Variation and co-variation of PM10, particle number concentration, NOx and NO2 in the urban air – Relationships with wind speed, vertical temperature gradient and weather type

Atmospheric Environment 120 (2015) 317e327

Contents lists available at ScienceDirect

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

Variation and co-variation of PM10, particle number concentration, NOx and NO2 in the urban air e Relationships with wind speed, vertical temperature gradient and weather type € m a, *, C. Hak b, d, D. Chen c, M. Hallquist d, H. Pleijel a M. Grundstro a

University of Gothenburg, Department of Biological and Environmental Sciences, P.O. Box 461, 405 30 Gothenburg, Sweden NILU e Norwegian Institute for Air Research, Urban Environment and Industry Department, P.O. Box 100, NO-2027 Kjeller, Norway University of Gothenburg, Department of Earth Science, P.O. Box 460, 405 30 Gothenburg, Sweden d University of Gothenburg, Department of Chemistry and Molecular Biology, 412 96 Gothenburg, Sweden b c

h i g h l i g h t s
NOx was a good proxy of particle number concentration (PNC) in calm conditions.
PM10 was a weak proxy for PNC.
Air pollution was strongly related to Lamb Weather Types (LWT).
The PNC-NOx relationship prevailed on an hourly, daily and LWT basis.
A seasonal variation existed for the PNC-NOx relationship, being weakest in summer.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 10 February 2015 Received in revised form 11 July 2015 Accepted 20 August 2015 Available online 21 August 2015

Atmospheric ultrafine particles (UFP; diameter < 0.1 mm) represent a growing global health concern in urban environments and has a strong link to traffic related emissions. UFP is usually the dominating fraction of atmospheric particle number concentrations (PNC) despite being a minor part of total particle mass. The aim of this study was to empirically investigate the relationship between PNC and other air pollutants (NOX, NO2 and PM10) in the urban environment and their dependence on meteorology and weather type, using the Lamb Weather Type (LWT) classification scheme. The study was carried out in Gothenburg, Sweden, at an urban background site during April 2007eMay 2008. It was found that daily average [PNC] correlated very well with [NOx] (R2 ¼ 0.73) during inversion days, to a lesser extent with [NO2] (R2 ¼ 0.58) and poorly with [PM10] (R2 ¼ 0.07). Both PNC and NOx had similar response patterns to wind speed and to the strength of temperature inversions. PNC displayed two regimes, one strongly correlated to NOx and a second poorly correlated to NOx which was characterised by high wind speed. For concentration averages based on LWTs, the PNC-[NOx] relationship remained strong (R2 ¼ 0.70) where the windy LWT W deviated noticeably. Exclusion of observations with wind speed >5 ms1 or DT < 0  C from LWTs produced more uniform and stronger relationships (R2 ¼ 0.90; R2 ¼ 0.93). Low wind speeds and positive vertical temperature gradients were most common during LWTs A, NW, N and NE. These weather types were also associated with the highest daily means of NOx (~30 ppb) and PNC (~10 000 # cm3). A conclusion from this study is that NOx (but not PM10) is a good proxy for PNC especially during calm and stable conditions and that LWTs A, NW, N and NE are high risk weather types for elevated NOx and PNC. © 2015 Published by Elsevier Ltd.

Keywords: Urban air pollution NOx NO2 PM10 Proxy PNC Wind Atmospheric stability Lamb weather types

1. Introduction

* Corresponding author. €m). E-mail address: [email protected] (M. Grundstro http://dx.doi.org/10.1016/j.atmosenv.2015.08.057 1352-2310/© 2015 Published by Elsevier Ltd.

Urban air contains a mix of pollutants from various sources, where two important constituents are particulate matter (PM) and nitrogen oxides (NOx ¼ NO þ NO2). In the case of nitrogen oxides,

318

€m et al. / Atmospheric Environment 120 (2015) 317e327 M. Grundstro

nitrogen dioxide (NO2) is more often regulated, but sometimes air quality standards (AQS) are based on NOx. Nitrogen dioxide is known to cause adverse effects both on human health (WHO, 2006; Samoli et al., 2006; Chiusolo et al., 2011) and vegetation, e.g. epiphytic lichens (Hultengren et al., 2004; Gadsdon et al., 2010). In many European urban environments traffic is a dominant source for NOx and NO2 exposure (Hak et al., 2010). PM pollution is more complex, covering a large size range. It consists of a multi-component matrix originating from various sources and is subjected to several parallel atmospheric processes resulting in a dynamic quantity (Heal et al., 2012). In many urban areas, particle number concentration (PNC) is normally dominated by the ultrafine particles (UFP; diameter < 0.1 mm) fraction (Molnar et al., 2002; Charron and Harrison, 2003; Kittleson et al., 2004; Kumar et al., 2011). In Gothenburg, 90% of the PNC was attributed €ll and Hallquist, 2005). Generally, PNC to the UFP size fraction (Janha or UFP in urban environments is dominated by particles from traffic exhaust emissions, especially diesel engines (HEI, 2013), while PM10 often has large contributions from many other sources such as nonexhaust direct emissions from clutches, brakes and re-suspension of road dust from road gritting, road and tyre wear (Harrison et al., 2001; Ketzel et al., 2007; Johansson et al., 2007). The health effects of particles are clearly recognized but the mechanisms and how they link to size, composition and physical properties are unclear (Cassee et al., 2013). It has been suggested that UFPs formed by combustion in vehicle engines, may cause enhanced health effects. WHO recently upgraded diesel exhausts from ‘probably carcinogenic’ to ‘carcinogenic to humans’ (IARC, 2012), where UFPs make up an important component of the pollution mix and act as carriers of adsorbed carcinogenic substances (IARC, 2013). UFPs are known to cause oxidative stress in the lung (Li et al., 2008) and have been suggested to pose a higher risk for cardiovascular disease due to their specific properties such as high surface area, large particle number, metal and organic carbon content (Delfino et al., 2005). Also an increased number concentration of UFPs has been associ€ lzel et al., 2007). ated with increased cardiovascular mortality (Sto Furthermore, Seaton and Dennekamp (2003) proposed UFP exposure, proxied by NO2, to be the cause of cardiovascular effects, even at low NO2 concentrations. To the extent that PNC (UFP) and NOx emanate from the same combustion processes (in vehicle engines), one could expect that atmospheric concentrations of PNC (UFP) and NOx are strongly related. In fact, it has been suggested to use NOx, NO or NO2 as “proxies” for particle pollution levels (Ketzel et al., 2003; Seaton and Dennekamp, 2003; Sardar et al., 2012; Beevers et al., 2012). This assumption of a close relationship between PM and NOx or NO2 needs more empirical evidence from different environments, and can be questioned in the case of any mass based measure, such as PM10, while number based measures such as UFP can be expected to be more closely related to NOx or NO2. Even if NO2, rather than NOx, is a more frequently and easily measured compound (e.g. using passive diffusion samplers) using it as the proxy for PM adds further uncertainty to any association with PM, including UFP. NO and NO2 concentration is highly sensitive to the NOeNO2eO3 relationship (Leighton, 1961). The oxidation of NO with O3, and the photolysis of NO2, can strongly affect the urban NO2 concentration, which thus varies with respect to variables that do not necessarily affect PM concentration. In addition, the fraction of NO2 in the exhaust NOx emissions varies with the composition of the vehicle fleet (Carslaw et al., 2011). Diesel engines have a higher fraction of NO2 and certain types of particle filters fitted to large diesel vehicles emit high fractions of NO2 (Alvarez et al., 2008). Thus, there are several reasons to believe that NOx, rather than NO and NO2, is a more reliable proxy for PM emanating from combustion in vehicle engines.

It is well established that weather conditions, such as wind speed and temperature inversion, strongly affect the degree of accumulation of air pollutants near emission sources such as traffic in urban environments (Rodriguez et al., 2007; Jones et al., 2010; € m et al., 2011; Pearce et al., 2011). For PM pollution the Grundstro relationship is not as evident. The coarse particle fraction included in PM10, can be enhanced during higher wind speed and atmospheric turbulence by facilitating re-suspension and reducing sedimentation (Harrison et al., 2001; Gradon, 2009). Lamb Weather Types (LWTs) represent an effective method of characterising local meteorological conditions based on synoptic scale sea level pressure (Chen, 2000). The distribution of the sea level pressure provides information about the geostrophic wind and vorticity, which has been proven to be a useful summary of the local meteorological conditions (e.g. Tang et al., 2009). The principal aim of this study was to investigate the relationships between PNC, NOx, NO2 and PM10 in the urban environment and how the relationships are affected by specific meteorological variables and the general weather conditions represented by LWT. A number of hypotheses were postulated:
The first hypothesis was that NOx is a better proxy for PNC compared to NO2 and especially PM10, which can be motivated by the facts that's 1) NOx is less sensitive to the NOeNO2eO3 relationship which obviously is important for NO and NO2, 2) NOx and UFP represented by PNC are both mainly emitted from combustion, in contrast to PM10 which mostly dominated by coarse particles emanating from a number of different sources.
The second hypothesis was that NOx and PNC respond similarly to the variation of important meteorological variables such as wind speed and atmospheric stability. Both pollutants are expected to have high concentration during low wind speeds and strong atmospheric stability.
The third hypothesis was that the response of the air pollutant concentrations to changing meteorological variables can be associated to LWTs where LWTs with a large fraction of low wind speeds and strong atmospheric stability would more often result in high levels of NOx and PNC.

2. Material and methods 2.1. Data Monitoring of air quality and meteorological data was performed on a rooftop 30 m above ground level in the commercial district of Gothenburg city centre (“Femman”; 5742.5220 N, 11 58.2360 E). The site is located adjacent to the central terminal for bus and trains and approximately 300 m away from a busy traffic route (E45). Hourly measurements of NO and NO2 (Tecan CLD 700 AL chemiluminescence instrument), PM10 (TEOM - Tapered Element Oscillating Microbalance, Series 1400b), PNC (Condensation Particle Counter TSI 3775, size cut-off 4 nm), atmospheric pressure (Vaisala PA11A), air temperature and relative humidity (Campbell Rotronic MP101 thermometer/hygrometer), wind direction and wind speed (Gill ultrasonic anemometer) were carried out. The vertical air temperature gradient at two heights (3 m and 73 m above ground) was measured (RM Young platinum temperature probe model 41342) at a site located 8 km south of the city centre €rnbrott”; 57 38,84560 N, 11 55,60070 E). The difference in tem(“Ja perature between the two heights (ThighTlow) represents a measure of the atmospheric stability and is signified by DT throughout this article. The measurement period for the study lasted from April 2007 to May 2008.

€m et al. / Atmospheric Environment 120 (2015) 317e327 M. Grundstro

2.2. Lamb weather types Daily mean sea level pressure (MSLP) for 16 grid points centred over the Gothenburg City centre (57 70 N, 11970 E), were obtained from the NCEP/NCAR Reanalysis database 2.5  2.5 pressure fields (Kalnay et al., 1996). Circulation indices, u (westerly or zonal wind), v (southerly or meridional wind), V (combined wind speed), xu (meridional gradient of u), xv (zonal gradient of v) and x (total shear vorticity) describing the geostrophic winds and Lamb weather types (Jenkinson and Collison, 1977) were calculated following Chen (2000). The classification scheme has 26 weather types: anticyclone (A), cyclone (C), eight directional types (NE, E, SE, …), 16 hybrid types (ANE, AE, ASE, CNE, CE, CSE, …). In this study, the 26 weather types were consolidated into 10 LWTs according to the directions of the geostrophic wind, eight directional: NE, E, SE, S, SW, W, NW, N, and two rotational: A and C. 2.3. Calculations Air pollutant concentrations are expressed as mixing ratios (ppb) for NO2 and NOX, mass per volume (mg m3) for PM10 and number concentration (# cm3) for PNC. Hourly and daily averages of PNC were correlated with NO2, NOx and PM10 using regression analysis based on total least squares. Additionally, NO2, NOx, PM10 and PNC were averaged for intervals of wind speed (u) and DT, using steps of 0.5 ms1 and 0.5  C, respectively. To assess the individual response of each pollutant to the meteorological variables the pollution concentration averages for each interval were correlated with u and DT using the least squares method. Days with a daily average of DT exceeding 0  C, were defined as inversion days. Further analysis was also made of the effect from u and DT on the proxy relationship found between PNC and NOx (hourly values). For each LWT, averages of air pollutant concentrations were calculated and correlated with PNC. Additionally, time fractions of u below 1 ms1 and DT above 0  C were calculated for each LWT and correlated with each other using regression analysis. Finally, meteorological variables (atmospheric pressure, temperature, wind speed, relative humidity) were averaged for each LWT in order to characterise each weather type in terms of meteorology. 3. Results 3.1. Relationship of PNC with PM10, NO2 and NOX Hourly PNC had a weak (R2 ¼ 0.17) positive relationship with PM10 with large scatter (Fig. 1a). On a daily basis the corresponding relationship was even weaker (R2 ¼ 0.08; Fig. 1b) and considering only inversion days (black marks), defined as daily DT averages > 0  C, provided a relationship which was marginally weaker (R2 ¼ 0.11). PNC showed a stronger relationship (R2 ¼ 0.50) to [NO2] on an hourly basis (Fig. 1c) and during inversion days [NO2] explained 54% of the PNC variation (Fig. 1d). PNC showed the strongest positive relationship with [NOx], explaining 55% of the PNC variation on an hourly basis (Fig. 1e), but providing an even stronger relationship (R2 ¼ 0.73) for inversion days (Fig. 1f). Analysis was also conducted for PNC-NO which displayed similar but weaker relationship patterns in comparison with [NOx] (not shown). For hourly averages the R2 value for PNC-NO was 0.47 and for inversion days 0.67. At low [NOx] the PNC to [NOx] variation was larger for both daily and hourly observations (Fig. 1e and f), indicating the importance of partly different sources for PNC and NOx, and of chemical/physical transformation processes. At high [NOx] the relationship was more uniform and the highest [NOx] and PNC occurred during temperature inversions signified by the black marks of Fig. 1f. There seems

319

to exists a baseline relationship between [NOx] and PNC representing a level of PNC which it principally never falls below (dotted line in Fig. 1f). This can be interpreted as the minimum daily level of PNC that will be present in the city for any given NOx situation. 3.2. Dependence of NOx, NO2, PNC and PM10 on wind speed and atmospheric stability Wind speed (u) had a strong influence on the concentrations of the analysed air pollutants. Both [NO2] and [NOx] decreased monotonically with increasing u calculated on an interval basis (Fig. 2a and b). These relationships were negative and strongly nonlinear with much higher concentrations at u below 2 ms1. Relatively low concentrations were observed for u higher than approximately 3 ms1. A similar relationship was observed for PNC, where the highest concentrations were also observed for low u (Fig. 2c). At u > 3 ms1, PNC reached a stable level only fluctuating slightly between 7000 and 7300 # cm3. [PM10] showed a more complex relationship to u with initially decreasing concentrations at increasing u (Fig. 4d). However, concentrations increased again at u > 5 ms1, which demonstrates the importance of the capacity of strong winds associated with large mechanical turbulence to carry and re-suspend coarser particles such as dust from roads and other surfaces. Vertical mixing of the air is strongly linked to the temperature stratification near the ground. Fig. 3 show air pollutants and their relationship to the temperature difference between 73 m and 3 m (DT). Strong linear relationships were observed for NOx (R2 ¼ 0.96), NO2 (R2 ¼ 0.96) and PM10 (R2 ¼ 0.62) (Fig. 3a, b and d), while for [PNC] the relationship was more complex (Fig. 3c). Here, a negative relationship (R2 ¼ 0.95) was observed for negative temperature differences (DT < 0  C) and a positive relationship (R2 ¼ 0.98) at positive temperature difference (DT > 0  C). Observations with DT < 0.5  C were characterised by high temperatures (average 11  C) and u often (78% of the time) in the range 2e6 ms1 (average ¼ 4.5 ms1). Situations with DT > 0.5  C were characterised by lower temperatures (average 7  C) and u often (81% of the time) lower than 2 ms1 (average 1.6 ms1). High concentrations of all pollutants were observed when the temperature gradient was inverted (DT > 0  C) and showed large variation when the temperature gradient was very strong (DT > 3  C), indicating that extremely high levels of pollutants sometimes occur under strongly stable atmospheric conditions, during which the vertical dispersion is very limited. Fig. 4 illustrates the hourly proxy relationships between [PNC] and [NOX], [NO2] and [PM10], with consideration of the different degrees of dispersion expressed by DT or u. The relationship between PNC-[NOX] stood out and was most prominent at DT  0.5  C (R2 ¼ 0.85) and low u (u  2 ms1; R2 ¼ 0.82) (Fig. 4a and d). A second regime of high PNC (>20 000 # cm3) during low [NOX] (<70 ppb) was apparent at high u (u  5 ms1), and the association between PNC and [NOX] was weak (R2 ¼ 0.24). These situations were also always associated with very unstable conditions occurring when DT  0.5  C, and a frequently occurring (60%) wind flow from the west. At stable and calm conditions (DT > 0.5  C or u  2 ms1) the PNC-[NO2] relationship was also quite good (R2 ¼ 0.59 and R2 ¼ 0.62, Fig. 4b and e). PNC showed weak relationships to [PM10], with the highest R2 observed during stable (DT  0.5  C; R2 ¼ 0.27) and calm conditions (u  2 ms1; R2 ¼ 0.29). 3.3. LWT connection with meteorological and air quality parameters LWTs were analysed with respect to different meteorological

320

€m et al. / Atmospheric Environment 120 (2015) 317e327 M. Grundstro

Fig. 1. Relationships between hourly and daily averages of PNC with (a, b) PM10, (c, d) NO2 and (e, f) NOx. The black regression lines (b, d and f) illustrate days with temperature inversions, defined as days with an average difference in temperature between 73 and 3 m larger than 0.5  C. The grey regression lines illustrate the relationship for days with no temperature inversions. The dotted black line in Fig. 1f represents a linear relationship between PNC and NOx, which quantifies a level which PNC never falls below. All relationships are based on the total least squares method and statistical significance is based on the F-test.

variables prevailing on the surface/local scale in order to classify their meteorological character in greater detail (Table 1). The most common LWTs were A, SW, W, NW and C (combined representing ~75% of the time) and the least common were NE and SE (~5% combined). LWTs A, N and NW were generally calm and dry weather types, with high atmospheric pressure, low average u and moderate to low temperatures and humidity. LWTs C, SW and W on the other hand were wet and turbulent with low atmospheric pressure, high average u and mild temperatures. The fraction of low u (u < 2 ms1) was most common for LWTs A, NW, N, and NE. For each LWT, averages were calculated for [NOX], [NO2], PNC and [PM10] which can be seen in Table 2. The highest concentrations were normally found in LWTs A, NE and N with the exception of PM10 which showed high concentrations also in LWT W. Low concentration averages were generally found in LWTs SE, S, SW and

W for NOX and NO2. PNC was low in S and C conditions while [PM10] was low in C, SE, SW and NW. A correlation analysis was conducted for [NOX] and PNC averages based on LWT (Fig. 5). A strong (R2 ¼ 0.70) positive linear relationship (black line) was found (Fig. 5a). LWTs A, N and NE were associated with the highest levels for both pollutants. LWT W deviated from the fitted regression line and the relationship was tested with the exclusion of W (grey line) which resulted in a stronger linear relationship (R2 ¼ 0.86). The same analyses (not shown) was conducted for [NO2] and [PM10] and showed similar patterns but weaker relationships (R2 ¼ 0.58; R2 ¼ 0.36). Since windy and unstable situations resulted in a poorer association between PNC and [NOx], an analysis was conducted where such conditions were not included. In Fig. 5b, observations when u > 5 ms1 were excluded from all LWTs and the PNC-[NOx]

€m et al. / Atmospheric Environment 120 (2015) 317e327 M. Grundstro

a

160

y = 3.6 + 58.7x-1.28 R2 = 0.99 p < 0.005

140

50

y = 1.2 + 24.9x-0.76 R2 = 0.99 p < 0.005

45 40 Average [NO2], ppb

Average [NOX], ppb

120

b

100

80 60 40

35 30 25 20 15

20

10

0

5

-20

0 0

0 1 2 3 4 5 6 7 8 9 10 11

1

2

wind speed, m s-1

c 30000

3

4

5

6

7

8

9 10 11

wind speed, m s-1

y = 6405.7 + 7695.5x-1.43 R2 = 0.99 p < 0.005

d

60

y = 17.5 + 296.1x-0.65 - 307.1x0.58

50 Average [PM10], μg m-3

25000 Average [PNC], # cm-3

321

20000 15000 10000 5000

R2 = 0.95 p < 0.005

40 30 20 10

0

0 0

1

2

3

4

5

6

7

8

9 10 11

0

1

2

wind speed, m s-1

3

4

5

6

7

8

9 10 11

wind speed, m s-1

Fig. 2. Relationships between wind speed (u) and the concentration of (a) NOx, (b) NO2, (c) PNC and (d) PM10. Each point in the relationships is an average for a wind speed interval with steps 0e0.5, 0.5e1 ms1 and so on. The dotted lines show the standard deviation for each interval. The relationships are based on the least squares method and statistical significance is based on the F-test.

relationship was improved (R2 ¼ 0.90) and less scattered with LWT W better aligned. In Fig. 5c, observations with DT < 0 were excluded and again the PNC-NOx relationship was improved (R2 ¼ 0.93).

pollution levels, but it is hard to find a simple explanation for the pattern of intercept variation in the PNC-NOx relationships among LWTs. The correlation between PNC and NOx was strong and significant in all LWTs (Table 4).

3.4. Seasonal and LWT variation of the PNC-NOx relationship 4. Discussion A seasonal influence on air pollution levels is expected due to the annual variation in e.g. temperature. Table 3 shows the relationship between NOx and PNC during four seasons. The strongest relationship (R2 ¼ 0.74) was found during autumn (SON). The summer (JJA) months produced the poorest (R2 ¼ 0.27) PNC-NOx relationship when the average temperature was high (16.6  C). Evaporation of particles during warm temperatures (Bigi and Harrison, 2010) is a likely reason for reducing PNC in the summer hence reducing its association with [NOx]. Winter and spring produced relatively good relationships (R2 ¼ 0.57; R2 ¼ 0.52). The winter of the studied period was 2.3  C warmer and windier than the ten-year average. We analysed the regression statistics of the PNC-NOx relationships for individual LWTs (Table 4). For all LWTs except SE (containing few data and with a very small negative intercept), the intercepts varied between 2028 and 5441 # m3. This means that there was a substantial, but not very large variation in the background level of PNC not related to NOx. The largest intercepts were found for LWTs A and NW, these weather types generally had high

This study has shown that the variation in urban PNC had a strong positive relationship with [NOx] and the variation of both pollutants had strong and similar responses to the variation in u and also to DT when positive. The relationship between PNC and [NOx] was also stable for concentration averages based on different LWTs, representing different weather situations. All pollutants, apart from PM10, responded monotonically to increasing wind speed with non-linearly decreasing concentrations. [PM10] showed a negative relationship with u at low u, but a weak positive response at u > 3 ms1. Wind speed affects dispersion of PM from the sources at street-level (Jones et al., 2010) and associated mixing with urban background air (PNC ~2000 # cm3; €ll and Hallquist, 2005) causes lower averages of [NOX] Janha (~18 ppb) and PNC (~7000 # cm3) at higher u, while higher u in addition increases the strength of some of its sources for [PM10] (e.g. re-suspension of road dust). This explains why [PM10] can be elevated both at low u by poor dilution, and at high u by stronger re-suspension. In London, re-suspension of particles has been

€m et al. / Atmospheric Environment 120 (2015) 317e327 M. Grundstro

322

a

180

140

50

y = 3.3x + 13.1 R² = 0.96 p < 0.0001

45 40

120

Average [NO2], ppb

Average [NOX], ppb

b

y = 12.1x + 22.1 R² = 0.96 p < 0.0001

160

100 80 60

40

35 30 25 20

15

20

10

0

5

-20

0 -2

0

2

4

6

-2

0

∆T, °C

c

y = -3525.2x + 5157.9 R² = 0.95

25000 20000 15000 10000 5000

4

6

60 y = 0.9798x + 19.3 R² = 0.62 p < 0.0001

50 Average [PM10], μg m-3

Average [PNC], # cm-3

d

35000 30000

2 ∆T, °C

40 30 20 10 0

y = 2221.5x + 7705.5 R² = 0.98

0

-10 -2

0

2

4

∆T, °C

6

-2

0

2

4

6

∆T, °C

Fig. 3. Relationships between the vertical temperature difference between 73 m and 3 m (DT) and the concentrations of (a) NOx, (b) NO2, (c) PNC and (d) PM10. Each point of the relationships is an average for a temperature interval with steps 0e0.5, 0.5e1  C and so on. The dotted lines show the standard deviation for each interval. The relationships are based on the least squares method and statistical significance is based on the F-test.

observed (Jones et al., 2010) and in Stockholm 90% of the PM10 was attributed to road wear, i.e. not linked to vehicle exhaust emissions (Johansson et al., 2007). One may note that TEOM measurements of PM10 can include some humidity and composition bias (Muir, 2000; Charron et al., 2004; Gehrig et al., 2005). However, the effects described above cannot be explained by such artefacts. In response to DT, [NOx], [NO2] and [PM10] showed positive linear relationships, [PM10] with a weaker slope. For PNC levels two regimes were apparent in response to increasing DT, one regime where PNC decreased with DT (negative regime, at DT<0  C) and the second where PNC increased with DT (positive regime, at DT>0  C). Very high PNC levels were always observed in the positive regime coinciding with high [NOX], indicating a strong link to traffic emissions accumulating under strong atmospheric stability. The negative PNC regime was not related with [NOX] and was characterized by u mostly between 2 and 6 ms1 representing both moderately calm and windier conditions. Unstable conditions defined by a negative DT are driven by solar heating of the ground and the resulting thermal turbulence can assist vertical transport of emissions from ground level and could be a plausible explanation for the increase in PNC and PM10 observed at the rooftop. At negative DT [NOx] levelled out while [NO2] continued to decrease, which could potentially be due to increased photolysis of NO2 at stronger solar radiation. As hypothesised, the response of [NOx] and PNC was similar to u and partly similar to DT (at positive DT).

For the three categories of DT (Fig. 4a) the slope of the regression between PNC-NOX remained similar, despite their differing individual responses. This suggests that the proxy relationship per se is not very sensitive to DT, but during large negative DT, [NOx] was a poorer proxy of PNC indicated by the lower R2 value within this category. The high u category (Fig. 4d), however, did affect the slope and weakened the PNC-NOx relationship, implying that the proxy relationship is sensitive to u and that [NOx] is a poorer proxy of PNC during windy conditions. Both average of [PM10] (Fig. 2d) and the standard deviation of PNC (Fig. 2c) increased with stronger winds (u > 3 ms1) indicating a potential contribution from coarser particle fractions to the number count which are not related to NOx emissions. Another aspect further supporting this particle contribution is the fact that LWT W, obviously having a higher PNC:[NOx] relationship than the other LWTs (Fig. 5a), was also frequently associated with windy conditions suggesting a contribution of marine salt particles transported from the sea located directly west n (2000) demonstrated significant of the city. Gustafsson and Franze sea salt aerosol transport across southern Sweden during strong westerlies. This shows that a different PM source is sometimes observable in the PNC-NOX relationship at windy conditions which have a reducing effect on [NOx], but increasing effect on PM, which most probably is of a non-exhaust origin. The hypothesis that NOx is the best proxy of PNC is supported when stable and unventilated weather conditions prevail. At ventilated conditions none of the

€m et al. / Atmospheric Environment 120 (2015) 317e327 M. Grundstro

323

Fig. 4. Relationships between hourly averages of PNC and NOx (a and d), NO2 (b and e) and PM10 (c and f). The upper figures (a, b and c) show observations for three categories of the vertical temperature gradient. Light grey marks show hours when temperature difference between 73 m and 3 m (DT) is negative (DT < 0.5  C), e.g. well mixed situations. Black marks show hours with positive temperature difference (DT  0.5  C), e.g. limited vertical mixing situations. Dark grey marks show hours with both negative and positive temperature difference (0.5  C < DT < 0.5  C). The lower figures (d, e and f) show observations for three categories of wind speed (u). Light grey marks show observations during windy conditions (u  5 ms1), black marks show observations during calm conditions (u  2 ms1) and dark grey marks show observations during an intermediate wind speed category (2 ms1 < u < 5 ms1).

Table 1 Mean and standard deviation (sd) of meteorological variables: atmospheric pressure (P, hPa), wind speed (u, ms1), air temperature (T,  C), vertical temperature difference between 73 m and 3 m (DT,  C), relative humidity (RH, %), global solar radiation (Rad, W m2) of different Lamb Weather Types (LWTs). f denotes time fraction. u

DT

T

RH

Rad

LWT

P Mean

sd

Mean

sd

Mean

sd

Mean

sd

Mean

sd

Mean

sd

A NE E SE S SW W NW N C

1023 1012 1016 1016 1004 1005 1006 1006 1012 996

9.7 7.2 7.2 9.5 8.9 9.3 9.8 8.5 10.6 10.5

2.6 3.5 3.7 4.1 4 4.9 5.2 4.3 3.2 4

1.3 2.0 1.7 1.6 1.8 1.9 2.3 2.3 1.8 2.1

7.2 9.8 15.4 9.5 9.8 8.2 9.2 9.7 6 8.9

5.7 9.1 6.1 8.5 7.0 5.0 4.7 5.8 5.6 5.8

0.22 0.06 0.12 0.37 0.23 0.4 0.3 0.11 0.11 0.34

1.65 1.52 1.23 0.80 0.96 0.62 1.14 1.52 1.50 0.95

80 71 67 72 80 92 88 77 73 89

19 20 22 17 21 11 13 17 18 13

104 156 158 100 105 56 77 123 111 76

182 210 196 171 169 122 143 184 167 131

tested pollutants were a good proxy of PNC. The noise caused by the source complexity of PM as described above, was also observed in the PNC-NOX relationship based on LWT. When excluding observations with u > 5 ms1 from the regression analysis, the overall scatter was reduced and the LWT W observation deviated less from the regression line. A possible explanation is that a potential source contribution to PNC from

fu<2

fu5

f DT  0.5

f DT < 0.5

0.39 0.28 0.16 0.08 0.16 0.06 0.09 0.21 0.28 0.17

0.07 0.21 0.23 0.3 0.28 0.45 0.54 0.39 0.15 0.29

0.27 0.19 0.19 0.09 0.13 0.04 0.07 0.15 0.24 0.08

0.47 0.48 0.47 0.5 0.42 0.36 0.39 0.41 0.39 0.45

marine salt particles was reduced and thus improved the overall association between PNC and [NOx] on an LWT basis (Fig. 5b). Similar wind driven marine source attributions to particle number with PM size 13e800 nm have been made in Barcelona due to easterly wind directions (Pey et al., 2009). However, to definitely verify a marine salt particle contribution to PNC in our study, data on particle size fractions and composition would be necessary.

€m et al. / Atmospheric Environment 120 (2015) 317e327 M. Grundstro

324

Table 2 Number of hours (n), time fractions (f) and average concentrations of NOx (ppb), NO2 (ppb), PM10 (mg m3) and PNC (particle number concentration,# cm3) of different Lamb Weather Types (LWTs). sd denotes standard deviation.

A NE E SE S SW W NW N C

n hr LWT

f LWT

1368 312 600 216 528 1248 1584 1184 576 1464

NO2 (ppb)

0.15 0.03 0.07 0.02 0.06 0.14 0.17 0.13 0.06 0.16

a

PM10 (mg m3)

NOx (ppb)

PNC(# cm3)

Mean

sd

Mean

sd

Mean

sd

Mean

sd

14.5 18.3 13.6 8.6 9 9.3 10 11.5 17.9 11.1

10.2 10.6 8.9 5.7 6.1 6.2 7.6 10.1 12.0 7.8

29.5 31.5 20.1 11.6 13.6 11.7 14.2 20.6 34.1 16.2

37.9 29.0 18.2 8.7 17.2 12.4 21.3 33.1 38.7 17.4

19.8 24.4 19.8 17.7 19.3 18.5 21 17.9 21.7 14.5

11.5 13.6 12.4 9.9 13.5 7.5 10.7 10.0 20.6 8.2

10479 9709 7103 4833 5836 6577 9003 9187 10400 6548

6824 6410 4441 3802 3392 3418 5598 6111 6634 3774

14000 y = 238.3x + 3127 R² = 0.70 p = 0.002

12000

PNC, # cm-3

10000

A

NW

W

N NE

8000 E

C SW S SE

6000 4000

c

y = 242.3x + 2766 R² = 0.86 p = 0.0002

2000

10

20

30

12000 40

NOx, ppb

b

u < 5 m s-1

A NNW NE W

10000 E SE SW SC

8000 6000 4000

14000 y = 225.2x + 3149 R2 = 0.90 p < 0.0001

12000 10000 PNC, # cm-3

y = 203.9x + 3144 R2 = 0.93 p < 0.0001

14000

0 0

∆T ≥ 0°C

16000

PNC, # cm-3

LWT

8000 SW C S SE

6000

N A NE NW

W E

2000 0 0

20

40

60

NOx, ppb

4000 2000 0 0

10

20

30

40

NOx, ppb Fig. 5. Relationships between [PNC] and [NOx] for different LWTs. The grey regression line (a) excludes LWT W from the relationship, and the black regression line includes all LWTs. The relationship between [PNC] and [NOx] (b) only includes observations with low wind speed (u < 5 ms1) and (c) only includes observations with positive temperature gradient (DT > 0.5  C). All regressions are based on the total least squares method and statistical significance is based on the F-test.

Despite the deviation by LWT W, the positive relationship between PNC and [NOx] for averages based on LWTs remained strong. LWTs with a high frequency of low u and high degree of atmospheric stability (A, NE, N and NW) were associated with higher PNC and [NOx], while LWTs with a large frequency in windy and unstable conditions (C, SE, S, SW) appeared in the lower end. This supports the hypothesis that the meteorological properties governing high levels of both NOX and PNC are strongly linked to the occurrence of

calm and stable conditions also when aggregated as LWT averages. Had the frequency of low u and positive DT been more evenly distributed among LWTs, the PNC-NOx relationship would probably have been weaker on an LWT basis. At ground level stations in London (Bigi and Harrison, 2010) and Stockholm (Johansson et al., 2007), the relationship between PNC and [NOX] showed similar patterns to our investigation, with the exception of the wind driven PNC regime observed in this study.

€m et al. / Atmospheric Environment 120 (2015) 317e327 M. Grundstro

325

Table 3 Relationship between hourly PNC and [NOx], and averages of temperature (T) and wind speed (u) for winter (DJF), spring (MAM), summer (JJA) and autumn (SON) during the studied period and 2000e2009. Statistical significance (***p < 0.001) is based on the F-test. Studied period

2000e2009

Season

Slope

Intercept

R2

Significance

T ( C)

u (ms1)

T ( C)

u (ms1)

DJF MAM JJA SON

163.5 138.2 111.1 130.8

3569 4806 5711 5052

0.57 0.52 0.27 0.74

*** *** *** ***

3.8 7.6 16.6 8.4

4.8 3.9 3.9 3.7

1.5 7.4 16.8 9.4

4.0 3.7 3.6 3.8

Table 4 Statistics related to the relationship between hourly PNC and NOx during different LWTs. Statistical significance (***p < 0.001) is based on the F-test. LWT

Intercept

Std. error intercept

Slope

Std. error slope

R2

Significance

A NE E SE S SW W NW N C

5160 2795 2211 263.4 3165 3373 4802 5441 4552 2028

122.9 353 182.9 351.1 116.2 109.6 144 146.4 182.9 128.6

180.2 225.6 242.6 439.2 196.7 272.5 305.3 181.8 171.6 314.3

2.6 8.4 6.7 24.3 5.3 6.4 6.3 3.7 3.6 6.9

0.72 0.57 0.55 0.34 0.62 0.33 0.37 0.53 0.75 0.36

*** *** *** *** *** *** *** *** *** ***

Daily averages were used in Stockholm. This time resolution seems to hide a possible wind driven PNC regime, which was obvious in the present study when an hourly time resolution was used. Another possible explanation could be that the wind driven source of PNC influences particle concentrations to a lesser degree in Stockholm. Both intercept and slope of the PNC-NOx relationship was larger in Stockholm, most likely due to differences in the actual vehicle fleet composition affecting the two monitoring sites. Since no data for particle size distribution were available during the period of these measurements, the extent to which PNC can be attributed to traffic emissions could not be fully established. However, previous measurement at the monitoring site used in the present study during a short winter time period illustrated a pronounced UFP peak during morning rush hours which coincided €ll et al., 2006). The number-size with peaks of NOx and CO (Janha distributions peaked at 40 nm which is typical for combustion generated particles being the dominant source during winter and especially during temperature inversions (Olofson et al., 2009). Furthermore, measurements at numerous other places have confirmed that a majority of the PNC in urban environments are found in the ultrafine size range and originate from vehicle exhausts (Pey et al., 2009; Woo et al., 2010; Kumar et al., 2011). On most spatial scales (except for road tunnels), the dilution process has the fastest time scale for particle transformation in comparison with coagulation and deposition (Kumar et al., 2011), and is expected to have occurred to a considerable extent when the ground level emission plume has reached the roof-top. There are studies showing that PNC can be up to 5 times larger at street-level than €kev€ roof-top levels (Va a et al., 1999; Kumar et al., 2008). Photochemically induced nucleation of new particles can be of importance for UFP concentrations (Rodriguez et al., 2007) and can be promoted by high u at roof-top level (Kumar et al., 2008). Freshly nucleated particles (UFP) may have added to the PNC in our study but is most probably of minor importance when only considering the strong association between PNC and [NOx] at stable conditions. Urban air pollution monitoring normally only includes legally regulated gases and therefore long-term measurements of UFP or PNC are scarce. Indirectly, NOX gives an indication of vehicle exhaust particles, but different after-treatment equipment (with or without diesel particle filters) on diesel vehicles cause trade-offs

between emissions of NOx and particles (Hallquist et al., 2013). Thus, to create a strategy for mitigating PNC and or NOx and using the information presented in this paper the variability in exhaust composition between different vehicle types must be taken into consideration. The lack of UFP data in cities throughout the world prevents any direct assessment of the long term evolution of UFPs in the ambient air, which is problematic for studying epidemiological effects from UFP exposure. Incorporating situations giving rise to high [NOX] and PNC may be of help but still does not completely single out the direct UFP effect, which is important for setting up accurate exposure guidelines and legislation (Morawska et al., 2008). Furthermore, multi-pollutant exposure may be more relevant when determining health effects such as cardiovascular risk (Brook et al., 2010) and our study has shown that all pollutants analysed, although all having somewhat different relationships with u and DT, tend to reach the highest concentrations during stable atmospheric conditions represented by low u, positive DT and LWTs A, NE and N, which can be characterised as high risk weather types with respect to the air pollutants covered by this study.

5. Conclusions
PNC had a positive relationship with all pollutants studied, where hourly [NOx] was the better proxy of PNC (R2 ¼ 0.55) as compared to [NO2] (R2 ¼ 0.50) and especially [PM10] (R2 ¼ 0.17). This is of large significance for the use of proxies in studies of the effects of urban pollutants on health.
During inversion days (daily DT > 0  C) the relationship between daily PNC and [NOx] (R2 ¼ 0.73) and [NO2] (R2 ¼ 0.54), respectively, was stronger. Low wind speeds (<2 ms1) were associated with high levels of NOx, PNC and PM10. [PM10] differed from the other pollutants in that concentrations increased again with wind speed when exceeding 3 ms1.
LWTs associated with a high fraction of very low wind speeds (<2 ms1) were A, NW, N and NE. They all represented high levels of NOx, NO2, PNC and PM10. For PM10 the difference was small between low wind speed LWTs and the more turbulent LWTs.

326

€m et al. / Atmospheric Environment 120 (2015) 317e327 M. Grundstro


PNC had a weaker positive relationship (R2 ¼ 0.24) to [NOx] at stronger winds (u  5 ms1) frequently occurring during LWT W, possibly reflecting a substantial particle contribution from sea salt particles not associated with NOx emissions. Acknowledgements Thanks are due to the Environment Administration of the City of Gothenburg for providing air pollution and meteorological data €rnbrott. Thanks are also from the monitoring sites Femman and Ja due to the Swedish Environmental Protection Agency and Country €stra Go €taland for financial support of the particle Board of Va counter. References Alvarez, R., Weilenmann, M., Favez, J.-Y., 2008. Evidence of increased mass fraction of NO2 within real-world NOx emissions of modern light vehicles d derived from a reliable online measuring method. Atmos. Environ. 42, 4699e4707. Beevers, S.S., Westmoreland, E., de Jong, M.C., Williams, M.L., Carslaw, D.C., 2012. Trends in NOx and NO2 emissions from road traffic in Great Britain. Atmos. Environ. 54, 107e116. Bigi, A., Harrison, R.M., 2010. Analysis of the air pollution climate at a central urban background site. Atmos. Environ. 44, 2004e2012. Brook, R.D., Rajagopalan, S., Pope, C.A., Brook, J.R., Bhatnagar, A., Diez-Roux, A.V., Holguin, F., Hong, Y.L., Luepker, R.V., Mittleman, M.A., Peters, A., Siscovick, D., Smith, S.C., Whitsel, L., Kaufman, J.D., 2010. Particulate matter air pollution and cardiovascular disease an update to the scientific statement from the American Heart Association. Circulation 121, 2331e2378. Carslaw, D.C., Beevers, S.D., Tate, J.T., Westmoreland, E.J., Williams, M.L., 2011. Recent evidence concerning higher NOx emissions from passenger cars and light duty vehicles. Atmos. Environ. 45, 7053e7063. roux, M., Gerlofs-Nijland, M.E., Kelly, F.J., 2013. Particulate matter Cassee, F.M., He beyond mass: recent health evidence on the role of fractions, chemical constituents and sources of emission. Inhal. Toxicol. 25, 802e812. Charron, A., Harrison, R.M., 2003. Primary particle formation from vehicle emissions during exhaust dilution in the roadside atmosphere. Atmos. Environ. 37, 4109e4119. Charron, A., Harrison, R.M., Moorcroft, S., Booker, J., 2004. Quantitative interpretation of divergence between PM10 and PM2.5 mass measurement by TEOM and gravimetric (Partisol) instruments. Atmos. Environ. 38, 415e423. Chen, D., 2000. A monthly circulation climatology for Sweden and its application to a winter temperature case study. Int. J. Climatol. 20, 1067e1076. Chiusolo, M., Cadum, E., Stafoggia, M., Galassi, C., Berti, G., Faustini, A., Bisanti, L., Vigotti, M.A., Patrizia Dessì, M., Cernigliaro, A., Mallone, S., Pacelli, B., Minerba, S., Simonato, L., Forastiere, F., 2011. Short-term effects of nitrogen dioxide on mortality and susceptibility factors in 10 Italian cities: the EpiAir study. Environ. Health Perspect. 119, 1233e1238. Delfino, R.J., Sioutas, C., Malik, S., 2005. Potential role of ultrafine particles in associations between airborne particle mass and cardiovascular health. Environ. Health Perspect. 113, 934e946. Gadsdon, S.R., Dagley, J.R., Wolseley, P.A., Power, S.A., 2010. Relationships between lichen community composition and concentration of NO2 and NH3. Environ. Pollut. 158, 2553e2560. Gradon, L., 2009. Resuspension of particles from surfaces: technological, environmental and pharmaceutical aspects. Adv. Powder Technol. 20, 17e28. Gehrig, R., Hueglin, C., Schwarzenbach, B., Seitz, T., Buchmann, B., 2005. A new method to link PM10 concentrations from automatic monitors to the manual gravimetric reference method according to EN12341. Atmos. Environ. 39, 2213e2223. € m, M., Linderholm, H.W., Klingberg, J., Pleijel, H., 2011. Urban NO2 and NO Grundstro pollution in relation to the North Atlantic oscillation NAO. Atmos. Environ. 45, 883e888. n, L.G., 2000. Inland transport of marine aerosols in Gustafsson, M.E.R., Franze southern Sweden. Atmos. Environ. 34, 313e325. Hak, C., Larssen, S., Randall, S., Guerreiro, C., Denby, B., Hor alek, J., 2010. Traffic and Air Quality - Contribution of Traffic to Urban Air Quality in European Cities. European Topic Centre on Air and Climate Change, Bilthoven (ETC/ACC Technical paper, 2009/12). €, M., Fallgren, H., Westerlund, J., Sjo € din, Å., 2013. Particle and Hallquist, Å.M., Jerksjo gaseous emissions from individual diesel and cng buses. Atmos. Chem. Phys. 13, 5337e5350. Harrison, R.M., Yin, J., Mark, D., Stedman, J., Appleby, R.S., Booker, J., Moorcroft, S., 2001. Studies of the coarse particle (2.5e10m m) component in UK urban atmospheres. Atmos. Environ. 35, 3667e3679. Heal, M., Kumar, P., Harrison, R.M., 2012. Particle, air quality, policy and health. R. Soc. Chem. 41, 6606e6630. HEI Review Panel on Ultrafine Particles, 2013. Understanding the Health Effects of Ambient Ultrafine Particles. Health Effects Institute, Boston, MA. n, H., Pleijel, H., 2004. Recovery of the epiphytic lichen flora Hultengren, S., Grale

following air quality improvement in south-west Sweden. Water, Air Soil Pollut. 154, 203e211. IARC, 2012. Diesel Engine Exhaust Carcinogenic. Press release No 213, June 2012. International Agency for Research on Cancer, Lyon. IARC, 2013. Diesel and Gasoline Engine Exhausts and Some Nitroarenes. In: IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, vol. 105. IARC Working Group on the Evaluation of Carcinogenic Risks to Humans, Lyon. Janh€ all, S., Hallquist, M., 2005. A novel method for determination of size-resolved, submicrometer particle traffic emission factors. Environ. Sci. Technol. 39, 7609e7615. Janh€ all, S., Olofson, Frans G., Andersson, P.U., Pettersson, J.B.C., Hallquist, M., 2006. Evolution of the urban aerosol during winter temperature inversion episodes. Atmos. Environ. 40, 5255e5266. Jenkinson, A.F., Collison, B.P., 1977. An initial climatology of gales of the North Sea. Synop. Climatol. Branch Memo. 62. Johansson, C., Norman, M., Gidhagen, L., 2007. Spatial and temporal variations in PM10 and particle number concentration in urban air. Environ. Monit. Assess. 127, 477e487. Jones, A.M., Harrison, R.M., Baker, J., 2010. The wind speed dependence of the concentrations of airborne particulate matter and NOx. Atmos. Environ. 44, 1682e1690. 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, K.C., Ropelewski, C., Wang, J., Leetmaa, A., Reynolds, R., Roy, J., Joeseph, D., 1996. The NCEP/NCAR 40-Year reanalysis project. Am. Meteorological Soc. 77, 437e471. Ketzel, M., Wåhlin, P., Berkowicz, R., Palmgren, F., 2003. Particle and trace gas emission factors under urban driving conditions in Copenhagen based on street and roof-level observations. Atmos. Environ. 37, 2735e2749. Ketzel, M., Omstedt, G., Johansson, C., Düring, I., Pohjola, M., Oettl, D., Gidhagen, L., Wåhlin, P., Lohmeyer, A., Haakana, M., Berkowicz, R., 2007. Estimation and validation of PM2.5/PM10 exhaust and non-exhaust emission factors for practical street pollution modeling. Atmos. Environ. 41, 9370e9385. Kittleson, D.B., Watts, W.F., Johnson, J.P., 2004. Nanoparticle emissions on Minnesota highways. Atmos. Environ. 38, 9e19. Kumar, P., Fennell, P., Langley, D., Britter, R., 2008. Pseudo-simultaneous measurements for the vertical variation of coarse, fine and ultrafine particles in an urban street canyon. Atmos. Environ. 42, 4304e4319. Kumar, P., Ketzel, M., Vardoulakis, S., Pirjola, L., Britter, R., 2011. Dynamics and dispersion in the urban atmospheric environment-A review. Atmos. Environ. 42, 580e603. Leighton, P.A., 1961. Photochemistry of Air Pollution. Academic Press, New York. Li, N., Xia, T., Nel, A.E., 2008. The role of oxidative stress in ambient particulate matter-induced lung diseases and its implications in the toxicity of engineered nanoparticles. Free Radic. Biol. Med. 44, 1689e1699. €ll, S., Hallquist, M., 2002. Roadside measurements of fine and ulMolnar, P., Janha trafine particles at a major road north of Gothenburg. Atmos. Environ. 36, 4115e4123. Morawska, L., Ristovski, Z., Jayaratne, E.R., Keogh, D.U., Ling, X., 2008. Ambient nano and ultrafine particles from motor vehicle emissions: characteristics, ambient processing and implications on human exposure. Atmos. Environ. 42, 8113e8138. Muir, D., 2000. New directions: the suitability of tapered element oscillating microbalances (TEOMs) for PM10 monitoring in Europe. The use of PM10 data as measured by TEOM for compliance with the European air quality standard. Atmos. Environ. 34, 3209e3211. € m, E., Tang, L., Chen, D., Olofson, K.F.G., Andersson, P.I., Hallquist, M., Ljungstro Pettersson, J.B.C., 2009. Urban aerosol evolution and particle formation during wintertime temperature inversions. Atmos. Environ. 43, 340e346. Pearce, J.L., Beringer, J., Niocholls, N., Hyndman, R.J., Taspper, N.J., 2011. Quantifying the influence of local meteorology on air quality using generalized additive models. Atmos. Environ. 45, 1328e1336. Pey, J., Querol, X., Alastuey, A., Rodriguez, S., Putaud, J.P., van Dingnenen, R., 2009. Source apportionment of urban fine and ultra-fine particle number concentration in a Western Mediterranean city. Atmos. Environ. 43, 4407e4415. Rodriguez, S., Van Dingenen, R., Putaud, J.-P., Dell'Acqua, A., Pey, J., Querol, X., Alastuey, A., Chenery, S., Ho, K.-F., Harrison, R., Tardivo, R., Scarnato, B., Gemelli, V., 2007. A study on the relationship between mass concentrations, chemistry and number size distribution of urban fine aerosols in Milan, Barcelona and London. Atmos. Chem. Phys. 7, 2217e2232. Samoli, E., Aga, E., Touloumi, G., Nisiotis, K., Forsberg, B., Lefranc, A., Pekkanen, J., Wojtyniak, B., Schindler, C., Niciu, E., Brunstein, R., Dodi c Fikfak, M., Schwartz, J., Katsouyanni, K., 2006. Short-term effects of nitrogen dioxide on mortality: an analysis within the APHEA project. Eur. Respir. J. 27, 1129e1137. Sardar, S.B., Fine, P.M., Yoon, H., Sioutas, C., 2012. Associations between particle number and gaseous co-pollutant concentrations in the Los Angeles Basin. J. Air & Waste Manag. Assoc. 54, 992e1005. Seaton, A., Dennekamp, M., 2003. Hypothesis: Ill health associated with low concentrations of nitrogen dioxide e an effect of ultrafine particles? Thorax 58, 1012e1015. € lzel, M., Breitner, S., Cyrys, J., Pitz, M., Woelke, G., Kreyling, W., Heinrich, J., Sto Weichmann, H.E., Peters, A., 2007. Daily mortality and particulate matter in different size classes in Erfurt, Germany. J. Expo. Sci. Environ. Epidemiol. 17, 458e467. Tang, L., Chen, D., Karlsson, P.E., Gu, Y., Ou, T., 2009. Synoptic circulation and its

€m et al. / Atmospheric Environment 120 (2015) 317e327 M. Grundstro influence on spring and summer surface ozone concentrations in southern Sweden. Boreal Environ. Res. 14, 889e902. €, M., Ha €meri, K., Kulmala, M., Lahdes, R., Ruuskanen, J., Laitinen, T., 1999. V€ akeva Street level versus rooftop concentrations of submicron aerosol particles and gaseous pollutants in an urban street canyon. Atmos. Environ. 33, 1385e1397.

327

WHO, 2006. WHO Air Quality Guidelines for Particulate Matter, Ozone, Nitrogen Dioxide and Sulphur Dioxoide. Gobal update 2005. Woo, K.S., Chen, D.R., Pui, D.Y.H., McMurry, P.H., 2010. Measurement of Atlanta aerosol size distributions: observation of ultrafine particle events. Aerosol Sci. Technol. 34, 75e87.