Effect of seasonal variability and land use on particle number and CO2 exchange in Helsinki, Finland

Urban Climate 13 (2015) 94–109

Contents lists available at ScienceDirect

Urban Climate journal homepage: www.elsevier.com/locate/uclim

Effect of seasonal variability and land use on particle number and CO2 exchange in Helsinki, Finland Mona Kurppa, Annika Nordbo, Sami Haapanala, Leena Järvi ⇑ Department of Physics, P.O. Box 48, 00014, University of Helsinki, Finland

a r t i c l e

i n f o

Article history: Received 26 August 2014 Revised 7 July 2015 Accepted 22 July 2015

Keywords: Aerosol particles Carbon dioxide Eddy covariance Street canyon Turbulent transfer Urban

a b s t r a c t Turbulent fluxes of particle number and CO2 were analysed in Helsinki between July 2011 and June 2013. The fluxes were measured using the eddy covariance method in a dense city centre and suburban location next to a large road allowing the study of the mutual connections of the two fluxes and their dependencies between different high intensity road traffic areas. In the city centre, the median particle (Fp) and CO2 fluxes (Fc) were 0.18  109 m2 s1 and 9.8 lmol m2 s1, and at the suburban site 0.17  109 m2 s1 and 5.7 lmol m2 s1, respectively. Fc was larger in the city centre than at the suburban site whereas particles were emitted with a similar strength from a single large road as from the city centre. Fp had the largest net fluxes in winter and the smallest in summer, whereas seasonal variability in Fc was minor. Partly this can be explained by increasing particle emissions in colder temperatures. Also the different vertical transfer efficiency of the two scalars affects the different behaviour. This study demonstrates how the behaviour of two seemingly similar urban pollutants vary already at a kilometre scale and with different meteorological conditions. Ó 2015 Elsevier B.V. All rights reserved.

1. Introduction Aerosol particles are identified as a major health risk especially in urban areas. Long-term exposure to particulate matter has been observed to be in close relationship with adverse severe health effects, particularly in the pulmonary and cardiovascular systems (Chow et al., 2006). According to the World Health Organization, urban air pollution was estimated to have caused up to 3.7 million premature deaths worldwide in 2012 and the main cause was a constant exposure to particulate matter (WHO, 2014). Fine particles (aerodynamic diameter < 2.5 lm) have been regarded as most harmful, but ultrafine particles (UFP, aerodynamic diameter < 0.1 lm), apart from growing to fine particles through condensation and coagulation, seem to have potential to damage other organs as well (Kreyling et al., 2006). Carbon dioxide (CO2), instead, is one of the most important and well known greenhouse gases that has a strong warming effect on our atmosphere. Aerosol particles also modulate Earth’s energy budget both directly by scattering and absorbing light, and indirectly by acting as cloud condensation nuclei. However, the cooling effect of anthropogenic aerosol particles is still far more uncertain compared to the warming effect of CO2 (IPCC, 2013). Studies of vertical exchange of these particles and CO2 between the surface and the atmosphere provide important knowledge for improving both numerical air quality and climate models.

⇑ Corresponding author. E-mail address: leena.jarvi@helsinki.fi (L. Järvi). http://dx.doi.org/10.1016/j.uclim.2015.07.006 2212-0955/Ó 2015 Elsevier B.V. All rights reserved.

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

95

Over half of the world’s population live in urban areas (53% in 2012) and the number is constantly growing (World Bank, 2014). In these high intensity traffic and densely built surroundings, particle and CO2 emissions are elevated and the cities act as net sources (e.g. Järvi et al. (2009b) for particle and Nordbo et al. (2012a) and Lietzke and Vogt (2013) for CO2 fluxes). A major source of particles and CO2 in city centres is road traffic, which produces fine and ultra fine particles and CO2 in combustion processes, and coarse particles mechanically as road, tyres and break-linings wear (Vogt et al., 2011b). UFP are additionally formed in atmospheric photochemical reactions from precursor vapours and fine particles as transformation products of UFP. Large amount of particles are formed in engines running with diesel compared to gasoline due to the higher burning temperatures, and especially heavy-duty vehicles contribute to particle emissions (Lähde et al., 2014). The combustion emissions are highest in cold ambient temperatures and fast accelerations (Kittelson et al., 2004; Virtanen et al., 2006). The most direct way to measure the vertical trace gas and particle fluxes from areas of neighbourhood or local scales is the eddy covariance (EC) method (Aubinet et al., 2012). The method has been widely used in vegetated homogeneous environments, but within the past few decades a growing number of EC measurements have been carried out at urban measurement sites with heterogeneous surface covers. An increasing number of articles have reported urban EC flux measurements of trace gases and particles over extended periods (i.e. >1 year) (Järvi et al., 2009b; Ripamonti et al., 2013; Vogt et al., 2011a), but still only a few studies have examined simultaneous flux measurements of particles and CO2 and similarities in their sources and sinks (Contini et al., 2012; Vogt et al., 2011a). Furthermore, the intra-city variation of particle fluxes and its dependencies from different urban surface covers has not been quantified before. In cities, air quality is usually monitored by measuring mass concentrations PM2.5 or PM10 (particles with aerodynamic diameter smaller than 2.5 lm or 10 lm, respectively) whereas particle number flux and concentration measurements with the EC method are less common despite their ability to observe UFP, in particular. In Helsinki, UFP comprise only a minor fraction of particle mass concentrations but 70–90% of total particle number concentrations when no long-range transport or road dust re-suspension episodes are present (Hussein et al., 2014; Ripamonti et al., 2013), and hence particle number fluxes measure mainly the vertical transport of UFP. In this paper we provide a study of particle and CO2 fluxes measured with the EC method in a densely built city centre (Hotel Torni, Nordbo et al. (2013)) and at a suburban site (SMEAR III Kumpula, e.g. Järvi et al. (2009a)) in Helsinki over two years. The emphasis is on temporal and spatial variations of the two scalars and in their mutual behaviour, which can be used to get information about the emission sources. In addition, we will analyse the effect of different factors, such as storage flux, stability, presumable sources within the source areas and turbulent transport efficiency on fluxes. The main focus is on the city centre site from where no previous particle flux measurement studies exist.

2. Measurements 2.1. Site description Helsinki, the capital of Finland, is the biggest city in the country with a population of 615,000 in 2013 (PRC, 2014). The population of the Helsinki metropolitan area (770 km2) reaches 1.1 million when the nearest surrounding towns are included (City of Helsinki, 2013). Located at high latitudes, but on the coast of the Gulf of Finland, the weather in the city is either maritime, coastal or mixed depending on the air mass history (Hussein et al., 2014). Especially the winter temperatures are higher than the latitudinal average mainly due to the influence of North Atlantic Drift. In central Helsinki, the measurements of turbulent particle number and CO2 fluxes were carried out on the tower of Hotel Torni building (Fig. 1). The tower (60°100 04.0900 N, 24°560 19.2800 E, 15.2 m above sea level (a.s.l.)) is a 57.7 m tall structure in the highly busy central Helsinki (Table 1). The site belongs to the local climate zone 2 as classified by Stewart and Oke (2012), and the surroundings are densely built urban area with a mean building height of 24.1 m. Within a 1 km radius circle around the tower, buildings cover 55%, paved area 42% and vegetation 3% of the total area (Nordbo et al., 2013). The central railway station is located 400 m northeast, and an year-round passenger and cargo harbour West Harbour is located 1.5 km south-west from the site. Despite a bulky structure of the building tower, Hotel Torni was chosen as the measurement site in order to reach a measurement height suitable for EC measurements. The site (hereafter called Torni) is situated 120 m south-west from the main road of central Helsinki, Mannerheimintie. Traffic rates decrease towards the centre, being around 20,000 motor vehicles per day next to the site (HCPD, 2013). The EC measurements of particle number and CO2 were also performed at the SMEAR III (Station for measuring the ecosystem-atmosphere relationship) Kumpula site (Järvi et al., 2009a). The site (hereafter Kumpula) is located outside city centre, 4.1 km northeast from Torni, and it is representative for local climate zone 6. The university campus is located in the area with a lot of green areas around the station. The surrounding area can be divided into three surface cover sectors, with the road sector in east and southeast, where one of the main roads leading to city centre passes the station. The other two sectors are the vegetation sector in west and northwest and the building sector in north and northeast, but in this study we only focused on measurements from the road sector due to similar particle and CO2 sources. In this sector, the mean building height is 11.5 m within a 1 km radius circle and the area is covered 15% by buildings, 39% by paved surfaces and 46% of vegetation (Table 1, Järvi et al. (2014)). A band of vegetation between the measurement site and the road might influence the particle fluxes by acting as a buffer for the road emissions.

96

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

Fig. 1. Satellite image of the Helsinki city center. The urban site at Hotel Torni is indicated by a white star. Overlaid on Google Earth map are the 80% isopleths of cumulative particle flux footprint areas in thermal winter and summer. Lönnrotinkatu, where traffic rates were measured, is indicated by a yellow line and the main road, Mannerheimintie, by a green line. Data were omitted from the wind direction of 50–180° due to flow distortion. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Table 1 The characteristics of the study sites. Surface characteristics were calculated as means for 1 km radius circles around the measurement towers and ± show standard deviations.

a b d e c

Variables

Unit

Torni

Kumpula-Road

Measurement height (a.g.l.) Building height (a.g.l.) Aerodynamic roughness length Displacement height Surface fraction of buildings Surface fraction of paved surface Surface fraction of vegetation Population densityd Local climate zone (LCZ)e

m m m m – – – (# ha1) –

60.9 24.1 ± 4.9a 1.9 ± 0.5a 14.9 ± 3.0a 0.55a 0.42a 0.03a 81 2

31.0 11.5 ± 2b 1.2 ± 0.23c 7.7 ± 1.33 0.15b 0.39b 0.46b 37 6

Nordbo et al. (2013). Järvi et al. (2014). HSY (2011). Stewart and Oke (2012). Rule of thumb (roughness length = 0.1zh and displacement height = 2/3zh).

At both sites, most of the particle emissions are due to traffic activity, since energy production, industry and shipping activities are located relatively far, especially from the city centre. Traffic activity remains relatively constant in Helsinki, except during holiday seasons the volumes decrease (mainly in July). In central Helsinki, rush hours take place at 8 am and 4 pm, which is about one hour later than outside the centre (Fig. 2). Light-duty vehicles (cars and vans) represent 92% of the fleet and heavy-duty vehicles the remaining 8%, of which 25% are trucks and 75% buses. At Kumpula, the proportion of buses is 2% lower and that of heavy-duty vehicles 1% higher than in the city centre. In autumn 2011, major rail works took place 1 km and 1.5 km south-west and road works 700 m west from the Torni site (Lilleberg and Hellman, 2012, 2013). Activity in the construction sites and changes in the traffic rates may have affected the measurements.

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

97

Fig. 2. Median diurnal course of traffic rates (veh h1) near the Torni site (black) in the central Helsinki and on the Kumpula road (red) during traffic measurement campaigns in spring 2013. Quartiles are shown by shaded areas. Pay attention to different y-axes. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

The climate in Helsinki has four distinct seasons including cold winters when particle emissions can increase particularly in residential areas due to increased use of wood or oil for heating. The heating in Helsinki is mostly centralized, but also in city centre apartments can have fireplaces and additional heating using wood combustion takes place. Over half of the mass concentration of particles in the city are transported from eastern Europe (HSY, 2013). However, these particles associated with long-range transport are larger in size but smaller in number than the particles from local sources and thus do not significantly affect number concentrations of particles (Vallius et al., 2003). 2.2. Measurements At Torni, the EC measurement set-up consisted of a three-dimensional sonic anemometer (USA-1, Metek GmbH, Germany) to measure all three wind components (u, v and w) and the sonic temperature (Ts), and a water-based condensation particle counter (WCPC, TSI-3781, TSI Incorporated, USA) to measure the particle number concentration p (# cm3). The 50% cut-off size of the WCPC is 6 nm but the cut-off size also slightly depends on the particle composition (Hering et al., 2005). The maximum particle sizes that the instrument can measure are around 1000 nm. In addition, CO2 mixing ratio c (ppm) and water mixing ratio q (kg kg1) were simultaneously measured using an enclosed-path infrared gas analyser (LI-7200, LICOR Environmental, USA). The EC equipment was mounted on a 2.3 m high mast at the north-western corner of the tower of Torni, resulting in a total measurement height of 60 m (Nordbo et al., 2013). This is a sufficient height for the EC setup as the measurements are made 2.5 times the mean building height. The centre of the tower causes flow distortion when the wind direction is 50–180°, which was observed by visual inspection of averaged wind speed and scalar power spectra and co-spectra (not shown). In a previous study, the quality of the flux measurements has been proven to be good outside the flow distortion area by means of spectra and co-spectra, integral turbulence characteristics and deflection angles (Nordbo et al., 2013). The Torni site has been operating since the 28th of September 2010 and the measurement set-up was changed on the 3rd of November 2011 to decrease the effect of tube attenuation. This was also taken into account in the post-processing of fluxes. The air inlet was positioned 0.30 m below the anemometer (previously 0.15 m below) and air flow to the WCPC was drawn through a 7.6 m (previously 5.9 m) long stainless steel tube and 1.0 m (previously 3.0 m) long PVC tube which had inner diameters of 8 mm and 4 mm (previously both 4 mm). The air flow to the gas analyser was drawn through a tube of 1.9 m (before 1.94 m) in length and 6.5 mm in inner diameter. The flow rate to the WCPC was 12 l min1 and 16 l min1 to the gas analyser. All tubes were heated with a power of 9 W per m2 to avoid condensation of water vapour on their walls. The sampling frequency of the raw EC data was 10 Hz. To link the Torni EC measurements to street level conditions, particle number concentration p and particle size distribution measurements from a street canyon in central Helsinki were utilized in the analysis. The street canyon measurements

98

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

were monitored by HSY (Helsinki Region Environmental Services Authority) at a height of 4 m and the site lies by Mannerheimintie 180 m north-east from Torni. The p and size distributions in the size 6–1000 nm were measured with a differential mobility particle sizer (DMPS) using a Vienna-type differential mobility analyser (DMA, length 0.28 m) and a condensation particle counter (CPC, Model 5.401, GRIMM, Germany). The overall amount of 60-min data points, when measurements of Fp and p from both sites were available, was 547. These data points date from fall 2011 (81%), spring 2013 (6%) and early summer 2013 (13%). Although the instrumentations at Torni and HSY site were not identical, the detection limits of particles were similar and the particle number concentration measurements can be compared. The EC measurements at Kumpula were performed from the top level of a 31-m-high lattice tower (26 m a.s.l.). The setup included the same anemometer and particle counter as at the Torni site and the CO2 flux measurements using a closed-path infrared gas analyser (LI-7000). In addition, meteorological measurements (temperature T, global radiation G, relative humidity RH and precipitation) from the rooftop of a university building (24 m a.g.l.), 100 m north-east from the measurement tower, were used in the analysis. A more detailed description of the measurements can be found e.g. from Järvi et al. (2009a). The data presented in this study cover the period between the beginning of July 2011 and the end of June 2013. The period was divided into thermal seasons according to 5-day running mean temperature. In Finland, thermal winter (26% of the measurement period) and summer (40%) are defined as periods when daily average temperatures are below 0 °C and above 10 °C over 5 days, respectively. Thermal fall (20%) and spring (14%) are between them. Climatological seasons are hereafter referred as thermal seasons. The measurements will be presented in Eastern European Time (EET, UTC+2) with a correction to daylight saving time. 2.3. Data processing and data quality The flux calculation procedures were performed according to Nordbo et al. (2012b) at both EC flux sites. Before the calculation of 30-min fluxes spikes in the data were replaced by previous measurement points and periods of clear noise were omitted to improve the quality of the data. The data were linearly de-trended and the 2-D rotation method, where the x-axis is set parallel to the mean wind direction and mean vertical velocity is set to zero, was applied. The lag times between the vertical velocity w and measured scalars (p and c) were defined using the maximum cross-covariance method (Moncrieff et al., 1997). The 30-min averaged fluxes were calculated in an iteration loop in which spectral corrections were applied to all co-spectra and fluxes with an assumption of scalar similarity (Fig. 3a). A transfer function TF, which gets values from 0 to 1, describes the amount by which a co-spectrum was corrected to account for loss of flux due to e.g. tube attenuation and instrument response time (Fig. 3b) (Massman, 2000). The average and standard deviation of spectral correction for Fp and Fc

Fig. 3. For the EC measurements at Torni carried out between the 1st of July and the 31st of December 2012: (a) Median normalized cospectra (fC/cov = cospectra function divided by covariance) as a function of frequency for w0 T0 , w0 p0 and w0 c0 . (b) Transfer function (TF) for the cospetra of w0 p0 and w0 c0 as a function of frequency. Measured TF are in black and a fitted TF using the formula by Moncrieff et al. (1997) is in grey (response time s is given as a fitted variable).

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

99

were (17 ± 10)% and (5 ± 1)%, respectively, which proves a good instrumental response. The longer inlet tube of the particle counter is the main reason for the larger spectral correction of Fp. A cross-wind correction and a sonic heat correction (Liu et al., 2001) were applied when calculating the sensible heat flux H. All fluxes were filtered applying a stationarity test (Foken and Wichura, 1996), according to which the 30-min flux should not deviate more than 30% from the mean of the fluxes of 5-min subperiods. The flux values with an inadequate stationarity in the turbulent signal were omitted with a stationarity limit of 60.3. The stationarity test omitted twice as much 30-min Fp data as Fc data at Torni, which is a typical difference between the two fluxes (e.g. Cava et al. (2014)). Furthermore, 32% of the flux data were eliminated to avoid flow distortion induced by tower structures at Torni. Eventually, Fp and Fc were quantified for 21% and 52%, and 19% and 25% of all 30-min periods of the measurement period at Torni and Kumpula, respectively. To compare, Dahlkötter et al. (2010) reported a 30-min flux data coverage of 24% for Fp and 30% for Fc during a measurement period of 78 days. The storage term S of a scalar defines the change of mass or number in the storage volume below the EC measurement height. This storage may be horizontally flushed away and stay undetected in the flux measurements, or cause abrupt spikes in concentration and fluxes when suddenly transported upwards (Burba, 2013). Hence, S can be used as an indicator for the uncertainty of the EC flux measurements. It can be calculated using concentration measurements from multiple heights, but in practice, due to the lack of profile measurements, it is often calculated using concentration measurements from a single level from the height of the EC measurements. This method assumes the vertical concentration distribution to be constant below the EC measurement level and therefore this method causes uncertainty to the estimation of S. However, a recent study by Crawford and Christen (2014) from Vancouver reported differences between a single and a multiple level estimations of CO2 storage flux to be on average 5.2%. At Torni, S was calculated using a single level measurement at the height of the EC setup according to:

S¼

     z t t  s z;  ; s z; t 2 2

ð1Þ



where t is the flux averaging period, s z; 2t and s z;  2t are the scalar concentrations at height z at the beginning and the end of the averaging period t and h is the measurement height. In place of the instantaneous concentration measurements, we used 60-s averaged concentration values in order to avoid large random uncertainty (e.g. Rannik et al. (2009)). In addition, transfer efficiencies RwX for particles, CO2 and heat were calculated as correlation coefficients between w0 and scalar X0

Rwx ¼

w0 X 0

rw r0X

;

ð2Þ

where w0 X 0 is the covariance between fluctuations of vertical velocity w and scalar X and rw and rX are standard deviations. Transfer efficiency provides information about the effectiveness of turbulent eddies to produce fluxes and can be used to assess the uncertainty of the EC method. To evaluate the fluxes and other variables in all seasons, 30- or 60-min medians and quartiles were calculated and classified by time of day. The data were also separated into weekdays and weekends and all public holidays were classified as weekends due to similar traffic activity. Medians and quartiles for less than five data points were omitted. To analyse the variability within a day, measurements between 8 am–4 pm and 11 pm–4 am were selected to represent day- and night-time values, respectively. Only the data points when both Fp and Fc at each site were available were chosen for the analysis. One should note that the flux observations from the two sites are not from the same 30-min periods, since from Kumpula only the road sector is considered.

2.4. Traffic monitoring Traffic rates in Helsinki are measured by the Helsinki City Planning Department. The nearest traffic monitoring point to Torni is on Lönnrotinkatu (130 m SE from the site, Fig. 1) and to Kumpula on the Hämeentie bridge (300 m S–SE from the site). Traffic rates are logged every hour except during rush hours, when they are logged four times per hour. Unfortunately, measurements are performed only as short campaigns in spring and fall, not at the same time at the two sites. In the centre, 9-day traffic monitoring campaigns were conducted in October 2011, October 2012 and April–May 2013, but at Kumpula only one 10-day campaign was conducted in April 2013. The diurnal traffic rates from both sites during spring 2013 are shown in Fig. 2. The traffic rates are five times higher at the traffic monitoring point in Kumpula than at the traffic monitoring point in the city centre. One should notice that in the city centre there are several roads with traffic and therefore the traffic rates from a single street are not representative for the whole area. However, the timing of rush hour peaks and other changes in traffic rates are considered to take place at the same time on other streets also. Despite the limited amount of traffic data, during a flux measurement period of two years the number of simultaneous traffic rate measurements is sufficiently high.

100

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

2.5. Footprint estimation Flux footprint, i.e. the source area of flux, for Torni site was estimated using the analytical model by Kormann and Meixner (2001) that has also been previously used at the Kumpula site. The extent of the footprint depends on the meteorology, such as stability and wind speed, the effective measurement height and the roughness length. The data were divided into 45° wind sectors and eight stability classes and mean parameter values were calculated for each set. The model returned a footprint function of each set describing the relative contribution of a certain area around the site to the calculated flux value (Burba, 2013). We present footprints here as isopleths of certain percentage of the total footprint since, theoretically, the function extends to infinity. Isopleths were calculated as volume integrals of a contribution function and they show the area from which a certain percentage of the measured flux originates. For Torni, the 80% isopleths of particle flux footprints in thermal winter and summer are shown in Fig. 1. Footprints are calculated for the 30-min measurement periods when the atmospheric stratification was neutral or unstable (89% of all data). The cumulative footprints in winter and summer did not notably differ from each other and therefore observed seasonal differences in the fluxes are not likely to be caused by changes in the source areas. In spring, footprint was slightly smaller from the direction of Mannerheimintie than on other seasons (not shown). Also at Kumpula, seasonal changes in the source areas have been found to be small (e.g. Ripamonti et al. (2013)). Diurnal variability in the footprints was small, but somewhat less contribution was observed from Southwest at night when compared to daytime (Figs. A1 and A2). The nocturnal summertime footprint extended a bit further North and Northeast than the daytime one. However, an analytical model for defining a footprint on a homogeneous area is only approximate on a heterogeneous urban measurement area. Moreover, in the case of particle flux, coagulation, deposition and phase changes of particles complicate the estimation of the footprint (Vesala et al., 2008). Thus the footprints are only indicative and should be interpreted with caution. 3. Results 3.1. Meteorological conditions For the analysed period, the 5-day running means of T, G, RH and the sum of daily precipitation are presented in Fig. 4. The mean T of 2012 in Helsinki was 5.9 °C which is the same as the 30-year (1981–2011) average. Year 2011, on the other hand, was notably warmer than the average with a mean temperature of 7.2 °C (FMI, 2013a). The average 30-min temperature during the analysing period from July 2011 to June 2013 was 6.6 ± 9.2 (std) °C. The daily mean temperature varied between

Fig. 4. Meteorological conditions at Kumpula, Helsinki from July 2011 to June 2013. (a) 5-day running mean temperature T in black (°C) and intensity of global radiation G in grey (W m2). (b) Daily precipitation in black (mm day1) and 5-day running mean relative humidity RH in grey (%).

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

101

19.3 °C and +24.6 °C. The daily mean intensity of G (as 5-day running mean) ranged from less than 10 W m2 in winter to 350 W m2 in summer and the day length from less than 6 h to 19 h (FMI, 2013b). The winter 2011–2012 was only 64 days long, whereas in 2012–2013 winter extended over 126 days. On the other hand, the fall 2011 and the spring 2012 were much longer (90 and 65 days) than the fall 2012 and spring 2013, which were 53 and 35 days long, respectively. The average wind speed at Torni was 4.5 ± 2.3 m s1 and the prominent wind direction was south-west. In winter however, the average wind was from south-east as the eastern and north-eastern winds were more frequent most likely due to a Siberian High. Typically, air is observed to be relatively cold and dry in Helsinki when the flow is from these directions. Sea-breeze also affects the wind speed and direction in Helsinki in spring and summer (Järvi et al., 2009a). The rainfall in 2012 was 883 mm, which is more than the 30-year average. The fall 2011 was rainier whereas the spring 2013 was drier than on average (175 mm and 115 mm in the fall and spring 1981–2010, respectively) (FMI, 2013b). 3.2. Data quality 3.2.1. Storage flux The storage terms of particle number concentration (Sp) and CO2 (Sc) were calculated for the analysing period using Equation (1). Both terms were found to be the largest in early morning, which is consistent with Sc calculated for Basel, Switzerland (Feigenwinter et al., 2012). Surprisingly nocturnal Sp was found to be negative and the storage increased the measured particle fluxes, whereas Sc remained positive. Sp was relatively small and Sc negligible when compared to the absolute fluxes (median Sp/Fp  0.1%, Fig. 5 and median Sc/Fc  0.0002%, not shown). Hence, according to our calculations, the storage term of both particles and CO2 have minor impact on our measurements and we can assume to be measuring the net exchange at the surface. Sp has been estimated to be small also at Kumpula (Ripamonti et al., 2013), whereas at a boreal forest site in southern Finland, Sp was frequently observed to be of the same magnitude or even larger than the measured flux (Rannik et al., 2009). Our Sc are clearly less than previously obtained for urban CO2 flux measurements (Feigenwinter et al., 2012; Nemitz et al., 2002), but unfortunately no values for Sp from other cities than Helsinki have been published. 3.2.2. Transfer efficiency Transfer efficiencies RwX of particles, CO2 and heat were calculated according to Eq. (2) to analyse the vertical transport and the Monin–Obukhov similarity theory (MOST) of fluxes at Torni and Kumpula sites. If MOST applies, scalars (mass, heat and particle number) would be transported by the same mechanism and the transfer efficiencies would get the similar values as a function of stability f (e.g. Roth and Oke (1995)). In urban environments, the heterogeneous sources and sinks may produce considerable variations with a large scatter in turbulence and concentrations. However, time-averaged fluxes may remain small if the transfer efficiency is low. Transfer efficiencies for p, c and Ts are depicted as a function of f in Fig. 6 for periods when both Fp and Fc from the site were available (19% of all measurements). Note that 30-min data points are plotted only for Torni site and for transfer efficiency of Ts we use RwT. The RwT at both sites decreased towards neutral stratification, which is a result of w0 T 0s approaching zero as the standard deviation rT does not. Instead, Rwp and Rwc only slightly decreased towards neutral stratification. Constant Rwc in unstable conditions was previously observed in Tokyo by Moriwaki and Kanda (2006). Furthermore, they reported a smaller transfer efficiency of CO2 than that of heat when f < 0, which was also observed for scalars when f < 0.4 at Torni and f < 0.05 at Kumpula. In stable conditions RwT < Rwc at both sites, which is in accordance with previous study from Helsinki (Nordbo et al., 2013). The transfer of heat was constantly more effective at Kumpula than at Torni. No intra-city variation was observed for Rwp in unstable condition, whereas Rwc was constantly around 15% larger at Torni. It

Fig. 5. Frequency distribution of particle storage fluxes Sp relative to particle fluxes Fp measured using the EC method at Torni. The Sp was estimated by averaging the concentration over a 60-s window at the beginning and end of the concentration time series at the height of EC measurements. Y-axis scale is relative to the maximum value.

102

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

Fig. 6. Transfer efficiencies of particles Rwp (black), CO2 Rwc (blue) and sensible heat RwT (red) as a function of atmospheric stability f at the Torni and Kumpula sites. Measurements are separated to unstable (f < 0) and stable (f > 0) situations. Data are from the entire 2-years measurements period. Medians, shown by solid lines for the Torni and dashed lines with triangles for the Kumpula site, were calculated for 40 stability classes with minimum of 20 data points in each. All measurements from Torni are shown with light coloured dots. Note the non-continuous x-axis. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

should be noted that transfer efficiencies for neutral and stable conditions should be treated with caution due to a small number of measurements from Torni (f > 0.1 represent only 28% of the data). On the whole, transfer of passive scalars, CO2 and particles was alike and the efficiencies remained nearly constant with changing stability, whereas transfer of temperature was stability dependent. The difference was presumably due to the active role of temperature, uneven sources and sinks of scalars and heat, and to inherent heterogeneity of urban surface structures because the MOST was initially established for homogeneous surfaces (Katul et al., 1995; Moriwaki and Kanda, 2006). Even though horizontal homogeneity increases with increasing measurement height, turbulent transport may additionally be affected by exchange between surface layer and mixed-layer due to large-scale convective structures (Roth and Oke, 1995). However at Torni site, the difference in transfer efficiencies was small when f < 0.1, which represents 72% of the flux measurements. Thus, scalar similarity can be applied when performing spectral corrections (see Section 2.3). At Kumpula, sources and sinks of particles and CO2 were likely equally heterogeneous according to the similarity of Rwp and Rwc, whereas the slightly different efficiencies at Torni indicate differences in the sources of the two scalars. Unfortunately, no earlier urban studies have reported transfer efficiencies, and therefore we cannot tell whether these are city specific characteristics.

3.3. Particle number fluxes above two land uses Over the analysing period, the median particle flux Fp was 0.18  109 m2 s1 at Torni, and the 5 and 95 percentiles were 0.01  109 m2 s1 and 0.82  109 m2 s1, respectively. Negative fluxes were observed only for 4.4% of the available 30-min particle flux data. These negative values were mostly random spikes indicating uncertainty of the EC measurements rather than actual physical behaviour. At the Kumpula road sector, the median Fp was of the same order of magnitude as in the city centre with a median value of 0.17  109 m2 s1 (5 and 95 percentiles 0.02  109 m2 s1 and 0.75  109 m2 s1, respectively). The median Fp varied by season and at both sites the lowest net fluxes were measured in summer and the highest in winter (Table 2). The difference between the two seasons was 61% at Torni and 50% at Kumpula. The seasonal variability is a combined effect from several different factors. In summer, lower Fp can be affected by on average 20% lower traffic rates during the holiday season in Helsinki (HCPD, 2013). On the other hand, also particle deposition on vegetation is higher in summer than outside the growing season (e.g. Rannik et al. (2009)) and this might impact the particle fluxes particularly at Kumpula. Furthermore, particle number concentrations close to large roads are commonly found to be notably higher in winter than in summer as the proportion of ultra fine particles emitted by motor vehicles increases with decreasing temperature (Virtanen et al., 2006). Also, in Finland streets are gritted in winter and the use of snow tires on icy streets wears the blacktop and increases particle number concentrations at ground level in late fall, winter and early spring (HSY, 2013; Lilleberg and Hellman, 2013). A previous study from Stockholm (Vogt et al., 2011b) reported elevated spring-time fluxes of particles larger than 1 lm, which were linked to road dust re-suspension. At the street canyon site in the centre of Helsinki, the size distribution of particles showed slightly elevated concentrations of particles > 0.5 lm in spring (not shown). Hence, road dust emissions can also contribute to Fp at our sites. Helsinki uses central heating, but some particle emissions can originate from small-scale wood combustion. Seasonal variations in particle flux footprints were so small (Fig. 1) that they can hardly explain the seasonal variations. At both measurement sites, particles were emitted with similar

103

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

Table 2 Median values of the particle flux (Fp, 109 m2 s1) and CO2 flux (Fc, lmol m2 s1) at Torni and Kumpula separated according to season and weekday/weekend. Torni

Fall

Winter

Spring

Summer

All Weekday Weekend All Weekday Weekend All Weekday Weekend All Weekday Weekend

Kumpula

Fp

Fc

Fp

Fc

0.23 0.27 0.12 0.28 0.33 0.21 0.23 0.30 0.15 0.11 0.11 0.10

11.2 13.2 7.9 12.0 14.0 9.2 10.1 11.4 8.0 10.0 11.1 8.8

0.18 0.24 0.11 0.26 0.33 0.17 0.20 0.24 0.10 0.13 0.18 0.07

6.5 8.1 5.1 5.4 6.9 3.8 8.0 9.1 4.1 5.6 7.4 3.5

magnitude implying that a single large road can have a same impact to particle emissions than a dense network of smaller roads in a city centre. Both on weekdays and weekends, Fp had the largest values at daytime and the smallest in the early morning around 2–5 am (Fig. 7a and b). The two-peaked diurnal traffic pattern was observed at neither of the sites, which has also been reported earlier studies from Helsinki (Järvi et al., 2009b; Ripamonti et al., 2013) as well as from other cities (e.g. Martin et al. (2009)). The reason has been claimed to be enhanced atmospheric mixing at noon and in the early afternoon (Contini et al., 2012; Dorsey et al., 2002; Mårtensson et al., 2006; Martin et al., 2009; Ripamonti et al., 2013), which increases the efficiency of turbulent transport (Barlow et al., 2011). On weekdays, Fp increased from its nocturnal minimum values one

Fig. 7. Median diurnal course of particle number flux Fp (109 m2 s1) on weekdays (a) and weekends (b), CO2 flux Fc (lmol m2 s1) on weekdays (c) and weekends (d) and atmospheric stability f (e) at Torni and Kumpula sites in all seasons. The shaded areas show the 25 and 75 percentiles. Percentiles are not represented for the stability parameter f since focus is on the median stability. Only medians with more than five data points are plotted. On weekends in spring, data gaps can be seen as missing data.

104

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

hour earlier at Kumpula than in the city centre following the one hour lag in traffic rates (Fig. 2; Lilleberg and Hellman (2013)). The particle emissions remained high at Torni (between 18 and 21 EET) after the afternoon rush hour peak, which might indicate existence of other emission sources than traffic (e.g. small wood burning, cooking). On the other hand, the atmosphere is less stable in the city centre than at Kumpula particularly in spring and summer evenings, which results in more developed turbulence and thus higher Fp. Despite that Fp values on average were distinctly smaller on weekends than on weekdays (Fig. 7b), the nocturnal fluxes were greater particularly in the city centre. This stems from different activity of people during weekend nights when compared to weekday nights. The nocturnal Fp was the greatest in winter, which might be an indication of small scale heating related emissions. Direct comparison of Fp with previous studies is difficult since measurement installations and locations as well as particle detection sizes vary widely between sites. In addition, measurement periods in most previous studies were of the order of weeks and thus are not representative for all seasons. Despite, measurements at Torni are of the same order of magnitude as those earlier reported from high intense traffic locations. In previous studies Fp have been reported to vary between 0.09 and 2.10  109 m2 s1 (Contini et al., 2012; Dahlkötter et al., 2010; Dorsey et al., 2002; Harrison et al., 2012; Järvi et al., 2009b; Mårtensson et al., 2006; Martin et al., 2009; Nemitz et al., 2002). 3.4. CO2 fluxes above two land uses The median CO2 flux (Fc) over the 2-year period at Torni was 9.8 lmol m2 s1 and the 5 and 95 percentiles were 2.1 lmol m2 s1 and 28.3 lmol m2 s1. Negative fluxes occurred only for 4.8% of the available 30-min Fc, and these were only occasionally observed similarly to Fp. The median Fc at Kumpula was smaller with 5.7 lmol m2 s1 (5 and 95 percentiles 0.5 lmol m2 s1 and 19.5 lmol m2 s1, respectively). Fc varied less with season than Fp. At Torni the daily median in summer was up to 17% and at Kumpula in winter up to 33% lower when compared to other seasons (Table 2). The greatest daily median Fc at Torni was measured in winter, but at Kumpula in spring. The smaller seasonal variability of Fc is surprising as the sources for both particles and CO2 are considered to be similar in urban areas and carbon uptake is expected to decrease the net exchange of CO2 during the growing season. This suggests that the other controlling factors are more important and particularly in the city centre vegetation uptake has minimal effect to Fc. Fc was constantly larger at Torni than at Kumpula and also the daytime pattern of Fc varied between the two locations. At Kumpula, Fc had a two-peaked pattern within the year, whereas in the city centre the fluxes were elevated along the day similarly to Fp at both sites. In spring, summer and fall the different behaviour at Kumpula can be explained by the carbon uptake of vegetation (66% of the surface area), as at Torni only less than 10% of the surface is covered with vegetation. However in winter, carbon uptake is an unlikely reason for the lower Fc around midday indicating differences in their emissions or turbulent transport of these two scalars. As Fp, also Fc started to increase and decrease slightly earlier at Kumpula than at Torni in the morning and afternoon and the fluxes remained high after the afternoon rush hour peak. Nocturnal Fc were slightly larger at Torni than at Kumpula on weekdays, whereas on weekends the difference was two to threefold. As no pronounced difference was observed in nocturnal Fp, the relatively large nocturnal Fc on weekends cannot be solely explained by traffic or other combustion due to similarity in the sources of the studied variables. A plausible explanation might be that the increased Fc on weekend nights are caused by increased number of people that increase the rate of respiration in the city centre. According to Prairie and Duarte (2007), people emit CO2 around 251 g C d1 which would , produce Fc of 1.95 lmol m2 s1 around the Torni site (with a population density of 8090 persons km2, HSY (2011)). This is however a lower limit estimation as the population density statistics covers people living in the area and not those only visiting. The CO2 fluxes in Helsinki centre are similar in magnitude to those measured in the centre of Basel, where daytime medians range between 15 and 20 lmol m2 s1 (Lietzke and Vogt, 2013). On the other hand, twice as high Fc has been observed in Florence, Italy (Matese et al., 2009) and in Edinburgh, UK (Nemitz et al., 2002). In the megacity of London, the difference is even higher due to the large amount of anthropogenic CO2 emissions (Kotthaus and Grimmond, 2012). A previous study has shown larger spatial variability between suburban and urban areas in Montreal (Bergeron and Strachan, 2011), but in the current study only the direction of the large road at Kumpula was analysed. 3.5. Spatial variability of Fp and Fc in the centre of Helsinki As it is clear from the above analysis, the two scalar fluxes do not behave similarly despite the similar sources. To get a better understanding where the differences may originate, the directional variability of both Fp and Fc were examined at Torni (Fig. 8). Most of the time both Fp and Fc experienced similar directional variability. Throughout the year, both fluxes were greater than on average from the wind direction 330–340°, which is parallel to the main street of Helsinki centre, the Mannerheimintie. The increased fluxes likely stemmed from the blending effect of turbulence lifting particles from the street canyon to the EC measurement level. In fall, spring and summer elevated fluxes were also observed from direction 240°, which is parallel to the cross streets of the main street. In winter, spring and fall, Fp and Fc were pronounced in direction 40–50°, which is normal to the main street Mannerheimintie (Fig. 1). Especially in spring, the distance of the particle flux footprint in this direction was shorter than on average (not shown), which implies that the greater particle fluxes were due to notably higher emissions on Mannerheimintie. For all seasons, both fluxes were notably small from the direction

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

105

Fig. 8. (a) Particle number flux Fp (109 m2 s1) and b) CO2 flux Fc (lmol m2 s1) as a function of wind direction (°) at Torni in all seasons. The thick line is the median and the dashed lines show the 25 and 75 percentiles. Flux measurements were omitted due to flow distortion when the wind direction was 50–185°.

10–30°, where the railway station is located. Open-air rail traffic has only a small influence on local particle emissions compared to road traffic (Abbasi et al., 2013) which is supported by our study. The spatial variability of Fp and Fc were particularly similar in fall and winter, but in spring and summer to somewhat different behaviour was seen. Fc had similar strength in all directions, whereas Fp only peaked in the maxima directions. The different behaviour can at least partly be explained by human metabolism particularly in summer, which would affect CO2, but not particles. Similarly soil respiration would increase Fc without affecting Fp, but as the effect of vegetation uptake is small on the net-fluxes, also the respiration can be assumed to be negligible. Some of the difference can also be explained by more efficient turbulent transfer of CO2, which was most pronounced in direction 190–240° when compared to the other directions (not shown).

3.6. Drivers of the fluxes The general behaviour of Fp and Fc suggests that the main driver for the emissions at both sites is road traffic. However, the fluxes respond very differently to seasonal changes as well as to the land use. Next we examine their response to different factors with main focus at the unexplored Torni site.

3.6.1. Surface cover In previous studies, strong exponential dependence between annual and daily Fc and fraction of vegetation has been found (Nordbo et al., 2012a; Velasco and Roth, 2010), but in central Helsinki no clear connection between either of the fluxes with the surface fractions was seen. This emphasizes the importance of emission sources as the surface can e.g. be covered

106

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

with large impervious areas with little traffic. Thus, the derived relationships do not necessarily apply in a smaller scale in a dense city centre. 3.6.2. Blending effect An important factor in the spatial variability of Fp and Fc can be orientation of the buildings and roads with respective to prevailing wind as this can affect how efficiently particles and CO2 are blended from the urban canopy layer to the constant flux layer. Partly the directional variability of both fluxes can be linked to this as elevated fluxes were observed when the wind was parallel to the street canyon. When wind is perpendicular to the street canyon, both scalars are likely to be less effectively blended to the above air and formation of wind vortices may cause increased concentrations on leeward side of the street, as Dos Santos-Juusela et al. (2013) earlier reported from a different street canyon site in Helsinki in the case of p. This is also supported by the particle concentration measurements made at the main street canyon as p were highest when the wind direction was normal to the main street. Generally around 150–200% higher p were measured in street level than at Torni when calculated as hourly medians from the particle concentration measurements. The difference increased to 250% before noon, after which it started to decrease, presumably due to enhanced vertical mixing, to a nearly constant evening value of 150%. The vertical particle concentration differences in central Helsinki did not show a distinct relationship with the diurnal course of traffic rates like the vertical CO2 concentration differences did in Basel, Switzerland (Lietzke and Vogt, 2013). Similarly in Vancoucer, Canada (Crawford and Christen, 2014), the largest vertical difference in CO2 was seen during the transition periods which is contrary to our observations in the case of particles. Unfortunately we do not have any vertical profile of CO2 concentrations available, so it is not clear whether the difference between this study and those listed above is due to the city or the scalar itself. 3.6.3. Traffic Despite the measured traffic rates in the city centre are measured only on a single road, we compare the correlations of traffic with the two studied scalars with the assumption that the changes in the traffic rates apply in larger area. Hourly traffic rates correlate linearly better with Fc (R2  0.67 and RMSE  4.6 lmol m2 s1) than with Fp (R2  0.33 and RMSE  0.25  109 m2 s1) during the traffic measurement campaigns in spring and autumn in the city centre traffic monitoring point. In previous studies, an exponential curve has been fitted to the traffic data despite it is still unclear should the relationship be a linear or a non-linear. In our case, the non-linear fitting (Fp = aebtraffic, where a and b are constants) provided better correlations with R2  0.67 and RMSE  1.5 lmol m2 s1 for Fc, and R2  0.53 and RMSE  1.9  109 m2 s1 for Fp. The better correlation with Fc could be due to the fact that CO2 emissions directly relate to the amount of fuel burnt, whereas particle emissions are more dependent on temperature, filter and engine techniques (e.g. Kittelson et al. (2004)). 3.6.4. Air temperature A previous study from Kumpula reported a negative correlation between particle emission factor and T due to enhanced particle emissions in cold temperatures from road traffic (Ripamonti et al., 2013). In our case a clear decrease in binned Fp as a function of T can be seen both at Kumpula and Torni (Fig. 9). At both sites, the daytime increase in Fp was fivefold when temperature dropped from 22 to 8 °C. In the case of Fc, the daytime increase was only 1.5 at Torni and negligible at Kumpula between the same temperature range. Hence, at least part of the different seasonal variability of Fp and Fc can be explained with increased particle emissions when compared to CO2 emissions in cold temperatures due to cold engines and road dust. Commonly, emissions of other greenhouse gases and particles are expressed as emission factors relative to CO2 (Vogt et al., 2011b). Based on our result, changes in air temperature should be considered in such studies at least in north European urban and suburban areas. The highest nocturnal net CO2 emissions in the city centre are seen above 20 °C. These are likely related to human respiration on warm summer nights as soil respiration is assumed to have only a small role similarly to CO2 uptake in daytime. This would also be supported by the footprint analysis as during summer nights the source area extends to the Railway Station Square where lot of people congregate in summer time. In Helsinki, the nocturnal Fc were not as sensitive to T than e.g. at urban and suburban sites in Montreal, Canada (Bergeron and Strachan, 2011). This might be do to the more intense traffic locations in Helsinki, whereas at their both sites the role of vegetation has greater importance. 3.6.5. Atmospheric stability Atmospheric stability also affects Fp and is a probable factor for not observing a two-peaked diurnal pattern for both particles and CO2. At Torni, Fp and Fc decreased with increasing stability (Fig. A3), a behaviour that has previously also been reported at Kumpula (Järvi et al., 2009b). This causes the highest fluxes to occur around the midday. The atmospheric stabilities also vary by land use. The median stability f at Torni was 0.2 (quartiles 0.5 and 0.1) (Fig. 7e), whereas at Kumpula the stratification remained more stable. In the city centre, the anthropogenic heat emissions from traffic and buildings maintain the atmosphere on average unstably stratified throughout the year. For example, the median difference in sensible heat fluxes between Torni and Kumpula has been observed to be over 50 W m2 (Nordbo et al., 2013). The atmosphere is more stable in fall and most unstable in summer. The stability varies less at Kumpula, where the stratification in winter and fall are on average stable or neutral (Fig. 7e). Thus, the stability differences between the two sites can explain some of the observed differences, like the elevated evening fluxes of both particles and CO2. This emphasizes the complex processes, where not only the emission sources matter but also the local meteorological conditions needs to be considered.

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

107

Fig. 9. Daytime and nocturnal dependence of aerosol particle number (Fp) and CO2 (Fc) flux against air temperature (T) at Torni and Kumpula. Values are calculated as medians for 3 °C bins and only those values with over 20 data points are considered.

Besides, the above-mentioned factors some of the difference in the fluxes of these two scalars can be caused by their different reactivity and transformation ability. CO2 is a passive scalar whereas particles are reactive species and can form as secondary pollutants. Thus, the chemical and physical transformation will affect particularly the number of ultrafine particles. In this context, some of the differences observed between CO2 and number of particles may be explained by the heterogeneous chemistry occurring during this time.

4. Conclusions Aerosol particle number and CO2 flux measurements using the EC method were performed at two locations in Helsinki, Finland, between July 2011 and June 2013. The observations were made in central Helsinki at Torni site, which is representative for a highly built-up urban surface, and at the suburban Kumpula site located outside the city centre. From the latter, only the direction of one of the main roads leading to the centre of Helsinki was included in the analysis of spatial and temporal variation of fluxes. For the first time, we were able to quantify the intra-city variation of surface particle fluxes simultaneously with CO2 fluxes. The EC method was shown to be an appropriate technique to measure the particle and CO2 fluxes also at complex dense city centre. The similarity theory, also used to correct spectral losses, was shown to apply sufficiently well supporting previous studies of momentum and heat flux from the site (Nordbo et al., 2013). In addition, the storage fluxes for both scalars were shown to be negligible when compared to the actual EC fluxes (0.1% and 0.0002% of fluxes, respectively). Both sites acted as a net source for particles (Fp, median 0.18  109 m2 s1 and 0.17  109 m2 s1 at Torni and Kumpula) and CO2 (Fc, median 9.8 lmol m2 s1 and 5.7 lmol m2 s1 at Torni and Kumpula) mainly due to emissions from traffic. However, their seasonal and spatial variability was different. Fp was rather independent on the land use and a large road emitted particles in a similar strength than a dense urban centre. At the same time Fc was constantly larger at Torni than at Kumpula where the vegetation plays a minor role in the carbon cycle. Especially, nocturnal CO2 emissions were two to threefold at Torni on weekends likely due to nightlife activity and human respiration. Fp varied more strongly by season than Fc with distinctly lower net fluxes in summer than in other seasons. One reason for the different seasonal variability of the two scalar fluxes is increased particle emissions from traffic in colder temperatures due to road dust and direct emissions. This effect seemed to be greater than the effect of vegetation particularly in the city centre where the fraction of vegetation cover is less than 10%. In the city centre, both scalars behaved similarly for most directions and the greatest fluxes were observed when the wind was parallel to the street canyons. The surface fraction of vegetation or impervious surface could not explain the spatial

108

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

variability of either fluxes. At the street level, p was 150–200% larger than at the Torni site, with the strongest differences before noon. Particle concentrations were highest when the wind direction was normal to the main street, which was ascribed to accumulation of particles on leeward side of the street canyon. These results should be taken into account when forecasting poor air quality events. Analysis of transfer efficiencies indicated that the transport of particles and CO2 was similar at Kumpula, whereas CO2 was more effectively transported at Torni, which explained relatively larger Fc in the city centre to some extent. This should be considered when examining differences between the two scalars. This study demonstrates that despite aerosol particles and CO2 are emitted from same sources in urban areas, these two important variables for urban atmosphere behave very differently due to the different emission mechanisms, chemical transformation, meteorological conditions as well as turbulent transport efficiencies. Thus, deriving simple relationships between these two scalars is not appropriate for different applications such as emission factor estimation, but rather the environmental factors also need to be considered. Acknowledgements For funding, we thank the Academy of Finland (Projects nos. 138328, 1127756 and ICOS-Finland 263149), the Academy of Finland Centre of Excellence (Project no. 1118615), and the Nordic Centre of Excellence DEFROST. The additional particle number concentration and traffic rate measurements were provided by Helsinki Region Environmental Services Authority and Helsinki City Planning Department, respectively. In addition, we thank Sokos Hotel Torni for providing the measurement platform for our city centre measurements. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.uclim. 2015.07.006. References Abbasi, S., Jansson, A., Sellgren, U., Olofsson, U., 2013. Particle emissions from rail traffic: a literature review. Crit. Rev. Env. Sci. Tec. 43 (23), 2511–2544. http://dx.doi.org/10.1080/10643389.2012.685348. Aubinet, M., Vesala, T., Papale, D., 2012. Eddy Covariance: A Practical Guide to Measurement and Data Analysis. Springer Atmospheric Sciences, Netherlands. Barlow, J.F., Dunbar, T.M., Nemitz, E.G., Wood, C.R., Gallagher, M.W., Davies, F., O’Connor, E., Harrison, R.M., 2011. Boundary layer dynamics over London, UK, as observed using Doppler lidar during REPARTEE-II. Atmos. Chem. Phys. 11 (5), 2111–2125. http://dx.doi.org/10.5194/acp-11-2111-2011. Bergeron, O., Strachan, I.B., 2011. CO2 sources and sinks in urban and suburban areas of a northern mid-latitude city. Atmos. Environ. 45 (8), 1564–1573. http://dx.doi.org/10.1016/j.atmosenv.2010.12.043. Burba, G., 2013. Eddy Covariance Method for Scientific, Industrial, Agricultural, and Regulatory Applications: A Field Book on Measuring Ecosystem Gas Exchange and Areal Emission Rates. LI-COR Biosciences, Lincoln, NE, USA. Cava, D., Donateo, A., Contini, D., 2014. Combined stationarity index for the estimation of turbulent fluxes of scalars and particles in the atmospheric surface layer. Agr. Forest Meteorol. 194, 88–103. http://dx.doi.org/10.1016/j.agrformet.2014.03.021. Chow, J.C., Watson, J.G., Mauderly, J.L., Costa, D.L., Wyzga, R.E., Vedal, S., Hidy, G.M., Altshuler, S.L., Marrack, D., Heuss, J.M., Wolff, G.T., Pope III, C.A., Dockery, D.W., 2006. Health effects of fine particulate air pollution: lines that connect. J. Air Waste Manage. 56 (10), 1368–1380. City, H., 2013. Urban Facts: Statistical Yearbook of Helsinki. City of Helsinki, Keuruu, Finland. Contini, D., Donateo, A., Elefante, C., Grasso, F.M., 2012. Analysis of particles and carbon dioxide concentrations and fluxes in an urban area: correlation with traffic rate and local micrometeorology. Atmos. Environ. 46, 25–35. http://dx.doi.org/10.1016/j.atmosenv.2011.10.039. Crawford, B., Christen, A., 2014. Spatial variability of carbon dioxide in the urban canopy layer and implications for flux measurements. Atmos. Environ. 98, 308–322. http://dx.doi.org/10.1016/j.atmosenv.2014.08.052. Dahlkötter, F., Griessbaum, F., Schmidt, A., Klemm, O., 2010. Direct measurement of CO2 and particle emissions from an urban area. Meteorol. Z. 19 (6), 565– 575. http://dx.doi.org/10.1127/0941-2948/2010/0492. 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), 791–800. http://dx.doi.org/10.1016/s1352-2310(01)00526-x. Dos Santos-Juusela, V., Petäjä, T., Kousa, A., Hämeri, K., 2013. Spatial-temporal variations of particle number concentrations between a busy street and the urban background. Atmos. Environ. 79, 324–333. http://dx.doi.org/10.1016/j.atmosenv.2013.05.077. Feigenwinter, C., Vogt, R., Christen, A., 2012. Eddy covariance measurements over urban areas. In: Aubinet, M., Vesala, T., Papale, D. (Eds.), Eddy Covariance: A Practical Guide to Measurement and Data Analysis. Springer, pp. 377–397. FMI, 2013a. Annual statistics. Finnish Meteorological Institute. (cited 15.08.13). FMI, 2013b. Seasonal statistics. Finnish Meteorological Institute. (cited 15.08.13). Foken, T., Wichura, B., 1996. Tools for quality assessment of surface-based flux measurements. Agr. Forest Meteorol. 78 (1–2), 83–105. http://dx.doi.org/ 10.1016/0168-1923(95)02248-1. Harrison, R.M., Dall’Osto, M., Beddows, D.C.S., Thorpe, A.J., Bloss, W.J., Allan, J.D., Coe, H., Dorsey, J.R., Gallagher, M., Martin, C., Whitehead, J., Williams, P.I., Jones, R.L., Langridge, J.M., Benton, A.K., Ball, S.M., Langford, B., Hewitt, C.N., Davison, B., Martin, D., Petersson, K., Henshaw, S.J., White, I.R., Shallcross, D.E., Barlow, J.F., Dunbar, T., Davies, F., Nemitz, E., Phillips, G.J., Helfter, C., Di Marco, C.F., Smith, S., 2012. Atmospheric chemistry and physics in the atmosphere of a developed megacity (London): an overview of the REPARTEE experiment and its conclusions. Atmos. Chem. Phys. 12, 3065–3114. http://dx.doi.org/10.5194/acp-12-3065-2012. HCPD, 2013. Traffic volumes on main street of Helsinki: September 2013. HCPD, Helsinki, Finland. Hering, S.V., Stolzenburg, M.R., Quant, F.R., Oberreit, D.R., Keady, P.B., 2005. A laminar-flow, water-based condensation particle counter (WCPC). Aerosol Sci. Tech. 39 (7), 659–672. http://dx.doi.org/10.1080/02786820500182123. HSY, 2011. SeutuCD’11. Helsinki Region Environmental Services Authority. HSY, 2013. Air quality in the Helsinki Metropolitan Area in 2013 (in Finnish). HSY, Helsinki Region Environmental Services Authority, Helsinki, Finland. Hussein, T., Mølgaard, B., Hannuniemi, H., Martikainen, J., Järvi, L., Wegner, T., Ripamonti, G., Weber, S., Vesala, T., Hämeri, K., 2014. Fingerprints of the urban particle number size distribution in Helsinki, Finland: local versus regional characteristics. Boreal Environ. Res. 19 (1), 1–20.

M. Kurppa et al. / Urban Climate 13 (2015) 94–109

109

IPCC, 2013. Climate Change 2013: The Physical Science Basis. Contribution of Working group I to the Fifth Assesment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK, and New York, NY, USA. Järvi, L., Grimmond, C.S.B., Taka, M., Nordbo, A., Setälä, H., Strachan, I.B., 2014. Development of the Surface Urban Energy and Water Balance Scheme (SUEWS) for cold climate cities. Geosci. Model Dev. 7 (4), 1691–1711. http://dx.doi.org/10.5194/gmd-7-1691-2014. Järvi, L., Hannuniemi, H., Hussein, T., Junninen, H., Aalto, P.P., Hillamo, R., Mäkelä, T., Keronen, P., Siivola, E., Vesala, T., Kulmala, M., 2009a. The urban measurement station SMEAR III: continuous monitoring of air pollution and surface–atmosphere interactions in Helsinki, Finland. Boreal Environ. Res. 14, 86–109. Järvi, L., Rannik, Ü., Mammarella, I., Sogachev, A., Aalto, P.P., Keronen, P., Siivola, E., Kulmala, M., Vesala, T., 2009b. Annual particle flux observations over a heterogeneous urban area. Atmos. Chem. Phys. 9 (20), 7847–7856. Katul, G., Goltz, S.M., Hsieh, C.I., Cheng, Y., Mowry, F., Sigmon, J., 1995. Esrimation of surface heat and momentum fluxes using the flux-variance method above uniform and nonuniform terrain. Bound.-Lay. Meteorol. 74 (3), 237–260. http://dx.doi.org/10.1007/bf00712120. Kittelson, D.B., Watts, W.F., Johnson, J.P., 2004. Nanoparticle emissions on Minnesota highways. Atmos. Environ. 38 (1), 9–19. http://dx.doi.org/10.1016/ j.atmosenv.2003.09.037. Kormann, R., Meixner, F.X., 2001. An analytical footprint model for non-neutral stratification. Bound.-Lay. Meteorol. 99 (2), 207–224. http://dx.doi.org/ 10.1023/a:1018991015119. Kotthaus, S., Grimmond, C.S.B., 2012. Identification of micro-scale anthropogenic CO2, heat and moisture sources – processing eddy covariance fluxes for a dense urban environment. 57, 301–316, doi: 10.1016/j.atmosenv.2012.04.024. Kreyling, W.G., Semmler-Behnke, M., Moller, W., 2006. Ultrafine particle-lung interactions: does size matter? J. Aerosol Med. 19 (1), 74–83. http:// dx.doi.org/10.1089/jam.2006.19.74. Lähde, T., Niemi, J.V., Kousa, A., Ronkko, T., Karjalainen, P., Keskinen, J., Frey, A., Hillamo, R., Pirjola, L., 2014. Mobile particle and NOx emission characterization at Helsinki Downtown: comparison of different traffic flow areas. Aerosol Air Qual. Res. 14 (5), 1372–1382. http://dx.doi.org/10.4209/ aaqr.2013.10.0311. Lietzke, B., Vogt, R., 2013. Variability of CO2 concentrations and fluxes in and above an urban street canyon. Atmos. Environ. 74, 60–72. http://dx.doi.org/ 10.1016/j.atmosenv.2013.03.030. Lilleberg, I., Hellman, T., 2012. The development of traffic in Helsinki in 2011 (in Finnish). Helsinki City Planning Department, Helsinki, Finland. Lilleberg, I., Hellman, T., 2013. The development of traffic in Helsinki in 2012 (in Finnish). Helsinki City Planning Department, Helsinki, Finland. Liu, H.P., Peters, G., Foken, T., 2001. New equations for sonic temperature variance and buoyancy heat flux with an omnidirectional sonic anemometer. Bound.-Lay. Meteorol. 100 (3), 459–468. http://dx.doi.org/10.1023/a:1019207031397. 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, 769–785. Martin, C.L., Longley, I.D., Dorsey, J.R., Thomas, R.M., Gallagher, M.W., Nemitz, E., 2009. Ultrafine particle fluxes above four major European cities. Atmos. Environ. 43 (31), 4714–4721. http://dx.doi.org/10.1016/j.atmosenv.2008.10.009. Massman, W.J., 2000. A simple method for estimating frequency response corrections for eddy covariance systems. Agr. Forest Meteorol. 104 (3), 185–198. http://dx.doi.org/10.1016/s0168-1923(00)00164-7. Matese, A., Gioli, B., Vaccari, F.P., Zaldei, A., Miglietta, F., 2009. Carbon dioxide emissions of the city center of Firenze, Italy: measurement, evaluation, and source partitioning. J. Appl. Meteorol. Clim. 48 (9), 1940–1947. http://dx.doi.org/10.1175/2009jamc1945.1. Moncrieff, J.B., Massheder, J.M., deBruin, H., Elbers, J., Friborg, T., Heusinkveld, B., Kabat, P., Scott, S., Soegaard, H., Verhoef, A., 1997. A system to measure surface fluxes of momentum, sensible heat, water vapour and carbon dioxide. J. Hydrol. 189 (1–4), 589–611. Moriwaki, R., Kanda, M., 2006. Local and global similarity in turbulent transfer of heat, water vapour, and CO2 in the dynamic convective sublayer over a suburban area. Bound.-Lay. Meteorol. 120 (1), 163–179. http://dx.doi.org/10.1007/s10546-005-9034-4. Nemitz, E., Hargreaves, K.J., McDonald, A.G., Dorsey, J.R., Fowler, D., 2002. Meteorological measurements of the urban heat budget and CO2 emissions on a city scale. Environ. Sci. Technol. 36 (14), 3139–3146. http://dx.doi.org/10.1021/es010277e. Nordbo, A., Järvi, L., Haapanala, S., Moilanen, J., Vesala, T., 2013. Intra-city variation in urban morphology and turbulence structure in Helsinki, Finland. Bound.-Lay. Meteorol. 146 (3), 469–496. http://dx.doi.org/10.1007/s10546-012-9773-y. Nordbo, A., Järvi, L., Haapanala, S., Wood, C.R., Vesala, T., 2012a. Fraction of natural area as main predictor of net CO2 emissions from cities. Geophys. Res. Lett. 39 (20), 20802. http://dx.doi.org/10.1029/2012gl053087. Nordbo, A., Järvi, L., Vesala, T., 2012b. Revised eddy covariance flux calculation methodologies – effect on urban energy balance. Tellus B 64, 18184. http:// dx.doi.org/10.3402/tellusb.v64i0.18184. Prairie, Y.T., Duarte, C.M., 2007. Direct and indirect metabolic CO2 release by humanity. Biogeosciences 4 (2), 215–217. PRC, 2014. Population statistics, Population Register Center of Finland. PRC. (cited 12.05.14). Rannik, Ü., Mammarella, I., Aalto, P., Keronen, P., Vesala, T., Kulmala, M., 2009. Long-term aerosol particle flux observations part I: Uncertainties and timeaverage statistics. Atmos. Environ. 43 (21), 3431–3439. http://dx.doi.org/10.1016/j.atmosenv.2009.02.049. Ripamonti, G., Järvi, L., Mølgaard, B., Hussein, T., Nordbo, A., Hämeri, K., 2013. The effect of local sources on aerosol particle number size distribution, concentrations and fluxes in Helsinki, Finland. Tellus B 65, 19786. http://dx.doi.org/10.3402/tellusb.v65i0.19786. Roth, M., Oke, T.R., 1995. Relative efficiencies of turbulent transfer of heat, mass, and momentum over a patchy urban surface. J. Atmos. Sci. 52 (11), 1863– 1874. http://dx.doi.org/10.1175/1520-0469(1995) 052<1863:reotto>2.0.co;2. Stewart, I.D., Oke, T.R., 2012. Local climate zones for urban temperature studies. B. Am. Meteorol. Soc. 93 (12), 1879–1900. http://dx.doi.org/10.1175/bamsd-11-00019.1. Vallius, M., Lanki, T., Tiittanen, P., Koistinen, K., Ruuskanen, J., Pekkanen, J., 2003. Source apportionment of urban ambient PM2.5 in two successive measurement campaigns in Helsinki, Finland. Atmos. Environ. 37 (5), 615–623. http://dx.doi.org/10.1016/s1352-2310(02)00925-1. Velasco, E., Roth, M., 2010. Cities as net sources of CO2: review of atmospheric CO2 exchange in urban environments measured by eddy covariance technique. Geograph. Compass 4 (9), 1238–1259. Vesala, T., Kljun, N., Rannik, Ü., Rinne, J., Sogachev, A., Markkanen, T., Sabelfeld, K., Foken, T., Leclerc, M.Y., 2008. Flux and concentration footprint modelling: state of the art. Environ. Poll. 152 (3), 653–666. http://dx.doi.org/10.1016/j.envpol.2007.06.070. Virtanen, A., Rönkkö, T., Kannosto, J., Ristimäki, J., Mäkelä, J.M., Keskinen, J., Pakkanen, T., Hillamo, R., Pirjola, L., Hämeri, K., 2006. Winter and summer time size distributions and densities of traffic-related aerosol particles at a busy highway in Helsinki. Atmos. Chem. Phys. 6, 2411–2421. Vogt, M., Nilsson, E.D., Ahlm, L., Mårtensson, E.M., Johansson, C., 2011a. Seasonal and diurnal cycles of 0.25–2.5 lm aerosol fluxes over urban Stockholm, Sweden. Tellus B 63 (5), 935–951. http://dx.doi.org/10.1111/j.1600-0889.2011.00551.x. Vogt, M., Nilsson, E.D., Ahlm, L., Mårtensson, E.M., Struthers, H., Johansson, C., 2011b. Traffic aerosol emission velocity derived from direct flux measurements over urban Stockholm, Sweden. Atmos. Environ. 45 (32), 5725–5731. http://dx.doi.org/10.1016/j.atmosenv.2011.07.026. WHO, 2014. World health organization: Ambient (outdoor) air quality and health: Fact sheet 313. WHO. (cited 12.05.14). World Bank, 2014. Urban Development. World Bank. (cited 12.05.14).