S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 5 2 7 – 5 4 0

ava i l ab l e a t www.sc i enced i r ec t . com

www.e l sev i e r. com/ loca te / sc i to tenv

Interpretation of the variability of levels of regional backgroundaerosols in the Western Mediterranean

N. Pérez, J. Pey, S. Castillo, M. Viana, A. Alastuey⁎, X. QuerolInstitute of Earth Sciences ‘Jaume Almera’, CSIC, C/Lluis Solé i Sabarís s/n, 08028, Barcelona, SpainInstitute of Environmental Assessment and Water Research, IDAEA, CSIC, C/ Jordi Girona, 18-26, 08034, Barcelona, Spain

A R T I C L E D A T A

⁎ Corresponding author. Fax: +34 93411 0012.E-mail address: aalastuey@ija.csic.es (A. Al

0048-9697/$ – see front matter © 2008 Elsevidoi:10.1016/j.scitotenv.2008.09.006

A B S T R A C T

Article history:Received 30 May 2008Received in revised form1 September 2008Accepted 4 September 2008Available online 25 October 2008

Results on interpretation of the variability of regional background PM levels in the WesternMediterranean basin (WMB) are presented. Mean PM levels recorded at Montseny, MSY(North-Eastern Spain) in the 2002–2007 period reached 17, 13 and 11 µg/m3 of PM10, PM2.5 andPM1, respectively.The daily evolution of PM levels is regulated by the breeze circulation (mountain and seabreezes). PM levels are lower at the rural sites at night owing to the nocturnal drainage flowsand to the lowering of the mixing layer height below the MSY high. These nocturnal lowlevels allowed us to estimate the continental background PM levels. At midday, theatmospheric pollutants accumulated in the pre-coastal depression are transported upwardsby the breeze, increasing PM levels.Maximum PM10 levels were recorded in summer, and February–March and November, andminimum values in the rest of the year coinciding with the highest frequency of Atlanticadvection. PM peak episodes attributed to Saharan dust outbreaks were recorded in summerand February–March. In addition, anticyclonic situations (February–March and November)may impact in elevated rural areas by increasing hourly levels of PM1 up to 75 µg/m3. Thisscenario induces the stagnation of pollutants in the pre-coastal depression. Solar radiationactivates mountain winds, transporting polluted air masses from the valleys to elevatedareas resulting in an increase of fine PM levels in areas outside the boundary layer.A significant decrease in PM annual means (40% and 34% for the entire monitoring period,7 µgPM10/m3 and 5 µgPM2.5/m3) was recorded at MSY between 2002 and 2007. There appearsto be no single cause behind these trends. This could partially be ascribed to the varyingfrequency and intensity of Saharan dust episodes, but also to large-scale meteorologicalprocesses or cycles, and/or to local or meso-scale processes such as nearby anthropogenicemission sources.

© 2008 Elsevier B.V. All rights reserved.

Keywords:PM10

PM2.5

PM1

Saharan dustAnticyclonic pollution episodesAnnual trendsWestern Mediterranean Basin

1. Introduction

Tropospheric aerosols are a cause for major concern becauseof their impact on health (Pope and Dockery, 2006), the Earth'sclimate (IPCC, 2007), visibility, continental and maritime eco-systems, and building materials.

astuey).

er B.V. All rights reserved

Research over the last decades has highlighted the need fora more comprehensive approach to atmospheric aerosolcomposition and processes. An improved understanding ofregional and intercontinental transport of aerosols is advisa-ble to implement efficient monitoring and emission abate-ment strategies. Aerosol observational data are currently

.

528 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 5 2 7 – 5 4 0

available from different European monitoring networks,mainly measured at urban monitoring stations. Moreover,the different networks do not adequately cover all types ofrepresentative environments to characterize the Europeancontinent, which results in a very heterogeneous spatial cov-erage of the different EU countries.

Air quality in polluted areas is characterized by high levelsof background aerosol and intense episodes of pollution. Mostof these episodes are associated with synoptic meteorologicalscenarios that induce the accumulation and the formation ofaerosol pollutants at regional or even continental scales. Thus,measurements performed at local scales aremost of the timesnot well suited to adequately interpret the origin of thepollution episodes. Aerosol measurements performed atregional background (RB) sites, at a sufficient distance fromlarge sources of pollutants, are clearly the best way toaccurately document both aerosol long-term trends andsynoptic features in air quality.

TheWesternMediterranean Basin (WMB) presents peculiarand complex atmospheric dynamics when compared with theAtlantic regions of the Iberian Peninsula (IP). The atmosphericdynamics influencing air quality in this region are consider-ably affected bymesoscale and localmeteorological processes,and also regional factors as described by Millán et al. (1997),Soriano et al. (2001), Gangoiti et al. (2001), Rodríguez et al. (2002and 2003), Jorba et al. (2004) and Pérez et al. (2004), amongothers. Based on these studies the peculiar atmosphericdynamics are conditioned by a number of factors: 1) theconsiderable influence of the Azores high-pressure system inthe meteorology of the IP; 2) the coastal ranges surroundingthe Mediterranean coast (Fig. 1); 3) the influence of the Iberianand Saharan thermal lows causing low pressure gradients in

Fig. 1 –Detailed location of the Montseny site an

the Mediterranean, and 4) the intense breeze action along theMediterranean coast favoured by the prevailing low advectiveconditions; and the scarce summer precipitation.

It should be pointed out that the atmospheric anthropo-genic emissions produced in this region are large. Theymainlyarise from densely populated areas (such as Barcelona,Marseille, Valencia and Tarragona) from large industrialestates around these cities and from the dense road traffic(urban areas and intense truck flow on highways).

All the above factors give rise to a scenario with a complexaerosol phenomenology, with large anthropogenic and naturalemissions, a high rate of secondary aerosol formation andtransformationand intensive interactionof aerosols andgaseouspollutants. The interpretation of PM levels and compositionresulting from such a complex scenario can only be accom-plishedby focusingonRBsites, comprising ideal locations to fulfilthe growing demand for an integrated atmospheric monitoringsystem for air quality and climate studies. The importance of thistype of environments for the study of regional-scale aerosols andlong-termPMtrendswashighlightedby the recentcreationof theEU-wide EUSAARnetwork (European Supersites for AtmosphericAerosol Research, http://www.eusaar.net/), which seeks tointegrate the measurements of atmospheric aerosol propertiesat 21 high quality European ground-based stations. One of theseis the Montseny site (MSY), in North-Eastern Spain, which wasselected to be part of the network owing to its unique settingrepresenting Southern European environments (data availablesince March 2002). Other RB locations along the Eastern coast ofSpain included in this study are Monagrega (ENDESA, Teruel,PM10 data available since July 1995), Coratxar (GeneralitatValenciana, TSP since January 1995), Cabo de Creus and Zarra(both EMEP, PM10 and PM2.5 since March 2001).

d of the other monitoring stations selected.

529S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 5 2 7 – 5 4 0

This paper summarizes the results of the interpretation ofthe variability of RB PM levels in the WMB. Special attention isgiven to the study of inter-annual trends of PM10, PM2.5 andPM1 levels, the identification and interpretation of seasonalpatterns, the influence of atmospheric transport scenarios onthe PM levels, and the detailed characterisation of specific PMepisodes, such as African dust outbreaks and regional pollu-tion episodes. Although data from most of the abovemonitoring sites are evaluated, the interpretations are mainlybased on data available from the MSY EUSAAR site.

2. Methodology

2.1. Monitoring sites and measurements

This study is based on PM measurements performed at RBstations in Eastern Spain. This region is characterised by anabrupt topography (Fig. 1): from North to South the Catalancoastal range (NE–SW) and the Iberian range (NW–SE). Theformer is crossedby deepvalleys that reach the flat coastal area.Situated between these two ranges is the Ebro basin (Fig. 1),which is characterisedbyasemi-aridsoil. The rangesaremainlycovered by a typical Mediterranean forest. The main urban andindustrial settlements in Eastern Spain are located along thecoastal plain (Fig. 1). The following monitoring stations wereselected for the present study (Table 1 and Fig. 1):

2.1.1. MonagregaThis station is located on the South-Western border of theEbro basin and forms part of the air quality monitoringnetwork belonging to Empresa Nacional de Electricidad S.A.(ENDESA-Teruel). Levels of PM10 are continuously measuredusing a real time TEOM (Rupprecht and Patashnick). Data areavailable since July 1995.

2.1.2. CoratxarThis site is situated on the Eastern side of the Iberian rangeand forms part of the air quality monitoring network belong-ing to the Generalitat Valenciana. Levels of TSP are continu-ously measured using a real time BETA monitor. Data areavailable since January 1995.

Table 1 – Location of the monitoring sites. m.a.s.l.: metersabove sea level; EMEP: Co-operative programme formonitoring and evaluation of the long range transmissionof air pollutants in Europe; EUSAAR: European Supersitesfor AtmosphericAerosol Research; RB: regional background

Sites Latitude Longitude m.a.s.l Type

Cabo de Creus 42°19′ N 03°19′ E 23 AQ-EMEPCoratxar 40°41′ N 00°05′ E 1200 AQ-RBMonagrega 40°57′ N 00°16′ E 600 AQ-RBMontseny 41°46′ N 02°21′ E 720 AQ-RB-

EUSAARZarra 39°05′ N 01°06′ W 885 AQ-EMEPSM de Palautordera 41°41′ N 02°26′ E 215 MeteorologyTagamanent 41°44′ N 02°16′ E 990 Meteorology

2.1.3. Cabo de Creus and ZarraThese stations are located on the North-Eastern coast close totheFrenchborder, and inamountainareanear theEasterncoast,respectively and belong to the Spanish Ministry of Environmentin the context of Co-operative Programme for Monitoring andEvaluation of the Long-Range Transmission of Air pollutants inEurope (EMEP). Levels of PM10 and PM2.5 are continuouslymeasured using high volume samplers (30 m3/h) withMCVPM1025 cut off inlets. Data are available since March 2001.

2.1.4. Montseny (MSY)This station is in the Montseny Natural Park 40 km to the NNEof the city of Barcelona, and 25 km from the Mediterraneancoast. The station is located in the top of a valley perpendi-cular to the coast in the Catalan Pre-Coastal Ranges. Levels ofPM10, PM2.5 and PM1 are continuously measured using realtime laser spectrometers (GRIMM 1107) and corrected with thefactors obtained by comparison with ‘in situ’ simultaneousPM10 and PM2.5 gravimetricmeasurements. 24h PM10 and PM2.5

samples were collected on quartz filters using high volumesamplers (30 m3/h) and DIGITEL cut-off inlets. Real time PM1

data were corrected with the optical/gravimetric PM2.5 factors.Hourly levels of wind direction and velocity, temperature,relatively humidity, pressure, precipitation and solar radiationare available fromnearbymeteorological stations belonging toMETEOCAT (Fig. 1). Even though the station is relatively farfrom urban and industrial agglomerations, this site could beaffected by anthropogenic emissions under specific meteor-ological scenarios given that the adjacent regions are denselypopulated and industrialized. Data are available since March2002.

PM levels are daily measured at Cabo de Creus and Zarra (aswell as at the other 8 EMEPmonitoring sites reported in Table 2)with gravimetry.AtMontseny this isperformedwithandopticalcounter, but the data are weekly corrected by comparison within situ and simultaneous PM gravimetric data. Finally, atMonagrega and Coratxar measurements are performed withTEOMandBeta attenuation instruments, but in situ comparisonwith gravimetric data performed during one an a half yearshowed very low differences. Consequently, differences on PMlevels due to the instrumentation are probably much reduced.

2.2. Additional analyses

In order to characterise the daily atmospheric scenarios withincidence on PM levels, complementary tools were used:

• NCEPmeteorologicalmaps (Kalnayetal., 1996) anddaily back-trajectories calculated byHYSPLIT4model (Draxler andRolph,2003) enabledus to interpret the different source regions of airmasses in the study area (see detailed methodology inEscudero et al., 2005). Daily 5 days back-trajectories werecalculated at 12h GMT for receptor points of 750, 1500 and2500 masl, also by modelling vertical velocity.

• The occurrence of African dust outbreaks was detected withthe aforementioned tools coupled with the informationobtained from different aerosol maps: Marine MeteorologyDivision of the Naval research Laboratory, USA (NRL) (http://www.nrlmry.navy.mil/aerosol); SKIRON surface dust concen-tration maps and simulations (http://forecast.uoa.gr, Kallos

Table 2 – PM2.5, PM10 and TSP mean annual levels at EMEP stations, Monagrega and Montseny, for 1996–2006, number ofmeasurements available for each year (N) and exceedances of the daily limit value of 50 µgPM10/m3 and arbitrary daily values of35 µgPM2.5/m3 and 70 µgTSP/m3

Station 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Mean

TSP N N70 µgTSP/m3

Viznar 28 30 36 42 44 41 39 37 18–49/yearNiembro 29 28 26 28 28 2–7/yearCampisábalos 18 15 20 22 17 18 2–14/yearCabo de Creus 33 41 37 40 35 38 6–35/yearBarcarrota 30 29 28 25 27 5–13/yearZarra 24 26 22 21 23 1–7/yearPeñausende 17 19 18 18 1–4/yearEls Torms 32 23 27 0–4/yearRiscollano 17⁎⁎ 20⁎⁎ 24⁎⁎ 20⁎⁎ 20⁎⁎ 23 23 21 4–7/yearO Saviñao 22 20 21 1–5/year

RatioTSP/PM10

PM10 N N50 µgPM10/m3

Viznar 1.8 15⁎ 17⁎ 20⁎ 23⁎ 24⁎ 24 21 21 24 22 20 21 21 12–28/yearNiembro 1.4 21⁎ 20⁎ 20 19 20 16 17 18 20 19 1–10/yearCampisábalos 1.5 11⁎ 9⁎ 13⁎ 14 11 12 13 12 12 12 0–10/yearCabo de Creus 1.8 21⁎ 19⁎ 20 19 25 21 21 19 19 21 0–15/yearBarcarrota 1.5 18⁎ 18⁎ 19 16 17 19 19 16 18 18 0–10/yearZarra 1.4 19⁎ 14⁎ 16 15 16 17 15 14 14 16 1–7/yearPeñausende 1.4 12⁎ 15 12 13 13 13 11 11 13 1–6/yearEls Torms 1.5 19 15 20 22 17 17 17 19 1–16/yearRiscollano 1.6 10⁎⁎ 12⁎⁎ 14⁎⁎ 12⁎⁎ 12⁎⁎ 15 12 14 16 15 13 13 2–11/yearO Saviñao 1.4 16 14 15 14 14 13 12 15 1–5/yearMonagrega 17 18 18 17 17 19 13 16 15 19 19 16 17 1–11/yearMontseny 21 21 19 14 14 14 17 0–8/year

Ratio PM10/2.5 PM2.5 N N35 µgPM2.5/m3

Viznar 1.9 12 10 9 11 11 10 11 11 0–3/yearNiembro 1.9 11 10 11 10 9 9 12 10 1–4/yearCampisábalos 1.5 9 7 7 8 8 8 8 0/yearCabo de Creus 1.6 12 13 17 13 12 10 10 13 2–15/yearBarcarrota 1.6 11 12 8 11 10 9 8 11 0–7/yearZarra 2.0 9 8 8 8 8 8 9 8 0/yearPeñausende 1.6 10 8 8 8 8 7 6 8 1–3/yearEls Torms 1.6 12 10 13 13 10 10 12 12 1–8/yearRiscollano 1.6 9 7 7 8 8 9 8 0–2/yearO Saviñao 1.7 12 9 9 9 10 9 8 9 1–2/yearMonagregaMontseny 1.3 16 15 15 11 11 11 13 0–3/year

⁎ Extrapolation of PM10 with a PST/PM10 ratio corresponding to the one measured at each site during 2001–2007.⁎⁎ S. Pablo de los Montes.

530 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 5 2 7 – 5 4 0

et al., 1997); BSC-ICoD/DREAM dust maps (http://www.bsc.es/projects/earthscience/DREAM/, Pérez et al., 2006); and fromthe satellite imagery provided by NASA SeaWiFS project(http://seawifs.gsfc.nasa.gov/SEAWIFS.html, McClain et al.,1998).

• Mixing Layer height was daily estimated at MSY by using thetools provided by NOAA-ARL available at: http://www.arl.noaa.gov/ready/amet.html.

Temporal trends were analysed at the 5 monitoring sitesand for all the size fractions available (TSP, PM10, PM2.5, PM1) bymeans of the nonparametric Mann–Kendall test for the trend

and the nonparametric Sen's method for themagnitude of thetrend. To this end, the MAKESENS template application (Salmiet al., 2002) was employed. The Mann–Kendall test is applic-able to the detection of amonotonic trend of a time series withno seasonal or other cycles, and therefore our test was appliedto mean annual concentrations of the different PM sizefractions. Moreover, to better understand the seasonality ofthe data, the test was also applied to datasets consistingseparately of mean PM concentrations for each month (e.g., adataset containing PM concentrations for January 1995,January 1996, January 1997, etc.). Comparison of the samemonth over the different years confirmed the monotonic

Fig. 2 –Meanmonthly temperature (°C) and precipitation (mm) registered at Tagamanent-PN-Montseny (METEOCAT) from 2002to 2007.

531S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 5 2 7 – 5 4 0

nature of the data and therefore the applicability of the Mann–Kendall test.

3. Atmospheric dynamics at MSY

The MSY area presents a typical Mediterranean climate withwarm summers, temperatewinters and irregular precipitationrates (Fig. 2). The average monthly temperatures registeredduring the studyperiod (2002–2007) ranged from1–7 °C inwinter(December, January and February) and 17–23 °C in summer(June, July andAugust). Annualmeanprecipitations varied from676 mm in 2004 to 882 mm in 2002, and also mean monthlyprecipitation levels varied from0mmin January 2005 to210mmin October 2005. The highest mean temperatures and lowestprecipitation rates were recorded in summer 2003, drier andwarmer than usual (1, 43 and 60 mm of rainfall and 21, 21 and23 °C were registered in June, July and August, respectively).Conversely, 2002 summer was characterized by the oppositeconditions (56, 49 and 123 mm of rainfall and 18, 18 and 17 °Cwere recorded in June, July and August, respectively).

In winter, the Azores anticyclone presents its lowestintensity and is usually placed to the West (Millán et al.,

Fig. 3 –Mean daily evolution of the mixing layer height (modelleTagamanent (METEOCAT).

1997; Martín-Vide & Olcina, 2001), favoring the entry ofAtlantic air masses to the WMB. Owing to the renewal of airmasses, the levels of atmospheric pollutants may be lower(Rodríguez et al., 2002, Escudero et al., 2007a).

In summer, the Azores high undergoes its highest intensityand is placed to the East and North (Millán et al., 1997), whereasthermal lows are developed over Iberia and the Sahara. Thisscenario favours a veryweak pressure gradient in theWMB andconsequently the local circulations dominate the atmosphericdynamics (Millán et al., 1997). The interaction of the sea andmountain breezes, the abrupt topography, the dominant North-Western flows at high atmospheric levels, the uplift of airmasses in the central IP and the compensatory subsidence overthe sea induce recirculation of air masses and the consequentageing and accumulation of pollutants. Furthermore, additionalfactors enhanced thehigher PMpollution levels across theWMBsuch as the high frequency of African dust outbreaks (Rodríguezet al., 2001, 2002; Escudero et al., 2005), the low rainfall, the highphotochemistry and the increased convective dynamics thatfavors resuspension.

During the colder months, the MSY station (720 masl) isgenerally located outside of the mixing layer and is thereforeaffected by the regional anthropogenic pollution with a low

d) at Montseny and wind speed at ground level measured at

532 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 5 2 7 – 5 4 0

frequency. Nevertheless, during several anticyclonic scenarios,highly polluted air masses from the coast and valleys aretransported towards MSY by the mountain/sea breeze. Duringthe warmer months, owing to the maximum height of themixing layer (Fig. 3), the MSY is frequently located within themixing layer beingaffectedby the regional pollution that reachesthestationveryoften. Inanycase thehighestdevelopmentof themixing layer occurs during the central hours of the day (Fig. 3).

Two main wind directions were recorded; the southerly(28%) and the westerly (17%). Both are conditioned by thetopography of the valley where the monitoring site is located.The southerly direction is registered when the mountain andsea breezes are developed, entering the valley from the South.The westerly direction corresponds to the intense advectionsfrom the North and North-West, channelled into the valleywith aWestern direction, and to the drainage flows during thenight (frequent in summer). Fig. 4 shows the typical dailycycles of wind direction and speed for each month. In winterthe main wind component is westerly, with relatively lowwind speeds. From April to October two wind directions arepredominant. During daytime, the southerly component isdominant corresponding to the breeze development and thehighest wind speed. Conversely, the westerly direction pre-vails during the night due to the drainage flows.

4. Results

4.1. Mean PM levels and comparison with the regionalbackground in Europe

Mean annual PM10 RB levels on the Mediterranean side of Iberiareached 16, 17, 17 and 21 µg/m3 at Zarra (period 2001–2007),

Fig. 4 –Annual evolution of hourlymeanwind direction (degrees)Palautordera (METEOCAT).

Monagrega (period 1996–2007), MSY (period 2002–2007) and Cabode Creus period (period 2001–2007). The levels recorded at Cabode Creus were relatively higher due to regional contribution ofpollutants from the urban and industrial hotspots surroundingBarcelona, Girona and towns in South-Eastern France. The otherthree sites showed very similar PM10 levels (16–17 µg/m3);however inter-annual variations, probably caused by inter-annual meteorological differences, are very important (14–17,13–19, 14–21 µg/m3 at Zarra, Monagrega and MSY, respectively).

When comparing theabove PM10 levelswith thosemeasuredat the Spanish EMEP stations (Table 2 and Fig. 4) the aforemen-tioned may be considered as intermediate. Thus, the stationslocated in the Atlantic and Central regions of Spain presentedrelatively low PM10 levels (12–15 µg/m3 as mean levels for OSaviñao, Riscollano, Campisábalos and Peñausende stations),whereas those measured in the Southern regions were higher(18 and 21 µg/m3 for Barcarrota and Viznar). Levels registered onthe Eastern side of Iberia were intermediate (16–17 µg/m3,excludingCabodeCreus). Thisdistribution ispossibly causedbythe increasingly higher frequency and intensity of African dustoutbreaks and the decreasing rainfall from the Atlantic regionsto the Eastern and to Southern regions of Iberia.

PM10 levels in the study area may also be regarded asintermediate when compared with those in other rural areasin Europe. Thus, taking the 2006 data from Airbase PM10 levelsrange from 15 to 27 µg/m3 in most rural areas in mostcountries of Central, Western and Eastern Europe (Fig. 5)whereas they range from 7 to 13 µg/m3 at rural sites inScandinavian countries (excluding DK, Fig. 5). The data fromthe study area are also intermediate when compared withthose from rural sites in the central (Italy and Cyprus) andeastern (Macedonia) Mediterranean countries, with PM10

ranging from 34 (Cyprus) to 14 (Macedonia) µg/m3 (Fig. 5).

andwind speed (m/s). 2002 and 2003 data fromSantaMaria de

Fig. 5 –Annual PM10 ranges and mean values measured at the study sites compared with the data available in Airbase 2006(http://air-climate.eionet.europa.eu/databases/airbase/airview/index_html) for rural sites of several countries of Europe.

533S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 5 2 7 – 5 4 0

As regards PM2.5 (Table 2), the mean annual levels rangedfrom 8 to 10 µg/m3 in the Atlantic and central Spain, from 10 to11 µg/m3 in southern Spain, and from 8 to 13 µg/m3 in easternSpain. It is clear that PM10 and PM2.5 do not show the samespatial variations over Spain. Thus, the highest PM2.5 levels arerecorded at MSY, Cabo Creus and Els Torms (Eastern Iberia),where the highest regional anthropogenic emissions arepresent, followed by Viznar and Barcarrota (Southern Iberia),where probably African dust has a higher incidence in PM2.5

levels. Most of the remaining monitoring sites recorded levelsclose to 8 µg/m3.

The ratio PM2.5/PM10 reaches the lowest values in theCanary Islands (Querol et al., 2008) and Southern IP, with 0.4and 0.5, respectively. In most of the other areas of the IP, thePM2.5/PM10 ratio ranges from 0.6 to 0.7, with the exception ofthe industrialized regions (such the Barcelona region), wherethe RB ratio reaches 0.8 (MSY). Again the coarser grain sizedistribution trend shown in Table 2 for the Southern sites isprobably the result of the highermineral load in PM10 owing tothe drier climate (especially in summer) and the proximity tothe African desert regions. Although sea spray may reachrelatively high levels in the coastal areas, at MSY sea spraylevels are expected to be relatively low due to the distance tothe coast, thus the coarse features are attributed to highermineral dust levels in this specific location.

PM1 levels have been continuously measured at MSY since2002. Mean annual levels were 11 µg/m3, ranging from 9 to13 µg/m3. The PM1/PM2.5 ratio reached 0.8 constantly as anannual mean value for each of the 6 years of measurement.Similar PM2.5 and PM1 levels and PM1/2.5 ratios were reportedfor the rural site of Bemantes in Northwest Spain for the year2001 by Salvador et al. (2007). PM1 levels measured at MonteCimone in Italy were 7 µg/m3, which were 30% lower than

those recorded at MSY (Marenco et al., 2006). This is probablybecause Monte Cimone can be considered as a remote site(2165 masl), but MSY represents the RB.

4.2. Inter annual trends

Temporal trend analyses were applied to the annual andmonthly datasets available between 1995 and 2007 from thefive monitoring sites (Cabo de Creus, Zarra, Coratxar, Mona-grega and MSY), with different data coverage (6 to 13 years)depending on the site (Table 2).

The Mann–Kendall tests showed no significant temporaltrends for annual or monthly PM concentrations for Cabo deCreus (PM10 and PM2.5), Zarra (PST, PM10, PM2.5) and Monagrega(PM10). Detectable, although not highly significant trends (levelof significance α=0.1) were observed for Cabo de Creus (TSP) forthe month of June and Coratxar (TSP) for March.

Conversely, more significant and robust results wereobtained for the MSY site, despite the relatively shorter dataset(2002–2007). Steadily decreasing trendsweredetected for annualmean concentrations but also for different months. These wereconsistent for the different PM size fractions (PM10, PM2.5, PM1).Mean annual PM10 and PM2.5 concentrations showed an averagedecrease of 40% and 34%, respectively, for the entiremonitoringperiod (at 0.05 significance level). In absolute terms this decreaseis equivalent to 7 µgPM10/m3 and 5 µgPM2.5/m3 between 2002 and2007, and these trends may be considered as significant (Fig. 6).In the fine fraction (PM1) a somewhat smaller decrease wasdetected (28%) even though it was not significant (αN0.1). Basedon the observation of Aerosol Optical Depth index obtained byMODIS, Papadimas et al. (2008) detected similar decreasingtrends for the North-West of the IP. The different trendsobserved in this study of the coarser (PM10, PM2.5) and fine

Fig. 7 –Mean summer and winter daily evolution of meanhourly PM levels measured at Montseny.

Fig. 6 –Temporal trends for PM10 and PM2.5 detected at Montseny by means of the Mann–Kendall test and Sen's method usingMAKESENS (Salmi et al., 2002).

534 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 5 2 7 – 5 4 0

(PM1) fractions suggest a variety of causes related to large-scalemeteorological processesor cycles, and/or to local ormeso-scaleprocesses such as nearby anthropogenic emission sources.

The analysis of the monthly datasets provided evidence ofdecreasing temporal trends for specificmonths of the year. PMlevels of all grain size fractions at MSY showed the strongestand most consistent decrease during March for the entirestudy period, with a reduction of 66% in PM10, 65% in PM2.5 and60% in PM1 (α=0.05). June and July also showed similar trendsin PM10 and PM2.5 although not in PM1 (an average decrease of48–51%, α=0.05). No significant trends were observed for thewintermonths (November through January) for any of the sizefractions studied. The fact that significant decreasing trendswere detected during different seasons of the year (March,June–July), and that no trends were observed in others(winter), indicates that there is no single cause behind thesetrends, but rather a combination of factors. The differentdegree of influence of African dust outbreaks, the influence ofatmospheric processes such as sea-breeze circulations (typicalof summer) or pollutant concentration on the decreasingboundary layer height (winter) would generate differenttrends in different seasons. In addition, the influence oflocal-scale anthropogenic emission sources could be deducedfor the winter period, given that the no clear trends weredetected for any of the size fractions.

4.3. Daily variability of PM levels

The daily evolution of PM levels at MSY depends strongly onthe breeze circulation (mountain and sea breezes) thatdominates the atmospheric dynamics at this site (Fig. 7).During the night (00–07 h GMT), relatively low PM levels wererecorded, coinciding with the western flow, with mean levelsfor the period of 12, 9 and 6 µg/m3 for PM10, PM2.5 and PM1, inwinter and 15, 12 and 9 µg/m3 in summer. From early morning(7–8 h GMT in summer, 9–10 h GMT inwinter) to afternoon, PMlevels increase progressively, coinciding with the Southernflow, reaching the highest values at around 13–14 h GMT insummer and 15–16 h GMT in winter (Fig. 7). Mean maximumhourly levels of 20, 16 and 13 µg/m3 in winter and of 23, 16 and13 µg/m3 in summer were registered for PM10, PM2.5 and PM1,respectively. Fig. 7 shows that the hourly levels of the differentgrain size fraction ranges correlate relatively well, although

during African episodes the correlation may significantlychange from the mean values (Fig. 7). The later occur with arelatively low frequency (12 to 24% of the days depending ofthe year) compared to other days (76 to 88%) when PM fromregional and local sources prevail.

The above described daily trend may be attributed to thefollowing processes: early in the day atmospheric pollutantsare accumulated in the pre-coastal depression (highly popu-lated and industrialized, crossed by important main roads);subsequently the diurnal breeze development (activated byinsolation) increases the PM levels at MSY by transporting theaged air masses upwards from the valley. The influence ofcleaner nocturnal drainage flows and the decrease of themixing layer height result in the lower nocturnal PM levels.These lower levels are measured in the nocturnal to early

Fig. 8 –Annual relative frequencies of each airmass origin selected for the study region for the period 2002–2007. Airmass originconsidered: Atlantic advection (ATL); African dust outbreaks (NAF); Mediterranean (MED); European (EU); Regional recirculation(REG); and Anticyclonic (ANT).

535S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 5 2 7 – 5 4 0

morning period due to the combination of the 720 masl of theMSY station and the very lowmixing boundary layer height ofthis period; these causing, for this period of the dayconsidered, the decoupling of MSY site from the mixingboundary layer. Consequently, the low levels measured maybe representative of the continental background. In manyaerosol studies performed in mountain areas PM levels aremeasured only during the nocturnal period to diminish theregional/local influence from mountain breezes and by thegrowth of themixing boundary layer. Formost days of the yearthis situation is prevailing, however for a few days reservestrata with aged pollution aerosols or African dust outbreaksmay increase also the nocturnal aerosol levels, but these arealso influencing the continental background of the area. Weconsider in this paper the regional–continental background asthe mean background levels of aerosols of a region with adiameter of around 500 km (or European PM baseline levels inthis region), whereas the regional–local scale would have a50 km size.

Thus, the above continental background is characterised bylevels of PM10, PM2.5 and PM1 reaching annual mean values of13, 10 and 8 µg/m3 (with slightly higher and lower summer andwinter levels). The results are compatible with the mea-surements performed at remote sites in Central and North-ern Spain (EMEP stations), with mean levels of 12–13 and 8–10 µg/m3 for PM10 and PM2.5. If these values are subtractedfrom the annual mean levels, the annual mean regionalcontributions may be quantified in around 4, 3 and 2 µg/m3 forPM10, PM2.5 and PM1. For the fine fractions these contributionsin winter are twice as large as those in summer probablybecause of the occurrence of the anticyclonic pollutionepisodes (see description below).

Table 3 – Mean levels of PM fractions at Montseny (period 20advection (NA, North; NWA, Northwest; WA, West; SWA, SouthEuropean (EU); Regional recirculation (REG); and Anticyclonic (A

µg/m3 NA NWA WA S

Montseny PM10 14 15 12PM2.5 12 12 9PM1 10 10 8

The inter-annual evolution of the continental backgroundlevels show a clear decrease from 18, 13 and 9 µg/m3 (2002) to12, 9 and 7 µg/m3 (2007) for PM10, PM2.5 and PM1, respectively,whereas the regional contributions do not show this decreas-ing trend. The Mann–Kendall test was applied to the con-tinental background mean levels calculated for MSY, showingimportant results at 0.05 significance level for the PM10

fraction (average decrease of 36% for the entire monitoringperiod). In absolute terms this decrease is equivalent to6 µgPM10/m3 between 2002 and 2007. PM2.5 and PM1 showedalso decreasing, but not significant, trends of 51 and 28%,respectively (equivalent to 4 µgPM2.5/m3 and 2 µgPM1/m3). Thistrend is significant for PM10 (60%) and PM2.5 (38%) if only thesummer months are considered. In the fine fraction (PM1) asimilar decrease was detected (38%) even though it was notsignificant (αN0.1).

Day to day variation of PM levels at MSY is driven by theconcatenation of different meteorological scenarios that mayincrease or reduce the levels and size of atmosphericparticulate matter in North-Eastern Iberia. In line with priorstudies these scenarios are as follows: Atlantic advectiveconditions, African dust episodes, regional recirculationepisodes, European and Mediterranean transport episodes,and local pollution episodes (Millán et al., 1997; Rodríguezet al., 2002, 2003; Escudero et al., 2005, 2007a). Escudero et al.(2007a) defined the summer anticyclonic episodes, equivalentto the regional recirculation episodes, with impact on the PM10

levels at RB sites, and the winter anticyclonic episodes,equivalent to the local pollution episodes. The latter have amajor influence in urban and industrial areas but a reducedinfluence at rural sites owing to the high stability of thesescenarios, giving rise to the frequent formation of near ground

02 to 2007) for each air mass origin considered: Atlanticwest); African dust outbreaks (NAF); Mediterranean (MED);NT)

WA NAF MED EU REG ANT

11 24 13 16 20 199 16 11 13 15 167 12 9 11 12 14

536 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 5 2 7 – 5 4 0

inversion layers (Rodríguez et al., 2003; Escudero et al., 2007a).Nevertheless, in this study the winter anticyclonic pollutionepisodes were taken into account given the relative proximityof large urban and industrial agglomerations. Frequencies ofthe episodes that affect the study area were calculated for the2002–2007 period (Fig. 8).

In order to evaluate the influence of the different meteor-ological scenarios affecting PM RB levels in the WMB, the timeseries of PM10, PM2.5 and PM1 were classified in the aboveatmospheric scenarios according to the origin of the airmasses. Mean PM levels obtained for each episode consideredare reported in Table 3. Atlantic and Mediterranean episodesare related to low PM levels because of the renovation of airmasses and the occurrence of precipitations. In contrast,African dust outbreaks, European episodes and regional re-circulations may increase PM concentrations at RB sites. Aswill be shown below, the winter anticyclonic episodes maymarkedly increase the PM levels even in regional areas.

The Atlantic advection prevailed along the year at MSY,with a range of frequency between 36% (2006) and 53% (2002)days/year (mainly from NW Atlantic regions, 17%), withepisodes lasting from 2 to 10 consecutive days. Annual meanlevels (2002–2007) registered under Atlantic advection at MSYreached 14, 11 and 9 µg/m3 of PM10, PM2.5 and PM1, re-spectively. PM under this scenario has a typical fine grain size(with both PM2.5/PM10 and PM1/PM2.5 ratios reaching valuesclose to 0.8).

The transport ofMediterraneanairmasses towardsMSYwasthe least frequent scenario with a frequency of 2% (2002) to 7%(2006) of thedays/year and ameandurationof 2–3days/episode.This scenario is normally associated with strong winds andrainfalls, causing the decrease in PM levels. Consequently,annual mean levels were 14, 12 and 10 µg/m3 of PM10, PM2.5 andPM1, respectively, with prevalent fine grain size particles (PM2.5/PM10 and PM1/PM2.5 ratios close to 0.8).

Africandust airmassesmay transportmineral dust fromaridareas locatedoverNorthAfrica, resulting inamarked increase inPM levels at the IP (Querol et al., 1998a,b). Dust outbreaksfrequently occur in February–March and June–October (Rodrí-guez et al., 2001; Escudero et al., 2005), although sporadicepisodes were detected along the entire year. The inter-annualfrequency of this kind of episode varied considerably at MSY,ranging from 12% (2005) to 24% (2003) days/year (Fig. 8), with amean duration from 2 to 5 days/episode. Annual mean levels

Fig. 9 –PM10, PM2.5 and PM1 levels registered at Montseny duringmap from 4/02/2004.

registered under this scenario were 24, 16 and 12 µg/m3 of PM10,PM2.5 and PM1, respectively. Owing to the typical coarse size ofthe mineral dust particles, coarse grain size distribution wasobtained (PM2.5/PM10 ratios close to 0.6 and PM1/PM2.5 close to0.8). The highest PM10 levels were registered under this scenario(Table 3), and high PM2.5 and PM1 levels were also recorded. TheAfrican dust contribution to the annual mean PM levels wasquantified by applying the method proposed by Escudero et al.(2007b). To this end, the daily PM10 RB levels were obtained byapplying a monthly moving 30th percentile to the PM10 timeseries at MSY for days without African dust transport. There-after, the daily PM RB levels obtained were subtracted from thedaily PM levels recorded at MSY only on days when African dustoutbreaks occurred, the difference being the daily net Africandust load. Thus, the annual mean contributions of the Africandust outbreaks calculated for 2002–2007 in PM10, PM2.5 and PM1

were 1.7, 0.7 and 0.4 µg/m3, respectively. The highest contribu-tionsweredetermined in2003,with4.9, 2.4 and1.7 µg/m3andthelowest from2005 to2007,with0.9, 0.3 and0.2 µg/m3 inPM10, PM2.5

and PM1, respectively. These results reflect the variability of thefrequency and intensity of African dust episodes. Thus thehighest frequencywas obtained in 2003 (24%of thedays) and thelowest in 2005 (12%of the days). Themean dust contribution perday/African episode was calculated for each year, with amaximum contribution of 15 µgPM10/m3 for 2003 and 2004, anda minimum of 10 µgPM10/m3 for 2006 and 2007.

The regional re-circulation episodes are developed fromMay to October, with the highest frequency in mid summer(Millán et al., 1997; Rodríguez et al., 2002). These episodesoccur over the WMB during low pressure gradient situationsand are characterised by the re-circulation and ageing of theair masses over the WMB (Rodríguez et al., 2002). Theseepisodes account for between 14% (2004 and 2007) and 20%(2006) days/year, lasting from 2–3 days to more than a week.Annual mean levels registered under this scenario wererelatively high, with mean values of 20, 15 and 12 µg/m3 ofPM10, PM2.5 and PM1, respectively (Table 3). Fine particles werepredominant, with PM2.5/PM10 and PM1/PM2.5 mean ratios of0.8, given that the formation of secondary aerosols isenhanced under this meteorological scenario.

European air masses may occasionally increase PM levelsat MSY due to the long range transport of pollutants fromCentral Europe. The frequency of these episodes rangedbetween 5% (2006) and 9% (2005) days/year, with a duration

an anticyclonic episode in February 2004. Sea level pressure

537S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 5 2 7 – 5 4 0

of a 2–4 days. Mean PM levels were very close to the annualmean levels with mean values of 16, 13 and 11 µg/m3 of PM10,PM2.5 and PM1, respectively (Table 3). The fine grain sizedistribution (PM2.5/PM10 ratio 0.8 and PM1/PM2.5 0.9) is aconsequence of the relatively high proportion of secondaryinorganic aerosol content in these air masses.

Weak gradient anticyclonic conditions in wintertime wererecurrent over the WMB with a frequency from 7% (2004) to16% (2007) days/year, and a mean duration of 4–8 days. Theabsence of intense advections favours the stagnation ofpollutants around populated and industrialized areas (mostlyconcentrated in the coastal and pre-coastal depressions in the

Fig. 10 –Top: Daily PM10, PM2.5 and PM1 recorded at Montseny forline) and the occurrence of African dust outbreaks (NAF, rhombuMSY. Bottom: Daily (thin line) and 30 daysmoving average (thickof African dust outbreaks (NAF, rhombus).

study area) inducing a sharp increase in PM concentrationsover these areas. Owing to the thermal inversions associatedwith these anticyclonic scenarios higher temperatures aremore frequent in the mountain areas (i.e. at MSY) than in thedepressions. After some days under an anticyclonic situationlocal slope breezes can be activated by solar radiation pushingpolluted air masses from the valley (Vallès industrial area andBarcelona metropolitan area) towards rural areas, therebymarkedly increasing the PM levels. At MSY mean PM10, PM2.5

and PM1 levels under anticyclonic scenarios reached 19, 16 and14 µg/m3, respectively. These anticyclonic episodes arecharacterized by a fine PM grain size (PM2.5/PM10 ratio 0.9

2002–2007, showing the EU daily PM10 limit value (horizontals). Middle: Mean monthly PM10, PM2.5 and PM1 recorded atline) PM2.5/10 ratios measured at MSY showing the occurrence

538 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 5 2 7 – 5 4 0

and PM1/PM2.5 0.9) because of the predominance of secondaryinorganic compounds and anthropogenic carbonaceous spe-cies. Fig. 9 shows the variability of PM levels at MSY under ananticyclonic scenario in February 2004. Very low backgroundlevels (close to 5 µgPM10/m3, hourly basis) are recorded duringthe late evening and night. At midday, levels of fine PMparticles show a sharp increase (up to 100 µgPM10/m3 and70 µgPM1/m3, hourly basis) owing to the transport of PMpollutants from the pre-coastal depression. During some days,PM levels may be high even during the night, probably due tothe influence of stable polluted layers upwards injected duringthe day (see data from the end of the episode in Fig. 9). Theseepisodes account for the highest PM1 concentrations recordedat MSY, both in hourly and daily basis, and consequently havea great impact on PM in the elevated rural areas of the WMB.

4.4. Seasonal evolution of PM levels

PM levels at MSY follow a clear seasonal trend. The highestlevelsare registeredduring the summermonths (especially Juneand July). This summer increase is associated with the highestfrequency of African dust outbreaks, the recirculation of air

Fig. 11 –Daily PM10 recorded at Montseny for 2004, showing theAfrican dust outbreaks (NAF), Atlantic advection (ATL), MediterraAnticyclonic (ANT) episodes. Moving average PM10 (grey line) refair masses in the bottom figure.

masses that prevent air renovation, the low precipitationsregistered, the highest resuspension owing to the dryness ofsoils during this period and the formation of secondary aerosolfrom gaseous precursors caused by the maximum solarradiation (Querol et al., 2001a,b; Viana et al., 2002; Escuderoet al., 2005). The summer increase is more pronounced in thecoarse fraction (PM2.5–10) than in the finer fractions (PM1–2.5 andPM1) because of the coarser size of the mineral dust and theoccurrence of coarse summer-nitrate particles (Fig. 10). Even iflevels of nitrate are usually higher in winter, levels of coarse Caand Na nitrate are usually higher in summer as a consequenceof the interactionofgaseousnitricacidwithsea salt andmineraldust, asdescribedbyQuerol et al. (2001a,b) for urbanareasof thestudy area. In the remaining months of the year the levels arerelatively low owing to the high frequency of Atlantic advectionand precipitation rates. However, a less pronounced seasonalmaximum is usually recorded in the winter months (fromNovember to March) when PM levels rise because of intensepollution episodes of anthropogenic (winter anticyclonic sce-narios) andnatural (Africandust) origins. Theseasonal variationof PM levels at MSY is also influenced by the evolution of thethicknessof theboundary layer. Thus,when themonitoring site

EU daily PM10 limit value (bold line) and the occurrence ofnean (MED), European (EU), Regional recirculation (REG) andlects the background PM10 levels. See sectors for the origin of

539S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 5 2 7 – 5 4 0

is outside the boundary layer (especially in winter), it is lessaffected by regional anthropogenic emissions. In contrast, thehigh vertical development of the boundary layer in summerallows PM pollutants to reach the monitoring site.

The occurrence of the above seasonal trend (characterizedby summer and February–March PM maxima) may vary fromone year to another (Fig. 10), depending on the intensity of thepollution and African dust episodes in winter, and on thedryness and frequency of dust episodes in summer. Thus, thehigh PM levels recorded in the summer 2003 were due to theheat wave and to the very frequent and intense African dustepisodes (Figs. 8 and 10). The relatively low PM levels recordedin the summers 2002 and 2005 were related to the highprecipitation (2002) and to the low frequency and intensity ofdust outbreaks (2005) (Figs. 8 and 10). However, 2004 and 2005underwent intense February–March dust outbreaks, consider-ably raising the PM levels in winter (Fig. 10).

This seasonal evolution of atmospheric scenarios also playsa major role in the PM2.5/10 and PM1/2.5 ratios. Thus, the typicalwinter PM2.5/10 reach values of 0.9–0.8 (Fig. 10) owing to the finePM size of the low PM episodes under Atlantic advectionscenarios and to the high (and fine) PM pollution episodes inwinter. Sporadically, these ratios decrease to 0.3 in accordancewith the occurrence of intensewinter dust outbreaks (Fig. 10). Insummer, PM2.5/10 fall to mean values of 0.7 (due to higher levelsof background dust and coarse nitrate), with values reaching 0.3in intense summer dust outbreaks. Summer PM2.5/10 maydecrease to mean values of 0.5 for specific years such as the2003 summerheatwaveormay reach0.7 inyearswithvery lightsummer conditions (2005) (Fig. 10).

5. Conclusions

PM10 RB levels along the Spanish Mediterranean coast may beregarded as intermediate when compared with those obtainedat other backgroundsites in the IP, NorthernandEastern Europeand the East Mediterranean basin. However, PM2.5 levels arehigher at the Mediterranean sites, reflecting a greater influenceof anthropogenic emissions. Regarding the mean PM levelsregistered in the 2002–2007 period (17, 13 and 11 µg/m3 of PM10,PM2.5 andPM1, respectively), theMSY site could be considered asrepresentative of the RB PM pollution in the WMB.

Atmospheric dynamics at MSY (elevated rural areas in theWMB) are dominated by mountain and sea breezes whichregulate the daily evolution of PM levels. During the night thenocturnal drainage flows and the decrease in the mixing layerheight result in lower PM levels at the RB sites. PM levelsmeasured during the night allowed us to estimate the con-tinental background levels in 12–13, 8–10 and 6–8 µg/m3 of PM10,PM2.5 and PM1, respectively.

Day to day variation of PM levels in theWMB is governed bythe concatenation of different meteorological scenarios, affect-ing the PM concentrations and size distribution of PM. Theseasonal distribution of these episodes together with theclimatic patterns of the WMB gives rise to a marked seasonalpattern for PM background (see 2004 as an example, Fig. 11).Maximum PM10 background levels are recorded in summer,associated with the elevated frequency of African dust out-breaks, the recirculation of airmasses over theWMB, the lowest

precipitations, the intense resuspension due to the dryness ofsoils, the enhanced formation of coarse secondary aerosols andthe highest development of the boundary layer. Secondarymaximums are observed in February–March and Novemberattributed to Saharan dust episodes and winter anticyclonicscenarios. Minimum values are recorded during the rest of theyear owing to the higher frequency of Atlantic advection andprecipitation rates. This background trend is overlapped by PMpeak episodes,mainly attributed to Saharan dust outbreaks andanticyclonic episodes.

This study shows that the winter anticyclonic scenarios,with a known impact on PM levels in urban and industrialareas, could have a notable influence on the PM levelsrecorded at elevated RB areas in the WMB. These pollutionepisodes mainly occur from November to March when lowtemperatures are registered at night and relatively hightemperatures are reached at noon, favouring the activationof mountain winds that transport polluted air masses fromnearby urban and industrialized areas. Under these scenarios,the hourly levels of PM1 at MSY may reach 75 µg/m3.

A significantdecrease inPM10 andPM2.5was recordedatMSY(in contrast to PM1) from 2002 to 2007. Decreasing temporaltrends were registered for March and June and July, but not forthe winter period. Mean annual PM10 and PM2.5 concentrationsshowed an average 40% and 34% decrease for the entiremonitoring period. In absolute terms this decrease is equivalentto 7µgPM10/m3and5µgPM2.5/m3. Similardecreasing trendsweredetected by Papadimas et al. (2008) for the North-West of theIberian Peninsula based on MODIS data. The different trendsobserved in our study for the coarse (PM10, PM2.5) and fine (PM1)fractions suggest a variety of causes related to large-scalemeteorological processesor cycles, and/or to local ormeso-scaleprocesses such as nearby anthropogenic emission sources.

The inter-annual evolution of the continental backgroundlevels (night levels at MSY) show a clear decrease from 18, 13and 9 µg/m3 (2002) to 12, 9 and 7 µg/m3 (2007) for PM10, PM2.5

and PM1, respectively, whereas the regional contributions donot show this decreasing trend. This trend could be partiallyattributed to the variability of the frequency and intensity ofthe Saharan dust episodes during this relatively short period(2002–2007). Thus, the highest mean annual African dustcontribution was registered in 2003, and the lowest from 2005to 2007. Nevertheless, this trend to lower contributions ofSaharan dust does notwholly account for the decreasing trendobserved for PM10 and PM2.5. Chemical characterisation ofPM10 and PM2.5 carried out at the MSY site since 2002 couldprovide some insight into the causes of this trend.

Acknowledgements

This study was supported by the Ministry of Science andInnovation (CGL2005-03428-C04-03/CLI, CGL2007-62505/CLI,GRACCIE-CSD2007-00067), the European Commission (6thframework CIRCE IP, 036961, EUSAAR RII3-CT-2006-026140).The authors would also like to acknowledge NASA/GoddardSpace Flight Center, SeaWIFS-NASA Project, University ofAthens, Navy Research Laboratory — USA and the BarcelonaSuper-Computing Centre for their contribution with TOMSmaps, satellite images, SKIRON dust maps, NAAPs aerosol

540 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 4 0 7 ( 2 0 0 8 ) 5 2 7 – 5 4 0

maps, and DREAM dust maps, respectively. The authorsgratefully acknowledge METEOCAT for the meteorologicaldata and the NOAA Air Resources Laboratory (ARL) for theprovision of the HYSPLIT transport and dispersionmodel and/or READY website (http://www.arl.noaa.gov/ready.html ) usedin this publication. Finally, wewould like express our gratitudeto Jesús Parga for his technical support.

R E F E R E N C E S

Draxler R.R., Rolph G.D. (2003). HYSPLIT (HYbrid Single-ParticleLagrangian Integrated Trajectory) Model access via NOAA ARLREADY Website (http://www.arl.noaa.gov/ready/hysplit4.html). NOAA Air Resources Laboratory, Silver Spring, MD.

Escudero M, Castillo S, Querol X, Avila A, Alarcón M, Viana MM,Alastuey A, Cuevas E, Rodríguez S. Wet and dry African dustepisodes over Eastern Spain. J Geophys Res 2005;110(D18S08):10.1029.

Escudero M, Querol X, Ávila A, Cuevas E. Origin of the exceedancesof the European daily PM limit value in regional backgroundareas of Spain. Atmos Environ 2007a;41:730–44.

Escudero M, Querol X, Pey J, Alastuey A, Pérez N, Ferreira F, AlonsoS, Rodríguez S, Cuevas E. A methodology for the quantificationof the net African dust load in air quality monitoring networks.Atmos Environ, (Technical note) 2007b;41:5516–24.

Gangoiti G, Millán MM, Salvador R, Mantilla E. Long rangetransport and re-circulation of pollutants in the WesternMediterranean during the RECAPMA Project. Atmos Environ2001;35:6267–76.

IPCC. Climate Change 2007: The Physical Science Basis.Contribution of Working Group I to the Fourth AssessmentReport of the IPCC (ISBN 978 0521 88009-1 Hardback; 978 052170596-7 Paperback); 2007.

Jorba O, Pérez C, Rocadenbosch F, Baldasano JM. Cluster analysis of4-day back trajectories arriving in the Barcelona Area (Spain)from 1997 to 2002. J Appl Meteorol 2004;43(6):887–901.

Kallos G, Kotroni V, Lagouvardos K. Proceedings of the Symposiumon Regional Weather Prediction on Parallel computerEnvironments. Greece: University of Athens; 1997. p. 109–22.

Kalnay E, Kanamitsu M, Kistler R, Collins W, Deaven D, Gandin L,Iredell M, Saha S, White G, Woollen J, Zhu Y, Chelliah M,Ebisuzaki W, Higgins W, Janowiak J, Mo KC, Ropelewski C,Wang J, Leetmaa A, Reynolds R, Jenne R, Joseph D. The NCEP/NCAR 40-Year Reanalysis Project. Bull Amer Meteor Soc1996;77:437–71.

Marenco P, Bonasoni F, Calzolari M, Ceriani M, Chiari CristofanelliP. Characterization of atmospheric aerosols at Monte Cimone,Italy, during summer 2004: source apportionment andtransport mechanisms. J Geophys Res 2006;111:D24202.

Martín-Vide J, Olcina J. Climas y tiempos de España. AlianzaEditorial; 2001. 258 p.

McClain CR, Cleave ML, Feldman GC, Gregg WW, Hooker SB,Kuring N. Sea Technol 1998;39:10–5.

MillánM, Salvador R, Mantilla E, Kallos G. Photo-oxidant dynamicsin the Mediterranean basin in summer: results from Europeanresearch projects. J Geophys Res 1997;102:8811–23.

Papadimas CD, Hatzianastassiou N, Mihalopoulos N, Querol X,Vardavas I. Spatial and temporal variability in aerosolproperties over the Mediterranean basin based on 6-year(2000–2006) MODIS data. J Geophys Res-Atmos 2008;113:D11205. doi:10.1029/2007JD009189.

Pérez C, Sicard M, Jorba O, Comeron A, Baldasano JM. Summertimere-recirculations of air pollutants over the North-EasternIberian coast observed from systematic EARLINET lidarmeasurements in Barcelona. Atmos Environ2004;38:3983–4000.

Pérez C, Nickovic S, Pejanovic G, Baldasano JM, Özsoy E. Interactivedust-radiation modelling: a step to improve weather forecasts.J Geophys Res 2006;111:D16206. doi:10.1029/2005JD006717.

Pope CA, Dockery DW. Health effects of fine particulate airpollution: lines that connect. J Air Waste Manage Assoc2006;56(6):709–42.

Querol X, Alastuey A, Puicercus JA, Mantilla E, Ruiz CR, Lopez-SolerA, Plana F, Juan R. Seasonal evolution of suspended particlesaround a large coal-fired power station: chemicalcharacterisation. Atmos Environ 1998a;32:719–31.

Querol X, Alastuey A, Puicercus JA, Mantilla E, Miro JV, Lopez-SolerA, Plana F, Artíñano B. Seasonal evolution of suspendedparticles around a large coal-fired power station: particulatelevels and sources. Atmos Environ 1998b;32:1963–78.

Querol X, Alastuey A, Rodríguez S, Plana F, Mantilla E, Ruiz CR.Monitoring of PM10 and PM2.5 around primary particulateanthropogenic emission sources. Atmos Environ2001a;35:845–58.

Querol X, Alastuey A, Rodríguez S, Plana F, Ruiz CR, Cots N,Massagué G, Puig O. PM10 and PM2.5 source apportionment inthe Barcelona Metropolitan Area, Catalonia, Spain. AtmosEnviron 2001b;35/36:6407–19.

Querol X, Alastuey A, Moreno T, Viana MM, Castillo S, Pey J,Rodríguez S, Artiñano B, Salvador P, Sánchez M, Garcia DosSantos S, Herce Garraleta MD, Fernandez-Patier R,Moreno-Grau S, Minguillón MC, Monfort E, Sanz MJ,Palomo-Marín R, Pinilla-Gil E, Cuevas E, de La Rosa J, Sanchezde La Campa A. Spatial and temporal variations in airborneparticulate matter (PM10 and PM2.5) across Spain 1999–2005.Atmos Environ 2008;42:3964–79.

Rodríguez S, Querol X, Alastuey A, Kallos G, Kakaliagou O. Saharandust contributions to PM10 and TSP levels in Southern andEastern Spain. Atmos Environ 2001;35:2433–47.

Rodríguez S, Querol X, Alastuey A, Mantilla E. Origin of high PM10

and TSP concentrations in summer in Estern Spain. AtmosEnviron 2002;36(19):3101–12.

Rodríguez S, Querol X, Alastuey A, Viana MM, Mantilla E. Eventsaffecting levels and seasonal evolution of airborne particulatematter concentrations in the Western Mediterranean. EnvironSci Technol 2003;37:216–22.

Salmi T, Määttä A, Anttila P, Ruoho-Airola T, Amnell T. Detectingtrends of annual values of atmospheric pollutants by theMann–Kendall test and Sen's slope estimates — the Exceltemplate applicationMAKESENS. In Publications on Air QualityNo. 31 (ed. FinnishMeteorological Institute); 2002. p. 35. FinnishMeteorological Institute.

Salvador P, Artíñano B, Querol X, Alastuey A, Costoya M.Characterisation of local and external contributions ofatmospheric particulate matter at a background coastal site.Atmos Environ 2007;41:1–17.

Soriano C, Baldasano JM, ButtlerWT,Moore K. Circulatory patternsof air pollutants within the Barcelona Air Basin in asummertime situation: lidar and numerical approaches.Boundary-Layer Meteorol 2001;98(1):33–55.

Viana M, Querol X, Alastuey A, Cuevas E, Rodríguez S. Influence ofAfrican dust on the levels of atmospheric particulates in theCanary Islands air quality network. Atmos Environ2002;36:5861–75.