Long-term study of cloud condensation nuclei (CCN) activation of the atmospheric aerosol in Vienna

Abstract

During a total of 11 months, cloud condensation nuclei (CCN at super-saturation S 0.5%) and condensation nuclei (CN) concentrations were measured in the urban background aerosol of Vienna, Austria. For several months, number size distributions between 13.22 nm and 929 nm were also measured with a scanning mobility particle spectrometer (SMPS). Activation ratios (i.e. CCN/CN ratios) were calculated and apparent activation diameters obtained by integrating the SMPS size distributions. Variations in all CCN parameters (concentration, activation ratio, apparent activation diameter) are quite large on timescales of days to weeks. Passages of fronts influenced CCN parameters. Concentrations decreased with the passage of a front. No significant differences were found for fronts from different sectors (for Vienna mainly north to west and south to east). CCN concentrations at 0.5% S ranged from 160 cm⁻³ to 3600 cm⁻³ with a campaign average of 820 cm⁻³. Activation ratios were quite low (0.02–0.47, average: 0.13) and comparable to activation ratios found in other polluted regions (e.g. Cubison et al., 2008). Apparent activation diameters were found to be much larger (campaign average: 169 nm, range: (69–370) nm) than activation diameters for single-salt particles (around 50 nm depending on the salt). Contrary to CN concentrations, which are influenced by source patterns, CCN concentrations did not exhibit distinct diurnal patterns. Activation ratios showed diurnal variations counter-current to the variations of CN concentrations.

Highlights

► CCN concentrations (at 0.5% super-saturation) are highly variable. ► No seasonal differences were found. ► CN show typical diurnal trends, while CCN do not. ► Activation ratios were found to be low (0.02–0.47, average: 0.13) but equally highly variable.

1. Introduction

Cloud condensation nuclei (CCN) are an important fraction of the atmospheric aerosol, because they influence cloud microphysical and radiative properties, and consequently aerosol indirect radiative forcing (IPCC, 2007). Despite intensive efforts by many international groups, which have greatly increased in the past few years (see Andreae and Rosenfeld, 2008, for an overview), information on real-world cloud condensation nuclei (CCN) is inadequate, and the review of uncertainties and open questions in the area of CCN activation to cloud droplets by McFiggans et al. (2006) is still to the point.

Field studies have been conducted both in remote and polluted areas under different meteorological and seasonal conditions, but usually only in campaigns lasting some weeks at most. Long-term studies on CCN were conducted more than 20 years ago (e.g. by Alofs and Liu, 1981; Hudson and Frisbie, 1991). In these studies, CCN concentrations and sometimes also the chemical composition of the total aerosol are measured, but the supporting information about the aerosol (size distribution, total particle concentration) depends on the study. Recently aerosol mass spectrometry data on single particle composition have become available, but even this information together with data on humidity growth seems to be inadequate for CCN modelling (Ervens et al., 2010). No information on the chemical composition of the CCN themselves is available.

Laboratory studies are focused on CCN activation of aerosols produced from single substances, mixtures of two or three substances (usually an inorganic salt mixed with a single organic substance or a combination of organic substances), or coated particles. In this approach, the vast complexity of atmospheric aerosol particles and their composition cannot be assessed satisfactorily. An earlier laboratory study conducted by our group on aircraft exhaust aerosol (Hitzenberger et al., 2003) also showed that modelled and measured activation ratios (AR = CCN/CN, with CN the total particle concentration) were quite diverse even though measured data on hygroscopic particle growth at sub-saturated conditions were available.

Some years ago, the question was raised whether size is more important than chemistry for CCN activation (Dusek et al., 2006a). In that field study at a background site in Western Europe, the CCN activation and chemical composition were investigated for aerosols with various origin. No AR were obtained, however, and even “polluted” air masses had originated several tens to hundreds of kilometres upwind and had ageing times of at least a day. In this case, variations in the size distributions were found to explain 84–96% of the measured CCN concentrations. Hudson (2007) pointed out that in several studies conducted during the past decade, the variability of activation is much larger than that found by Dusek et al. (2006a) and that chemistry does play an important role in CCN activation. Recently, a limited number of short field studies were also conducted in polluted urban regions (e.g. Broekhuizen et al., 2006; Furutani et al., 2008; Cubison et al., 2008; Wang et al., 2010) and again large variations of CCN activation were found.

As CCN are important for radiative forcing, a parameterisation of CCN properties of aerosols would be highly desirable, and several models were developed to estimate CCN concentrations from size distributions or even CN concentrations (e.g. Fitzgerald et al., 1982; Covert et al., 1998; Ervens et al., 2007; Medina et al., 2007; Petters and Kreidenweis, 2008; Lance et al., 2009). From these studies, however, no general way has emerged to predict CCN concentrations.

The at the time even more limited knowledge of CCN especially in urban regions combined with the failure of the models led us to decide to perform a long-term (>1 year) field study of CCN activation in a source region in order to gain insight into the variability of CCN concentrations and AR both on short and long timescales. As central Europe is a comparatively small but heavily industrialized region with high population densities and strong traffic and seasonal space heating sources, an extrapolation of our results from the city of Vienna to the aerosol in the larger Central European area may be possible.

2. Experimental section

2.1. Site description and measurement set-up

All measurements were performed at the roof laboratory of the Physics Building in downtown Vienna (population in the conurbation ca. 2 million) at approximately 35 m above ground. The lab is situated inside the attic and aerosol was drawn in from the courtyard side which is shielded from direct traffic emissions by interconnected buildings and other courtyards. As roof heights in the area are rather homogeneous and lower than the roof of the Physics building, the station has uninterrupted fetch in all directions except due west (Vienna General Hospital building) and is above chimney heights in the area. Heating in the vicinity is provided by either natural gas or district heating. The aerosol at the site can be characterized as well mixed urban background aerosol.

From July 2007 to October 2008, measurements of CCN and CN concentrations as well as number size distributions were performed for a total of 11 months. Table 1 gives an overview of the measurement months and instrument runtime.

Table 1.

The table shows the months when the CCNC and/or the SMPS System were operated.

Month	CCNC	SMPS
July 2007	x	x
August 2007	x	x
September 2007	x
October 2007	x
November 2007	x
December 2007	x	x
January 2008	x	x
February 2008	x	x
July 2008	x	x
September 2008	x	x
October 2008	x	x

All instruments were placed inside the lab close to a window and aerosol was drawn in with sampling tubes of ca. 2 m length. The tube for the CCN/CN counters was a fabric-enforced PVC tube with 10 mm inner diameter and the tube for the SMPS a polyurethane tube with 6 mm inner diameter. Flow rates of 1.5 L min⁻¹ (CCN and CN counters) and 7.5 L min⁻¹ (SMPS) were controlled by critical orifices.

2.2. Instrument description

The University of Vienna CCN counter (Giebl et al., 2002) is a static thermal diffusion chamber. Two parallel horizontal plates separated by a glass ring and wetted with ultra-pure water on their inner surfaces are held at different temperatures. When the chamber is closed, linear temperature and water vapour pressure gradients are established. Due to the non-linear dependence of saturation vapour pressure on temperature, the air in the chamber is super-saturated with maximum S near the centre plane. S_max is controlled by the temperature difference between the plates and was set to 0.5% in this study. CCN concentrations are measured on a time scale of 5 min. In a measurement cycle, the chamber is flushed for 4.5 min, then the chamber is closed (static mode) and the flow directed through a bypass to a CN counter (TSI model 3010, working fluid: butanol). In the static mode, the formation of droplets in the CCN chamber is observed for 30 s. The droplets are illuminated by a laser beam passing the centre line of the chamber. Pictures taken by a CCD camera are stored on a computer and droplets are counted by an image analysis routine. Activation ratios are calculated from the CCN and CN concentrations. Detailed descriptions of the set-up and calibration procedure of the CCN counter are given by Giebl et al. (2002) and Dusek et al. (2006b). Diffusion losses for 10 nm particles were ca. 10%.

During part of the study, a closed-loop Vienna type differential mobility analyzer (DMA, Winklmayr et al., 1991; closed-loop arrangement: Jokinen and Mäkelä, 1997) was operated as a scanning mobility particle spectrometer (SMPS) in combination with a TSI CPC 3760A (working fluid: butanol) to obtain a scan of the number size distribution every 10 min. The instrument has a central electrode with a length of 570 mm and a radius 15.75 mm (radius of outer electrode: 22.2 mm). At a sheath air flow rate of 7.5 L min⁻¹ (aerosol flow rate: 1.5 L min⁻¹) and voltages between 9.76 V and 10,000 V, the size range of the particles is (13.22–929) nm. Outside aerosol was drawn in with 7.5 L min⁻¹ as well to minimize diffusion losses of particles, and the flow rate was reduced to 1.5 L min⁻¹ by an overflow right at the inlet of the SMPS. Diffusion losses for the smallest particles were ca. 5% for the whole set-up of the SMPS.

These number size distributions were used to estimate an apparent activation diameter d_act,a (i.e. the lower size limit of an integral of the size distribution yielding the same number concentration as the measured CCN concentration), as actual measurements of activation diameters with the CCN counter were not possible due to low particle concentrations in the narrow size channels provided by this SMPS.

3. Results

3.1. Monthly mean values of CCN and CN concentrations and AR

To investigate possible seasonal trends, time series of CN and CCN concentrations and AR were evaluated statistically and monthly mean and median values were calculated. Fig. 1 shows these data for CN concentrations (Fig. 1a), CCN concentrations (Fig. 1b) and the AR (Fig. 1c). The vertical bars represent the range of variation (standard deviation). All measured quantities are far from constant over the measurement period, although no seasonal trend is visible. The arithmetic mean in all cases is larger than the corresponding median indicating that the distribution of the values is slightly skewed.

Fig. 1.

Monthly means and medians of CN concentration (a), CCN concentration at 0.5% S (b) and activation ratio (c). The vertical bars represent the range of variation (standard deviation) to each calculated mean value.

In Table 2 mean, maximum and minimum values for the whole measurement period are listed. The differences between maximum and minimum values are remarkably large giving an idea of how much the CCN and CN concentrations and AR can vary at the site. CN concentrations ranged from 910 cm⁻³ to over 50,000 cm⁻³ which is typical for urban aerosols. The low minimum CN concentrations are surprising, but similarly low concentrations have been observed occasionally after frontal passages advecting air masses from the North Atlantic (Reischl, personal communication). The CCN concentration at 0.5% S ranged from 160 cm⁻³ to approximately 3600 cm⁻³ with a mean value of 820 cm⁻³.

Table 2.

Mean, maximum and minimum values of CN and CCN concentration and activation ratio at 0.5% S for the measurement period.

	CN concentration [1 cm−3]	CCN concentration [1 cm−3]	Activation ratio [−]
Mean	7300	820	0.13
Maximum	51,600	3600	0.47
Minimum	910	160	0.02

In an earlier study with two campaigns of two weeks each in winter and spring 1996 in Vienna (Hitzenberger et al., 1999) the maximum and minimum values of the CCN concentration (0.5% S) were 3080 cm⁻³ and 500 cm⁻³ in winter and 2630 cm⁻³ and 170 cm⁻³ in spring. Those values compare well with our current study. The mean concentrations then, however, were significantly higher (winter: 2448 cm⁻³, spring: 1353 cm⁻³) than the mean concentration (11-month average: 820 cm⁻³) in the current study. The earlier values, however, had been obtained in two short periods while the current mean value is calculated by averaging over 11 months. As shown before the CCN concentrations vary strongly, so short measurement periods might not be representative.

As expected, the CCN concentrations in this study are much higher than concentrations measured by Hitzenberger et al. (1999) in two-week campaigns in 1995 and 1996 at a mountain site in Austria (Mt. Sonnblick 3106 m asl) where mean and maximum values ranged around a few hundreds per cm³ (fall: 243 and 570 cm⁻³, summer: 402 and 786 cm⁻³) while the minimum values were 80 cm⁻³ in fall and 29 cm⁻³ in summer. No CN data were obtained on Sonnblick, but in March 2000, a three week measurement campaign with a more complete set of instruments was conducted at another mountain site in Austria (Mt. Rax, 1680 m asl). Out-of-cloud CCN concentrations (at 0.5% S) there were between <100 and ca. 2300 cm⁻³ with an average of 388 cm⁻³ (Hitzenberger et al., 2000). CN concentrations were between 125 and 6640 cm⁻³ with an average around 1850 cm⁻³ (Giebl, personal communication).

The annual mean AR in Vienna was 0.13 with maximum and minimum values of 0.43 and 0.02. Again the arithmetic mean in all cases is larger than its corresponding median. At Mt. Rax, out-of-cloud AR were found to be between 0.02 and 0.76, with an average of 0.24, which corresponds also to the data given by van Ekeren et al. (2005) for measurements at Jungfraujoch (0.4–0.6).

In the literature, several other studies are described where CCN concentrations and AR were measured (see e.g. Andreae and Rosenfeld, 2008; Ervens et al., 2010, and references given by McFiggans et al., 2006). Published AR data, however, have to be treated with care because the lower cut size of CN counters crucially influences total CN concentrations. Different CN counters were used with different (and sometimes undeclared) cut sizes. At 0.5% S very low CCN concentrations (<100 or a few hundred cm⁻³) are found in arctic air masses (Yum and Hudson, 2001), over the Atlantic Ocean, the Southern Ocean and the Eastern Pacific (Hudson and Xie, 1999). These concentrations seem to have been stable over the last decades. Pruppacher and Klett (1997) summarize data from older studies and confirm that CCN concentrations in maritime air masses rarely exceed 300 cm⁻³ even at 1% S. The highest CCN concentrations (several thousands cm⁻³) occur within highly polluted continental air masses, e.g. in the Amazon Basin during the fire season (Vestin et al., 2007), or in Beijing and Guangzhou (Southern China, Rose et al., 2010).

As pointed out by Andreae and Rosenfeld (2008) AR almost reaches 1 in aged background aerosols (characterized by particles with sizes mainly above ∼50 nm) for 1% S while in fresh aerosols (large fraction of particles in the lower Aitken and nucleation modes, e.g. urban aerosols) AR at the same S has quite small values which agrees well with our observations. An earlier evaluation of the number size distributions by Burkart et al. (2010) revealed that on average 75% of the particles are smaller than 100 nm, which indicates the presence of large fractions of freshly emitted aerosol particles or maybe fresh secondary particles, although new particle formation events are rare in Vienna (Borsós et al., in preparation).

AR and CCN concentrations at different S are available for other urban locations such as Toronto (Broekhuizen et al., 2006), Riverside (Cubison et al., 2008), Guangzhou (Rose et al., 2010) or Mexico City (Wang et al., 2010). The time spans of the studies, however, ranged again from a few days to a few weeks, so they might not be representative for the range and averages of AR and CCN concentrations at these sites.

In downtown Toronto (Broekhuizen et al., 2006) the range of CCN concentrations ((144–3291) cm⁻³, 0.58% S) is very similar to our data while in highly polluted Guangzhou (Rose et al., 2010), CCN concentrations are much higher ((9649 ± 5214) cm⁻³, S: 0.47%). No AR are given for Toronto. At Guangzhou the average AR was 0.53 ± 0.19 at 0.47% S, which is much higher than AR determined for Vienna at 0.5% S. In Riverside at the eastern edge of the Los Angeles basin, Cubison et al. (2008) found quite constant weekday CCN concentrations around 2000 cm⁻³ at 0.5% S. AR was found to be 0.08 ± 0.03 at S = (0.27 ± 0.05)% which is fairly similar to Vienna taking into account the lower S in the Riverside measurements. In Mexico City (Wang et al., 2010) average CCN concentrations were (2000–4000) cm⁻³ and the average AR was 0.41 ± 0.15 (at 0.29% S), which again is higher than the AR in Vienna even though the Mexico City data were determined at lower S. Wang et al. (2010) attribute this high AR to rapid photochemical ageing processes.

3.2. Apparent activation diameters

If all particles had identical chemical composition, the activation diameter of an aerosol could be obtained by integrating the size distribution from large particles towards smaller ones until the integral equals measured CCN concentrations at a specific S. The lower size limit of such an integration is commonly referred to as the activation diameter (e.g. Furutani et al., 2008) and corresponds to the smallest size for CCN active at a specific S. As atmospheric aerosols, however, usually are complex mixtures of internally and externally mixed particles, such a calculation can yield only an “apparent activation diameter” d_act,a.

For the Vienna study, d_act,a were calculated for 0.5% S from the number size distributions. Fig. 2 gives a typical time series, and average values are listed in Table 3 for those months where SMPS data coverage was sufficient. It can be seen that d_act,a varied strongly and seldom had values below 100 nm, which would correspond to the Kelvin diameter of insoluble but wettable particles at 0.5% S. Generally d_act,a was larger than the Kelvin diameter and much larger than the activation diameter of soluble single-salt particles (ca. 50 nm depending on the salt). The average value of d_act,a for the entire measurement period was 169 nm (minimum: 69 nm; maximum: 368 nm).

Fig. 2.

Typical time series of the apparent activation diameter in July 2007.

Table 3.

Monthly means of apparent activation diameters at 0.5% S obtained by integrating the number size distributions.

	Jul.07	Dec.07	Jan.08	Feb.08	Jul.08	Sep.08	Oct.08	Nov.08
Mean act. diam [nm]	136	181	188	185	170	161	180	157
Minimum	69	162	134	91	69	111	69	111
Maximum	204	204	370	363	308	241	308	241

Under the assumption of completely internally mixed particles, such large d_act,a would indicate a highly hydrophobic aerosol such as fresh Diesel combustion exhaust. Similarly high and even higher d_act,a were found by Hitzenberger et al. (2003) for the aerosol produced by an aircraft engine combustor. In that study, d_act,a at 0.7% S were between 301 nm and 146 nm depending on operating conditions of the combustor indicating highly hydrophobic particles. Similar results for freshly emitted exhaust particles were found also by Hallett et al. (1989) and Pitchford et al. (1991).

In the study by Quinn et al. (2008) d_act,a for the maritime aerosol in the Gulf of Mexico ranged between 70 nm and 90 nm, whereas in the Houston ship channel with high marine traffic densities close to industrial and anthropogenic sources values were between 90 nm and 170 nm. Furutani et al. (2008) obtained similar results along the southern coast of California, where d_act,a was ca. 110 nm (at 0.6% S) for fresh ship exhaust (sampling downwind of the exhaust chimney of the ship’s diesel engines), while for fresh anthropogenic aerosol d_act,a ranged from 70 nm to 110 nm, and was around 50 nm for aged anthropogenic and clean maritime aerosol.

The urban aerosol of Vienna shows significantly higher values of d_act,a than found in coastal regions even in the presence of fresh ship engine exhaust. These large d_act,a indicate that as expected, the assumption of a homogeneously internally mixed aerosol is not true in urban areas and confirm the conclusions of the studies e.g. by Cubison et al. (2008) and Ervens et al. (2010) that fresher pollution aerosols require a more realistic treatment of the mixing state and knowledge of size-resolved chemical composition. Knowledge of the size distribution alone definitely is insufficient to accurately predict CCN concentrations (or d_act,a).

Furutani et al. (2008), however, found quite good correlations between AR and d_act,a for the different air masses sampled during their cruise along the Californian coast. Fig. 3 shows the relationship between AR and d_act,a for our study for summer (July 2008) and winter (November 2008) including fitted curves. The relation seems to be non-linear rather than linear, but regression coefficients are not high in either case. The variation in this relationship is too large to deduce even “rules of thumb”.

Fig. 3.

Apparent activation diameter [nm] vs. AR for summer (July 2008, left) and winter (November 2008, right) conditions in Vienna at 0.5% S.

3.3. Time series of CN and CCN concentrations and AR

As discussed no seasonal trends in AR, CCN or CN concentrations were found. A closer look at the time series reveals that especially CN concentration and AR show strong diurnal or weekly variations.

Particle concentrations depend generally not only on production and transformation processes, but also on the meteorological situation (air mass origin, mixing height, atmospheric stability, etc.). Meteorological influences are discussed in Section 3.4.

Fig. 4 shows a typical time series of CN and CCN concentrations and AR under wintry conditions (January 2008). Fig. 5 shows the same for a typical late summer period. Such time series were obtained for all 11 measurement months. In all cases we observed a strong diurnal variation of CN concentrations which was sometimes, however, interrupted by a frontal passage (see Section 3.4).

Fig. 4.

Typical time series of the CN concentrations and AR during wintry conditions (January 2008). The dashed arrows indicate situations when the total particle concentration finds a minimum/maximum and AR shows a concurrent maximum/minimum.

Fig. 5.

Typical time series of the CN concentrations and AR during stable conditions in late summer (September 2008). The dashed arrows indicate situations when the total particle concentration finds a minimum/maximum and AR shows a concurrent maximum/minimum.

During a typical weekday several maxima and minima of CN concentration were observed (see Figs. 6 and 7 for average diurnal patterns). Such patterns are typical for urban aerosols and detailed descriptions of diurnal cycles and differences between weekday and weekend patterns are found abundantly in the literature (e.g. Ruellan and Cachier (2000), Johansson et al. (2007), Pey et al. (2008), Costabile et al., 2009; Liu et al., 2011). CCN concentrations showed much less diurnal variation than CN concentrations, which corresponds to the diurnal variations of aged aerosol found e.g. by Liu et al. (2011) at Holme Moss (a background station near Manchester, UK).

Fig. 6.

Average CN and CCN concentrations on all weekdays and all Sundays for the whole measurement period. The CCN concentration is rather constant while the CN concentration clearly shows a diurnal pattern on weekdays as well as on Sundays.

Fig. 7.

The averaged AR on weekdays and on Sundays. On Sundays AR is fairly constant while on weekdays a minimum during the morning hour is observed coinciding with the morning rush hour traffic.

To further examine the differences in CN and CCN concentrations during weekdays and weekends, average concentrations for weekdays and Sundays were calculated. Saturdays were excluded because of more complicated traffic patterns (regulations prohibit heavy duty traffic on Saturdays from 3 p.m. onwards). Fig. 6 shows averaged weekday and Sunday CN and CCN concentrations. On weekdays CN concentrations rise in the early morning, reach a peak around 8 a.m. coinciding with traffic rush-hour, decrease during the day and clearly reach a second peak between 7 p.m. and 9 p.m. Sunday CN concentrations show a different pattern. The morning peak is much weaker (by a factor of 2) and shifted to later hours (between 9 and 10 a.m.). During the Sunday evening rush-hour CN concentrations are not significantly different from weekday values.

CCN concentrations showed diurnal patterns neither on weekdays nor on Sundays, which agrees with the findings for the aged aerosol at Holme Moss (Liu et al., 2011) or CCN at Riverside (Cubison et al., 2008). Weekday average concentrations in Vienna were higher on Sundays (weekdays: 948 cm⁻³, Sundays: 628 cm⁻³). This difference can at least be partially explained by the higher weekday CN concentration (weekdays: 7953 cm⁻³, Sundays: 6177 cm⁻³).

Also for AR weekday and Sunday averages were calculated (see Fig. 7). On Sundays AR was fairly constant while on weekdays a minimum during the morning hours was found coinciding with the morning maximum of CN concentrations. AR has highest values in the early morning hours and later drops to 0.09. A drop in AR with increasing CN concentrations is found also on single days and when CN concentrations are quite low, AR shows a local maximum. In Figs. 4 and 5, examples of such situations are indicated by dashed arrows.

For the CCN concentration no dependence on traffic patterns was found. Rush-hour traffic creates mainly small particles (below 100 nm; see Fig. 8 for the evolution of the size distribution during a typical weekday). Fresh traffic-generated particles are known to be hydrophobic (e.g. Smith and Chughtai, 1995) and had been found earlier to be inefficient CCN (e.g. Lammel and Novakov, 1995; Weingartner et al., 1997; Chughtai et al., 1996). Aged accumulation mode aerosol has much less diurnal variation and contains particles that can easily be activated at 0.5% S (e.g. Andreae and Rosenfeld, 2008; McFiggans et al., 2006). The highly variable CN concentrations in the range below 100 nm have little influence on the measured CCN concentrations.

Fig. 8.

Example of the evolution of an urban aerosol size distribution during a weekday (Dec. 19, 2007). Most of the variation takes place in the size range below 100 nm.

Contrary to the situation in Vienna, a diurnal variation of CCN concentrations was observed during weekdays in Mexico City (Wang et al., 2010), when CCN concentrations reached a maximum (>4000 cm⁻³) at around 10:30 a.m., while during the rest of the day concentrations were much lower (∼2000 cm⁻³). The maximum coincided with a substantial increase in nitrate and oxygenated organic aerosols indicating a conversion of hydrophobic primary organic aerosols and black carbon to hydrophilic particles by photochemical processes known to be dominant in Mexico City. The rise in the CCN concentration a few hours after the morning traffic rush hour is explained by the fast photochemical processes and the dilution in the rapidly rising boundary layer. Such conditions are found in Vienna only on summer days with very stable meteorological situation so it is not surprising that the diurnal pattern of CCN concentration in Vienna is different and does not show a peak after the morning traffic rush hour.

In Riverside (Cubison et al., 2008), however, CCN and AR patterns are more similar to Vienna. In this highly polluted area, a diurnal variation of AR was observed. At 0.5% S, AR reached a maximum (∼0.27) around noon, which is about twice the lowest value observed during the morning rush hour. CCN concentrations showed almost no diurnal variation and ranged quite constantly around 2000 cm⁻³. Although simultaneous size distribution measurements revealed the presence of large particles (>100 nm) which have a clear peak around noon they obviously did not activate as CCN.

3.4. CCN characteristics and meteorological conditions

Stable weather conditions are characterized by low wind speeds and no significant change in wind direction. During such periods the CN concentrations exhibited their typical source-dominated pattern while CCN concentrations remained rather constant and variations in AR were dominated by the variation in CN concentrations and therefore show some diurnal patterns. Under rapidly changing meteorological conditions (frontal passages), CCN and CN concentrations change with the change in air mass. During the whole measurement campaign, 51 frontal passages were identified (34 from the N–W sector, and 17 from the S–E sector).

Fig. 9 shows an example of the variability of CN and CCN concentrations and AR during changing weather conditions. At the beginning, meteorological conditions were stable with weak winds from the south-eastern sector. Starting in the early morning hours of Oct. 18, 2007, a front arrived from the northwest. Average wind speed changed from 2 m s⁻¹ to 7.5 m s⁻¹ and the wind direction from southeast to northwest. A rapid decrease in CCN and CN concentrations is clearly observed. The following days were characterized by westerly winds and rain and CCN concentrations stayed low. CN concentrations also decreased (by a factor of 5). CCN concentrations dropped by a factor of 10, and AR consequently by a factor of 2. The emission-dominated CN concentrations resumed their usual diurnal patterns shortly after the frontal passage. No significant differences were found for fronts arriving from the S–E sector.

Fig. 9.

a) (top panel) the variation of the CCN concentration and AR before, during and after the passage of a front. b) (bottom panel) same for the CN concentration.

Similar time series patterns were observed in an earlier study (Hitzenberger, 1985) for the scattering coefficient σ_sc in Vienna under conditions of frontal passages, when an increase of σ_sc occurred just before the front arrived, which was linked quantitatively to the decrease in mixing height associated with the low wind speeds just before the passage. Depending on the origin of the advected air mass, σ_sc returned to previous levels (air mass from the south to east sector) or decreased significantly (air mass from the north to west sector).

4. Summary and conclusions

From this large data base obtained in the Central European urban area of Vienna we conclude that variations in CCN parameters (concentration, AR, d_act,a) are quite large on timescales of days to weeks, so the typical two-week field campaigns might not yield representative pictures of CCN. No seasonal differences were found, though the meteorological situation (mixing height, frontal passages) of course influences CCN parameters. CCN concentrations at 0.5% S ranged from 160 cm⁻³ to 3600 cm⁻³ with a campaign average of 820 cm⁻³. The lower values compare quite well to those found in the European alpine background aerosol (Sonnblick: 29 cm⁻³–786 cm⁻³, Hitzenberger et al., 1999, 2000; Rax: 100 cm⁻³–2300 cm⁻³). AR were quite low (0.02–0.47, average: 0.13) and comparable to AR found in other polluted regions (e.g. Cubison et al., 2008: AR = 0.08 ± 0.03 at (0.27 ± 0.05)% S in Riverside, CA), and d_act,a were much larger (campaign average: 162 nm, range: 69 nm–370 nm) than activation diameters for single-salt particles (around 50 nm depending on the salt). Contrary to CN concentrations, which are influenced by source patterns, CCN concentrations do not exhibit distinct diurnal patterns. According to its definition, AR varies counter-currently to the CN concentrations. At least in the complex externally/internally mixed aerosol investigated in this study, predictions of CCN concentrations from information on the size distributions is not possible. Contrary to the hypothesis by Dusek et al. (2006a), accurate pictures of the CCN situation and CCN relevant parameters can only be obtained by direct measurements under these conditions.