Traffic aerosol emission velocity derived from direct flux measurements over urban Stockholm, Sweden

Traffic aerosol emission velocity derived from direct flux measurements over urban Stockholm, Sweden

Atmospheric Environment 45 (2011) 5725e5731 Contents lists available at ScienceDirect Atmospheric Environment journal homepage: www.elsevier.com/loc...

1MB Sizes 0 Downloads 53 Views

Atmospheric Environment 45 (2011) 5725e5731

Contents lists available at ScienceDirect

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

Traffic aerosol emission velocity derived from direct flux measurements over urban Stockholm, Sweden M. Vogt a, *, E.D. Nilsson a, L. Ahlm a, E.M. Mårtensson a, c, H. Struthers a, C. Johansson a, b a

Department of Applied Environmental Science (ITM), Stockholm University, SE-10691 Stockholm, Sweden City of Stockholm Environment and Health Administration, Box 8136, SE-10420 Stockholm, Sweden c Department of Earth Sciences, Uppsala University, SE-752 36 Uppsala, Sweden b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 7 April 2011 Received in revised form 13 July 2011 Accepted 14 July 2011

Size-resolved aerosol vertical number fluxes were measured using the eddy covariance method, 105 meters above the ground over the city of Stockholm, Sweden, between 1st April 2008 and 15th April 2009. The size range of the measurements cover particles from 0.25 to 2.5 mm diameter (Dp). Emission velocities (ve) were calculated for the same size range and were found to be well correlated with friction velocity (u* ) and CO2 fluxes (FCO2 ). These variables were used to parameterize the emission velocity as

Keywords: Primary aerosol emissions Carbon dioxide emissions Traffic aerosol Urban aerosol Traffic activity Emission factors Eddy covariance Aerosol flux

    ve ¼ 129:14 0:01D2p þ 0:06Dp þ 0:008 þ 1:22 u* FCO2 where ve and u* are given in [m s1], Dp in [mm], and FCO2 in [mmol m2s1]. The parameterization reproduces the average diurnal cycle from the observations well for particles sizes up to 0.6 mm Dp. For larger particles the parameterization tends to over predict the emission velocity. These larger particles are not believed to be produced by combustion and therefore not well represented by FCO2 , which represents the traffic source through its fossil fuel consumption and the related CO2 emissions. Published by Elsevier Ltd.

1. Introduction Aerosols significantly influence air quality, and regional and global climate; the later due to aerosols influence on the radiative properties of the atmosphere (IPCC, 2007). In the last 20 years, the adverse health effects of atmospheric particulate matter have been highlighted, in particular the fine particle fraction (<2.5 mm in diameter) has been recognized as a serious hazard due to the fact that these particles can penetrate deep into the lunges (WHO, 2005). A large research effort has been focused on urban air quality. Cities and conurbations are net sources of aerosols, the urban plume consisting of a characteristic mixture of aerosols as well as volatile, semi-volatile gases and radical species. The so called airborne particulate matter (PM) is expressed in terms of number, surface area, mass or volume (Harrison et al., 2000). Most research and regulations are devoted to PM10 and PM2.5 concentrations. The PM10 fraction consists of particles with an aerodynamic diameter of

* Corresponding author. E-mail address: [email protected] (M. Vogt). 1352-2310/$ e see front matter Published by Elsevier Ltd. doi:10.1016/j.atmosenv.2011.07.026

less than 10 micrometers and the PM2.5 of particles with an aerodynamic diameter of less than 2.5 micrometers. Particle concentrations have been measured on street and roof levels in several European cities (Norman and Johansson, 2006; Longley et al., 2004a,b; Ketzel et al., 2007). There are however, very few studies regarding direct emission measurements in cities using micrometeorological techniques like the eddy covariance method. Such measurements provide direct, quantitative information about the sources and sinks of particles (Mårtensson et al., 2006). Aerosol concentration measurements contain the information required for population exposure studies however, emission information is a fundamental requirement for accurate aerosol modeling and the prediction of future pollution levels. With regard to the question of determining particle emissions from vehicles, the eddy covariance method is the only practical method able to provide emission factors for an entire vehicle fleet under real world conditions. Using this method, Dorsey et al. (2002), Mårtensson et al. (2006) and Järvi et al. (2009) found positive correlations between aerosol flux and traffic flow, which were used to parameterize particle emissions for each data set. The measurements were able to discern specific differences in the emissions for different seasons, city transport patterns, as well as

5726

M. Vogt et al. / Atmospheric Environment 45 (2011) 5725e5731

vehicle and fuel mixtures however, little work has been completed on the aerodynamic transport behavior or mechanisms (Dorsey et al., 2002; Longley et al., 2004b). Due to computational constraints, complex nonlinear sub-grid scale aerosol processes including turbulent vertical transport are typically treated in a rudimentary way in regional (Ketzel and Berkowicz, 2004) and air quality models, and there is a need for better quantification and parameterization of this process based on micrometeorological and traffic intensity parameters. Dorsey et al. (2002) parameterized the total aerosol number (>10 nm diameter) emissions using the concept of emission velocity. While emission velocity is defined in a similar way to deposition velocity (the aerosol flux normalized by the aerosol concentration), it differs in one important aspect; Deposition fluxes can be considered to be driven by the aerosol concentration i.e. the more particles in the air, the larger the deposition flux. The deposition velocity works as a scale velocity for the deposition process. On the other hand, in the real world emissions should be independent of the atmospheric aerosol concentration. The emission velocity can however provide important insight regarding the processes that are most important for quantifying the emissions. This paper presents measurements of particle vertical number fluxes and transport velocities over Stockholm, the capital of Sweden with about 1,000,000 inhabitants. Previous studies have shown that the majority, by number, of aerosol particles in Stockholm lie within the size range 10e1000 nm Dp and the dominant source of these particles is road traffic exhaust emissions (e.g. Gidhagen et al., 2005). The results shown here concern size resolved aerosol vertical fluxes in the range of 0.25 mm to 2.5 mm dry diameter. We parameterize the aerosol emission velocity in the size range of (Dp ¼ 0.25e2.5 mm) over an urban area, where the aerosol emissions are dominated by traffic sources. 2. Measurement site and instrumentation The measurements were made in Stockholm (Sweden), from the top of a telecommunication tower in the southern central part of the city (Latitude North 59 180 0.4300 , Longitude: East 18 50 53.1700 ). The tower is constructed from concrete, 105 m tall with the base 28 m above sea level. On the top of the tower (90 m) is an elevator machine room and on top of that is an 11 m high metal frame with a 2.5  2.5 m platform at the top. The actual measurement height was 4 m above this platform and 105 m above the surrounding ground. The platform enables us to extend the flux measurements far enough from the bulkier concrete construction to avoid any flow distortion caused by the tower. Central Stockholm, with its associated high level of traffic activity, is situated to the north of the tower. A large forest area dominates in the easterly direction and significant green sectors mixed with residential areas can be found from the east through to the south-west (Fig. 1). The site has been previously described by Mårtensson et al. (2006) and Vogt et al. (in press, 2011). 2.1. Instruments and measurement setup The instrumentation consists of a Gill R3 ultrasonic anemometer, an open path infrared CO2/H2O analyzer LI-COR 7500 (LI-COR, Inc., Lincoln, Nebraska 68504, USA), and two identical Optical Particle Counters (OPC) (Model, 1.109, Grimm Ainring, Bayern, Germany) in a housing with a system to heat and dry the sampled air (Grimm Model 265, special version going up to 300  C). The sampled air was dried by 1:1 dilution with close to 0% humidity particle free air, minimizing the risk of unwanted loss of semivolatile compounds, which occurs when the air is simply heated in order to dry it. In order to sample data from the 1.109 OPC at the

Fig. 1. The location of the tower where the measurements took place, and its surroundings. Blue ¼ open water surfaces, green ¼ forest/park areas, brown/orange/ grey ¼ built-up areas (mainly residential areas), public buildings (schools, sport arenas etc), black ¼ roads.

maximum rate of 1 Hz we could only sample either the 15 smallest (0.25e2.5 mm) or 15 largest channels (2.5 mme32 mm). In this study we operated one OPC at (Dp ¼ 0.25e2.5 mm), dried and unheated (detailed setup information about the instruments can be found in Vogt et al., in press). 2.2. Eddy covariance method, data processing corrections and errors The vertical aerosol number flux was calculated using the eddy covariance method (EC). For this study, the flux W 0 N0 was calculated over periods of 30 min. The coordinate system was rotated to zero average vertical wind. The fluctuations w0 and N0 were separated from the mean by linear de-trending, which also removes the influence of low frequency trends. The validity of the EC technique at the measurement location was confirmed in earlier studies (Mårtensson et al., 2006; Vogt et al., in press, 2011) where it was also shown, based on the turbulent spectra, that the measurements are within the surface layer during daytime. The fluxes have been corrected for the limited time response of the sensor and attenuation of turbulent fluctuations in the sampling line. The response time constant sc for both OPC and sampling line was estimated to be 1.0 s using transfer equations for damping of particle fluctuations in laminar flow (Lenshow and Raupach, 1991) and in a sensor (Horst, 1997). The typical magnitude of these corrections varied, corresponding to an underestimation prior to the correction of between 12 and 32%, depending on wind speed and stability conditions. The median relative counting error in the aerosol flux was 15% at the smallest and largest OPC sizes with a peak of 35% at Dp ¼ 0.7 mm (see Vogt et al., in press). In addition, the aerosol fluxes and concentrations were corrected for tube losses in the sampling line, which amounted to w5% for the largest size class in the OPC (Dp ¼ 2 mm to 2.5 mm) and much less for the smaller size classes. The CO2 flux was corrected for variations in air density due to fluctuations in water vapor and heat fluxes in accordance with

M. Vogt et al. / Atmospheric Environment 45 (2011) 5725e5731

5727

of 0.0132 m s1 (s ¼ 0.0204 m s1). For the total particle flux (Dp ¼ 0.25e2.5 mm) there are net emissions 90% of the time indicating that emissions dominate over deposition in this data set. The particle emission velocity is defined as

ve ¼

F N

(1)

where F is the vertical aerosol number flux (positive upward) and N is the aerosol number concentration. In this study the term emission velocity is used when the sign of ve is positive. 3.1. Diurnal cycles

Fig. 2. Frequency distribution of the derived particle emission velocities within the size range Dp ¼ 0.25e2.5 mm. Positive values represent upward (emission) velocities and negative represents downward (deposition) velocities.

Webb et al. (1980). This resulted in a maximum increase around noon for CO2 flux of w37%. For more details see Vogt et al. (in press). To ensure the analysis was performed on emissions dominated by traffic emissions, we only used data from the northern sector (from 270 to 90 ). This selection criterion leaves us with 4393 individual quality checked, half hour data points. 3. Results and discussion Fig. 2 shows the frequency distribution of the emission velocities (Dp ¼ 0.25e2.5 mm), from which we can conclude that the city is on average, a source of particles of accumulation mode size, with an median emission velocity of 0.009 m s1 (with the 25 and 75 percentiles at 0.0036 and 0.019 m s1, respectively) and mean

The observed diurnal cycles of particle flux, particle number concentration, CO2 flux, friction velocity, particle emission velocity and atmospheric stability are shown as half hourly median values in Fig. 3. Particle fluxes begin to increase around 6.00 LT (Local Time) and show a broad peak during the day before they start to decline around 19.00 LT (Fig. 3a). The aerosol number concentration does not vary significantly over the diurnal cycle (Fig. 3b) whereas the CO2 flux has a pronounced diurnal cycle (Fig. 3c), with the highest emissions during daytime and lower emissions during nighttime. The friction velocity also has a broad peak between 9.00 LT and 19.00 LT (Fig. 3d). The median ve ranges from 4 mm s1 during night to approximately 19 mm s1 at noon and the diurnal pattern is similar for all aerosol size classes included in this study (not shown). The median stability changes sign near sunrise and sunset (Fig. 3f). Unstable conditions (L < 0) can be observed between 7.00 LT to about 19.00 LT. The most obvious explanation for the observed diurnal cycles in aerosol flux and emission velocity is the diurnal cycle in vehicle traffic within the city (Dorsey et al., 2002; Mårtensson et al., 2006; Vogt et al., in press, 2011). Traffic activity is higher during the day which leads to a stronger emission of aerosol particles, in turn resulting in an increase in the emission velocity. Dorsey et al. (2002) showed ve derived from measurements made in Edinburgh, Scotland and observed that the average diurnal cycle for ve peaked at about

Fig. 3. Median diurnal cycles of a) particle number flux, b) particle concentration, c) CO2 flux, d) friction velocity e) emission velocity and f) atmospheric stability. The vertical bars represent the 25 and 75 percentiles.

5728

M. Vogt et al. / Atmospheric Environment 45 (2011) 5725e5731

Fig. 4. Median emission velocities ve for Dp ¼ 0.25e2.5 mm within constant intervals of friction velocity u*. The dashed lines represent the 25 and 75 percentiles.

65e70 mm s1 during the daytime with a minimum of about 20 mm s1 in early morning. Although the Dorsey et al. (2002) results were for the total number emission of aerosol particles (>10 nm Dp), they should primarily originate from the same source (traffic) as this study. The two studies show consistency in the phase of the diurnal cycle in ve and a similar relationship to traffic activity. 3.2. Parameterization of friction velocity ve values (for Dp ¼ 0.25e2.5 mm) were aggregated into 0.05 ms1 wide u* intervals with the median values presented in Fig. 4 (dashed curves showing the 25e75 percentiles of observed data). It can be seen that the emission velocity increases with increasing friction velocity. Increasing u* can be interpreted as increasing the mechanical turbulence (increasing momentum flux), or increasing the wind speed and/or decreasing the atmospheric stability. Initially attempting to parameterize ve as function of Dp and u* did not produce satisfactory results as it appeared that at least one important parameter was missing. In the current study, the particle flux can to a large extent be attributed to traffic related combustion processes (Vogt et al., in press, 2011; Gidhagen et al., 2005). Therefore, it is expected that ve should be related to the upward CO2 flux (emissions) which is also strongly associated with traffic combustion. Vogt et al. (2011) already established the close relationship between traffic related aerosol and CO2 emissions and so, to determine whether this constitutes an additional key parameter, we plotted the ratio of ve =u* against the CO2 flux. Fig. 5 shows the median ve =u* values as a function of the measured CO2 flux. It can be seen that with increasing CO2 flux, ve =u* increases with a near linear trend. It is notable that this strong dependence is only found for particle sizes between 0.25 and

Fig. 5. Median values for the ratio of emission velocity to friction velocity ve/u* for Dp ¼ 0.25e2.5 mm within constant intervals of CO2 flux. The red dashed lines represent the 25 and 75 percentiles.

1.45 mm. The last 2 size classes of the OPC are less dependent of CO2 flux. The reason for this may be that the majority of the larger particles are produced by mechanical (wind driven) processes rather than combustion processes (for example compare the wind dependent emission factors from Vogt et al., 2011). Based on the analysis of the measurement results, for Dp ¼ 0.25e1.45 mm, a linear dependence on CO2 flux and a second order polynomial dependence on the size was assumed for the derivation of a emission parameterization. Merging the information on the size resolved particle emission measurements and Figs. 4 and 5 makes it is possible to derive a parameterization of emission velocity as a function of particle size, friction velocity and CO2 flux. The emission velocity can be expressed in the form ve ¼ ðaðd1 D2p þ d2 Dp þ d3 Þ þ bÞu* FCO2 , where FCO2 is the CO2 flux and a, b, d1, d2 and d3 are empirical constants. Based on a least-square fit, the following relationship was found

    ve ¼ 129:14 0:01D2p þ 0:06Dp þ 0:008 þ 1:22 u* FCO2

(2)

where ve is given in [ms1]. Dp is expressed in units of [mm], u* in [ms1], and FCO2 in units of [mmol m2 s1]. In this form, the parameterization reproduces the average conditions and the average variability reasonably well (see Fig. 6), but Equation (2) does appear to under-predict the emission velocity at nighttime and late evening (see Fig. 6). Considering the physical meaning of Equation (2), we see that the emission velocity increase with increasing turbulence (friction velocity), and increasing CO2 flux. The latter may indicate increasing CO2 emissions and/or increasing turbulence i.e. the CO2 flux and the friction velocity are not independent. Changes in friction velocity could imply both a change in wind speed and stability (u* increases with increasing wind speed and peaks at neutral stratification):

8   1 > z > > ln þ 4:7ðz  z Þ z=L>0 Stable > 0 > > z0 > >    > 1 < z ln z=L ¼ 0 Neutral u* ¼ kU z > 0 > ! ! >     >  1  2 þ 1 ðh þ 1Þ2 > > h z 0 > 1 1 0 > þ 2 tan h  tan h0 >  : ln z þ  2 h þ 1 ðh þ 1Þ2 0

(3) z=L < 0 Unstable

M. Vogt et al. / Atmospheric Environment 45 (2011) 5725e5731

5729

Fig. 6. Median diurnal cycles of observed and predicted emission velocity ve for a) Dp ¼ 0.26 mm b) Dp ¼ 0.375 mm c) Dp ¼ 0.68 mm d) Dp ¼ 1.15 mm size range. Solid black lines represent observed ve, and the green lines represent the parameterization. The dashed lines represent the 25 and 75 percentiles of the observations.

where k (¼0.4) is von Karmans constant, U is the mean wind speed, z is the measurement height, z0 is the surface roughness, h ¼ ð1  15ðz=LÞÞ1=4 ; h0 ¼ ð1  15ðz0 =LÞÞ1=4 , and L is the MonineObukov length (see for example Seinfeld and Pandis, 1998). It is

easy to see that increasing wind speed increases the friction velocity and vice verse. A more careful examination of the equations also reveals that friction velocity will decrease when the stability deviates from neutral stratification.

Fig. 7. Test of parameterized emission velocity ve against observations between 12th to 17th of September 2008 for the size range of a) Dp ¼ 0.26 mm b) Dp ¼ 0.375 mm c) Dp ¼ 0.68 mm d) Dp ¼ 1.15 mm. Solid black line represents observed emission velocity, red the parameterization and the corresponding linear fits and correlation coefficients in the size range of e) Dp ¼ 0.26 mm f) Dp ¼ 0.375 mm g) Dp ¼ 0.68 mm h) Dp ¼ 1.15 mm.

5730

M. Vogt et al. / Atmospheric Environment 45 (2011) 5725e5731

Fig. 8. a) Correlation coefficients of modeled and measured emission velocity for weekends (red), weekday (green), wind speed <2 ms1 (blue), wind speed >8 ms1 (black), stable (purple) and unstable (brown) atmospheric stratification. b) The same as Fig. 8a but for the slopes of the linear fit between modeled and measured emission velocity.

Despite the fact that our measurements cover very different particle size ranges than the Dorsey et al. (2002) parameterization, the aerosol vertical transport process may at least partly be attributed to the same meteorological parameters and so it is relevant to compare with the two emission velocity parameterizations. The Dorsey equations for ve are independent of any traffic measures, while our Equation (2) includesFCO2 . Furthermore, the only meteorological parameter Dorsey et al. (2002) includes is the stability z ¼ (z d)/L, where d is the displacement height, z the measurement height and L is the MonineObukov length 0:3

ve ¼ 11eðzÞ

ve ¼ 80  45z

3

0:36

z < 0ðunstableÞ

z  0ðstableÞ

(4) (5)

Our Equation (2) is simply a linear function of u*. However, through the stability dependent version of the logarithmic wind law, in unstable conditions u* is proportional to arctanð1  15zÞ which has 0:3 a similar shape to eðzÞ . In stable conditions u* is proportional to 1 z , which has a similar shape to 80  45z0:36 . 3.3. Testing the parameterization The parameterization has been tested using actual data for the period of September 12 to September 17 (Fig. 7). The parameterization reproduces the data reasonably well, even for individual days. In addition to the test period, the behavior of the parameterization for data on weekday versus weekend, periods of stable versus unstable stratification, and for data in which periods of low and high wind speeds were excluded, are examined in Fig. 8, which plots the correlation coefficients and slopes of the linear fits between observed and predicted ve. It can be seen that the correlation coefficient, which represent the accuracy and the slope of the linear fit, which represent the precision, decreases with increasing particle diameter. The rapid decrease of the slopes with particles sizes larger than 0.6 mm indicates that the parameterization tends to over-predict the emission velocity for the larger particle sizes.

Ketzel et al. (2007) assumed that particles smaller than 0.6 mm diameter mainly originated from exhaust emissions and particles larger than 0.6 mm were mechanically produced. As CO2 flux, which originates from combustion, was introduced into the parameterization, the parameterization is likely to be biased toward particles sizes which originate from combustion. Therefore the parameterized emission velocity might fail for larger sizes because these particles are more likely to be mechanically produced by cars and/ or resuspended from the street and are not directly linked to the CO2 emissions from combustion. The best accuracy and precision of the parameterization can be seen on the weekend, which cannot be explained by the authors. High and low wind speeds have contradictory effect on the parameterization. For high wind speeds (>8 m s1) the smallest sizes (Dp < 0.45 mm) show the best correlation, but the parameterization overpredicts the emission velocities. This might be explained by the increased dilution of the aerosol concentration at increasing wind speeds. For low wind speeds the opposite effect is evident in the size interval (Dp < 0.45 mm). With less wind speed, less dilution takes place, particle concentrations increase and the parameterization under predicts the emission velocity. With regard to atmospheric stability, it appears that the parameterization works better in stably stratified conditions. Tests on the normalized emission velocity (ve =u* ) show constant values for stable stratification, but increasing values as the atmosphere becomes more unstable (not shown). This relationship resembles the observations of the normalized deposition velocity by Wesely et al. (1985) and Gallagher et al. (1997). 4. Summary and conclusion Size-resolved aerosol number fluxes and emission velocities of particles with Dp ¼ 0.25e2.5 mm were measured with the eddy covariance method 105 ms above the ground over the city of Stockholm, Sweden. The emission velocity was observed to be well correlated with friction velocity and CO2 fluxes. These parameters, along with the particle size were used to express the emission velocity as

M. Vogt et al. / Atmospheric Environment 45 (2011) 5725e5731

    ve ¼ 129:14 0:01D2p þ 0:06Dp þ 0:008 þ 1:22 u* FCO2 where ve is given in [m s1], Dp in [mm], u* in [m s1], and FCO2 in [mmol m2 s1]. This parameterization was found to reproduce the observed average diurnal cycle and individual half hourly data reasonably well for particles sizes up to 0.6 mm diameter. For larger particles the parameterization tends to over-predict the emission velocity. This may be caused by the fact that these larger particles are not produced by combustion and are therefore misrepresented by the CO2 flux, which is associated with the traffic combustion source. Acknowledgements We would like to thank the Swedish Research Council for Environment, Agricultural Science and Spatial Planning (FORMAS) and the Swedish Research Council (VR) for supporting this project. We also acknowledge Leif Bäcklin and Kai Rosman for technical assistance and Peter Tunved for useful discussions. We would also like to thank Telia for using the communication tower. References Dorsey, J.R., Nemitz, E., Gallagher, M.W., Fowler, D., Williams, P.I., Bower, K.N., Beswick, K.M., 2002. Direct measurements and parameterisation of aerosol flux, concentration and emission velocity above a city. Atmos. Environ. 36 (5), 791e800. Gallagher, M.W., Beswick, K.M., Duyzer, J., Westrate, H., Choularton, T.W., Hummelshøj, P., 1997. Measurements of aerosol fluxes to Speulder forest using a micrometeorological technique. Atmos. Environ. 31, 359e373. Gidhagen, L., Johansson, C., Langner, J., Foltescu, V.L., 2005. Urban scale modeling of particle number concentration in Stockholm. Atmos. Environ. 39 (9), 1711e1725. Harrison, R.M., Shi, J.P., Xi, S.H., Khan, A., Mark, D., Kinnersley, R., Yin, J.X., 2000. Measurement of number, mass and size distribution of particles in the atmosphere. Philos. Trans. R. Soc. A 358, 2567e2579. Horst, T.W., 1997. A simple formula for attenuation of eddy fluxes measured with first-order-response scalar sensors. Bound. Lay. Meteorol. 82 (2), 219e233.

5731

IPCC, 2007. Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assesment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Järvi, L., Rannik, U., Mammarella, I., Sogachev, A., Aalto, P.P., Keronen, P., Siivola, E., Kulmala, M., Vesala, T., 2009. Annual particle flux observations over a heterogeneous urban area. Atmos. Chem. Phys. 9, 7847e7856. Ketzel, M., Omstedt, G., Johansson, C., During, I., Pohjolar, M., Oettl, D., Gidhagen, L., Wahlin, 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. Ketzel, Berkowicz, 2004. Modelling the fate of ultrafine particles from exhaust pipe to rural background: an analysis of time scales for dilution, coagulation and deposition. Atmos. Environ. 38, 2639e2652. Lenshow, D.H., Raupach, M.R., 1991. The attenuation of fluctuations in scalar concentrations through sampling tubes. J. Geophys. Res. 96, 5259e5268. Longley, I.D., Gallagher, M.W., Dorsey, J.R., Flynn, M., 2004a. A case-study of fine particle concentrations and fluxes measured in a busy street canyon in Manchester, UK. Atmos. Environ. 38, 3595e3603. Longley, I.D., Gallagher, M.W., Dorsey, J.R., Flynn, M., Bower, K.N., Allan, J.D., 2004b. Street canyon aerosol pollutant transport measurements. Sci. Total Environ. 334e335, 327e336. Mårtensson, E.M., Nilsson, E.D., Buzorius, G., Johansson, C., 2006. Eddy covariance measurements and parameterisation of traffic related particle emissions in an urban environment. Atmos. Chem. Phys. 6, 769e785. Norman, M., Johansson, C., 2006. Studies of some measures to reduce road dust emissions from paved roads in Scandinavia. Atmos. Environ. 40, 6154e6164. Seinfeld, J.A., Pandis, S.N., 1998. Atmospheric Chemistry and Physics. John Wiley & Sons, New York. Vogt, M., Nilsson, E.D., Ahlm, L., Mårtensson, M., Johansson, C. Seasonal and diurnal cycles of 0.25e2.5 mm aerosol and CO2 fluxes over urban Stockholm, Sweden. Tellus B, in press. Vogt, M., Nilsson, E.D., Ahlm, L., Mårtensson, E.M., Johansson, C., 2011. The relationship between 0.25e2.5 mm aerosol and CO2 emissions over a city. Atmos. Chem. Phys. Discuss. 10, 21521e21545. doi:10.5194/acpd-10-21521-2010. World Health Organization, 2005. Effects of Air Pollution on Children’s Health and DevelopmentdA Review of the Evidence. WHO Regional Office for Europe, Copenhagen. Webb, E.K., Pearman, G.I., Leuning, R., 1980. Correction of flux measurements for density effects due to heat and water vapour transfer. Q. J. Roy. Meteorol. Soc. 106, 85e100. Wesely, M.L., Cook, D.R., Hart, R.L., Speer, R.E., 1985. Measurements and parameterization of particulate sulfur dry deposition over grass. J. Geophys. Res. 90, 2131e2143.