Science of the Total Environment 490 (2014) 957–969

Contents lists available at ScienceDirect

Science of the Total Environment

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

2001–2012 trends on air quality in Spain

X. Querol a,⁎, A. Alastuey a, M. Pandolfi a, C. Reche a, N. Pérez a, M.C. Minguillón a, T. Moreno a, M. Viana a,
M. Escudero b, A. Orio c, M. Pallarés c, F. Reina c
a
Institute of Environmental Assessment and Water Research, IDAEA-CSIC, C/Jordi Girona 18-26, 08034 Barcelona, Spain
b
Centro Universitario de la Defensa de Zaragoza, Academia General Militar, Ctra. Huesca s/n, 50090 Zaragoza, Spain
c
Spanish Ministry of Agriculture, Food and Environment, G.D. Environmental Quality and Evaluation, Pl. San Juan de la Cruz s/n, 28071 Madrid, Spain

H I G H L I G H T S

• We interpret the 2001–2012 trends of major air pollutants in Spain.

• Marked downward concentration trends were evidenced for PM10, PM2.5 and CO.
• The effect of major policy actions on air quality was evidenced.
• The low effect of specific actions was also shown.
• The effects of meteorological variability and financial crisis are also discussed.

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

Article history: This study aims at interpreting the 2001–2012 trends of major air pollutants in Spain, with a major focus on evalu-
Received 31 March 2014 ating their relationship with those of the national emission inventories (NEI) and policy actions. Marked downward
Received in revised form 15 May 2014 concentration trends were evidenced for PM10, PM2.5 and CO. Concentrations of NO2 and NOx also declined but in a
Accepted 16 May 2014
lesser proportion at rural and traffic sites. At rural sites O3 has been kept constant, whereas it clearly increased at
Available online 7 June 2014
urban and industrial sites. Comparison of the air quality trends and major inflection points with those from NEIs,
Editor: Pavlos Kassomenos the National Energy Consumption and the calendar of the implementation of major policy actions allowed us to
clearly identify major benefits of European directives on power generation and industrial sources (such as the
Keywords: Large Combustion Plants and the Integrated Pollution Prevention and Control Directives). This, together with a
PM10 sharp 2007–2008 decrease of coal consumption has probably caused the marked parallel decline of SO2, NOx and
PM2.5 for PM2.5 concentrations. Also the effect of the EURO 4 and 5 vehicle emission standards on decreasing emissions
NO2 of PM and CO from vehicles is noticeable. The smooth decline in NO2–NOx levels is mostly attributed to the low ef-
NOx ficiency of EURO 4 and 5 standards in reducing real life urban driving NO2 emissions. The low NOx decrease together
CO
with the complexity of the reactions of O3 formation is responsible for the constant O3 concentrations, or even the
O3
urban increase. The financial crisis has also contributed to the decrease of the ambient concentration of pollutants;
Time trends
Air quality policy however this caused a major reduction of the primary energy consumption from 2008 to 2009, and not from 2007 to
2008 when ambient air PM and SO2 sharply decreased. The meteorological influence was characterized by a 2008–
2012 period favorable to the dispersion of pollutants when compared to the 2001–2007.
© 2014 Published by Elsevier B.V.

1. Introduction
Improvement of air quality (AQ) is one of the major environmental
objectives of developed societies. According to this objective, the
European Union has made a large effort in the last decades by means of
the elaboration and implementation of both ambient air and emission di-
rectives. Furthermore, additional measures to abate pollution have been
applied by some member states, and regional and city governments. In
⁎ Corresponding author.
E-mail address: xavier.querol@idaea.csic.es (X. Querol).
these cases, the levels of ambition varied widely, from most Northern
and Central European countries where early measures were adopted be-
fore the entrance in force of the 2005 limit values, up to specific Southern
and Eastern European countries where measures were much delayed and
in most cases much less ambitious. Among the national and city scale
measures it is interesting to highlight the national German labeling (S2
and S4) of vehicles for the Low Emission Zones (LEZs), and the setting
up of a number of LEZs in the largest cities of Europe, such as London or
Berlin (Minguillón et al., 2013).
Out of the efficiently implemented standards in emission abatement
we would like to highlight the IPPC (Integrated Prevention and Pollu-
tion Control) Directive (1996/61/EC, Industrial Emissions Directive

http://dx.doi.org/10.1016/j.scitotenv.2014.05.074
0048-9697/© 2014 Published by Elsevier B.V.
958 X. Querol et al. / Science of the Total Environment 490 (2014) 957–969
2010/75/EC), the Large Combustion Plants Directive (2001/80/EC),
the National Emission Ceilings (NEC) Directive (2001/81/EC), and the
EURO standards on road traffic emission (1998/69/EC, 2002/80/EC,
2007/715/EC). Currently, the EC is in a process for evaluating the NEC
Directive and generating a new Medium Combustion Plants Directive.
Furthermore, during the last decades, several regulations have been
issued to control emissions from shipping (Adamo et al., 2014). IMO
(International Maritime Organization)/MARPOL and the EU have set
absolute limits on sulfur content in fuel, and SOx and NOx emissions
from ships (IMO, 2011; Directive 2005/33/EC).
In Spain a national AQ plan has been approved by the Council of
Ministers of the Government of Spain, the last update being Plan Aire
(April 2013). Furthermore, 45 regional and 3 local (city scale) AQ
plans have been implemented since 2004. Thus, most of the plans are
designed and approved by the Autonomous Regional Government but
the territory where these apply tends to be rather small, mostly focusing
on improving AQ at major city centers or specific industrial areas.
There is a large proportion of AQ zones across Europe meeting the AQ
limit values and objectives for CO, SO2, PM2.5 and metals. Conversely,
and in spite of the above policy efforts, an important proportion of
the European population lives in areas exceeding the AQ standards for
the annual limit value of NO2, the daily limit value of PM10 and the
health protection objective of O3 (EEA, 2013). For PM10 and NO2, causes
of these exceedances have already been discussed by Harrison et al.
(2008) and Williams and Carslaw (2011), among others. EEA (2013)
shows that PM10 and NO2 limit values are exceeded mostly in urban
areas, and especially at traffic sites. In specific countries such as Spain it
has been reported that N90% of the NO2 exceedances are attributed to
road traffic emissions (Querol et al., 2012).
It has been evidenced that the real-life NO2 emissions of diesel
passenger cars in urban driving conditions are much higher than
expected and in any case much higher than the emissions produced
for the EURO test driving cycles (Williams and Carslaw, 2011). EURO
standards have had a very clear benefit on PM emissions from diesel
passenger cars, especially since EURO 4 (2005) and EURO 5 (2009),
but not on abating urban ambient PM concentrations (Fuller and
Green, 2006 even found evidences of an increasing trend for near road
PM in London), neither for reducing NO2 emissions (Carslaw and
Rhys-Tyler, 2013). Furthermore, the proportion of diesel cars has mark-
edly increased in many European fleets due to the support given to the
use of diesel fuel by climate policy. Taking these two issues into account,
it is easy to understand the large and widespread problem that we have
in our cities to attain NO2 AQ standards. It becomes also evident that
non-technological measures focusing on diminishing the use of private
cars in cities are needed to attain these standards. The problem may be
even more serious in specific southern European cities due to the higher
inhabitants and car density and the widespread occurrence of street
canyons (Cyrys et al., 2012).
Ambient levels of PM10 and NO2 in urban areas largely exceed the
WHO guidelines in Europe (EEA, 2013). Concerning EC PM10 standards
attainment, the problem is less generalized than for NO2, but in many
cases exceedances are also occurring in urban areas. Eeftens et al.
(2012) showed evidences of a clear gradient of PM levels across
European cities, with an increase of both PM10 and PM2.5 from north-
ern to central and to southern and eastern regions, and that these dif-
ferences are more pronounced for PM10. The latter is highlighting the
large influence of road dust and anthropogenic resuspension on the
urban levels of PM10 in Southern Europe. This was already underlined
by Querol et al. (2004), by comparing PM speciation data from a number
of European countries, and nowadays it seems that this PM source may
become more relevant for attaining PM10 standards at traffic sites
(Harrison et al., 2008). It is evident that PM exhaust emissions have
clearly decreased in the last century (from 180 mg/km to 5 mg/km of
EURO1 (1992) to EURO5 (2009) for diesel passenger cars according to
EC standards) but non-exhaust emissions have been, and are, unaffected
by legislation during the same period.
Exceedances of the AQ standards may be also occurring around spe-
cific industrial hotspots, harbors and power plants. Important differ-
ences can be observed across the different regions of Europe according
to the regional industrial and power generation policies (EEA, 2013).
In spite of the above problems, there is a clear evidence for PM
concentrations having markedly decreased during the last decade in a
number of European regions (EEA, 2013; Barmpadimos et al., 2012;
Cusack et al., 2012), as a result of: a) the EU policy for reducing emissions;
b) the national and numerous regional and local AQ plans implemented,
and c) the favorable 2008–2012 meteorology for Southern Europe as
compared with 2005–2007 (Barmpadimos et al., 2012; Cusack et al.,
2012).
Cusack et al. (2012) reported decreases of PM2.5 concentrations for
remote/rural sites within the period 2001–2010, reaching up to 30–36%
in a number of European regions. Barmpadimos et al. (2012) found that
PM10 and PM2.5 concentrations decreased during 1998–2010 at a
number of urban and rural background stations in five European coun-
tries with an average decrease of the raw and meteorologically adjusted
data of −0.4 µg/m3/year for both PM10 and PM2.5 size fractions. How-
ever, the time trend observed for coarse PM, averaged over all the con-
sidered stations, was less marked and not statistically significant. EEA
(2013) reported stably high levels of O3 during the last decade, but ev-
idenced increases of O3 levels in a number of urban areas of Europe. EEA
(2013) also reported NO2 decreasing trends for a large number of sites
although both the relative number of sites and the quantitative decrease
were lower than for PM.
This study focuses on the interpretation of 2001–2012 trends for
major air pollutants at a number of urban, industrial and rural sites of
Spain with the aim of identifying the main drivers of these trends. The
data on national emissions and energy consumption are also evaluated
to interpret trends. Understanding past trends may be relevant for
devising new strategies for air pollution abatement. Regional back-
ground trends are firstly evaluated on the basis of the data yielded by
Spanish EMEP (The European Monitoring and Evaluation Programme;
http://www.emep.int/) monitoring sites. Concentration trends at urban
background and traffic sites in cities having more than 240,000 inhabi-
tants (to include the industrial area of A Coruña) and at industrial
hotspots are subsequently interpreted. In order to support the under-
standing of the major causes of the decreasing trends found for PM, a
similar 2002–2012 analysis was performed for a data set of PM2.5
chemical components available from the background site of Montseny,
in, northeast Spain.
2. Methodology
2.1. Measurement sites and techniques
Spanish monitoring stations used for AQ assessment reporting time
series of pollutants during no less than 7 years in cities having more
than 240,000 inhabitants (to include A Coruña, an important and indus-
trial city in NW Spain) were selected for this work, without including
those series interrupted by the closure of the station during the last
years (2010–2012). Furthermore data from 9 EMEP monitoring sites
were selected. Time series for PM10, PM2.5, SO2, NOx and NO2 are avail-
able for the period 2001–2012, while CO records started in 2003 and O3
records in 2004. Once selected, the stations were classified into regional
background (RB; EMEP sites), urban background (UB), traffic (TR) and
industrial (IN), as reported by the respective AQ networks. The IN
category contains those stations classified as industrial by the local AQ
network, and also those classified as urban or traffic sites but located
in an area with strong industrial activity. In the cities where more
than one station is available for a specific environment type, a single
data set was obtained as an average of those available. Table 1 and
Fig. 1 supply information on the location, environment and parameters
measured at the study sites selected.
 X. Querol et al. / Science of the Total Environment 490 (2014) 957–969 959

Table 1
Details of the industrial, urban background and traffic monitoring sites from cities having N240,000 inhabitants and with a data availability N75% for 2001–2012, and EMEP monitoring
sites whose data is evaluated in the present study. Lack of data availability for pollutants does not necessary mean that the site is not or were not measuring a given pollutant, but the
data coverage is not meeting the requirements of 75%. RB, regional background, UB, urban background, TR, traffic, IND, industrial.

City Station Location Pollutants measured

Name Code Type Lat Long Height SO2 O3 NO2 CO PM10 PM2.5

Urban background, traffic & industrial sites (dgr) (dgr) m.a.s.l.

ALGECIRAS E4:RINCONCILLO ES1637A IND 36.162 −5.443 1 x

ALGECIRAS ALGECIRAS EPS ES1664A IND 36.136 −5.453 24 x
ALICANTE/ALACANT ESTACIÓN RENFE ES1178A TR 38.342 −0.493 55 x x
ALICANTE/ALACANT ALACANT-EL PLÁ ES1635A TR 38.359 −0.472 45 x x x x
AVILÉS PLAZA DE LA GUITARRA ES1424A IND 43.559 −5.928 20 x x x x x
BARCELONA IJ-BARCELONA (GRACIA-S. GERVASI) ES1480A TR 41.399 2.153 57 x x x x x
BARCELONA PLAÇA UNIVERSITAT ES0559A TR 41.387 2.165 24 x
BARCELONA ID-BARCELONA ES1396A TR 41.379 2.133 35 x
BARCELONA IL-CIUTADELLA ES1679A UB 41.386 2.187 7 x x x
BARCELONA ZONA UNIVERSITARIA ES0567A UB 41.385 2.120 24 x
BARCELONA IN-BARCELONA(VALL. D'HEBRON) ES1856A UB 41.426 2.148 136 x
BARCELONA L.SOLÉ SABARIS-P.REIAL ES1782A UB 41.385 2.119 55 x X
BILBAO PARQUE EUROPA ES1713A UB 43.255 −2.902 76 x x x x x
BILBAO MAZARREDO ES1244A TR 43.268 −2.918 33 x x x x
CÓRDOBA ASOMADILLA ES1800A UB 37.903 −4.780 152 x x x
CORUÑA (A) CORLAB 1 ES1138A IND 43.368 −8.419 35 x x x
GIJÓN AVENIDA DE CASTILLA ES1358A IND 43.542 −5.650 7 x x x x x
GRANADA GRANADA — NORTE ES1560A TR 37.196 −3.613 689 x x x x x
HOSPITALET DE LL. (L') I3-L'HOSPITALET DE LLOBREGAT ES0692A TR 41.371 2.115 29 x x x
MADRID CASA DE CAMPO ES1193A UB 40.420 −3.749 645 x x x x x x
MADRID ARTURO SORIA ES0124A UB 40.440 −3.639 698 x
MADRID FAROLILLO ES0126A UB 40.395 −3.732 625 x x x
MADRID PLAZA DEL CARMEN ES1422A UB 40.419 −3.703 657 x
MADRID PUENTE DE VALLECAS ES1532A UB 40.388 −3.651 677 x x
MADRID MORATALAZ ES1426A TR 40.408 −3.645 685 x x x x
MADRID ESCUELAS AGUIRRE ES0118A TR 40.422 −3.682 672 x
MADRID BARRIO DEL PILAR ES1521A TR 40.478 −3.711 673 x x x
MÁLAGA EL ATABAL ES1751A UB 36.730 −4.466 86 x
MÁLAGA CARRANQUE ES1750A UB 36.720 −4.448 36 x
MURCIA SAN BASILIO ES1633A TR 37.994 −1.145 40 x x x x
OVIEDO PURIFICACIÓN TOMÁS ES1572A IND 43.373 −5.873 286 x x x x x
PALMA DE MALLORCA CASTILLO DE BELLVER ES1604A UB 39.563 2.621 117 x x x x x
PALMA DE MALLORCA FONERS ES1610A TR 39.570 2.656 23 x x x x x
PALMAS DE GRAN CANARIA (LAS) MERCADO CENTRAL ES1573A TR 28.134 −15.433 11 x x x x x x
PALMAS DE GRAN CANARIA (LAS) JINÁMAR III FASE ES0873A IND 28.035 −15.416 53 x x
PALMAS DE GRAN CANARIA (LAS) NÉSTOR ÁLAMO ES1451A IND 28.015 −15.243 33 x x x x
PAMPLONA/IRUÑA ITURRAMA ES1472A UB 42.807 −1.651 449 x x x x x
PAMPLONA/IRUÑA ROTXAPEA ES1747A UB 42.827 −1.649 418 x
PAMPLONA/IRUÑA PLAZA DE LA CRUZ ES1740A TR 42.812 −1.640 455 x x x
SANTA CRUZ DE TENERIFE TOME CANO ES1131A IND 28.462 −16.262 67 x x x x x
SANTA CRUZ DE TENERIFE LOS GLADIOLOS ES1132A IND 28.459 −16.268 95 x x x x x
SANTA CRUZ DE TENERIFE MERCA TENERIFE ES1757A IND 28.444 −16.277 130 x x
SANTA CRUZ DE TENERIFE CASA CUNA ES1759A IND 28.451 −16.278 183 x
SEVILLA PRINCIPES ES1425A UB 37.375 −6.006 9 x x
SEVILLA SANTA CLARA ES1450A UB 37.398 −5.952 29 x x x x
SEVILLA BERMEJALES ES1638A UB 37.347 −5.980 26 x x x x
SEVILLA CENTRO ES1644A UB 37.388 −5.991 19 x
SEVILLA LA RANILLA ES0817A TR 37.384 −5.960 29 x x x
SEVILLA TORNEO ES0890A TR 37.395 −6.003 17 x
SEVILLA SAN JERÓNIMO ES1519A IND 37.430 −5.980 21
VALENCIA VALÈNCIA-VIVERS ES1619A UB 39.478 −0.368 11 x x x x x x
VALENCIA VALÈNCIA-PISTA DE SILLA ES1239A TR 39.456 −0.376 11 x x x
VALLADOLID LA RUBIA ES1016A TR 41.632 −4.740 689 x x
VALLADOLID ARCO DE LADRILLO II ES1631A TR 41.646 −4.730 700 x x
VALLADOLID VEGA SICILIA ES1632A TR 41.621 −4.747 690 x
VALLADOLID PUENTE REGUERAL ES1636A IND 41.656 −4.734 691 x
VALLADOLID ENERGYWORKS-VA 1 ES1596A IND 41.666 −4.715 694 x x
VALLADOLID ENERGYWORKS-VA 2 ES1597A IND 41.683 −4.741 753 x
VIGO ARENAL ES1137A TR 42.219 −8.742 53 x x x
VIGO ESTACION 2 (OESTE) ES1439A IND 42.203 −8.747 26 x x
ZARAGOZA RENOVALES ES1641A UB 41.635 −0.894 220 x x x x x
ZARAGOZA ROGER DE FLOR ES1047A TR 41.651 −0.916 212 x x x x x
ZARAGOZA EL PICARRAL ES1044A TR 41.670 −0.871 195 x x x x

EMEP sites
BARCARROTA BARCARROTA ES0011R EMEP 38.473 −6.924 393 x x x x x
CADAQUÉS CABO DE CREUS ES0010R EMEP 42.319 3.316 23 x x x x x
VÍZNAR VÍZNAR ES0007R EMEP 37.237 −3.534 1230 x x x x
CAMPISÁBALOS CAMPISÁBALOS ES0009R EMEP 41.274 −3.143 1360 x x x x x

(continued on next page)

960 X. Querol et al. / Science of the Total Environment 490 (2014) 957–969

Table 1 (continued)
City Station Location Pollutants measured

Name Code Type Lat Long Height SO2 O3 NO2 CO PM10 PM2.5

Urban background, traffic & industrial sites (dgr) (dgr) m.a.s.l.

EMEP sites
TORMS (ELS) ELS TORMS ES0014R EMEP 41.394 0.735 470 x x x x x
SAVIÑAO (O) O SAVIÑAO ES0016R EMEP 42.635 −7.705 506 x x x x x
LLANES NIEMBRO ES0008R EMEP 43.439 −4.850 134 x x x x x
ZARRA ZARRA ES0012R EMEP 39.083 −1.101 885 x x x x x
PEÑAUSENDE PEÑAUSENDE ES0013R EMEP 41.239 −5.898 985 x x x x x

PM2.5 speciation x
MONTSENY MONTSENY ES1778A RB 41.779 2.358 693 x

2.1.1. Regional background, urban, traffic and industrial sites
The 2001–2012 daily and annual average concentrations of SO2, NOx,
NO2, O3, PM10 and PM2.5 from 9 EMEP Spanish regional background
(currently 13 sites are available, but only 9 have long records) and up
to 11 urban background, 16 traffic and 9 industrial monitoring sites
were compiled from the data set of the Spanish Ministry of Agriculture,
Food and Environment (SMAFE) (Fig. 1). The number of sites varies
depending on the pollutant (Table 1). At all Spanish EMEP sites the
same measurement instruments are used to determine the ambient
concentrations of the above pollutants. Also the measurement protocols
followed at the selected stations meet the requirements of the European
AQ directives and therefore the data can be considered as comparable.
The monitoring methods used for the measurements of air pollutants
in this study are summarized in Tables S1 and S2.
2.1.2. 2002–2012 PM speciation data from Montseny
In the case of the 2002–2012 PM2.5 chemical composition data set
from the Montseny regional background station, the analysis of trends
was done individually for around 35 major and trace PM components.
PM2.5 samples were collected on quartz fiber filters for 24-h periods
roughly once a week until 2007, and uninterruptedly every four days
from 2008, with high volume samplers (30 m3 h− 1) DIGITEL-DH80
and MCV-CAV, equipped with a PM2.5 cut off inlet. Filters were ana-
lyzed using different instrumental techniques to determine concentra-
tions of a range of elements and components, as described by Querol
et al. (2008). Briefly, 1/2 of each filter was acid digested (HF:HNO3:
HClO4) for the determination of major and trace elements. Major
components (Al, Ca, Na, Mg, Fe, K) concentrations were determined
by Inductively Coupled Plasma Atomic Emission Spectroscopy, ICP-
AES. Trace element concentrations (Pb, Cu, Zn, Cr, Mn, Cd, As and P
among others) were determined by means of Inductively Coupled Plas-
ma Mass Spectroscopy, ICP-MS. Water leachates from 1/4 of the filter
were used to analyze SO− 2 −
4 , NO3 and Cl
−
by Ion Chromatography
HPLC (High Performance Liquid Chromatography), and NH+ 4 by an ion
specific electrode. The remaining 1/4 of each filter was used for the
elemental analysis of organic and elemental carbon (OC and EC) by a

Regional Background Urban Background Traffic Industrial

Regional Background PM2.5 speciation

A Coruña Gijón
Niembro
Avilés Pamplona
Cabo de
Oviedo Bilbao
Vigo Creus
O Saviñao
Zaragoza Montseny
Campisábalos Barcelona
El Torms
Peñausende Valladolid
Madrid L’Hospitalet

P. Mallorca
Valencia

Barcarrota Zarra

Córdoba Alicante
Murcia
Sevilla
Granada Víznar
Santa Cruz
Las Palmas

Fig. 1. Location of the monitoring sites considered in this work: EMEP stations (circles), urban background (white rectangles), traffic (black rectangles), industrial (white triangles), and
Montseny regional background station for chemical data (white cross).
 X. Querol et al. / Science of the Total Environment 490 (2014) 957–969 961

thermal–optical transmission technique using a Sunset Laboratory patterns, a short evaluation of the NECo statistical data for Spain is pre-
OCEC Analyzer using the EUSAAR-2 protocol (Cavalli et al., 2010). sented here (MINETUR, 2013) as well as of the time trends evidenced
for the NEIs (MAGRAMA, 2013).
2.2. Data analysis The primary energy consumption increased by +15% from 2001 to
2007, slightly decreased from 2007 to 2008 (− 3%), and markedly
The time trends of the selected AQ parameters were evaluated using decreased in 2009 (− 12% with respect to 2007) (Fig. 2). Since 2009,
the Theil–Sen statistical estimator (Theil, 1950; Sen, 1968), available in the energy consumption indicator seems to remain constantly low.
the OpenAir software (Carslaw, 2012; Carslaw and Ropkins, 2012). This Thus the marked 2007–2008 decrease detected in many trends of air
estimator allows the calculation of the regression parameters of the data pollution described above does not seem to be related with a trend of
trends, including slope, uncertainty in the slope and p value. The applied emission patterns from energy consumption. Oil consumption was
method yields accurate confidence intervals even with non-normal data more constant, with only an increase of + 8% during 2001–2007, but
and it is less sensitive to outliers and missing values (Hollander and an important (− 24%) decrease during 2007–2012, without major
Wolfe, 1999). Regarding the single data set on chemical composition, changes from 2007 to 2008 (− 3%). This trend is probably governed
monthly averages of each of the PM2.5 major and trace component by the fuel consumption for traffic road. Natural gas consumption in-
concentrations were used for the statistical analysis, which required a creased by 119% from 2001 to 2008, and since then diminished by 20%
prior seasonal adjustment. Annual data averages were used for the until 2012. Coal consumption remained constantly high from 2001 to
rest of considered pollutants. Bootstrap resampling was used to esti- 2007 whereas, as for ambient air levels of PM and SO2, a sharp decrease
mate the resulting regression parameters. The symbols shown in the occurred from 2007 to 2008 (− 30%), continuing until 2010 (− 65%
figures for each trend estimate relate to how statistically significant 2007–2010 decrease). However, in the last 2 years there was an impor-
the trend estimate is: p b 0.001 = ***, p b 0.01 = **, p b 0.05 = * and tant increase, leading to a consumption only slightly lower than in 2001.
p b 0.1 = +; no symbol stands for no significant trend (p N 0.1). Nuclear energy consumption remained relatively constant along the
Official data on the annual National Energy Consumption (NECo) study period, whereas the hydroelectric generation had three maxima
from different energy sources (MINETUR, 2013) and from the National in 2001, 2003 and 2010, the latter coinciding with the fall of coal con-
Emission Inventories (NEIs, MAGRAMA, 2013) were also evaluated in sumption. Renewable energy consumption increased by 207% from
order to interpret trends of air pollution. 2001 to 2012, with a gradual growth in the NECo. The 2010 increase
of hydroelectrical consumption (125% over the minimum value of the
3. Results study period, 2005) is very remarkable and due to the high rainfall
rate of this year. Thus, 2010 was a favorable year for atmospheric disper-
Before the trend analysis for air pollutants, and in order to analyze sion and washout of pollution in Spain (Cusack et al., 2012), but air qual-
the possible relationship of the ambient air pollution with the emission ity also improved by lower atmospheric emissions due to the decrease
Million tons equivalent oil

150

145

140
Primary energy
135

130

125
Million tons equivalent oil

75
180
70 160
Arbitrary units

65 Oil 140
60 120
100
55 80 Primary energy + (iNAO*10)
50 60
45 40
20
40
0

25
Million tons equivalent oil

40
Arbitrary units

Natural gas 20
35
30 15
25
10
20 Coal
Coal + (iNAO*2)
15 5
10
5 0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
0
Million tons equivalent oil

18 Nuclear
16
14
12
10 Renewables
8
6 Hydroelectric
4
2
0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Fig. 2. Annual (2001–2012) energy consumption for Spain (in million tones equivalent oil). Data from the Spanish Ministry of Industry (MINETUR, 2013) The two right graphs illustrate
specific energy consumption data multiplied by 10× and 5× North Atlantic Oscillation Index (iNAO) to modulate energy consumption by meteorological factors.
962 X. Querol et al. / Science of the Total Environment 490 (2014) 957–969

of coal consumption in favor of a hydroelectrical growth. To adjust ener- rate of agricultural and farming emissions (87% of NH3 emissions,
gy consumption by the meteorological patterns of the specific years, we −0–6%/year, *).
added to Fig. 2 two graphs where the energy consumption is multiplied
by a factor depending on the winter North Atlantic Oscillation Index 3.1. Time trends of major air pollutants
(iNAO, defined as the pressure difference between the Azores High
and the Icelandic depression) obtained from Osborn (2013). These A marked 2001–2012 decrease was recorded for SO2 concentrations
data will be discussed together with air pollution data in the subsequent across Spain (Fig. 3). Concerning the RB EMEP sites, 2 types of trends are
sections. clearly evidenced. The northern sites, close to important coal power
The evaluation of the NEIs for Spain for the studied pollutants evi- generation plants (O Saviñao, Niembro and Peñausende; Sav, Nie,
denced the following patterns of temporal trends (Table S3): Peñ), show relatively high and constant 2001–2007 concentrations, a
sharp decrease in 2008–2009, and relatively low and constant levels
• A clear and high statistically significant decrease of the total SO2 emis- since 2009. This sharp decrease is parallel to the decrease of SO2 emis-
sion was evidenced (−6.1%/year, ***) driven by the trend from power sions from the power generation sector as shown by the NEIs (Fig. 3,
generation (accounting in 2001 70% of the national emissions) MAGRAMA, 2013); see later possible causes. The other EMEP sites,
(−7.2%/year, ***). spread across Spain (Barcarrota, Cabo de Creus, Viznar, Campisábalos,
• A less pronounced decrease was followed by total NOx emissions, with Els Torms and Zarra; Bar, CdC, Viz, Cam, Tor, Zar), also show a marked
a moderate statistical significance (*) and a much lower decreasing decrease in 2008–2009, although concentrations at the beginning of
rate (−2.5%/year). This trend is probably the combination of a statis- the study period were much lower and the trend is much more gradual.
tically significant smooth decreasing trend for road traffic (38% of The time trend analysis (Table S4) shows a statistically significant de-
NOx emissions, − 3.1%/year, ***) and those of power generation, crease (*** to *) of − 3.6 to − 7.7%/year, with the exception of Víznar
non-road transport and industrial processes with low or no statistical (in some cases very close to the trend found for national emissions
significance (15% of NOx emissions, − 0.4 to − 4.5%/year, no signifi- Table S3). Concerning the UB, TR and IN sites, relatively higher average
cance to +). SO2 concentrations (4, 5 and 8 μg/m3 as averages for the whole study
• A highly significant decreasing trend was obtained for primary period, respectively for UB, TR and IN, respectively) were observed com-
PM10 national emissions (− 2.5%/year, ***) mostly driven by the pared with EMEP stations. Statistically significant decreasing trends
decrease of three major emission sources (road transport, power were found in around 50% of the cases (Table S5.). Thus, mean decreas-
generation and non-industrial combustion, 60% of the primary ing rates of −3.6, −4.6 and −4.3%/year at UB, TR and IN sites, respec-
PM10 emissions, −0.4 to −6.9%/year, ***). A similar trend was evi- tively, were obtained when all sites were computed together, from 35
denced for PM2.5 (− 2.2%/year, ***), but in this case mostly driven to 50% lower than the decrease found for national emissions.
by the decrease of only two major emission sources (road transport The NO2 concentrations at RB sites decreased during the study
and non-industrial combustion, 52% of the primary PM2.5 emissions, period, with the sharpest decrease occurring in 2007-2008 (Fig. 3).
−0.4 to −3.5%/year, ***). This sharp decrease is again coincident with the decrease of NOx
• A significant decreasing trend was obtained for CO emissions emissions from the power generation shown by the NEIs (Fig. 3,
(− 3.1%/year, **) driven by the trend from road transport (39% of MAGRAMA, 2013). Here it is again interesting to distinguish between
the CO emissions, − 9.0%/year, ***). two groups of EMEP sites, but different from those described for SO2.
• Finally also a less significant and much lower decreasing trend was On the one hand, Cabo de Creus, Víznar, Els Torms, O Saviñao and
obtained for NH3 emissions (−0.5%/year, *) due to the low decreasing Niembro (CdC, Viz, Tor, Sav, Nie), show high concentrations from

20 60 120
18 SO2 NO2 NOx
50 100
16
14 40 80
12
µg/m3

µg/m3

µg/m3

10 30 60
8
6 20 40
4 10 20
2 Traffic Urban backgraound Industrial Traffic Urban backgraound Industrial Traffic Urban backgraound Industrial
0 0 0

3,5 7,0 8,0

3,0 7,0 NOx

Regional Background 6,0 Regional Background
(EMEP) (EMEP) 6,0
2,5 5,0
5,0
µg/m3

µg/m3
µg/m3

2,0 4,0
4,0
1,5 3,0
3,0
1,0 2,0 2,0
0,5 1,0 1,0
Mean Sav, Nie, Peñ Bar, CdC, Viz, Cam, Tor, Zar Media CdC, Viz,Tor, Sav, Nie Bar, Cam, Zar, Peñ Media CdC, Viz,Tor, Sav, Nie Bar, Cam, Zar, Peñ
0,0 0,0 0,0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

1.2E+6 6E+5
SO2NationalEmissions NOx National Emissions
1.0E+6 5E+5

8.0E+5 4E+5
Industrial processes with combustion
Mg
Mg

6.0E+5 Power generation and transformation 3E+5

Other transport and mobile machinery
4.0E+5 2E+5
Industrial processes with combustion
2.0E+5 1E+5 Power generation and transformation
Other transport and mobile machinery
Road transport
0.0E+0 0E+0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Fig. 3. Time trends of annual mean concentrations of SO2, NO2 and NOx at traffic, urban background and industrial (top) and regional background (middle) sites. Spanish national emission
of SO2 and NOX from different sectors (in Mg, bottom).
 X. Querol et al. / Science of the Total Environment 490 (2014) 957–969
2001 to 2007, a sharp 2007-2008 decrease, and constant lower levels
since 2008; on the other hand, Barcarrota, Campisábalos, Zarra and
Peñausende (Bar, Cam, Zar, Peñ) show little trend but with relatively
lower concentrations. These decreasing trends were statistically signifi-
cant (* to +, see Table S4) only for the first set of EMEP sites (closer to
urban and industrial anthropogenic emissions compared with the
others). Mean decreasing rates from −2.8 to −3.7%/year, much lower
than those for SO2, were obtained for the stations with statistically sig-
nificant decreasing rates. In the case of NOx the trend obtained is very
similar to the one of NO2 (Fig. 3) but this decreasing trend is slightly
higher −2.3 to −4.2%/year, and the statistical significance is also mark-
edly higher (*** to +, Table S7). This reflects probably the higher de-
creasing rate of NO vs NO2, even at these remote/rural sites. As far as
the UB, TR and IN sites are concerned, relatively high levels of NO2
(10–42, 25–61 and 11–38 μg/m3 on average for the whole period and
sites for UB, TR and IN, respectively) were recorded, and statistically sig-
nificant decreasing trends were also found in around 60% of the cases
(Table S6.). Mean decreasing rates of − 1.1, − 1.8 and −2.0%/year for
UB, TR and IN, respectively, were obtained when all sites were comput-
ed together. These trends are lower than the ones obtained for NOx in
the emission inventory (Table S3). Overall, the decrease of ambient
NO2 levels is progressive since 2001, with the 2007-2008 minimum
being much less evident than for RB sites (Fig. 3). In a few cases (Vigo-
industrial site and the traffic sites of Murcia and Pamplona) there was
a clear increasing statistically significant trend (Table S6). In the case
of NOx (TS7, Fig. 3) the trends are similar to NO2 in the case of UB
sites (where NO contribution to NOx is expected to be low with
respect to NO2), but much marked in the case of IN and specially TR
sites (closer to emission sources and where higher NO/NO2 ratios are
expected to occur). Thus, not only the statistical significance is higher
(for NOx compared to NO2) in a large proportion of sites, but also the de-
creasing trends are much more marked −0.8 to −4.4%/year, ***to * for
TR, and −3.8 to −4.9, *** to *, with the exception of the above tree in-
creasing sites. This shows very clearly the higher decreasing trend of NO
vs NO2 recorded close to the emission sources.
The concentration of O3 has been high at the RB sites in the last de-
cade without significant trends (Fig. 4). In this case, the RB sites were
also grouped into two sets: 1) Cabo de Creus, Víznar, Campisábalos,
Zarra and Peñausende, with higher concentrations and spread across
the Iberian peninsula, but far from anthropogenic emissions; and
2) Barcarrota, Els Torms, O Saviñao and Niembro, closer to emission
sources. Only in 3 of the 9 RB sites annual O3 levels follow statistically
60
50
40
µg/m3
30
20
10
Traffic Urban Background
0 0
0,7
0,6
0,5
mg/m3
0,4
0,3
0,2
0,1
Traffic Urban background
0,0
2001 2002 2003 2004 2005 2006 2007
Fig. 4. Time trends of annual mean concentrations of CO and O3 at traffic, urban background and industrial and regional background sites. Spanish national emission of CO from different
sectors (in Mg, bottom right).
963
significant trends (** to +), with a slight decrease in two cases
(− 1.9%/year) and a slight increase in the other case (+ 0.4%/year)
(Table S4). As expected, in the case of the UB, TR and IN sites, relatively
lower levels of O3 were measured (40–62, 35–59 and 35–52 μg/m3 on
average for UB, TR and IN, respectively), and statistically significant in-
creasing trends were also found in around 38% of the cases (Table S8.).
Thus, mean increasing rates of + 1.7, + 3.0 and + 1.2%/year for UB,
TR and IN, respectively, were obtained when all sites were computed
together. The trends are constantly characterized by a progressive
increase until 2009 and a slight decrease during 2009–2012 (Fig. 4).
CO concentrations were only available for UB, TR and IN sites. Rela-
tively low levels were recorded (0.2–0.6, 0.1–0.6 and 0.2–0.4 mg/m3
on average for UB, TR and IN, respectively), and statistically significant
decreasing trends were found in around 82% of the cases (Table S9.).
This decrease is parallel to the decrease of national CO emissions from
road transport (Fig. 4, MAGRAMA, 2013). Mean decreasing rates of
− 3.8%, − 6.4% and − 5.1%/year for UB, TR and IN sites, respectively,
were obtained when all sites were computed together. This decreasing
trend was progressive during 2003–2009, whereas from 2009 to 2012
slight changes were observed (Fig. 4). In absolute values the decrease
is intermediate between those found for the total (national) and road
traffic CO emissions (Table S3).
Fig. 5 and Tables S10 and S11 show the 2001–2012 trends for PM2.5
and PM10 concentrations. Although with a parallel trend two groups of
RB sites were identified based on the PM2.5 concentrations. On the one
hand, the southern and northeastern sites (Barcarrota, Cabo de Creus,
Víznar and Els Torms), with relatively higher concentrations, and on
the other hand, the central, northern and eastern sites (Campisábalos,
O Saviñao, Niembro, Zarra and Peñausende), with lower PM2.5 concen-
trations. Regarding PM10 concentrations, the distinguished groups are
similar, but the eastern site (Zarra) and one northern site are included
in the group with higher PM10 levels. The decreasing trends of PM2.5
concentrations are statistically significant (*** to +) in all sites, but
with a higher level (*** to **) reported for Els Torms, Campisábalos,
Cabo de Creus and Peñausende, and a lower level (+) for the southern
and eastern sites (Zarra, Niembro and Viznar) (Table S4). The de-
creasing rate of ambient PM10 concentrations varies from − 3.9 to
− 5.1%/year, and from − 1.7 to − 3.0%/year for the first and second
group respectively. In the case of PM10 the decreasing trends are char-
acterized by a high statistical significance level (*** to **) in most sites
(Table S4). However, the decrease rate is less pronounced than for
PM2.5 concentrations (− 2.5 to − 3.4%/year, and −1.7 to −2.4%/year
100
O3 90 O3
80
70
60
µg/m3
50
Regional Background
40 (EMEP)
30
20
10
Industrial Media CdC, Viz, Cam, Zar, Peñ Bar, Tor, Sav, Nie
1.2E+6
Industrial processes with combustion
CO Agriculture and farming CO
1.0E+6 Industrial processes without combustion
Road transport
Non industrial combustion
8.0E+5
Mg
6.0E+5
4.0E+5
2.0E+5
Industrial National Emissions
0.0E+0
2008 2009 2010 2011 2012 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
964 X. Querol et al. / Science of the Total Environment 490 (2014) 957–969
Fig. 5. Time trends of annual PM2.5 concentrations at EMEP sites across Spain.
for the above groups). As a whole, PM2.5 trends are characterized by a
gradual decrease, with a marked fall from 2007 to 2008 (as described
for SO2 and NO2), followed by a slight decrease or even nearly stable
average levels since 2008. For PM10 concentrations, the decreasing
trend is very similar, with also a less pronounced decline since 2008;
nevertheless, the sharp decrease from 2007 to 2008 is not observed so
clearly (Fig. 5).
As regard to the UB, TR and IN sites, the recorded PM10 concentra-
tions are relatively high (means of 18–35, 20–39 and 18–30 μg/m3 for
UB, TR and IN, respectively) with statistically significant decreasing
trends being reported for around 55% of the cases (Table S10.). Thereby,
mean decreasing rates of −1.9, −3.3, and −4.7%/year for UB, TR and IN,
respectively, were obtained when all sites were computed at once. At
the UB and TR sites, high and constant PM10 concentrations were
recorded during 2001–2007, undergoing a progressive decrease since
2008 (Fig. 6). Conversely, PM10 levels at the IN sites are characterized
by a drastic 2001–2008 decrease, remaining then almost constant dur-
ing 2009–2012 (Fig. 6). Focusing on PM2.5 levels, only 5 cities recorded
long time series (2003–2012), one traffic site and 4 urban background
sites. Mean concentrations range between 7 and 19 μg/m3. Statistically
significant decreasing trends were detected in all the sites (*** to +),
with a mean decreasing rate of −2.7%/year (Table S11). This decreasing
trend was progressive along the study period (Fig. 6).
The NEIs evidenced a marked decreasing trend for both primary
PM10 and PM2.5, but lower in absolute values (−2.2 and −2.5%/year,
Table S3), but closer to the decrease of emissions from the road traffic
sector (− 2.9- and − 3.9%/year, TS3 and Fig. 6). This is due to the fact
that secondary PM is also contributing to this decrease of ambient air
PM, and not only primary PM as it occurs in the NEIs. This also support-
ed by the fact that the sharp decrease observed in ambient air PM2.5
levels coincides also with the above described sharp decrease of
the national SO2 and NOx emissions from power generation (Fig. 3,
MAGRAMA, 2013).
The decreasing PM10 and PM2.5 trends reported for UB and TR sites
are parallel to that described for the RB sites. However, in the case of the
IN sites, the decrease is sharper until 2006 and from this year trends are
similar to ones from the UB, TR and RB sites.
3.2. Time trends of PM2.5 components
In order to support interpretations of major causes for the PM de-
creasing time trends found, a similar trend analysis was performed
for the chemical components of PM2.5, making use of the longest
available regional background data record. This data record has
been obtained at the Montseny regional background site (40 km
NNE of Barcelona, NE Spain). The Montseny rural background site
is relatively remote from urban and industrial areas, but given the
high population density and industrial development of the region,
the site can be affected by anthropogenic emissions under specific
meteorological conditions. The 2002–2010 PM2.5 trends at this
site are documented by Cusack et al. (2012). The re-analysis of the
2002–2012 trends using the methodology of the current paper is
shown in Table S12 and Figs. 7 and 8. The results show a statistically
significant (* to ***) decreasing trend for all PM2.5 components with
the exception of EC and sea salt.
 X. Querol et al. / Science of the Total Environment 490 (2014) 957–969 965

70 30

60 PM10 PM2.5
25

50
20
40

µg/m3
µg/m3
15
30
10
20

10 5
Traffic Urban background Industrial
Mean
0 0

25 14

12
20
10
15

µg/m3
µg/m3

6
10
Regional Background
Regional Background 4
(EMEP)
5 (EMEP)
2
Media Bar, CdC,Viz, Tor Cam, Sav, Nie, Zar, Peñ
Media Bar, CdC, Viz,Tor, Zar, Nie Cam, Sav, Peñ
0 0

3.5E+4 3.0E+4
PM10 National Emissions PM2.5 National Emissions
3.0E+4 2.5E+4

2.5E+4
2.0E+4
2.0E+4

Mg
Mg

1.5E+4
1.5E+4
Industrial processes with combustion 1.0E+4
1.0E+4 Agriculture and farming Industrial processes with combustion
Other transport and mobile machinery Other transport and mobile machinery
5.0E+3 5.0E+3 Road transport
Road transport
Non industrial combustion Non industrial combustion
0.0E+0 0.0E+0
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

Fig. 6. Time trends of annual mean concentrations of PM10 and PM2.5 at traffic, urban background and industrial (top) and regional background (middle) sites. Spanish national emission
of PM10 and PM2.5 from different sectors (in Mg, bottom).

5
4
Crustal Sea salt SO42-
1.0 4
3
[ng/m3]

[ng/m3]

[ng/m3]

3
2
0.5
2
1

1
0 0.0

8 6

NO3- OC 0.6 EC
5
6
0.5
4
[ng/m3]

[ng/m3]

[ng/m3]

0.4
4
3
0.3

2 2
0.2

1 0.1
0
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
year year year

Fig. 7. Time trends of concentrations of major PM2.5 components measured at Montseny site.
966 X. Querol et al. / Science of the Total Environment 490 (2014) 957–969

8 Pb 6
Cu
6
[ng/m3]

[ng/m3]
4

2
2

0
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2004 2005 2006 2007 2008 2009 2010 2011 2012

0.6
1.0

0.8
Cd 0.5 As
[ng/m3]

[ng/m3]
0.4
0.6
0.3
0.4
0.2
0.2
0.1
0.0
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

60
6
Mn 50 Zn
40
4
[ng/m3]

[ng/m3]

30

2 20

10

0 0
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012

1.2

3 Ni 1.0
Sb
0.8
[ng/m3]

[ng/m3]

2
0.6

0.4
1
0.2

0.0
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2004 2005 2006 2007 2008 2009 2010 2011 2012
year year

Fig. 8. Time trends of annual mean concentrations of trace elements in PM2.5 measured at Montseny site.

OC concentrations show a statistically significant decreasing rate
of − 5.2%/year (− 0.1 μg/m3/year). EC concentrations were relatively
low and trends are difficult to identify due to the noise of the analyt-
ical error, therefore the results do not show a statistically significant
trend. However, downward trends of secondary inorganic pollutants
concentrations were statistically significant in all cases (*** for SO24 −
−
and NH+ 4 and * for NO3 ) with rates of − 5.5, − 7.0 and − 5.3%/year
(− 0.1, − 0.1, − 0.05, μg/m3/year), respectively.
Crustal (mineral) matter (SiO2 + Al2O3 + Fe + Ca−+ Mg + CO23 +
Mn + Ti + K) concentrations decreased by around − 6.5%/year
 X. Querol et al. / Science of the Total Environment 490 (2014) 957–969
(− 0.06 μg/m3/year, ***), a component that at Montseny is primarily
associated with Saharan intrusion episodes and regional soil resuspen-
sion processes. This trend is mainly driven by the declines experienced
in the Ca and Fe concentrations (− 8%/year), especially in 2009 and
2010. Regarding Fe concentrations, there is a notable increase in 2011,
indicating that the former decrease is linked to meteorological factors
that reduced re-suspension of PM, since 2010 was characterized by
the most negative iNAO (North Atlantic Oscillation Index) from the last
150 years and this favored rainfall and dispersive conditions (Cusack
et al., 2012).
SO2−4 concentrations at Montseny are characterized by a marked de-
crease from 2007 to 2008 and a small increase in the last 2 years (Fig. 7).
This trend is similar to the one described for SO2 levels and for the coal
consumption adjusted for meteorology by using an equation in which
iNAO is included (Figs. 2 and 3) and for PM2.5 measured at EMEP sites
in regions where large coal fired power plants operate (Fig. 7). In this
northeastern Spanish region the use of heavy oils and petroleum coke
for power generation is forbidden since 2007 and this may also contrib-
uted to the ambient decrease of SO2− 4 concentrations and, as explained
above, consequently PM2.5 concentrations.
Regarding trace elements (Table S11 and Fig. 8), there is also a clear
downward trend with high statistical significance (*** for Pb, Cu, Zn, Mn,
Cd, As, Ti, Sr, Sn, Sb, and V; and also for Ni and Cr, ** and P *). The source
of trace elements at the Montseny environment is difficult to discern
due to the mixing and dilution of PM during transport. After the high
levels recorded between 2003 and 2004, trace elements related to
industrial production (smelters) at regional scale around Barcelona
(Cu, Sb, Mn, Pb, Zn, Cr, Cd, according to Cusack et al., 2012; Amato
et al., 2009) decreased significantly until the end of the study period. Be-
sides the industrial source, Cu, Sb, Sn and Zn have been also partially at-
tributed to non-exhaust road dust emissions in the study area (Amato
et al., 2009). The reduction of levels of these industrial-traffic markers
ranges between −4.3 (Cr) and −10.0 (Sb)%/year. The implementation
of the IPPC Directive in the study area is the most probable cause of this
downward trend, which was evident already before the financial crisis.
Typical tracers of fuel–oil combustion, V and Ni, experienced a signif-
icant gradual reduction since 2004, with a slope close to −5%/year, but
on the contrary concentrations appear to remain constant since 2010.
The 2007 ban on the use of heavy oils and petroleum coke for power
generation may also account for the drastic decrease of Ni and V since
that year.
For the mineral-related trace elements (Ti, Mn, Sr), the decline is
very evident since 2006, however, no pattern can be distinguished
since 2010. The lowest slope (−2.5%/year) among the mineral-related
species and statistically significant degree (*) was found for P concen-
trations, probably due to its partial biogenic origin.
4. Discussion
The data analysis shows that AQ in Spain has markedly improved
during the last 12 years. This improvement is evident at most RB, UB,
TR and IN sites. Levels of PM2.5, PM10, CO and SO2 decreased more
markedly than those of NO2. Indeed, these decreases reached total
values for the last 12 years of −28 to −43% for PM2.5, −18 to −48%
for PM10, − 42 to − 51% for CO (last 9 years in this case), − 38 to
−72% for SO2 and −18 to −27% for NO2. As regards O3, levels remained
constant across the last 9 years in 70% of the RB sites and increased by
around 15% at most UB, TR and IN sites.
Time series of SO2 levels are characterized by a continuous 2001–
2012 decrease. Nonetheless, in the northern regions, where most of
the coal power generation facilities are based, an abrupt 2007–2008 de-
crease is evident (Fig. 3). This sharp decrease is parallel to the decrease
of SO2 emissions from the power generation sector as shown by the NEI
(Fig. 3, MAGRAMA, 2013) and coincident with the implementation of
the EC Directive on Large Combustion Plants, which resulted in the ap-
plication of flue gas desulfurization (FGD) systems in a number of large
967
facilities in 2007–2008 in Spain. Furthermore, the increase of energy
produced from renewable sources and de decreease of coal consump-
tion contributed to lower the emissions. The 2009–2012 increase of
coal consumption described above (Fig. 2) did not result in an increase
of the SO2 ambient air levels and it might reflect the fact that FGD sys-
tems were operative in this period.
Mean levels of NO2 at IN and UB sites did not show clear trends,
whereas a slight decreasing trend can be observed in the TR sites. At the
RB sites, NO2 concentrations did not decrease as sharply as SO2 concentra-
tions with a decline being only recorded in a limited number of EMEP sta-
tions. The decreasing trends NOx are much more marked and statistically
significant, especially at TR and IN sites, although in 3 specific sites in-
creasing statistically significant trends were evidenced. This suggests a
more marked decrease of NO levels compared to NO2, probably to the in-
crease of the NO2/NO ratios of diesel exhaust emissions. As shown by the
MINETUR (2013) data presented above (Fig. 2), the oil consumption
trends did not show also a sharp variation as reported for other energy
consumption data, this being probably due to the lower impact of the fi-
nancial crisis, and meteorological features of a given year, on fuel con-
sumption by urban traffic. The sharp 2007–2008 decrease of ambient
NO2 and NOx concentrations is again coincident with the decrease of
NOx emissions from the power generation shown by the NEI (Fig. 3,
MAGRAMA, 2013). As for many EU member states, there is an inconsis-
tency with the clear downward emission trends of NOx reported by the
Spanish NEI (Fig. 3, MAGRAMA, 2013) and the much less marked trend
observed for ambient NO2 concentrations recorded at UB, TR and IN
sites. This is probably due to the higher real-life urban driving emissions
of EURO4 (2005–2008) and EURO5 (since 2009) diesel passenger cars
compared to the emission factors applied in the European emission in-
ventories (Williams and Carslaw, 2011). This demonstrated also the low
impact of the national, regional and local actions on the vehicle related
NO2 emissions. Furthermore, industrial NOx emissions in Spain have de-
clined much less than those of other pollutants such as SO2 and PM. In
Spain there are still very few De-NOx industrial plants. Also, in spite of
the reported reduction of NOx (NO + NO2) emissions, the decrease in am-
bient concentrations has been much more marked for NO than for NO2.
O3 concentrations remained relatively constant at RB sites, with
significant decreasing trends at only 2 sites, and non-significant trends
at the rest of the stations. However, at UB, TR and IN sites a statistically
significant slight increase was detected for 38% of the sites. These re-
verse patterns are attributed probably to a decrease of NOx emissions,
particularly NO. Under lower NO concentrations in an urban environ-
ment, less O3 is consumed to oxidize this pollutant into NO2. Further-
more, an increased urban formation of O3 can be due to a possible
VOCs/NOx disequilibrium induced by decreased NOx concentrations
(Jeffries and Crouse, 1990, Seinfeld and Pandis, 1998). Hopefully, if
actions are taken to drastically reduce VOCs and NO2 levels in the near
future, O3 levels will also be reduced, both at urban and regional scales.
Concerning CO concentrations, a decreasing trend was gradual
during 2003–2009, whereas from 2009 to 2012 slight changes were ob-
served (Fig. 4). This decrease coincides with the one of the national CO
emissions from road transport (Fig. 4, MAGRAMA, 2013). The decrease
is most likely related with the abatement of CO emissions induced by
the implementation of the EURO 4 and 5 emission standards, but also
by the progressive dieselization of the fleet, since diesel engines emit
much less CO than gasoline vehicles. The higher CO decreasing rate
at the TR sites compared to the UB sites (68% higher) points also to a
major road traffic origin.
The higher PM levels in the Southern and Eastern regions (Barcarrota,
Víznar and Zarra) are probably the result of the higher influence of
African dust outbreaks; whereas in the north and northeastern regions
(Cabo de Creus, Niembro and Els Torms) this is probably due to a higher
anthropogenic influence given their proximity to urban and industrial
areas. The decrease rate obtained for PM2.5 concentrations is by 70%
higher than that obtained for PM10 concentrations at those RB sites
with time trends reaching the highest statistical significance levels. This
968 X. Querol et al. / Science of the Total Environment 490 (2014) 957–969
is probably due to the fact that the African mineral dust influences more
heavily the PM10 concentrations (especially in the southern and eastern
regions) compared to PM2.5 concentrations, that are highly influenced
by the regional anthropogenic emissions. This is also supported by the
finding that, with the exception of Niembro (with very strong industrial
development in the region), the PM2.5 decreasing rates are much more
pronounced and with higher statistical significance levels in the northern
sector of the Iberian peninsula (Els Torms, Peñausende, O Saviñao, Cabo
de Creus, Campisábalos) than in the southern one. In the southern sector
a more constantly high influence of African dust hinders the decreasing
patterns probably arising from a progressive lowering of anthropogenic
emissions. In the case of the UB, TR and IN sites, the difference between
the decreasing rate of PM2.5 and PM10 concentrations is not as high as
at the RB sites, probably due to the PM2.5-10 fraction being mostly
anthropogenic in these environments (Amato et al., 2009). As stated
above the decreasing PM10 and PM2.5 trends reported for UB and TR
sites are parallel to that described for the RB sites. However, in the case
of the IN sites, the decrease is sharper until 2006 and from this year the
trend is parallel to those from the UB, TR and RB sites. This evidences
the effect of possible actions taken specifically at industrial sites, such
as the implementation of IPPC directive and other effects from environ-
mental policies.
The abatement of ambient PM in the last 12 years is very evident.
In 2005, around 35 AQ zones of Spain exceeded both the annual and
the daily PM10 limit values, whereas in 2012 only one zone exceeded
the annual limit and nine exceeded the daily one (MAGRAMA, 2013),
although the 2007–2008 sharp decrease is very evident in most sites.
This decrease has been in general gradual, even before the 2008-
starting financial crisis, and especially evident in the industrial areas
where the IPPC implementation had a very positive effect.
When considering the spatial variation of PM2.5 concentration
trends at EMEP RB sites across Spain, we observe a gradual decrease in
the central region, where industrial clusters are scarce. All the other re-
gions show the stated sharp 2007–2008 decrease (Fig. 5) observed also
for the trend of annual coal consumption adjusted by iNAO (Fig. 2).
PM2.5 concentrations at the northeastern and southern regions show
a clear 2002–2008 decreasing trend, remaining constant since then,
whereas in the northwestern sites a marked fall from 2008 to 2010 is
observed (parallel to the decrease on coal consumption and SO2 levels,
Figs. 2 and 3). Here PM levels increased from 2010 to 2012 (Fig. 5) as
also observed for the coal consumption adjusted by iNAO (Fig. 2). At
the eastern regions, 2 periods are clearly differentiated, with high levels
during 2001–2007, and low levels for 2008–2012 but a progressive
decrease was not observed (Fig. 5).
In summary a clear and gradual decline of SO2, CO, PM10 and PM2.5
concentrations during the study period can be concluded. The trend
shows a sharp decline from 2007 to 2008, which is followed by slow or
even absent decreasing trends from 2009 to 2012. The NEI evidenced a
marked decreasing trend for both primary PM10 and PM2.5 emissions
only from the road traffic sector (Fig. 6, MAGRAMA, 2013), but the
sharp decrease observed in ambient air PM2.5 levels coincides also with
the above described sharp decrease of the national SO2 and NOx emissions
from power generation (Fig. 3, MAGRAMA, 2013) probably caused by the
2008 implementation of the EC Directive on Large Combustion Plants. The
reduction of the industrial (power generation) emissions of these gaseous
precursors may account for a decrease of ambient levels of secondary PM
species, such as ammonium sulfate and nitrate. Also the clear decrease on
ambient concentrations of PM10 recorded for in industrial areas from
2003 to 2006 (Fig. 6) may be related with the progressive implementa-
tion of the IPPC Directive, and this is not fully accounted by the NEI.
The downward trend of PM2.5 concentrations at Montseny is char-
acterized by a slope of −2.7%/year (***, −0.39 μg/m3/year). In absolute
concentration values, this decrease is very similar to the total decrease
aggregately suffered by the mass concentrations of crustal matter, OC,
−
SO2− +
4 , NO3 and NH4 , accounting for a decrease of − 0.41 μg/m /year.
3
This total decrease is dominated by SO2− 4 , which accounted for 34% of
the bulk decrease. If we take into account that this component is mostly
present as ammonium sulfate, this accounts for 47% of the −0.41 μg/m3
total annual mean decrease. The OC decrease explains another 26% of
the bulk decrease, the ammonium nitrate 14% and the crustal matter
the remaining 12%.
5. Conclusions
Taking into account all the above discussion on time trends of the
levels of atmospheric pollutants, the NEI and NECo indicators, on the
major policy actions on atmospheric emissions and on the meteorolog-
ical scenarios, the observed decreasing trends of pollutants can be
explained based on the following factors:
1. The implementation of European directives affecting industrial and
power generation emissions (such as the Large Combustion Plants
and the IPPC Directives), and sulfur content in fuels (whether used
on ships and road traffic), as well as the increase of the proportion
of energy produced from renewable sources, among others, pro-
duced a significant reduction of SO2, and, in a lesser extent, NOx emis-
sions. Even though NH3 national emissions were only slightly
reduced in absolute values (−0.5%) in the same proportion accord-
ing to NEI, an associated reduction of secondary PM (mainly ammo-
nium sulfate and nitrate) is evidenced. This, together with the sharp
2007–2008 decrease on coal-based NECo would have likely been the
cause of the sharp decline observed for SO2, NOx and for PM2.5 con-
centrations (and emission inventories in the case of the two gaseous
pollutants) during 2007–2008 (47% of the PM2.5 decrease being at-
tributed to ammonium sulfate in the Montseny RB site). Although
the coal energy production increased in the last 3 years, the imple-
mentation of FGD systems accounted for the maintenance of SO2 at
low concentrations, even in the coal production regions (northwest-
ern sector: Sav, Nie, Peñ EMEP sites in Fig. 3).
2. Also the effect of the EURO 4 and 5 standards on decreasing emis-
sions of PM and CO from vehicles is noticeable, since filter traps are
required for new diesel cars since 2005. This is supported by the
progressive downward trends and the similarity of decreasing rates
observed for the road transport sector in the NEIs for PM10, PM2.5
and CO. In central Spain, where industrial emissions are reduced,
the progressive downward trends of ambient PM2.5 (Fig. 5) con-
centrations recorded at the EMEP sites follow the same pattern
than those of the emissions from the road traffic sector in the NEI
(Fig. 6). In the other regions of Spain the downward RB trends
of PM2.5 (Fig. 5) evidence the influence of both the road traffic
(progressive downward trend) and industrial (sharp changes in
2007–2008) emission reduction of both primary PM10 and PM2.5
and gaseous precursors of secondary PM.
3. One national (Plan Aire from the Government of Spain), 45 regional
and 3 local action plans implemented to improve AQ applied in
most regions and cities since 2004 might also account for a local/
regional decrease of SO2, SO24 − and PM2.5 concentrations. An
example is the 2007 ban around Barcelona on the use of heavy oil
and petroleum coke for power generation, or the 2008–2012 imple-
mentation of SCRT (continuously regenerating PM traps with a selec-
tive catalytic reduction for NO2), hybridization and shift to natural
gas engines of the bus fleet of this city.
4. The financial crisis, whose impact on industrial production and use
of fuels is evident since October 2008. As shown by Fig. 2, the main
decrease of primary energy consumption was evidenced from 2008
to 2009, and not from 2007 to 2008 when a sharp decrease of ambi-
ent air PM and SO2 was evidenced.
5. The meteorological cycles, which were characterized by a 2008–
2012 period that eas very favorable to the dispersion of pollutants
when compared to the 2001–2007 weather conditions. This was
described by Cusack et al. (2012), evidencing a high iNAO during
2001–2007, while in 2010 the iNAO was the most negative of the
 X. Querol et al. / Science of the Total Environment 490 (2014) 957–969
last 150 years. When the iNAO is negative (less pressure difference
between the anticyclone and the depression), fronts can sweep the
Iberian Peninsula from west to east, with a lower chance of influence
for African air masses, and causing higher wind and hydroelectric
energy production that in turn induces a lower coal consumption.
However, when the iNAO is positive, major episodes of atmospheric
stagnation and a greater African dust influence can occur.
6. The smooth decline in ambient NO2 levels (more marked for NOx),
mostly due to the low effectiveness of EURO 4 and 5 standards in
reducing real life urban driving NO2 emissions. The dieselization of
the fleet has also probably contributed to increase NO2 emissions
and reduce those of CO.
7. This relatively low decrease of NO2, together with the complexity of
the reactions of O3 formation (with the intervention of natural
VOCs and insolation in the process), may be responsible for the
very soft change on O3 concentrations in the last 12 years.
The results demonstrate a clear beneficial effect of environmental
policies on air quality in recent years. However, for specific pollutants,
such as O3 and NO2 EC target and limit values are still exceeded. To
meet these values and the WHO guide levels (not only for these pollut-
ants but also for PM10 and PM2.5) important actions are still required
for the next decade. We also would like to highlight that the interpreta-
tion of past air quality trends may yield very relevant results for plan-
ning further cost-effective actions.
Acknowledgments
The research leading to these results has received funding from the
European LIFE + AIRUSE Project (LIFE11 ENV/ES/584) and the Spanish
Ministry of Agriculture, Food and the Environment (UCA2009020083).
The software OpenAir was used for data analysis and plotting in this
publication (http://www.openair-project.org/).
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://
dx.doi.org/10.1016/j.scitotenv.2014.05.074.
References
Adamo F, Andria G, Cavone G, De Capua C, Lanzolla AML, Morello R, et al. Estimation of
ship emissions in the port of Taranto. Measurement 2014;47:982–8.
Amato F, Pandolfi M, Escrig A, Querol X, Alastuey A, Pey J, et al. Quantifying road dust
resuspension in urban environment by Multilinear Engine: a comparison with
PMF2. Atmos Environ 2009;43:2770–80.
Barmpadimos I, Keller J, Oderbolz D, Hueglin C, Prevot ASH. One decade of parallel fine
(PM2.5) and coarse (PM10–PM2.5) particulate matter measurements in Europe:
trends and variability. Atmos Chem Phys 2012;12:3189–203. http://dx.doi.org/
10.5194/acp-12-3189-2012.
Carslaw DC. The OpenAir manual — open-source tools for analysing air pollution data.
Manual for version 0.5-16. London: King's College; 2012.
969
Carslaw DC, Rhys-Tyler G. New insights from comprehensive on-road measurements
of NOx, NO2 and NH3 from vehicle emission remote sensing in London, UK. Atmos
Environ 2013;81:339–47.
Carslaw DC, Ropkins K. OpenAir — an R package for air quality data analysis. Environ
Model Softw 2012;27–28:52–61.
Cavalli F, Viana M, Yttri KE, Genberg J, Putaud J-P. Toward a standardized thermal–optical
protocol for measuring atmospheric organic and elemental carbon: the EUSAAR
protocol. Atmos Meas Tech 2010;3:79–89.
Cusack M, Alastuey A, Perez N, Pey J, Querol X. Trends of particulate matter (PM2.5) and
chemical composition at a regional background site in the Western Mediterranean
over the last nine years (2002–2010). Atmos Chem Phys 2012;12:8341–57. http://dx.
doi.org/10.5194/acp-12-8341-2012. [www.atmos-chem-phys.net/12/8341/2012/].
Cyrys JM, Eeftens J, Heinrich C, Ampe A, Armengaud R, Beelen T, et al. Variation of NO2
and NO concentrations between and within 38 European study areas: results from
the ESCPAE study. Atmos Environ 2012;62:374–90.
EEA. European Environmental Agency. Air quality in Europe — 2013 reportEEA report 9/
2013, Copenhagen1725-9177; 2013. [107 pp. http://www.eea.europa.eu/publications/
air-quality-in-europe-2013].
Eeftens M, Tsai M-Y, Ampe C, Anwander B, Beelen R, Bellander T, et al. Variation of PM2.5,
PM10, PM2.5 absorbance and PM coarse concentrations between and within 20
European study areas — results of the ESCAPE project. Atmos Environ 2012;62:
303–17.
Harrison RM, Stedman J, Derwent D. New directions: why are PM10 concentrations in
Europe not falling? Atmos Environ 2008;42(3):603–6.
Hollander M, Wolfe DA. Nonparametric statistical methods. 2nd ed. New York: John
Wiley & Sons; 1999. p. 421–3.
IMO. International Maritime Organization. Amendments (Tier II/III)—Annex VI amend-
ments on new fuel quality. “International Convention on the Prevention of Pollution
from Ships” (MARPOL 73/78); 2011.
Jeffries HE, Crouse R. Scientific and technical issues related to the application of incremen-
tal reactivity. Report prepared for Western States Petroleum Association, Glendale,
CA, September 17; 1990.
MAGRAMA. Inventario Nacional de Emisiones de Contaminantes a la Atmósfera.
Ministerio de Agricultura, Alimentación y Medio Ambiente del Gobierno de España.
http://www.magrama.gob.es/ca/calidad-y-evaluacion-ambiental/temas/sistema-
espanol-de-inventario-sei-/, 2013.
MINETUR. Ministerio de Industria, Energía y Turismo. Gobierno de España: energy statis-
tics and balances. http://www.minetur.gob.es/energia/balances/Balances/Paginas/
CoyunturaTrimestral.aspx, 2013.
Minguillón MC, Viana M, Querol X, editors. Particulate matter: environmental monitoring
and mitigation. Future Science Ltd. 978-1-909453-14-2; 2013. [177 pp.].
Osborn T. North Atlantic Oscillation Index Data. http://www.cru.uea.ac.uk/~timo/
datapages/naoi.htm, 2013.
Querol X, Alastuey A, Ruiz CR, Artiñano B, Hansson HC, Harrison RM, et al. Speciation and
origin of PM10 and PM2.5 in selected European cities. Atmos Environ 2004;38:
6547–55.
Querol X, Alastuey A, Moreno T, Viana MM, Castillo S, Pey J, et al. Spatial and temporal
variations inairborne particulate matter (PM10 and PM2.5) across Spain 1999–
2005. Atmos Environ 2008;42:3694–979.
Querol X, Viana M, Moreno T, Alastuey A, Pey J, Amato F, et al. Scientific bases for a
National Air Quality Plan (in Spanish). Colección Informes CSIC, 3 978-84-00-09475-1;
2012 [349 pp.].
Seinfeld JH, Pandis SN. Atmospheric chemistry and physics — from air pollution to climate
change. John Wiley and Sons, Inc. 0-471-17816-0; 1998.
Sen PK. Estimates of regression coefficient based on Kendall's tau. J Am Stat Assoc 1968;
63:1379–89.
Theil H. A rank invariant method of linear and polynomial regression analysis, I, II, III.
Proceedings of the Koninklijke Nederlandse Akademie Wetenschappen, SeriesA —
Mathematical Sciences, 386–392, 521–525, 1397–1412; 1950.
Williams ML, Carslaw D. New directions: science and policy — out of step on NOx and
NO2? Atmos Environ 2011;45(23):3911–2.