Atmospheric Environment 42 (2008) 65236534

Contents lists available at ScienceDirect

Atmospheric Environment
journal homepage: www.elsevier.com/locate/atmosenv

Inuence of sea breeze circulation and road trafc emissions on the relationship between particle number, black carbon, PM1, PM2.5 and PM2.510 concentrations in a coastal city
guez a, b, *, Emilio Cuevas a, Yenny Gonza lez a, Ramo n Ramos a, Sergio Rodr Pe rez c, Xavier Querol c, Andre s Alastuey c Pedro Miguel Romero a, Noem
a a Atmospheric Research Centre, AEMet, Associated Unit to CSIC Studies on Atmospheric Pollution, C/ La Marina, 20, Izan 6a Planta, E38001 Santa Cruz de Tenerife, Canary Islands, Spain b Department of Geology, University of Huelva, Associated Unit to CSIC on Air Pollution, Campus El Carmen, E21071 Huelva, Spain c i Sabaris S/N, E08028 Barcelona, Spain Institute of Earth Science Jaume Almera, CSIC, Lluis Sole

a r t i c l e i n f o
Article history: Received 19 February 2008 Received in revised form 26 March 2008 Accepted 2 April 2008 Keywords: Ultrane particles New particle formation Black carbon Soot Vehicle exhausts

a b s t r a c t
The physical characterisation of metrics representative of ambient air particle concentration is becoming a topic of great interest for urban air quality monitoring and human exposure assessment. In this article, the inuence of sea breeze circulation and primary road trafc emissions on the relationship between the urban aerosol number (N3, particles >3 nm), black carbon <10 mm (BC), PM1, PM2.5, PM2.510 (PMx) concentrations was studied in a coastal city (Santa Cruz de Tenerife, Canary Islands). The daily cycles of sea and mountain breezes (inland during daylight and seaward at night) and road trafc emissions exerted a great, and well differentiated, inuence on the BC, N3 and PMx concentrations. In this scenario, the following major aerosol features were observed: (1) fresh vehicle exhaust emissions resulted in high BC and N3 concentrations, in such a way that these two metrics increased when the road trafc intensity (vehicles h1)/wind speed ratio increased, (2) PM1 and PM2.5 levels were lower during daylight (due to inland entry of relatively clean marine air masses) than at night (due to the seaward drainage airow resulting in the transport of aged particulate pollutants from the city), (3) although N3 and BC concentrations exhibited a signicant correlation during the whole study period, the N3/BC ratio experienced a daily evolution with a maximum during daylight. Thus, high N3 concentrations associated with high N3/BC ratios and high solar irradiance conditions were recorded during the daylight inland breeze period due to an enhancement of processes favouring new particle formation. Data analysis points out that this enhancement in the new particle formation processes is strongly related to the nucleation of photooxidized vapours under the relatively low PMx (and consequently low aerosol surface area) concentrations prompted by the inland entry of clean marine air due to the daylight breeze blowing. The results obtained show that, in addition to the vehicle exhaust emissions, new particle formation in coastal urban areas due to photo-oxidation processes may signicantly contribute to the ultrane particle concentration. Crown Copyright 2008 Published by Elsevier Ltd. All rights reserved.

1. Introduction
* Corresponding author. Izan a Atmospheric Research Centre, AEMet, Associated Unit to CSIC Studies on Atmospheric Pollution, C/ La Marina, 20, 6a Planta, E38001 Santa Cruz de Tenerife, Canary Islands, Spain. Tel.: 34 922151718; fax: 34 922574475. guez). E-mail address: srodriguez@inm.es (S. Rodr

Studies performed over the last two decades related exposure to urban ambient PM10 and PM2.5 (mass concentrations of particulate matter-PM <10 mm and <2.5 mm

1352-2310/$ see front matter Crown Copyright 2008 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.atmosenv.2008.04.022

6524

guez et al. / Atmospheric Environment 42 (2008) 65236534 S. Rodr

aerodynamic diameter, respectively) with an increased risk to sufferers of respiratory and cardiovascular diseases (Dockery and Pope, 1996). As a result of this, PM10 and PM2.5 are currently used as urban air quality reference metrics. More recent research has highlighted the high potential toxicity of some PM components. For example, ultrane particles (<0.1 mm) are considered especially detrimental to human health (Peters et al., 1997). These particles easily reach the alveoli, enter the blood circulatory system and may be distributed throughout the whole human body, with this distribution being greater for insoluble material. Moreover, fresh soot particles are particularly surface-reactive and may carry carcinogenic compounds (e.g. PAHs; Highwood and Kinnersley, 2006). The debate already open on whether additional (to PM10 and PM2.5) metrics on the ambient air particle concentration should be monitored to improve human health protection has prompted several studies focused on characterising aerosol physical properties. The PM Working Group of the European Commission recognised that more studies are necessary on other metrics of the aerosol concentration in order to consider a possible regulation of metrics complementary to PM10 and PM2.5 (Second Position Paper on PM, 2004). Recently, much of the attention has been paid to ultrane particles in cities. Ultrane particles typically account for 8090% of the total number concentration (e.g. Wehner and Wiedensohguez et al., 2007). In urban air, new ultrane ler, 2003; Rodr particles have two main origins: (1) Primary vehicle exhaust emissions. Primary particles emitted by vehicle exhausts tend to exhibit a size distribution with a nucleation mode (<30 nm) and a soot mode (5080 nm) (Burtscher, 2005; Rose et al., 2006). The nucleation mode is mainly attributed to binary H2OH2SO4 nucleation, nucleation of some organic compounds and subsequent particle growth (up to w30 nm) by condensation processes during the dilution and cooling of vehicle exhaust (Harris and Maricq, 2001; Casati et al., 2007). Low ambient air temperature, low dilution conditions or high relative humidity increase the formation and growth rates of nucleation particles during the dilution and cooling of vehicle exhaust (Olivares et al., 2007; Casati et al., 2007). The soot mode is mainly made up of elemental carbon, absorbed organic material and some other trace elements formed in the engine and directly emitted in the aerosol phase (Burtscher, 2005; Harris and Maricq, 2001). Soot emissions have classically been higher in diesel than in gasoline vehicles. (2) In situ new particle formation in ambient air. This may occur by nucleation of photo-oxidized species under high solar radiation conditions (Boy and Kulmala, 2002; Rodr guez et al., 2005; Hamed et al., 2007). Although it is known that these processes also occur in urban air (e.g. Wehner and Wiedensohler, 2003; Kulmala et al., 2004, 2005; Qian et al., 2007), they have not yet been investigated in depth owing to the fact that most of these studies on nucleation in ambient air have been performed in rural and remote areas. Processes affecting PM1, PM2.5 and PM2.510 concentrations are signicantly different from those affecting the ultrane particle number concentration. Accumulation

mode particles (0.11 mm) are the most important contributors to PM1 and PM2.5, even if they account for a marginal fraction of the number concentration (1015%). Although some primary emissions may contribute (e.g. soot due to heavy duty diesel engines or mineral dust), most of the accumulation mode mass is due to gas to particle conversion processes onto pre-existing particles and to ultrane particle growth (Cabada et al., 2004). In contrast, coarse particles (>1 mm) are mainly made up of mineral dust, abrasion products from tyre and brake wear, and sea salt at coastal sites. Ultrane, accumulation and coarse particles also differ in life time. Coarse particles are relatively rapidly removed through sedimentation, whereas ultrane particles rapidly grow through condensation and coagulation. In contrast, accumulation mode particles exhibit a long residence time (from days to a few weeks) and consequently, they may signicantly contribute to the urban and regional aerosol background. There is a growing interest in studying the processes affecting the relationship between metrics representative of the particle concentration in urban air. The relationship between PM10 and number concentration has been nkko nen studied in several cities (Ketzel et al., 2004; Mo et al., 2004; Harrison and Jones, 2005). Ruuskanen et al. (2001) performed a comparative study on PM2.5 lter blackness and concentrations of PM2.5, ultrane and accumulation mode particles in cities in Germany, Finguez et al. (2007) studied land and The Netherlands. Rodr the relationship between number size distribution and PM2.5 composition in Milan, Barcelona and London. More recently, Puustinen et al. (2007) investigated the spatial variations of PM10, PM2.5, soot and number concentration in Amsterdam, Athens, Birmingham and Helsinki. The relationship between number concentration and other aerosol concentration metrics may be complex for several reasons. For example, long range transport of (aged) aerosol may exhibit a much higher inuence on PM10 and on PM2.5 than in number concentration (Ketzel et al., 2004). On the other hand, processes contributing to the decrease in PM10 and/or PM2.5 (e.g. meteorology) may enhance the new particle formation rates owing to the fact that reductions in PMx may decrease the surface area acting as a condensation sink for the aerosol gaseous precursors (Burtscher, 2005; Hamed et al., 2007). In a recent study, Hoek et al. (2008) observed that in urban areas the number concentration exhibits a much higher spatial variability than PM2.5, soot and sulphate-aerosol. The purpose of this study is to investigate the inuence of sea breeze circulation and of road trafc emissions on the relationship between the particle number, black carbon, PM1, PM2.5 and PM2.510 concentrations in a coastal city. Studies performed in urban areas of Eastern Spain showed that sea breeze exerts a major inuence on PM10 and PM2.5 concentrations (Viana et al., 2005). The results of this study show how the great changes in the daylightnight meteorological conditions (including inland entries of clean air during daylight) exert a marked inuence on the relationship between number concentration and other aerosol concentration metrics.

guez et al. / Atmospheric Environment 42 (2008) 65236534 S. Rodr

6525

2. Methodology 2.1. Study city This study was performed in Santa Cruz de Tenerife (Canary Islands; 28 280 N, 16 150 W; Fig. 1). This is a 221,500 inhabitant city (distributed over 150 km2) located at the bottom of the southern slope of the Anaga ridge and the eastern slope of the SW to NE ridge crossing the Island.

This topographic setting protects the city from the Trade winds (NNE) that blow over the ocean (see details on the meteorology in Guerra et al., 2004). The main sources of anthropogenic aerosols in the city are: (1) road trafc (the two main city roads are highlighted in Fig. 1A), (2) a crude oil renery (COR) located on the SE shore of the city (see COR in Fig. 1C), and (3) ship emissions in the harbour. Natural Saharan dust and sea salt may signicantly contribute to the daily aerosol load (>100 mgPM10 m3 during dust events; Viana et al., 2002).

2.2. Measurements Measurements of PM were performed simultaneously (mid-March to mid-April 2006) at three sites in the city of Santa Cruz de Tenerife (Fig. 1C): (1) Santa Cruz Observatory (SCO), (2) Tome Cano (TC) and (3) Gladiolos (GL).

Fig. 1. (A)Topographic map of Tenerife for every 100 m. Santa Cruz de Tenerife city: (B) Vertical contour highlighting the direction of the breeze blowing (grey rectangles represent buildings), (C) picture of the city, and (D) regular daily evolution of wind speed and direction during the measurement campaign (mode of wind direction and mean of wind speed). Different parts of the city are indicated: Santa Cruz Observatory (SCO), Tome Cano (TC), Gladiolos (GL) and the Crude Oil Renery (COR).

(A) Santa Cruz Observatory (SCO; Fig. 1C) is a coastal urban background site located on the roof of an 8-oor building at 52 m above sea level, at about 45 m above ground. Measurements of meteorological parameters (wind speed and direction, temperature, relative humidity, pressure and global radiation), trace gases (CO, SO2 and O3) and particles at SCO are performed a GAW Observatory in the frameby the staff of Izan work of the GURME (GAW) program. The site is on the NE edge of the city, close to the shore and on the western side of Anaga Avenue (a 4-lane road running along the shore). The main data set used in this study was collected at this site, where the following measurements were performed: B PMx mass concentrations. One-minute concentrations of PM1, PM2.5 and PM10 have been obtained at this site since 2002 by multiplying the aerosol volume concentrations (monitored with an Optical Particle Counter OPC GRIMM) by experimentally determined volume-to-mass conversion factors. These factors were determined by cross correlating the aerosol (PM10 and PM2.5) mass (determined by 24-h lter sampling with high volume samplers) versus the volume (<10 mm and <2.5 mm) determined by the OPC measurements. These PM gravimetric versus OPC-volume measurements showed a good degree of agreement: slope 1.65 r2 0.94 for PM10 and 1.31 r2 0.75 for PM2.5. These values were obtained with 53 PM10 and 51 PM2.5 samples collected from January to September 2006. B The mass concentrations of black carbon (BC; the component of soot responsible for light absorption) less than 10 mm were monitored by placing a PM10 impactor in the inlet of a Multi Angle Absorption Photometer MAAP (Thermo Carusso 5052). These measurements were performed from 17th March to 10th April 2006. B The number concentration of particles coarser than 3 nm (N3) was continuously monitored with an Ultrane Condensation Particle Counter UCPC (TSI model 3025A). These measurements were performed from 17th March to 10th April 2006.

6526

guez et al. / Atmospheric Environment 42 (2008) 65236534 S. Rodr

(B) Gladiolos and Tome Cano are city centre sites located to the NW of the industrial area of the city (crude oil renery) and to the NW of the motorway (which links the city to the northern side of the island; Fig. 1C). At these sites measurements of PM10, PM2.5, gaseous pollutants (CO, NOx, SO2 and O3) and meteorology (wind speed and direction, temperature, relative humidity and pressure) are continuously performed by following standard procedures in air quality networks. Measurements of PM10 and PM2.5 were performed with beta and TEOM instruments, respectively. PM2.5 was not monitored at the Tome Cano site. Finally, measurements of road trafc intensity (number of vehicles h1) were also performed in several parts of the city by city council staff. In this article we will only show the road trafc intensity at the coastal Anaga Avenue (where the SCO site is located).

3. Results and discussion During the intensive eld measurement campaign, mean concentrations of 24 mg m3 of PM10, 14 mg m3 of PM2.5, 9 mg m3 of PM1, 1.9 mg m3 of BC and 26 305 cm3 of N3 were recorded at the urban background SCO site (Santa Cruz Observatory), 46 mg m3 of PM10 at TC (Tome Cano) and 43 mg m3 of PM10 and 13 mg m3 of PM2.5 at GL (Gladiolos) were recorded at the city centre sites. These mean PMx concentrations are slightly lower than the annual mean PMx concentrations typically recorded in this city because Saharan dust events (the main cause of high PMx concentrations in this city; Viana et al., 2002) did not occur during our campaign. These mean PMx concentrations are within the range of those typically observed in mid European cities (Harrison and guez et al., Jones, 2005; Puustinen et al., 2007; Rodr 2007; Ruuskanen et al., 2001).

morning rush hours, a decrease in the PMx concentrations such as that described above at SCO (09:0019:00) was not observed (Fig. 2GI). However, PM10 and PM2.5 at TC and GL showed an increasing trend until 21:00 (local time). This contrasting behaviour between the PMx concentrations at the coastal (SCO) and city centre (TC and GL) sites is a consequence of the easterly inland breeze blowing during daylight, which prompts the entry of clean marine air masses along the coast and favours the inland transport of the pollutants (emitted in the eastern part of the city) to the city centre. This is one of the reasons because PM10 concentrations are much lower at SCO coastal than at TC and GL city centre sites. In fact, this daylight breeze results in the inland transport of the (renery) SO2 plumes, giving rise to higher SO2 concentrations during daylight than at night at the TC and GL city centre sites (Fig. 2GI). At the SCO coastal site, N3 and BC concentrations also exhibited well dened daily cycles (Fig. 2AC). During working days, N3 and BC presented a maximum during the morning rush hours. The morning maximum in the vehicles h1/wind speed ratio indicates that the morning maximum in the particle concentration is favoured by the coupled effects of an increase in the road trafc intensity and low wind speed in the morning (owing to the fact that the inland breeze has not yet started blowing). Observe that the daily evolution of N3 and BC during weekdays is different from that recorded at weekends because of the change in road trafc patterns. This daily evolution in number concentration is similar to that typically observed in other cities (e.g.Wehner and Wiedensohler, 2003; Charron guez et al., 2007). In Helsinki and and Harrison, 2003; Rodr Liepzig, a similar daily evolution in the black carbon mass and soot number concentrations has been observed, respectively (Pakkanen et al., 2000; Rose et al., 2006).

3.2. Processes affecting urban coastal aerosol concentrations The relationship between road trafc intensity, meteorology and particle concentration at SCO (coastal urban background site) was examined using Principal Component Analysis (PCA) followed by a varimax rotation. This analysis was performed at several times of the day: morning (07:0009:00), central daylight (11:0017:00) and night (00:0005:00). Three principal components, accounting for 81, 66 and 72% of the total variance for the three daytime periods selected, were found (Table 1):
B

3.1. Mean daily evolution of aerosols The urban scale transport of air pollutants in Santa Cruz de Tenerife is mainly driven by breeze circulation (Fig. 1B and C). This breeze is characterised by inland (westward) airows during daylight (34 m s1) and a slight seaward (eastward) airow at night (w1 m s1). Inland breeze blowing started at 08:00 and was characterised by an abrupt shift in wind direction. Fig. 2 displays the regular daily evolution of the road trafc intensity and the particle and some gaseous pollutant concentrations at the SCO coastal (Fig. 2AF), TC and GL city centre (Fig. 2GI) sites. At the SCO coastal site, particle (PMx, BC and N3) and CO concentrations exhibited a maximum during the morning rush hours of working days, and a subsequent decrease (09:0019:00) associated with the easterly entry of marine air (Fig. 2AF). At night, a smooth increase in the PMx concentrations, associated with slight westerly winds bringing air masses from the city centre, was observed at SCO. At the TC and GL city centre sites, after the maximum in the NO and PMx concentrations recorded during the

principal component 1 is positively correlated with PMx and BC and negatively correlated with wind speed. The fact that this component does not exhibit a signicant correlation with road trafc intensity (number of vehicles h1) indicates that fresh road trafc emissions do not exhibit a major inuence on PMx variance. The signicant negative factor loading of wind speed indicates that this component accounts for the inuence on the aerosol mass (BC and PMx), of the dilution and the air mass renovation prompted by wind blowing over the study city. The negative association of the N3/ BC ratio in this component will be discussed later.

guez et al. / Atmospheric Environment 42 (2008) 65236534 S. Rodr

6527

50103

Monday-Friday

Saturdays

Sundays

2vehicles/h / wind speed, s/m

N3, cm-3 10BC, ng/m3

1500 1000

2CO, g/m3

vehicles / h

25103
500 SCO SCO

0 30

SCO

SCO

SCO

SCO

g/m3

15

0
0 4 9 14 19 0 4 9 14 19 0 4 9 14 19

PM10 PM2.5 PM1 PM2.5-10

60

30

PM10 -TC PM10 -GL 2xPM2.5 -GL NO -GL SO2 -GL TC & GL 4

g/m3

0
0 4

TC & GL

TC & GL 0 4

14

19

14

19

14

19

GMT
Fig. 2. Mean daily cycles. Hourly averaged concentrations of CO and the particle N3, BC and PMx concentrations at the Santa Cruz Observatory (SCO) coastal urban background site (AC) and of PM10, PM2.5 and gaseous pollutants at the Tome Cano (TC) and Gladiolos (GL) city centre sites (GI) during weekdays, Saturdays and Sundays.
B

principal component 2 is positively correlated with road trafc intensity, BC and N3. This component represents the inuence of fresh vehicle exhaust emissions on the particle concentration. principal component 3 is present during daylight (i.e. morning rush hours 07:0009:00 and central daylight period 11:0017:00). The positive correlation of the N3/BC ratio and solar radiation intensity suggests that nucleation of photo-oxidized vapours contributes to increasing the above ratio by enhancing new particle formation in ambient air. Further details are provided below.

3.3. Inuence of wind speed The results of the PCA shown above indicate that wind speed exerts: (1) a major inuence on PMx, (2) a moderate inuence on BC and (3) a relatively low inuence on N3. In fact, wind speed exhibited Pearsons correlation coefcients (r) of 0.50, 0.44, 0.41, 0.33 and 0.0 with PM2.510, PM2.5, PM1, BC and N3, respectively (values calculated using 3092 10-min observations). These results are corroborated when analysing the particle concentration versus wind speed plots shown in Fig. 3. The fact that high N3 concentrations (>45 103 cm3) may still be recorded under high wind speed conditions (>5 m s1) may be accounted for by new particle formation processes. At a roadside site in London, Charron and Harrison (2003) found that increases in

wind speed produced much higher decreases in the concentration of accumulation mode particles (0.11 mm) than in the concentration of nucleation mode particles (<30 nm). They argued that this was caused by the fact that clean atmospheres (favoured by high wind speeds) tend to favour new particle formation owing to the low aerosol surface area (mostly occurring in the accumulation mode) available for condensation of the aerosol gaseous precursors. They concluded that the lack of correlation between wind speed and number concentration is a feature of new particle forguez et al. mation. Similar results were found by Rodr (2005) at a rural site inuenced by nearby town emissions in Northern Italy. The analysis of the N3/BC ratios shown below supports this involvement of the new particle formation processes during periods of high wind speed (typically associated with inland breeze blowing). 3.4. Quantication of the impact of road trafc emissions The impact of fresh road trafc emissions on the particle concentration was quantied at several times of the day. Fig. 4A1F1 presents the mean BC, N3 and PMx concentrations measured for each 100 vehicles h1 bin during the morning (07:0009:00), the central part of daylight (11:0017:00) and at night (00:0005:00). The use of 100 vehicles h1 bins smoothes the variability in the concentrations due to other inuencing parameters (mainly meteorology). The concentrations of BC and N3 tend to increase when increasing road trafc intensity (Fig. 4A1 and B1).

6528

guez et al. / Atmospheric Environment 42 (2008) 65236534 S. Rodr

Table 1 Factor loading of the Principal Components Analysis performed with data from selected periods Morning rush hours (07:0009:00) Vehicles h1 BC N3 N3/BC PM1 PM2.5 PM2.510 Wind speed Radiation Explained variance PC 1 0.11 0.56 0.17 0.50 0.95 0.96 0.88 0.42 0.24 38% PC 2 0.89 0.70 0.90 0.05 0.06 0.04 0.16 0.15 0.03 24% PC 3 0.14 0.33 0.07 0.61 0.07 0.08 0.08 0.81 0.73 19% Central daylight (11:0017:00) PC 1 0.29 0.39 0.32 0.23 0.86 0.89 0.52 0.63 0.08 29% PC 2 0.51 0.82 0.81 0.18 0.27 0.21 0.16 0.07 0.20 20% PC 3 0.04 0.22 0.28 0.77 0.08 0.05 0.55 0.12 0.72 17% Night (00:0005:00) PC 1 0.01 0.75 0.30 0.77 0.86 0.86 0.85 0.49 46% PC 2 0.85 0.59 0.80 0.04 0.36 0.37 0.46 0.09 26%

This behaviour is not observed for PM2.5 and PM1 (Fig. 4C1 and D1). During the morning, the PM2.510 concentrations also tend to increase when road trafc intensity increases, this being most probably due to road dust re-suspension (Fig. 4E1). The much lower direct inuence of the road trafc emissions in PM2.5 and PM1 (Fig. 4C1 and D1) than in N3 and BC (Fig. 4A1 and B1) may be accounted for by the coming into force of the European Standards EURO 14 (progressively since 1992) for vehicle exhaust emissions. Because these standards are established in terms of particle mass (g km1), they have decreased the emissions of accumulation (0.11 mm) and coarse (>1 mm) mode particles, which are the main contributors to PMx. In contrast, N3 and BC tend to increase when the road trafc intensity increases. This is due to the fact that vehicle exhaust mostly emits ultrane and soot particles and these are the main contributors to N3 and BC (Burtscher, 2005; Harris and Maricq, 2001; Rose et al., 2006). Although BC concentrations tend to increase when road trafc intensity increases, different BC concentrations are observed for a given number of vehicles h1 bin at different times of the day (Fig. 4A1). For example, a road trafc intensity within the range of 11001200 vehicles h1 is associated with 3.7 mgBC m3 during the 07:0009:00 period and with 1.5 mgBC m3 during the 11:0017:00 period. The lower BC concentrations during the 11:0017:00 period are a consequence of the higher wind speeds during that period of daylight (because of the inland breeze blowing). For this reason we have normalized the road trafc intensity data by dividing them by wind speed (Fig. 4A3). Observe how BC concentrations tend to increase in a more regular way (monotonic growth) when increasing the vehicles h1/wind speed ratio than when increasing the vehicles h1 ratio (Fig. 4A1 and A3). The results show how increases in 200 vehicles h1 m1 s1 are associated with BC increases of about 0.50 0.29 mgBC m3 (on average) at any time of the day. This analysis of particle concentration versus vehicles h1/wind speed was also applied to the PMx and N3 concentrations (Fig. 4A3F3). The most important results are:

period and of about 6523 1437 cm3 every 200 vehicles h1 m.s1 in the 11:0017:00 period. These higher values in the 11:0017:00 period in comparison with the 07:0009:00 period suggest an enhancement in the particle formation processes at noon and during the afternoon. This interpretation is supported by the results shown in the following section. (2) PM2.5 and PM1 do not exhibit a strong association with the vehicles h1/wind speed ratio (as that described for BC and N3). (3) PM2.510 exhibits a clearly increasing trend of about 1.20 0.90 mgPM2.510 m3 every 200 vehicles h1 m.s1 during the morning rush hours (07:0009:00), most probably due to road dust re-suspension. The weaker association observed during the 11:0017:00 period (Fig. 4E3) is probably due to the inuence of sea salt transported inland during daylight.

3.5. The numbertoblack carbon ratios The relationship between N3 and BC concentrations at the SCO site has been used to identify periods of enhancement in the new particle formation rates, which would lead to an increase in the N3/BC ratio. At SCO, N3 and BC exhibited a signicant correlation because of the vehicle exhaust emissions: r 0.70 during the whole measurements campaign (based on 3092 10-min averages data). When plotting the N3 versus BC data, it can clearly be observed that all points are comprised between two well dened borders with slopes (S) Smin 4.8 106 particles ngBC1 and Smax 47 106 particles ngBC1, representing the minimum and maximum N3/BC ratios observed in ambient air at the SCO site, respectively (Fig. 5). The fact that the N3 versus BC data are comprised between these two slopes is observed at any time of the day. For the periods 00:0005:00, 07:0009:00 and 11:0017:00, Smin and Smax exhibit values of 4.5, 4.9 and 8.1 106 particles ngBC1 and 48, 43 and 57 particles ngBC1, respectively (plots not shown for the sake of brevity). The fact that the slope Smin is much higher during the central part of daylight, indicates that: (1) the slope Smin is strongly related to the minimum contribution of the vehicle exhaust emissions to the particle number

(1) N3 also exhibits an increasing trend with the vehicles h1/wind speed ratio of about 3250 2395 cm3 every 200 vehicles h1 m.s1 during the 07:0009:00

guez et al. / Atmospheric Environment 42 (2008) 65236534 S. Rodr

6529

A
BC, g/m3

12

y = 1.87x-0.38 R2 = 0.120

B
PM2.5-10, g/m3

20 y = 5.68x-0.59 R2 = 0.524 10

ratio exhibits a daily evolution with a maximum during daylight. Previous studies have shown that enhancements in the new particle formation rates during the dilution and cooling of the vehicle exhaust are favoured by low ambient air temperature (e.g. Olivares et al., 2007) and/or low wind speed leading to less dilution (Casati et al., 2007). However, these two reasons are not plausible at SCO, where the highest N3/BC ratios are observed during the period of highest temperature and wind speed 11:0017:00 (Fig. 6). At SCO the N3/BC ratio reaches the highest values during the noon and afternoon period when:

(1) solar radiation intensity reaches the highest values. In fact, in the PCA discussed above, it was shown how Principal Component 3 is correlated with the N3/BC ratio and the solar radiation (PC 3 in Table 1), (2) PM1 and PM2.5 concentrations (where most of the aerosol surface area is comprised) are minima (Figs. 2 and 6). The fact that the N3/BC ratio exhibited a much higher anti-correlation with PM1 and PM2.5 than with PM2.510 during the 11:0017:00 period (plot not shown), suggests a more active role of the accumulation mode particles (which typically accounts for most of the aerosol surface area) as condensation sink for the aerosol gaseous precursors, (3) it is positively correlated with wind speed for wind speed values between 0 and 4 m s1 (Figs. 3E and 4F3).

C
PM2.5, g/m3

40

y = 14.9x-0.21 R2 = 0.323

20

D
N3103, cm-3

90

45

E
N3/BC, 106, /ng

40

20

10

wind speed, m/s

Fig. 3. Concentration of particles (AD) and N3/BC ratio (E) versus wind speed at the SCO site. Red dots and line in plots D and E represent the mean N3 and N3/BC values for each 1 m s1 bin.

concentration (minimum amount of particles emitted and formed during the dilution and cooling of exhaust per nanogram of BC emitted by the vehicle exhaust), and (2) there is an enhancement in the new particle formation rates during daylight resulting in a higher N3/BC ratio (4.9 106 particles ngBC1 during the 07:0009:00 period and 8.1 106 particles ngBC1 during the 11:0017:00 period on average; Fig. 6). Observe in Fig. 6 how the N3/BC

These features suggest that, in addition to the new particle formation during the dilution and cooling of the vehicle exhaust, there is a non-negligible contribution to N3 concentrations due to new particle formation in ambient air. In fact, the features described above t very well with those of the new particle formation in ambient air described by Boy and Kulguez et al. (2005) and Hamed et al. mala (2002), Rodr (2007). The behaviour observed at the SCO site indicates that the decrease in PM1 and PM2.5 concentrations, prompted by the increase in wind speed (due to the inland breeze blowing), results in a decrease in the aerosol surface area concentrations. This would favour nucleation, rather than condensation onto pre-existing particles, for those aerosol gaseous precursors formed by photo-oxidation processes. Observe in Fig. 6 how the daily variation of the N3/BC ratio is much better modulated by the daily evolution of the inverse of PM1 (which typically comprises most of the aerosol surface) than by the daily evolution of PM2.5101. An example of one of these events (8th April 2006) is shown in Fig. 7: observe the correlated increases in the N3/BC ratio and in solar radiation intensity from 10:00 to 16:00. This contribution to N3 due to nucleation in ambient air during the central part of the daylight accounts for: (1) the faster decrease in the BC than in the N3 concentrations after the morning rush hours (Fig. 2 A), and (2) the higher increase in N3 concentrations for every 200 vehicles h1 m.s1 bin during the 11:00 17:00 period than during the 07:0009:00 period (described above; 3250 cm3 during the 07:0009:00 and about 6523 cm3 during the 11:0017:00 periods).

6530

guez et al. / Atmospheric Environment 42 (2008) 65236534 S. Rodr

BC, g/m3

7-9h 11-17h 0-5h

A1

A2

A3

0 60

103N3, m-3

30

0 15

B1

B2

B3

PM1, g/m3

10 5 0

C1

C2

C3

PM2.5, g/m3

20

10

D1 E1

D2 E2

D3 E3

PM2.5-10, g/m3 N3/BC, 106/ng

20

10

0 20

10

0 0 400 800 1200 1600

F1
0 2 4 6 2000

F2
0 8

F3
400 800 1200 1600 2000 2400

vehiclesh-1

wind speed, m/s

vehiclesh-1/ wind speed, s/m

Fig. 4. Particle concentration and N3/BC ratio averaged for each 100 vehicles h1 bin (A1F1), for each 1 m s1 bin (A2F2) and for each 200 vehicles h1 bin (A3F3).

3.6. Day-to-day particle events Fig. 8 shows the hourly BC, N3 and PMx concentrations at the SCO site and the road trafc intensity during a period of the measurement campaign. Fig. 9 shows the median of the BC, N3 and PMx concentrations for each wind direction for the 07:0009:00, 11:0017:00 and 00:0005:00 periods during all the measurement campaigns. These data (Figs. 8 and 9) clearly illustrate how N3, BC and PMx are inuenced in different ways by breeze circulation, vehicle exhaust emissions and ageing processes given that:

1. high BC and N3 concentrations are recorded during the morning rush hours due to fresh vehicle exhaust emissions (07:0009:00; Figs. 8A,B and 9A,B), the highest being BC and N3 concentrations registered when wind blows from the city centre (Southwest; Fig. 9A1) or from the coastal Anaga Avenue (East; Fig. 8A1); 2. high N3 concentrations (associated with high N3/BC ratios) during the central part of daylight (11:0017:00) are associated with easterly winds (inland breeze; Fig. 9B2) due to the above described new particle formation in ambient air (Figs. 6 and 9A2,B2). Examples of

guez et al. / Atmospheric Environment 42 (2008) 65236534 S. Rodr

6531

100103

Smax 10BC, ng/m3

70103

1000

Radiation, w/m2

N3, cm-3

50103

35103

500

Smin

0 0 0 6 12 70103

N3, cm-3

BC,

g/m3

30

N3/BC, 106/ng

Fig. 5. Hourly averaged values of the particle number N3 versus particle black carbon (BC) concentrations at the SCO site.

35103

15

these events occurred during the periods 2224 and 28 31 March 2006, when N3/BC ratios of 1525 106 particles ngBC1 and <10 106 particles ngBC1 were recorded during daylight and at night, respectively (Fig. 8; N3/BC ratios not shown); and 3. the highest PM1 and PM2.5 and the lowest N3 and BC concentrations are recorded at night when the wind blows from the city centre, bringing aged particulate pollutants to the shore owing to the seaward airow associated with nocturnal breeze circulation (Fig. 9C3 and D3). See examples of these PM1 and PM2.5 nocturnal events associated with the outows from the city centre during the periods 2326 March and 30 March2 April 2006 (Fig. 8). BC and N3 tend to exhibit patterns correlated with those of the road trafc intensity, evidencing the high inuence of fresh emissions (Fig. 8). In contrast, PM1 and PM2.5 tend to exhibit high concentrations at night, when road trafc emissions are not signicant. The nocturnal conditions (e.g. decrease in the boundary layer depth

0 0 6 12 18

time of day
Fig. 7. 10-min averaged values of the solar radiation, N3/BC ratio, black carbon (BC) and number N3 concentrations at the SCO site during 8th April 2006.

N3 / BC, 106/ng 50PM2.5-10-1, m3/g 125PM1-1, m3/g wind speed, m/s 1000 30

resulting in an increase of the concentrations of aerosols and its gaseous precursors, or decrease in temperature decreasing the equilibrium vapour pressure and consequently favouring condensation) favour particle growth by coagulation and condensation processes and this contributes to the increase in the particle mass in the accumulation mode and consequently to the increase in PM1 and PM2.5 concentrations. Observe in Fig. 10A how PM2.5 experiences low spatial variations across the city (slope and intercept of about 1.0). In contrast, PM10 experiences a much higher spatial variability across the city because of the local contribution of coarse particles in the city centre (w20 mg m3): see the high value of the intercept (1821 mg m3) of the scatter plots PM10: GL versus SCO (Fig. 10B) and TC versus SCO (Fig. 10C) and PM2.510: GL versus SCO (Fig. 10D). This supports the idea that PM2.5 is mostly representative of the urban background (accumulation mode aged aerosols formed by condensation/reaction of gases onto pre-existing particles) whereas PM2.510 is mostly affected by local (street) emissions. 4. Summary and conclusions

global radiation, w/m2

20 500 10

0 0 4 8 12 16 20

time of day
Fig. 6. Daily mean cycle (hourly averaged values) of the global radiation, wind speed, the N3/BC ratio, and the inverse of the PM1 (PM11) and PM2.510 (PM2.5101) concentrations.

In a coastal urban environment the daily cycles of breeze circulation (inland during daylight and seaward at night) and the road trafc emissions exert a great, but differentiated, inuence on the particle black carbon (BC), number (N3, >3 nm) and PMx (PM1, PM2.5 and PM2.510) concentrations. BC and N3 concentrations are greatly inuenced by fresh vehicle exhaust emissions and exhibit an increasing trend when the road trafc intensity (vehicles h1)/ wind speed ratio increases. In contrast, PMx (specially PM1 and PM2.5) reaches the highest values at night due to the seaward transport of aged particulate pollutants to the shore, and the lowest values during daylight due to the inland entry of clean marine air. Although N3 and BC concentrations exhibit a signicant correlation throughout

6532

guez et al. / Atmospheric Environment 42 (2008) 65236534 S. Rodr

SCO 2000 1000 0

10

BC, g/m3

vehicles / h

2000 1000 0 2000 1000 0

80103

N3, cm-3

40103

0 60

g/m3
PM10 PM2.5 PM1 PM2.5-10

30

21 22

23

24 25

26

27

28

29

30

31

March

April

Fig. 8. Hourly mean values of road trafc intensity at the coastal Anaga Avenue (vehicles h1) and of the particle BC, N3 and PMx concentrations at the SCO site from March 21st to April 3rd 2006.

the whole day, the N3/BC ratio experiences a daily evolution correlated with that of the solar irradiance and wind speed, in such a way that high N3 concentrations associated with high N3/BC ratios and high solar irradiance conditions are recorded during the daylight inland breeze period due to an enhancement of some new particle formation processes. The analysis performed indicates that this enhancement in the new particle formation processes is strongly related to the nucleation of photo-oxidized vapours under the relatively low PMx (and consequently low aerosol surface area) concentrations prompted by the inland entry of clean marine air due to the daylight breeze

blowing. The statistical tools used in this study corroborate the association between BC and N3 with road trafc intensity (fresh emissions), PMx and wind speed1 and N3/ BC ratio and solar radiation (enhancement in the new particle formation due to photo-oxidation processes). Although it is known that new particle formation due to photo-oxidation of aerosol gaseous precursors may occur in the urban atmosphere (e.g. Wehner and Wiedensohler, 2003; Kulmala et al., 2004; Qian et al., 2007), most studies on urban ultrane particles are currently focused mainly on primary vehicle exhaust emissions. The results obtained here indicate that this contribution due to photo-chemically

BC

N3 x103

PM1

PM2.5

PM2.5-10

A1
7-9h morning

12 6

B1

70 35

C1

20 10

D1

40 20

E1

20 10

A2
11-17h central daylight

12 6

B2

70 35

C2

20 10

D2

40 20

E2

20 10

A3
0-5h night

12 6

B3

70 35

C3

20 10

D3

40 20

E3

20 10

Fig. 9. Hourly mean concentrations of the particle BC, N3 and PMx for each wind direction at several times of the day at the SCO site.

guez et al. / Atmospheric Environment 42 (2008) 65236534 S. Rodr

6533

GL vs. SCO
40

GL vs. SCO
100

TC vs. SCO
100

GL vs. SCO
50

PM2.5

PM10

PM10

PM2.5-10

g/m3

20
y=1.0x + 1.0 r2= 0.88

50
y=0.9x + 20.7 r2= 0.82

50
y=1.2x + 17.7 r2= 0.89

25
y=0.78x + 21.3 r2= 0.54

0 0 20 40

0 0

0 50 100 0
3

0 50 100 0 25 50

PMx SCO, g/m

Fig. 10. Daily mean concentrations of PM2.5, PM2.510 and PM10 at city centre sites (GL and TC) versus at SCO coastal urban background.

induced nucleation may signicantly contribute to the urban ultrane particle concentration. The results of this study show how the strong changes in daylight to night meteorological conditions (which includes inland entries of clean air during daylight) exert a great/major inuence on the complex relationship between the number concentration and other aerosol concentration metrics. This, and other processes, not yet well characterised, may contribute to the large spatial variability in the particle number concentration as described by Hoek et al. (2008). According to Kulmala et al. (2004), new particle formation rates in a coastal environment as high as those in industrial plumes are frequently recorded. This could, in part, be due to the fact that some of the factors which may favour nucleation are present in the coastal environment: (i) enough high relative humidity to favour binary H2OH2SO4 nucleation (Easter and Peter, 1994; Olivares et al., 2007), (ii) relatively clean air conditions (low aerosol surface area concentration) due to the marine air entries) and (iii) the mixing of two air parcels with different temperature and different relative humidities (Nilsson and Kulmala, 1998). Acknowledgements This study was nanced by the Ministry for the Environ n General de Calidad Ambiental; ment of Spain (Direccio n integral del impacto Project B026/2007/3-10.1: Evaluacio culas de los automo viles en la calde las emisiones de part idad del aire urbano). The Council of Santa Cruz de Tenerife and the Departments of Environmental Health and of the Environment of the Canary Islands Government also contributed to this study. Measurements at SCO are performed within the framework of the GURME (GAW) program. Part of this study was performed into the framework of the project- GRACCIE (CONSOLIDER INGENIO; Ministry of Education and Culture of Spain). References
Boy, M., Kulmala, M., 2002. Nucleation events in the continental boundary layer: inuence of physical and meteorological parameters. Atmospheric Chemistry and Physics 2, 116. Burtscher, H., 2005. Physical characterization of particulate emissions from diesel engines: a review. Journal of Aerosol Science 36, 896932. Cabada, J.C., Rees, S., Takahama, S., Khlystov, S.A., Pandis, S.N., Davidson, C. I., Robinson, A.L., 2004. Mass size distributions and size resolved chemical composition of ne particulate matter at the Pittsburgh supersite. Atmospheric Environment 38, 31273141.

Casati, R., Scheer, V., Vogt, R., Benter, T., 2007. Measurement of nucleation and soot mode particle emission from a diesel passenger car in real world and laboratory in situ dilution. Atmospheric Environment 41, 21252135. Charron, A., Harrison, R.M., 2003. Primary particle formation from vehicle emissions during exhaust dilution in the roadside atmosphere. Atmospheric Environment 37, 41094119. Dockery, D., Pope, A., 1996. Epidemiology of acute health effects: summary of time-series studies. In: Wilson, R., Spengler, J.D. (Eds.), Particles in Our Air: Concentration and Health Effects. Harvard University Press, Cambridge, MA, USA, pp. 123147. Easter, R.C., Peters, L.K., 1994. Binary homogeneous nucleation: temperature and relative humidity uctuations, nonlinearity and aspects of new particle production in the atmosphere. Journal of Applied Meteorology 33, 775784. guez, S., Arencibia, M.T., Garc a, M.D., 2004. Study on the Guerra, J.C., Rodr formation and transport of ozone in relation to the air quality management and vegetation protection in Tenerife (Canary Islands). Chemosphere 56, 11571167. Hamed, A., Joutsensaari, J., Mikkonen, S., Sogacheva, L., Dal Maso, M., Kulmala, M., Cavalli, F., Fuzzi, S., Facchini, M.C., Decesari, S., Mircea, M., Lehtinen, K.E.J., Laaksonen, A., 2007. Nucleation and growth of new particles in Po Valley, Italy. Atmospheric Chemistry Physics 7, 355376. Harris, J.S., Maricq, M.M., 2001. Signature size distributions for diesel and gasoline engine exhaust particulate matter. Journal of Aerosol Science 32, 749764. Harrison, R.M., Jones, A.M., 2005. Multisite study of particle number concentrations in urban air. Environmental Science and Technology 39, 60636070. Highwood, E.J., Kinnersley, R.P., 2006. When smoke gets in your eyes: the multiple impact of atmospheric black carbon on climate, air quality and health. Environment International 32, 560566. Review article. Hoek, G., Kos, G., Harrison, R., de Hartog, J., Meliefste, K., ten Brink, H., Katsouyanni, K., Karakatsani, A., Lianou, M., Kotronarou, A., Kavouras, I., Pekkanen, J., Vallius, M., Kulmala, M., Puustinen, A., Thomas, S., Meddings, C., Ayres, J., van Wijnen, J., Hameri, K., 2008. Indooroutdoor relationships of particle number and mass in four European cities. Atmospheric Environment 42, 156169. Ketzel, M., WhlinKristensson, A., Swietlicki, E., Berkowicz, R., Nielsen, O.J., Palmgren, F., 2004. Particle size distribution and particle mass measurements at urban, near-city and rural level in the Copenhagen area and Southern Sweden. Atmospheric Chemistry and Physics 4, 281292. ja , T., Mo nkko nen, P., Koponen, I.K., Dal Maso, M., Kulmala, M., Peta Aalto, P.P., Lehtinen, K.E.J., Kerminen, V.M., 2005. On the growth of nucleation mode particles: source rates of condensable vapor in polluted and clean environments. Atmospheric Chemistry and Physics 5, 409416. ja , T., Dal Maso, M., Lauria, A., Kulmala, M., Vehkamaki, H., Peta Kerminen, V.M., Birmili, W., McMurry, P.H., 2004. Formation and growth rates of ultrane atmospheric particles: a review of observations. Journal of Aerosol Science 35, 143176. nkko nen, P., Umac, R., Srinivasan, D., Koponen, I.K., Lehtinen, K.E.J., Mo meri, K., Suresh, R., Sharma, V.P., Kulmala, M., 2004. Relationship Ha and variations of aerosol number and PM10 mass concentrations in a highly polluted urban environmentdNew Delhi, India. Atmospheric Environment 38, 425433. Nilsson, E.D., Kulmala, M., 1998. The potential for atmospheric mixing processes to enhance the binary nucleation rate. Journal of Geophysical Research 103, 13811389.

6534

guez et al. / Atmospheric Environment 42 (2008) 65236534 S. Rodr between mass concentrations, chemistry and number size distribution of urban ne aerosols in Milan, Barcelona & London. Atmospheric Chemistry and Physics 7, 22172232. nder, J., Tuch, T., Rose, D., Wehner, B., Ketzel, M., Engler, C., Voigtla Wiedensohler, A., 2006. Atmospheric number size distributions of soot particles and estimation of emission factors. Atmospheric Chemistry and Physics 6, 10211031. Ruuskanen, R., Tuch, Th., Ten Brink, H., Peters, A., Khlystov, A., Mirme, A., Kos, G.P.A., Brunekreef, B., Wichmann, H.E., Buzorius, G., Vallius, M., Kreyling, W.G., Pekkanem, J., 2001. Concentrations of ultrane, ne and PM2.5 particles in three European cities. Atmospheric Environment 35, 37293738. Second Position Paper on Particulate Matter, 2004. CAFE working group on particulate matter (accessed 20.12.04). guez, S., 2002. InuViana, M., Querol, X., Alastuey, A., Cuevas, E., Rodr ence of African dust on the levels of atmospheric particulates in the Canary Islands air quality network. Atmospheric Environment 36, 58615875. rez, C., Querol, X., Alastuey, A., Nickovic, S., Baldasano, J.M., Viana, M., Pe 2005. Spatial and temporal variability of PM levels and composition in a complex summer atmospheric scenario in Barcelona (NE Spain). Atmospheric Environment 39, 53435361. Wehner, B., Wiedensohler, A., 2003. Long term measurements of submicrometer urban aerosols: statistical analysis for correlations with meteorological conditions and trace gases. Atmospheric Chemistry and Physics 3, 867879.

m, J., Hansson, H.C., 2007. The role of amOlivares, G., Johansson, C., Stro bient temperature for particle number concentrations in a street canyon. Atmospheric Environment 41, 21452155. Pakkanen, T.A., Kerminen, V.M., Ojanen, C.H., Hillamo, R.E., Aarnio, P., Koskentalo, T., 2000. Atmospheric black carbon in Helsinki. Atmospheric Environment 34, 14971506. Peters, A., Wichmann, H.E., Tuch, T., Heinrich, J., Heyder, J., 1997. Respiratory effects are associated with the number of ultrane particles. American Journal of Respiratory and Critical Care Medicine 155, 13761383. meri, K., Pekkanen, J., Kulmala, M., de Hartog, J., Puustinen, A., Ha Meliefste, K., ten Brink, H., Kos, G., Katsouyanni, K., Karakatsani, A., Kotronarou, A., Kavouras, I., Meddings, G., Thomas, S., Harrison, R., Ayres, J.G., van der Zee, S., Hoek, G., 2007. Spatial variation of particle number and mass over four European cities. Atmospheric Environment 41, 66226636. Qian, S., Sakurai, H., McMurry, P.H., 2007. Characteristics of regional nucleation events in Urban East St. Louis. Atmospheric Environment 41, 41194127. guez, S., Van Dingenen, R., Putaud, J.P., Martins-Dos Santos, S., Rodr Roselli, D., 2005. Nucleation and growth of new particles in the rural atmosphere of Northern ItalydRelationship to air quality monitoring. Atmospheric Environment 39, 67346746. guez, S., Van Dingenen, R., Putaud, J.P., DellAcqua, A., Pey, J., Rodr Querol, X., Alastuey, A., Chenery, S., Kin-Fai, H., Harrison, R.M., Tardivo, R., Scarnato, B., Gianelle, V., 2007. A study on the relationship