CAMS Ensemble model: interim re-
analyses (left) and validated re-analyses (right) 41
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 6 of 42
Executive summary This report is the CAMS air quality annual assessment report for the year 2015 delivered by the air
quality re-analysis multi-model system implemented in the CAMS regional services. The air quality
assessment European maps presented and discussed in this report, result from an Ensemble model
built upon seven European regional air quality modelling and data assimilation systems. Such
approaches combine raw simulation outputs from chemistry-transport models (CTM) with validated
observations data issued from monitoring stations from regulatory air quality networks implemented
in the EU Member States to comply with air quality Directives (2008/50/EC).
For main air pollutants - ozone, nitrogen dioxide, and particulate matter (PM10 and PM2.5)- regulatory
and exposure indicators established on yearly and seasonal bases are presented. It is expected this
very comprehensive information on air pollution patterns and levels, can be considered as the “best
estimate” to describe status of air pollution in Europe in 2015. It should be noted that within CAMS
systematic evaluation of the model and data assimilated results against a relevant set of dedicated
observation data (not used for assimilation in the models) is conducted. It showed very satisfactory
performances confirming the relevance of the approach to support analysis of air pollution issues in
Europe and support decision making.
CAMS regional air quality re-analyses are relevant for assessing rural and urban background
concentrations and areas where they exceed limit values and quality objectives set in the Directive
on ambient air pollution and cleaner air in Europe (2008/50/EC). However, it should be noted that
the CAMS air quality regional assessments are not suited for mapping local exceedances (street
canyons, industrial sites). Models resolution (10 km) is too coarse to simulate correctly such
situations. Indicators fields mapped in this report are compared to air pollutants observations
reported to the European Environment Agency (EEA) by the Member States in compliance with the
implementation provision of the air quality Directive 2008/50/EC. The EEA has developed a data
viewer (http://eeadmz1-cws-wp-air.azurewebsites.net/products/data-viewers/statistical-viewer-
public/) which allows to display observed indicators at monitoring stations. This report includes a
number of snapshots of the EEA’s data viewer which show excellent consistency between CAMS re-
analyses and reported air quality measurements. The review of the status of air quality in Europe in
2015 has been published in fall 2017 by the EEA1.
The year 2015 revealed interesting issues in terms of air pollution and was comparable to 2014 in
term of air pollution patterns. Globally temperatures in Europe were high. Several heatwaves
occurred in summer time, and several countries broke temperatures records. Unusual high
temperatures were also recorded in November and December, especially in Western Europe but also
in Finland, Republic of Moldova and Estonia2.
1 Air Quality in Europe- 2017 report, EEA report N° 13/2017; https://www.eea.europa.eu/publications/air-
quality-in-europe-2017 2 WMO statement on the status of global climate in 2015; report WMO-N° 1167
http://www.wmo.int/pages/prog/wcp/wcdmp/documents/wmo_1167_en.pdf
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 7 of 42
Main conclusions and lessons learnt from this analysis are summarized below.
For ozone:
• Annual and seasonal ozone concentrations were high due to exceptionally high temperatures and sunrise in 2015, which was classified by the World Meteorological Organization as one of
the warmest one. Ozone annual average in Europe ranged from 35 to 90 µg/m3. If Southern Europe recorded logically the highest ozone levels (Portugal, Spain, Italy, Slovenia, Croatia,
Montenegro, Albania, Greece, Turkey) exceptional high ozone levels were also observed in
Nordic countries and in Central Europe;
• Spring average concentrations were remarkably high everywhere in Europe which confirms a trend to “early ozone episodes” already observed in 2014;
• Ecosystems and human health exposure indicators show a degradation compared to 2014 with larger areas impacted everywhere in Europe. This can be explained by high temperatures
recorded in 2015
• The number of hourly peaks also increased compared to 2014. The Pô Valley is the most
concerned area but exceedances of the information threshold value (180 µg/m3) were observed in other locations like Spain, Portugal, France and Western and Central Europe.
• No real decreasing trend, regarding ozone background concentrations in Europe can be
pointed out despite precursor emissions strategies implemented in European Union over the
past years. This results from sensitivity of ozone formation processes to meteorological
variability and increasing temperatures and by the influence of hemispheric transport of
ozone which limit efficiency of regional emissions control strategies.
For nitrogen dioxide:
• Background nitrogen dioxide concentrations exceeded in 2015 the annual limit value in a
limited number of places where NOx emissions are usually high. The Pô Valley is one of the
most exposed area, because of the convergence of high emission levels and non-dispersive
meteorological conditions in the valley.
• NO2 concentrations are the highest in the biggest cities, near main roads and in sea areas because of shipping emissions.
• Air concentrations distribution of NO2 in Europe and their seasonal variability remain
relatively stable compared to the previous years, but certainly because of meteorological
conditions, they were higher in 2015 than in 2014 in several countries, especially in Central
and Eastern Europe.
For particulate matter (PM10 and PM2.5):
• PM10 annual average of background concentrations ranged in 2015 from very low level
(5µg/m3) in a large part of Northern Europe to high values (higher than 30 µg/m3 in average)
in North of France, Benelux, Pô Valley and Eastern Europe (Poland, Czech Republic, Serbia in
particular). The annual limit value was exceeded in Eastern Europe (from Poland to the
Balkans) and in the Pô Valley. The conjunction of high emissions, especially in the residential
heating and wood burning sector together with cold and stable meteorological conditions can
partly explain increasing PM concentrations in those areas.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 8 of 42
• Western Europe (France, Benelux, Germany) has been exposed to exceedances of the daily
limit value for PM10, especially during spring episodes in March mainly due to ammonium
nitrate production because of ammonia from agriculture releases facilitated by mild
temperatures.
• Geographical distribution of PM10 concentrations in Europe in 2015 was very consistent with what was observed the previous years. Eastern part of Europe and Pô valley remain still the
most exposed areas.
• Exposure to fine particulate matter in Europe (PM.2.5) remains a sensitive issue. If the current
limit value for annual average set in the Air Quality Directive (25 µg/m3) is exceeded only in few areas in Northern Italy and Central and Eastern Europe, exposure to more stringent
thresholds (20 µg/m3 as the indicative limit value proposed in the air quality Directive to be
implemented in 2020, upon revision by the Commission and 10 µg/m3 according to WHO air quality guidelines) is worrying.
• Pô Valley, Central Europe, but also, in some periods, Western Europe are the most sensitive
areas. Concentrations may exceed 20 µg/m3 in several places in winter and in spring.
• Temporal variability of the concentrations is high. It is driven by meteorological conditions (dispersive conditions in summer reduce significantly PM levels) and by complex physico-
chemical processes that impact emissions of precursors. Sensitive emission sectors are
residential heating, agriculture (ammonia) and road traffic (nitrogen oxides).
.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 9 of 42
1. Rationale
This report is the Copernicus air quality assessment report for the year 2015 for Europe.
It presents and comments best estimates and maps of air quality indicators elaborated running air
quality models (regional chemistry-transport models) and assimilating validated or “official” air
quality observations from regulatory observation networks implemented in Europe. Such simulations
are called re-analyses. Within the CAMS framework, air quality regional re-analyses are built upon a
set of seven air quality models implemented with data assimilation systems to improve air pollution
patterns and levels. Their results are combined in a unique Ensemble model results that are more
robust and more accurate, in general than the individual models results themselves. Regional re-
analyses (or assessment) of air pollutant concentrations and metrics which describe the situation over
the past years are considered as the best estimates that can be achieved to describe background air
pollution in Europe
Thus, so-called CAMS European air quality assessment reports describe, with a yearly frequency, the
state and the evolution of background concentrations of air pollutants in European countries. Special
care is given to pollutants characterized by the influence of long range transport, correctly caught by
European scale modelling systems: ozone, nitrogen dioxide, particulate matter (PM10 and PM2.5).
The Copernicus annual air quality assessment reports aspire to become actual tools for supporting
European policy and decision makers in charge of air quality monitoring and management. For the
targeted pollutants, regulatory and exposure indicators built up from airborne hourly concentrations
are proposed. They are interpreted with respect to the limit and target values set in the Directive on
Ambient Air quality and Cleaner Air for Europe (the AQ 2008/50/EC of 21 May 2008). Moreover, the
Member states have to report to the European Commission air pollution levels and situations when
those threshold values are exceeded and to inform their citizens about which levels they are exposed
to. Situations (or episodes) when nitrogen dioxide and particulate matter concentrations exceed
regulatory limit values (or target values for ozone) must be carefully analyzed, described
(geographical extension, duration, intensity, population exposed...) and action plans to limit their
impact and to avoid their future development must be proposed. Member states usually base their
investigations on observations available from national air quality monitoring networks implemented
to comply with the Directives requirements, and national expertise.
This report provides complementary information with an accurate and reliable description of air
pollution patterns that developed throughout the European region. Maps resulting from modelling
and observation data assimilation processes are synthesized and interpreted for policy users as well
as general public. For each pollutant considered, annual and seasonal indicators (averages, exposure
indicators, number of exceedances of threshold values) are proposed and commented. It should be
noted that the spatial resolution of the proposed maps does not allow a fine description of very local
patterns (inside the city) or hot spots (near busy roads or industrial sites). Only background
concentrations in or outside cities are proposed. They are representative of the influence of both
local and regional sources.
Finally because this is the second year operational CAMS re-analyses modelling chains are run to feed
such a report, comparison between 2015 and 2014 re-analyses is possible and presented for some
indicators. Also technical annex II presents a preliminary analysis of performance indicators for
interim and validated re-analyses related to the year 2015.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 10 of 42
2. Ozone
2.1 Annual and seasonal averages
Figure 1 and Figure 3 present 2015 ozone annual and seasonal averages respectively. These metrics
are not directly used for air quality policy implementation in Europe but help in understanding:
1) the North-South gradient of concentrations
2) the seasonal variability
In 2015, ozone annual average in Europe (Figure 1) ranged from 40 to 90 µg/m3 what is quite high, and those highest values are more widespread in different parts of Europe than the levels observed
the previous year, in 2014 (Figure 2). However geographical distribution of ozone concentrations
fields are the same for both years. Logically, Southern Europe recorded the highest ozone levels
(South of Portugal, Spain, South of France Italy, Slovenia, Croatia, Montenegro, Albania, Greece,
Turkey…), but also Switzerland, Austria and some parts of the Czech Republic. As in 2014, the
expected gradient South/North that generally describes ozone distribution in Europe was disturbed
in 2015 by remarkable high levels of ozone (around 55-60 µg/m3 annual average) in Scandinavia.
The seasonal variability of ozone concentrations is consistent with what was expected: low levels in
fall and winter and high levels in summer and spring. However, spring ozone concentrations were
remarkably high in spring in Scandinavia (what drives the annual average in this region) and Eastern
part of Europe. Spring ozone concentrations reached 70-80 µg/m3 in Norway, Sweden, Finland, Baltic
countries, Russian Federation, Ukraine, while they stayed between 60 and 65 µg/m3 in summer in those countries.
The Mediterranean area is the most impacted by ozone with average concentrations exceeding 90
µg/m3 as an average in spring and summer. According to the World Meteorological Organization
(WMO 3 ), several countries in Europe, and especially Southern and Eastern Europe experienced
monthly average recorded in temperatures: Spain, Austria, Poland, Portugal, Slovenia and France
were particularly concerned.
3 WMO statement on the status of global climate in 2015; report WMO-N° 1167
http://www.wmo.int/pages/prog/wcp/wcdmp/documents/wmo_1167_en.pdf
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 11 of 42
Annual average of ozone concentrations in 2015
Annual average of ozone concentrations in 2014
(source : CAMS – 2014 air quality assessment report)
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 12 of 42
Seasonal averages of ozone concentrations in 2015: Spring (a), Summer (b),
autumn (c), winter (d)
(a) (b)
(c) (d)
Figure 4 series are issued from the C3S Copernicus Climate services and shows the global surface
temperature anomaly in 2015 compared to the 1980-2010 average for the full year and three months.
Red colors over Europe shows the whole continent was exposed to temperatures higher by 2 to 3 °C
than the average recorded over the last 3 decades. Exceptional high temperatures were recorded in
May in Far-Eastern Europe and in Spain and Portugal, with anomalies exceeding 5°C. The same
occurred in July when only Scandinavian countries recorded temperatures lower than the last 30
years average. But in August records of temperatures were observed everywhere in Europe,
especially in Central Europe with anomalies temperature exceeding 5°C. Relatively warm conditions
even replaced cooler July temperatures in Nordic countries.
The ozone production cycle in the atmosphere is driven by photochemical reactions that develop
under sunny and warm meteorological conditions. With higher and higher temperatures, ozone levels
increase in several parts of Europe.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 13 of 42
Surface air temperature in 2015 relative to its 1981-2010 average
(a) whole year, (b) May, (c) July, (d) August
(source: ECMWF, Copernicus Climate Change Service -C3S temperature re-analyses)
(a)
(b)
(c)
(d)
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 14 of 42
2.2 Exposure indicators
Three regulatory exposure indicators are presented in this section:
• AOT 40 (Accumulated dose over a threshold of 40 ppb) used to assess ozone impact on vegetation.
• Number of days when maximum daily eight hours mean exceeds 120 µg/m3 used as an
indicator for protection of human health
• SOMO35 (Sum of ozone mean over a threshold of 35 ppb) used to assess long term human exposure to ozone concentrations
The two first indicators are defined in the Air Quality Directive 2008/50/EC, while the third one is
recommended by the World Health Organization (WHO) for quantification of health impacts of ozone.
AOT 40 (Accumulated dose over a threshold of 40 ppb) requires for its calculation hourly ozone data.
It is the sum of the differences between the hourly ozone concentration (in ppb) and 40 ppb, for each
hour when the concentration exceeds 40 ppb, accumulated during daylight hours (8:00-20:00 UTC).
AOT40 has a dimension of (µg/m3)·hours. In the 2008 Air Quality Directive, the target value for the
protection of vegetation is defined as the AOT40 calculated from May to July and is set at 18.000
(µg/m3)·hours, with a long term objective of 6.000 (µg/m3)·hours.
Figure 5 presents AOT4O indicator calculated for the period from May to July 2015, and the same
indicator calculated for the year 2014 by CAMS is given on Figure 6 for comparison. Both maps
illustrate very well the fact that no improvement trends regarding impact of ozone in Europe can be
foreseen. It confirms several other studies on ozone. Total area where the target value was exceeded
in Europe in 2015 is significantly larger than in 2014. All Mediterranean countries are concerned
(Spain, South of France, Italy, Slovenia, Croatia, Greece), but also locally in South of Germany and
Central Europe (East of Austria, Hungary, Czech Republic, Slovakia).
More worrying is compliance with the long term objective of 6000 (µg/m3).hours. This objective is
exceeded in a large part of Europe, except in Nordic countries, in the United Kingdom, along the
Atlantic sea side and in Romania and far-East of Europe.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 15 of 42
AOT40 indicator in 2015
AOT40 indicator in 2014
(source : CAMS – 2014 air quality assessment report)
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 16 of 42
SOMO35 is the sum of the differences between maximum daily 8-hour running mean concentrations
greater than 35 ppb and 35 ppb. SOMO35 has a dimension of (µg/m3)·days. This indicator is calculated
for the whole year to be representative of long term exposure of human health to ozone background
concentrations. This metric has been defined and recommended by WHO4 and was used as an impact
constraint in impact assessment modelling used to define national emission ceilings set in the NEC
Directive (2001/81/EC) which has been revised in December 2016. There is no target value or long
term objective for this indicator. Its annual variation is monitored to assess its improvement or
degradation with implementation of emissions control strategies.
Figure 7 represents the SOMO35 calculated for the year 2015, and the same indicator in 2014 is
presented for comparison on Figure 8. South/North gradient appears very clearly for this health
indicator, showing Mediterranean countries are much more exposed that other countries. Same
conclusions as for AOT40 hold logically. Human exposure to ozone concentrations in 2015 was more
important than in 2014, especially in Southern countries. And, almost all countries in Western,
Southern and Central Europe had SOMO35 values higher than 4 000 (µg/m3).days in 2015. Only
Atlantic and North Sea sides were spared with the United Kingdom, Scandinavia and the far-East part
of Europe where photochemical air pollution is more limited.
In the 2008 Air Quality Directive, health protection target value refers to the number of days when
the maximum daily 8-hours ozone average exceeds 120 µµµµg/m3. This number should not be higher
than 25 as an average over 3 years. For information,this indicator is displayed for one year, the year
2015 on Figure 9. The target value was exceeded in a large area covering a large part of Italy, Spain,
South-East of France, Switzerland, Germany, Poland, Austria, Czech Republic, Hungary and Slovakia.
Only the UK, the Netherlands, the Baltic countries Denmark, Sweden, Norway and Finland were
spared from exceedance of the 25 days target value. It means that ozone issues are not solved yet in
Europe, especially when background average values to which we are exposed every day, are
considered. Figure 10 shows the seasonal distribution of this health indicator. The number of days
exceeding 120 µg/m3 (maximum eight hours average) was already above 25 in spring in Mediterranean regions in Spain, France, Italy demonstrating that significant impact of ozone
concentrations on health can occur early in the year and certainly earlier than in the past.
4 Health risk of ozone from long range transboundary air pollution, WHO 2008,
http://www.euro.who.int/__data/assets/pdf_file/0005/78647/E91843.pdf
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 17 of 42
SOMO35 indicator in 2015
SOMO35 indicator in 2014
(source : CAMS – 2014 air quality assessment report)
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 18 of 42
Number of days when 120 µg/m3 (maximum daily 8-hours average) was
exceeded in 2015
Number of days when 120 µg/m3 (maximum daily 8-hours average) was
exceeded in Spring (left) and summer (right) 2015
(a) (b)
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 19 of 42
Number of days when 120 µg/m3 (maximum daily 8-hours average) was
exceeded in 2015 at the monitoring stations reported to the EEA
(source: EEA data viewer)
It is interesting to note that maps of number of days exceeding the 25 target value computed by
CAMS (Figure 9) is remarkably consistent with data reported to the European Environment Agency
(EEA) in the framework of the air quality directive and that is available on the EEA data viewer
(http://eeadmz1-cws-wp-air.azurewebsites.net/products/data-viewers/statistical-viewer-public/). A
snapshot from this data viewer is proposed in Figure 11 for comparison.
2.3 Peaks indicators
The European legislations sets information and alert hourly thresholds values that should not be
exceeded. They describe situations when ozone concentrations become very high and justify
implementation of short term action plans or emergency measures to quickly decrease or stabilize
ozone levels. Those threshold values are 180 µg/m3 and 240 µg/m3 (hourly averages) for information
and alert values respectively.
Figure 12 shows the number of hours when the information threshold was exceeded in 2015. The
situation is degraded compared to the previous year with exceedances disseminated in several
countries whereas only the Pô Valley recorded such exceedances in 2014. Consequently, the Pô Valley
was still a hot spot in 2015. No exceedance of the alert threshold is detected, although such
exceedances were monitored locally by the AQ monitoring network, what can be explained by some
uncertainties in the model to detect highest values where they are very punctual.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 20 of 42
Number of hours when the information ozone threshold (180 µg/m3)
was exceeded in 2015
2.4 Conclusions for ozone
• Annual and seasonal ozone concentrations were high and partially due to exceptionally high temperatures and sunrise in 2015, which was classified by the World Meteorological
Organization as one of the warmest one. Ozone annual average in Europe ranged from 35 to
90 µg/m3. If Southern Europe recorded logically the highest ozone levels (Portugal, Spain,
Italy, Slovenia, Croatia, Montenegro, Albania, Greece, Turkey) exceptional high ozone levels
were also observed in Nordic countries and in Central Europe;
• Spring average concentrations were remarkably high everywhere in Europe which confirms a trend to “early ozone episodes” already observed in 2014;
• Ecosystems and human health exposure indicators show a degradation compared to 2014
with larger areas impacted everywhere in Europe. This can be partially explained by high
temperatures recorded in 2015
• The number of hourly peaks also increased compared to 2014. The Pô Valley is the most
concerned area but exceedances of the information threshold value (180 µg/m3) were observed in other locations like Spain, Portugal, France and Western and Central Europe.
• No real decreasing trend, regarding ozone background concentrations in Europe can be pointed out despite precursor emissions strategies implemented in European Union over the
past years. This results from sensitivity of ozone formation processes to meteorological
variability and increasing temperatures and by the influence of hemispheric transport of
ozone which limit efficiency of regional emissions control strategies.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 21 of 42
3. Nitrogen dioxide Warning note: It should be noted that the CAMS re-analyses mapping system is not fitted to deal with
local hot spot situations as those that develop near busy or on industrial sites. Actually, the models
resolution is about 10*10 km, and is not sufficient to catch actual NO2 concentrations at traffic and
industrial sites. This is also the reason no map of indicators related to the number of situations when
the limit hourly value of the Air Quality Directive is exceeded is proposed in this report. In most of the
cases such situations occur at traffic sites, near busy roads, and cannot be described by the European-
wide re-analysis system implemented in CAMS.
3.1 Annual and seasonal averages
Annual and seasonal averages of NO2 concentrations are presented on Figure 13 and Figure 16. The
2008 Air Quality Directive sets a limit value for the annual average, and it must not be exceed 40
µµµµg/m3.
Figure 13 shows that background concentrations5 exceeded this limit in Milan and Moscow areas.
Cities footprints appear very clearly as main roads and maritime routes. Highest concentrations are
found in Madrid area, Benelux, South-East of the United Kingdom, Western Germany, Poland, Paris
and Ile de France, and in the Pô Valley. In those regions they generally exceed 30 µg/m3. Logically, this pattern is driven by the location of main sources of nitrogen oxides (NOx), NO2 behaving
as a “local pollutant” regarding its time life in the atmosphere. Figure 14 displays, for comparison,
NO2 annual average concentrations in 2014. The geographical distribution of the concentrations is
the same (because they are mainly influenced by the same sources), but the concentrations seem
higher in 2015 than in 2014. Variability between 2014 and 2015 can be explained by meteorological
conditions.
Figure 15 is a snapshot of the data viewer web site proposed by the EEA and presents NO2 average
concentrations in 2015 monitored at background stations (industrial and traffic stations are excluded
since they are out of scope of CAMS products). It illustrates good consistency with CAMS re-analyses
regarding concentrations levels and their spatial distribution.
5 Which means that hot spot values are not taken into account
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 22 of 42
Annual average concentrations of NO2 in 2015
Annual average concentrations of NO2 in 2014
(source : CAMS – 2014 air quality assessment report)
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 23 of 42
NO2 annual average in 2015 at the monitoring background stations reported
to the EEA
(source: EEA data viewer)
Maps on Figure 16 propose geographical distribution of seasonal NO2 concentrations. They are
significantly lower in spring and summer than in fall and winter. Two main reason explain this
difference: higher NOx emissions from residential heating that usually go along with temperatures
decrease on one side and stable and cold meteorological conditions leading to thermal inversions
that stick ground level emissions near the surface on the other side.
Biggest cities (Paris, Madrid, Brussels, Berlin, London, Moscow), the Benelux and the Pô valley are
clearly impacted by such effects with seasonal concentration averages above 40 µg/m3 in winter time.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 24 of 42
Seasonal averages of NO2 concentrations in 2015; Spring (a), Summer (b),
autumn (c), winter (d)
(a) (b)
(c) (d)
3.2 Conclusions for nitrogen dioxide
Background nitrogen dioxide concentrations exceeded in 2015 the annual limit value in a limited
number of places where NOx emissions are usually high. The Pô Valley is one of the most exposed
area, because of the convergence of high emission levels and non-dispersive meteorological
conditions in the valley. NO2 concentrations are the highest in the biggest cities, near main roads and
in sea areas because of shipping emissions.
Air concentrations distribution of NO2 in Europe and their seasonal variability remain relatively stable
compared to the past years, but certainly because of meteorological conditions, they were higher in
2015 than in 2014 in several countries, especially in Central and Eastern Europe.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 25 of 42
4. Particulate Matter (PM10)
4.1 Annual and seasonal averages
Annual and seasonal averages of PM10 concentrations in 2015 are presented in Figure 17 and Figure
20 respectively. Map of annual PM10 concentrations in 2014 is presented in Figure 18 for comparison.
The air quality Directive 2008/50/EC sets a limit value for PM10 annual average to 40 µµµµg/m3.
In 2015, annual average ranged from 5 µg/m3 in Northern Europe (Iceland, Scandinavia), to more
than 30 µg/m3 locally, and especially in the Eastern part of Europe (in Poland, Serbia) and in the Pô Valley. This diagnostic is confirmed by the comparison of the 2015 CAMS re-analyzed map with
background observation data reported to the EEA in accordance to the air quality Directive (Figure
19). However, it should be noted an overall underestimation of CAMS PM values in some parts of
Europe (Pô Valley, Poland) which is much more pronounced in some countries in the South-East of
Europe where annual concentrations are underestimated (Bulgaria, Montenegro, Albania) by CAMS.
Seasonal analyses show that PM concentrations were generally quite low everywhere in summer
while concentrations in fall and winter drove the annual average.
Winter average was high in the Pô Valley, and in a large part of Central and Eastern Europe with values
higher than 30 µg/m3 and sometimes exceeding the 50 µg/m3 value (Poland, Serbia). However, it should be noted that such exceedances stayed very local. Winter concentrations were also quite high
in several countries in western Europe (North of France, Benelux, Germany).
High PM levels in winter and fall can be attributed to residential heating activities (with a large part
of primary particulate matter) and more particularly wood combustion, in many countries.
Combination of increasing emissions and stable and cold meteorological conditions that generally
develop at the same time can explain increasing in ambient concentrations, even if winter 2015 was
not considered as a very cold period compared to the 30 previous years.
Analysis of PM10 concentrations at the end of winter (typically March) shows as usually higher values
in the North of France, Benelux, Northern Italy, and Poland. The influence of meteorological
conditions can be a driver factor to explain this situation. Daily temperature gradient may be very
important (cold nights and mild days) which favor thermal inversions that stuck air pollutants near
the ground. Moreover, sunny conditions can increase secondary PM formation with volatilization of
ammonia from soils resulting from fertilizers spreading. Ammonia emissions combined with NOx
emitted by road traffic and combustion activities are responsible for ammonium nitrate formation
which is one of the most widespread chemical compounds in PM observed in February-March-April
in several European countries. This phenomenon is very documented and explain a number of PM10
exceedances in this period.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 26 of 42
Annual average of PM10 concentrations in 2015
Annual average of PM10 concentrations in 2014
(source : CAMS – 2014 air quality assessment report)
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 27 of 42
PM10 annual average in 2015 at the monitoring background stations reported
to the EEA (source: EEA data viewer)
Seasonal averages of PM10 concentrations in 2015; Spring (a), Summer (b),
autumn (c), winter (d)
(a) (b)
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 28 of 42
(c) (d)
4.2 Daily exceedances
PM10 episodes are generally characterized by exceedances, over several days, of the daily limit value
set in the Air Quality Directive: the threshold of 50 µg/m3 (daily average) should not be exceeded more than 35 times a year. Figure 21 and Figure 23 show respectively the annual and seasonal number
of days when PM10 background concentrations exceeded the 50 µg/m3 daily limit value. Because of the resolution of the models, maps are not representative of local exceedances that can typically
happen at traffic stations, but relate to the background level of air pollution most citizens are exposed
to. Therefore, CAMS re-analyses can be compared to the number of exceedances at background
monitoring stations reported to the EEA and displayed by the EEA’s data viewer (Figure 22).
Good consistency between CAMS re-analyses and observations can be noted, with highest values in
the Pô Valley and in Poland, except in Bulgaria and Western Balkans countries where it seems that
observations have not been considered. Over the year, the number of days when the daily PM10
average exceeded 50 µg/m3 was higher than 35 in some places in Italy, Czech Republic, Poland, Serbia
and in the far-east of Europe (Ukraine, Russian Federation). Number of exceedances ranged between
15 and 35 in a large part of Western Europe (Benelux, North of France, Germany) and in a large part
of Czech Republic.
In depth analysis of seasonal variability of this indicator helps in understanding the main air pollution
drivers in the different countries. Investigating the seasonal distribution of the number of
exceedances on Figure 23, it can be noted that limited number of PM10 peaks occurred from April to
September 2015, except at the Southern boundary of the domain impacted by Saharan dusts
episodes, more frequent in this dry period.
Different patterns can be drawn for October/November/December and January/February/March
period. In the first one, PM10 daily limit value was exceeded several days in Northern Italy, in Eastern
Europe: Poland, Czech Republic, Hungary, and Serbia, and barely in Austria.
In the second period, Northern Italy was still impacted, but also North part of France and Benelux
(because of a huge PM episode that occurred in March), Germany, Poland and Czech Republic. The
reason why PM10 levels raised might be different for each period. In winter months, increasing is
rather attributed to residential heating while in March, it is the influence of agriculture ammonia
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 29 of 42
emissions in Western Europe that combined with NOx emissions to create ammonium nitrate in
secondary inorganic aerosols. Indeed, a remarkable PM episode occurred in 2015 in Western Europe
during several days in March.
Such processes and the dichotomy between the influence of different primary PM and secondary
PM’s precursors sources is now well-known and demonstrated by a large number of observations and
scientific studies dedicated to the analysis of the chemical composition of PM in Europe.
As in 2014, the number of PM exceedance situations was rather low in 2015. It is explained by quite
high temperatures over a large part of the year in Europe that limited the effect of residential sector
PM emissions, and facilitated dispersion. Figure 24 shows that temperatures in December 2015 were
exceptionally high in Europe compared to the average over the past 30 years. It can also be the result
of PM emission reduction strategies that developed in all European countries in compliance with air
pollution European legislation.
Number of days when the daily limit PM10 value was exceeded in 2015
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 30 of 42
Number of days when PM10 daily average exceeded the daily limit value in
2015 at the monitoring background stations reported to the EEA
(source: EEA data viewer)
Number of days when the daily limit PM10 value was exceeded by season in
2015; spring (a), summer (b), autumn (c), winter (d)
(a) (b)
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 31 of 42
(c) (d)
Surface air temperature in December 2015 relative to its 1981-2010 average
(source: ECMWF, Copernicus Climate Change Service -C3S temperature re-analyses)
4.3 Conclusions for PM10
The main points that can summarize the PM10 diagnostic in 2015 are the following:
• PM10 annual average of background concentrations ranged in 2015 from very low level
(5µg/m3) in a large part of Northern Europe to high values (higher than 30 µg/m3 in average)
in North of France, Benelux, Pô Valley and Eastern Europe (Poland, Czech Republic, Serbia in
particular). The annual limit value was exceeded in Eastern Europe (from Poland to the
Balkans) and in the Pô Valley. The conjunction of high emissions, especially in the residential
heating and wood burning sector together with cold and stable meteorological conditions can
partly explain increasing PM concentrations in those areas.
• Western Europe (France, Benelux, Germany) has been exposed to exceedances of the daily limit value for PM10, especially during spring episodes in March mainly due to ammonium
nitrate production because of ammonia from agriculture releases facilitated by mild
temperatures.
• Geographical distribution of PM10 concentrations in Europe in 2015 was very consistent with what was observed the previous years. Eastern part of Europe and Pô valley remain still the
most exposed areas.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 32 of 42
5. Particulate Matter (PM2.5)
5.1 Annual and seasonal averages
Annual and seasonal averages of PM2.5 concentrations in 2015 are presented in Figure 25 and Figure
28. The 2008 Air Quality Directive sets an annual limit value of 25 µµµµg/m3 that may be reinforced in
2020 to 20 µµµµg/m3 which is the indicative limit value proposed in the air quality Directive to be
implemented in 2020, upon revision by the Commission.
As for PM10, considering background concentration levels6, the current limit value is only locally
exceeded in Northern Italy, Poland, Serbia, Montenegro, Turkey and Ukraine. However, the situation
worsens when the more stringent 2020 target value applies, with larger areas concerned by
exceedances in these countries.
The situation can even be more critical if the guideline value recommended by the WHO for
controlling human health exposure to fine particulate matter is considered. This value is 10 µg/m3
(annual average). Except in the Northern part of Europe and in few regions in Portugal, Spain, France,
Switzerland, and Austria, this value is likely to be exceeded everywhere.
Figure 26 reminds the situation in 2014. Both annual maps are remarkably comparable regarding
geographical patterns. The same areas are the most or least exposed. But concentrations were higher
in 2015. Diagnostic is confirmed by the observations reported to the EEA in compliance with the Air
Quality Directive and presented on the EEA’s data viewer (Figure 27).
Seasonal analysis shows that the situation is more critical in autumn and winter than in spring and
summer. The meteorological conditions, more stable and colder in these periods partly explain this
fact. Emissions of primary PM (residential wood combustion for instance) and of precursors of
secondary particles issued from chemical formation processes are the other part of the explanation.
March is included in the winter period, and ammonium nitrate episodes that occurred in 2015 in this
period because of ammonia emissions attributable to fertilizers or manure spreading in a large part
of Western Europe can explain increasing PM2.5 ambient concentrations.
Therefore, highest exposure concerns Northern Italy and all countries of Central Europe. Western
Europe (France, Benelux) show high PM2.5 concentrations in winter as well.
6 Local exceedances that can occur near traffic sites cannot be represented on this map because of model
resolution.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 33 of 42
Annual average of PM2.5 concentrations in 2015
Annual average of PM2.5 concentrations in 2014 (source: CAMS – 2014 air quality assessment report)
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 34 of 42
PM2.5 annual average in 2015 at the monitoring background stations
reported to the EEA (source: EEA data viewer)
Seasonal averages of PM2.5 concentrations in 2015; Spring (a), Summer (b),
autumn (c), winter (d)
(a) (b)
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 35 of 42
(c) (d)
Even if this is not a regulatory indicator, the number of days when daily average exceeded 25 µg/m3
in 2015 gives an instructive overview of fine particulate pollution status in Europe (Figure 29). The
most sensitive areas are drawn: North of Italy, Poland, Czech Republic, Hungary, Slovakia, Serbia,
Romania, Ukraine but also Turkey experienced largest number of exceedance days (more than 50).
Western Europe (France, Benelux, Germany) has a significant number of exceedance days as well
(between 30 and 40). Southern boundaries of the domain are largely influenced by transport of
Saharan dusts which are, however, rather coarse particles.
Number of days when daily average exceeded 25 µg/m3 in 2015
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 36 of 42
5.2 Conclusions for PM2.5
• Exposure to fine particulate matter in Europe (PM.2.5) remains a sensitive issue. If the current
limit value for annual average set in the Air Quality Directive (25 µg/m3) is exceeded only in few areas in Northern Italy and Central and Eastern Europe, exposure to more stringent
thresholds (20 µg/m3 as the indicative limit value proposed in the air quality Directive to be
implemented in 2020, upon revision by the Commission and 10 µg/m3 according to WHO recommendations) is worrying.
• Pô Valley, Central Europe, but also, in some periods, Western Europe are the most sensitive
areas. Daily concentrations may exceed 20 µg/m3 in several places in winter and in spring.
• Temporal variability of the concentrations is high. It is driven by meteorological conditions (dispersive conditions in summer reduce significantly PM levels) and by complex physico-
chemical processes that impact emissions of precursors (ammonium nitrate formation in
spring when ammonia is available in the atmosphere due to agriculture activities). Sensitive
emission sectors are residential heating, agriculture (ammonia) and road traffic (nitrogen
oxides).
.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 37 of 42
Technical annex I
The Copernicus Atmosphere Service component dedicated to regional air quality delivers evaluations
and forecasts of air quality in Europe. Evaluations are based on available observation data-sets (in-
situ and satellite observations) that are used to correct model results with data assimilation
techniques. The combined product is named “analyses” if delivered in a routine daily way using
available data whatever their validation status, or “re-analyses” if delivered within a longer period
which allows verification and validation of the observations.
The service CAMS-50 aims at developing and running operational air quality forecasting platforms
coupled with data assimilation systems to produce on a daily basis up to 4 days forecasts of air
pollutant concentrations (ozone, nitrogen dioxide, PM10 and PM2.5), analyses for the previous day and
re-analyses for the two previous years. Seven models are run in that perspective and their results
(daily concentrations of the targeted pollutants) combined in a composite median model called the
“Ensemble”. The present report is based on Ensemble results that are considered as the best estimate
of air pollutants background concentrations in Europe over the targeted year (2015). The chemistry-
transport models were run with a spatial resolution of about 10 km throughout Europe, and with
similar input datasets. Emissions were issued from the MACC/TNO emission inventory elaborated
within the MACC-suite projects. The 2011 version was available and used. Meteorological inputs were
provided by ECMWF, as chemical boundary conditions from global scale Copernicus atmosphere
services. Fire emissions were taken into account and were available from the Global Fire Assimilation
System (GFAS) developed as a new CAMS service.
The models involved in the European air quality re-analyses production are the following:
Model Origin
CHIMERE INERIS (France)
EMEP Met.No (Norway)
EURAD University of Köln (Germany)
LOTOS-EUROS TNO and KNMI (the
Netherlands)
MATCH SMHI (Sweden)
MOCAGE Meteo France (France)
SILAM FMI (Finland)
Finally, it is essential to note that although the model resolution used (10km*10km) is quite high and
challenging to perform simulations at the European scale, it does not allow to simulate and catch very
local air pollution patterns. Hot spots near emissions sources (busy roads, industrial sites, working
urban areas) cannot be taken into account in this analysis. The figures proposed relate to background
concentrations, whatever the pollutant. Nevertheless, this is a relevant information for national and
local authorities that have to look for control measures impacting most of their territory and
population living in. Hot spots management requires other tools and approaches that are out of the
scope of the Copernicus Atmosphere services.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 38 of 42
Technical annex II: The present report was based on validated CAMS European air quality re-analyses produced by
regional CAMS services thanks to an Ensemble modelling approach. Re-analyses are built upon data
assimilation techniques that allow to account for observations when mapping air pollutant fields. The
validated assessments use validated observation datasets (according to complex quality assurance
protocols compliant with the air quality directives) while interim re-analyses use near-real-time or
up-to-date datasets that are not fully validated, that are generally incomplete (because not all
countries report up-to-date observation data with the same frequency). Interim re-analyses are
generally also issued from older model versions than the validated ones. For these reasons, validated
re-analyses are considered as more robust and more accurate than interim re-analyses.
For the year 2015, both reports have been produced within the framework of the CAMS services. The
2015 interim assessment report for air quality in Europe was published in September 2016
(CAMS71_2016SC1_D71.1.1.2_201609).
We propose below few quality indicators showing the differences between the interim (IRA) and the
validated (VRA) production. It appears to be worthwhile to maintain both production channels since
the VRA production is significantly better, but comes also significantly later. Therefore, the IRA
production should be considered as a first guess or analysis of what happened during the targeted
year. The proposed indicators are Taylor diagrams. Taylor diagrams synthesize on a unique quadrant,
various statistical indicators for several models: the radii correspond to the correlation coefficient
values, the x-axis and the y-axis delimits arcs with bias values and the internal semi-circles correspond
to the RMSE values. Therefore, this is a very pedagogic way to present an overview of the relative
performances of a set of models, often used in model intercomparison exercises.
This is a preliminary analysis to compare IRA2015 and VRA2015, which is performed for three metrics:
- Ozone daily maximum
- PM10 daily average
Ozone daily Maximum:
Figure 30, Figure 31, Figure 32 show comparison of CAMS models performances and results (ozone
summer average) for interim and validated simulation. Differences are significant in the air pollutant
patterns (Figure 32) and the Taylor diagrams confirm better performance of both individual and
ensemble (light blue in the figures below) models for the validated re-analyses. Correlation coefficient
is slightly improved, and root mean square error. Standard deviation of the model values is closer to
standard deviation of observed values as well. This can be explained by the quality of observation
data, the fact that a number of observations is missing in the interim datasets (because they are not
reported by the Member states) and by the quality of the model itself.
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 39 of 42
IRA 2015: Taylor diagram showing the CAMS regional air quality models
performances for the simulation of ozone daily maximum – Interim production (2016)
VRA 2015: Taylor diagram showing the CAMS regional air quality models
performances for the simulation of ozone daily maximum – validated production
(2016)
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 40 of 42
Ozone summer average simulated by CAMS Ensemble model : interim
simulation (left) and validated simulation (right)
PM10 Daily average The difference between interim and validated re-analyses is much more significant for PM10
indicators than for ozone. This is explained by the fact that PM10 near real time reporting process is
less mature for PM10 than for other pollutants (the reference gravimetric method standardized for
PM monitoring does not provide hourly averages). This explain the large differences in the Taylor
diagrams presented on Figure 33 and Figure 34 (caution: the scales are not the same). Nevertheless,
VRA results are also significantly better and the validated re-analysed map displayed Figure 35 is
certainly more reliable than the interim one.
IRA 2015: Taylor diagram showing the CAMS regional air quality models
performances for the simulation of PM10 daily average – Interim production (2016)
CAMS71_2016SC2 – Annual air quality assessment report for 2015 Page 41 of 42
VRA 2015: Taylor diagram showing the CAMS regional air quality models
performances for the simulation of PM10 daily average – validated production (2016)
Number of days when daily average exceeded 35 mg/m3 simulated by CAMS
Ensemble model: interim re-analyses (left) and validated re-analyses (right)
Copernicus Atmosphere Monitoring Service
atmosphere.copernicus.eu copernicus.eu ecmwf.int
ECMWF - Shinfield Park, Reading RG2 9AX, UK
Contact: [email protected]