Lead-210 observations within CARBOSOL: A diagnostic tool for assessing the spatiotemporal variability of related chemical aerosol species?

Authors


Abstract

[1] We report on observations of atmospheric 210Pb, coregistered with inorganic and organic aerosol species, during 2002–2004 at six European sites. This network reaches from the Azores to the Hungarian plain to represent marine, coastal, mountain and continental conditions. The motivation for observing this natural secondary aerosol tracer was to give insight to what extent it might assist in understanding the more complex aerosol chemistry changes. Synopsis of the 210Pb variability revealed a continental increase, up to a factor of three, from west to east. During the three winter months, we find a variation on nearly the same order in the 210Pb concentration between low- and high-altitude sites. Seasonal 210Pb cycles exhibit summer/winter ratios of around 2–3 at high-altitude sites, but remain damped at low-altitude stations. However, all sites show distinct 210Pb changes of around ±50% independent of season on the synoptic timescale. Comparison of concentration variations of organic carbon (OC) and anthropogenic sulphate with the 210Pb variations show largest differences associated with the seasonal cycle at the low-altitude sites. In contrast, significant covariations of all three components are seen on the synoptic timescale for these sites. At high altitudes, clear covariations of OC and anthropogenic sulphate with 210Pb are seen on both seasonal and synoptic timescales. At two mountain sites with comparable elevation, all three aerosol compounds show strong intersite correlations along with systematic enhancements at the downwind site. Attributing these offsets to a common continental pileup, simple 1-box model calculations yielded OC- and anthropogenic sulphate-related emission flux densities, which are broadly in agreement with the expected values.

1. Introduction

[2] Because of a short residence time of only a few days the aerosol concentration in the lower continental troposphere is highly variable, both in time and space. This makes the observed spatiotemporal variability of chemical aerosol components difficult to interpret and complicates attempts to disentangle changes driven by air mass transport processes from those tied to variations of the various involved sources and sinks strengths.

[3] Observation of the terrigenous radioisotope pair 210Pb and 222Rn may help to reduce this deficit including attempts to validate aerosol-related Chemical Transport Models (CTMs), in which especially the simulation of convective transport, aerosol emission and removal remain a critical issue [Considine et al., 2005; Liu et al., 2001; Rehfeld and Heimann, 1995]. 210Pb constitutes the decay product of 222Rn with the longest radioactive half-life of 22.3 years, rendering its radioactive decay rate negligible compared to its atmospheric deposition. The 210Pb precursor gas 222Rn is almost entirely produced from Radium ubiquitously contained in soils. It is emanating from the soils at a relatively uniform and constant rate into the atmospheric boundary layer [Dörr and Münnich, 1990; Turekian et al., 1977]. The controlling atmospheric sink of the noble gas 222Rn is fixed by its radioactive decay with a mean life time of 5.5 d. This life time corresponds to a transfer rate of 18% per day producing 210Pb atoms which become immediately attached to submicron aerosol particles [Whittlestone, 1990; Sanak et al., 1981].

[4] Therefore 210Pb may be regarded as a proxy for a secondary submicron aerosol body, produced by a rather simple continental precursor gas system. This system stands out by a well-known, almost constant and uniformly distributed emission rate as well as by a fixed gas-to-particle transfer rate. In this respect, it differs greatly from the relevant but much less well constrained secondary aerosol systems like sulphate or organic species (i.e., secondary organic aerosol (SOA)). Various continental sources of chemical aerosol species generally show a more patchy spatial distribution with emission rates frequently changing with season. In addition, the controlling gas-to-particle transfer rate can vary substantially both in time and space (e.g., with season, altitude and atmospheric pollution levels).

[5] The 210Pb-related applications in atmospheric research mainly comprise validations of global CTMs by globally distributed observations [Hongyu et al., 2001; Considine et al., 2005; Liu et al., 2001; Rehfeld and Heimann, 1995] and studies of long-range transport of continental chemical species by airborne sampling over very remote marine areas [Dibb et al., 1996, 1997; Virkkula et al., 2006]. Furthermore there is the assessment of mean life time of submicron aerosol via the ratio of different 222Rn decay products [Nancy et al., 2000] and the dissection of continental and stratospheric air masses, via the 7Be/210Pb ratio related to ozone source apportionment studies [Graustein and Turekian, 1996]. However, so far less effort has been put into dedicated 210Pb applications in the context of ground-based aerosol sampling networks, where the spatiotemporal variability of the various chemical aerosol components is regularly monitored (e.g., EMEP and various national networks).

[6] Here, we report on 210Pb observations performed in the frame of the EC-CARBOSOL project, whose main goal was constraining the role of aerosol with respect to its past and present radiative forcing in western Europe (see CARBOSOL concept outlined by Legrand and Puxbaum [2007]). Attempting to quantify the present-day situation, a dedicated aerosol observation network was established for two years in remote and rural regions of western Europe. Analyses in this network included major inorganic and organic chemical species as well as source-related tracers [see Pio et al., 2007; T. S. Oliveira et al., unpublished manuscript, 2007]. The primary objective of this field study, which was supported by regional CTMs and emission models [Simpson et al., 2007; Marmer and Langmann, 2007], was the characterization of the relevant organic aerosol components, with respect to their spatial and seasonal variability and their related source apportionment [Gelencsér et al., 2007].

[7] In this context, the 210Pb analysis coregistered with the whole suite of chemical aerosol species was expected to reveal to which extent the observed variability of chemical aerosol components may be attributed to air mass transport versus source-related phenomena. A further motivation was the validation of the regional CTM deployed in CARBOSOL with 210Pb data and concurrent 222Rn monitoring (see section 2.2).

[8] In this paper we present the 210Pb analysis focusing on the spatial (including altitudinal) and seasonal patterns and its possible relationship to the pattern of other chemical aerosol components like organic carbon (OC), for example. We assess the basic method of using 210Pb for backing up the interpretation of chemical aerosol data on regional scales from sites largely differing in their geographical characteristics.

2. Methods

2.1. Sampling Sites

[9] The network of the CARBOSOL aerosol sampling stations is described in detail by Pio et al. [2007]. As illustrated in Figure 1, the five western European sites are arranged on a line roughly following a southwest to northeast direction, corresponding to one of the major routes of the zonal water vapor advection.

Figure 1.

Geographical distribution of CARBOSOL aerosol sampling sites: AZO, Azores Island, Atlantic Ocean; AVE, Aveiro, Portugal; PDD, Puy de Dome, France; SIL, Schauinsland, Germany; SBO, Sonnblick Observatory, Austria; KPZ, K-puszta, Hungary.

[10] Table 1 lists the geographical position, altitude and climatological classification of the sites. The network is characterized by a systematic increase of the continental fetch of about 2300 km (from the Portuguese coast to the Hungarian plain) as well as by a large altitude range of more than 3000 m.

Table 1. Overview of the CARBOSOL Aerosol Sampling Network
StationPositionAltitude, m aslSampling PeriodSite Classification
Azores (AZO), Atlantic38°41′N, 27°21′W50Jul 2002 to Jun 2004maritime background
Aveiro (AVE), Portugal40°34′N, 8°38′W47Jul 2002 to Jun 2004costal
Puy de Dome (PDD), France45°46′N, 2°57′E1450Oct 2002 to Sep 2004low mountain range
Schauinsland (SIL), Germany47°55′N, 07°54′E1205Oct 2002 to Dec 2004low mountain range
Sonnblick (SBO), Austria47°03′N, 12°57′E3106Oct 2002 to Sep 2004high alpine
K-puszta (KPZ), Hungary46°58′N, 19°35′E126Jul 2002 to May 2004continental low land

[11] A short classification of the sites with respect to their typical climatology and their relevant aerosol sources is given in the following:

[12] 1. The marine background station Azores (AZO) (Terceira Island, 400 km2), situated about 1600 km west of the Portuguese coast, is expected to sample maritime air masses of the westerlies with additional episodic continental influences.

[13] 2. The coastal station Aveiro (AVE), lies about 10 km inland on almost flat terrain. At this site the air masses originate from different regimes, on some days pure maritime and on others pure continental air masses, but including mixed conditions as well, driven by the offshore entrainment of continental air into the sea breeze.

[14] 3. The clean air, medium-elevation mountain stations, which are both significantly influenced by westerly air streams include (1) Puy de Dôme (PDD), at an isolated mountain top in central France and (2) Schauinsland (SIL), situated on an mountain ridge of the Black Forest in southwestern Germany.

[15] 4. The continental background station is represented by the high-altitude Alpine Sonnblick (SBO) Observatory. As it is situated at 3106 m asl. on an isolated peak of the glaciered East Alpine ridge, the site is expected to sample mostly free tropospheric air. This site is the least influenced by ground-level aerosol (and precursor gas) sources during winter.

[16] 5. The continental ground-level station K-puszta (KPZ) is situated in the central Hungarian plain in a flat rural area, where a more continental weather type prevails.

2.2. Aerosol Collection

[17] During the CARBOSOL campaign, continuous high-volume aerosol sampling was performed at all sites on a weekly base in the years 2002 to 2004, though at each station at slightly different periods (see Table 1). A further site-specific difference lies in the aerosol size cutoff of the sampler inlets. These were fixed at around 2.5 μm at AZO, AVE, SBO and KPZ, but the SIL and PDD samplers were equipped with a heated sampling head (protecting against rain and snow) which did not provide a well-defined size cutoff. A rough estimate based on the inlet geometry and airflow rate gave an upper particle size limit of ∼10 μm, collected by the two identical sampling systems at SIL and PDD. In the case of 210Pb, which is mainly attached to the submicron accumulation mode [Winkler et al., 1998], no significant influence is expected from the site-specific sampler cutoff.

[18] In contrast to some of the chemical aerosol species regularly analyzed in the filters, no sampling artifacts are expected for 210Pb (for an overview on the entire analysis scheme see Pio et al. [2007]). It should be noted that at the AVE, SBO and KPZ sites additional continuous observations of the atmospheric 222Rn activity were performed (using the “Heidelberg” 222Rn monitoring described by Levin et al. [2002]), while at SIL such records are commonly obtained by the German Federal Office for Radiation Protection (BfS) and at PDD by the French Laboratoire des Sciences du Climat et de l'Environnement (LSCE). The 222Rn results will be reported elsewhere.

2.3. Analysis

[19] The filter activities for 210Pb (along with 7Be) were assayed at the low-level laboratory of the Institut für Umweltphysik (University Heidelberg) by nondestructive γ-spectrometry using solid state Ge-detectors [Wagenbach et al., 1988]. To minimize the effect of a nonuniform areal aerosol density distribution, half of the quartz-filter samples were used. Loaded filters were wrapped in aluminum foil to avoid contamination with carbon species and then folded to achieve a well-defined counting geometry (but not compressed to ensure filter integrity for subsequent chemical analyses). Calibration of the 210Pb counting efficiency traceable to German Office of Standards (PTB) was associated with a systematic uncertainty of ∼4%. The statistical counting errors (corresponding to typical counting times of one to two days and air volumes on the order of 5000 m3) are around 5% for the low-activity samples (i.e., mainly from the AZO and SBO sites) and 3% for ones with higher activity from the KPZ, PDD, SIL and AVE sites. Thus the overall uncertainty of the 210Pb results, including the maximal uncertainty of the sampled air volume was estimated to range in general between 6% and 12% [Hammer, 2003].

[20] No correction on the 210Pb filter-collection efficiency was applied (which was experimentally found to be larger than 98% for the 210Pb-carrying aerosol); although visual filter inspection revealed small holes (diameter 1–2 mm), arising from the impact of ice crystals on winter samples at the PDD, SIL and SBO sites. The analyses procedures of the various chemical aerosol species, including sulphate and OC reported in this paper, are detailed by Pio et al. [2007]. For sulphate 1 to 3 cm2 of the filter was extracted and analyzed by ion chromatography with a Dionex 600 system. The analysis for OC was based on a thermo-optical technique, using temperature criteria to distinguish the different carbon fractions, along with an optical method to correct for pyrolysis artifacts.

3. Data Overview

[21] The 210Pb data were taken without any significant gaps at all sites, except for KPZ, where in the winter of 2002/2003 no samples were taken. The common sampling period ranges from October 2002 to May 2004, however, no strict coherence of the weekly sampling intervals could be achieved, leading to intersite time offsets of up to three days.

[22] Note, that aeolian soil dust, which provides the only primary aerosol source of 210Pb, does not significantly add to the observed values at any time. Assuming an upper mineral dust level of around 0.4 μg/m3 derived from Ca2+ data at KPZ and a maximum 210Pb content in uncultivated topmost soil of 150–220 mBq 210Pb per gram [Matisoff et al., 2002] (including the contributions from accumulation of deposited 210Pb), mineral dust aerosol of the filter samples can make at most 7% to 10% of the 210Pb load derived from airborne 222Rn.

[23] The general feature of the weekly 210Pb data set is presented in Table 2 and illustrated in Figure 2, which highlights its major spatial and temporal variability (with upper and lower percentiles serving as surrogate for the upper and lower concentration envelopes, respectively).

Figure 2.

Annual 210Pb means (squares) and associated core summer (i.e., June to August) means (circles) and core winter (i.e., December to February) means (stars). The sites are arranged along their geographical longitude, with the second scale indicating the altitudes asl.

Table 2. Descriptive Statistics of Weekly 210Pb Data of the CARBOSOL Network
SiteAnnual Mean, mBq/m3Weekly Variability,a %90% Percentile, mBq/m310% Percentile, mBq/m3Core Winter Mean (Dec–Feb),b mBq/m3Core Summer Mean (Jun–Aug),b mBq/m3
  • a

    Relative 1σ variability of single values deviations from mean (seasonal) cycle.

  • b

    1σ variability.

AZO0.26570.500.110.15 ± 0.060.31 ± 0.08
AVE0.53580.970.210.50 ± 0.260.53 ± 0.21
PDD0.36490.680.090.18 ± 0.100.52 ± 0.12
SIL0.54450.930.170.36 ± 0.110.70 ± 0.26
SBO0.41450.650.140.22 ± 0.080.57 ± 0.27
KPZ0.80401.170.430.84 ± 0.310.79 ± 0.22

[24] The following main findings manifest the strongly varying imprint of secondary continental aerosol on the CARBOSOL sites:

[25] As was to be expected, a systematic increase in the mean annual 210Pb level is observed from mostly maritime to more continental sites. At the lowest levels of the 210Pb concentration, marine background values observed on the Azores are typically on the order of 0.11 mBq/m3 (see 10% percentiles in Table 2), that, however, is still a factor of 2 to 3 higher than the mean levels of the remote South Pacific (0.06 mBq/m3 [Larsen et al., 1995]) or coastal Antarctica (0.03 mBq/m3 [Wagenbach et al., 1988]). Therefore a significant influence of long-range transported continental aerosol from North America, North Africa or Europe may be expected at the Azores. At the highest levels of the 210Pb concentration, the Hungarian low-altitude site K-puszta exhibits values of ∼0.8 mBq/m3, pointing toward a pileup factor of around 3 compared to the maritime air masses of the westerlies. However, this continental pileup can be nearly counteracted by the concentration decrease with altitude observed at the high-altitude sites, particularly during winter. This is responsible for the fact that continental background 210Pb level at the high Alpine Sonnblick site (∼0.14 mBq/m3) does not differ much from that seen in the mid-North Atlantic region.

[26] Restricting the evaluation to core summer (i.e., June to August) and core winter means (i.e., December to February) to closely reflect the insolation cycle, a prominent seasonal 210Pb contrast with summer/winter ratios of 2 to 3 is observed at all high-altitude sites. This difference is basically driven by the seasonality of the vertical mixing intensity. In contrary, only a subdued or slightly reversed summer/winter contrast is seen at the low-altitude sites.

4. Discussion

[27] Regarding the spatiotemporal 210Pb variability observed within the CARBOSOL network, the basic question arises if the coregistered change of aerosol species may be already explained by this variability (i.e., by atmospheric transport processes rather than by their spatiotemporal source and sinks properties). For this reason we broadly base our discussion of the 210Pb findings and their possible relationship to other aerosol components on a simple one-box model. The formal details dealing with the principal constraints of spatiotemporal 210Pb changes are outlined in Appendix A of this paper. For the sake of simplicity, we address among the various analyzed organic species only the bulk of organic carbon compounds. This analytically defined quantity includes the relatively insignificant primary cellulose fraction [Sánchez-Ochoa et al., 2007], but no elemental carbon (EC) [Pio et al., 2007]. We refer additionally to secondary inorganic aerosol via non-sea-salt, nondust, sulphate (hereafter denoted as anthropogenic sulphate), since the sources and sinks of this major aerosol component are thought to be relatively well constrained.

4.1. Synoptic Changes

[28] As illustrated in Figure 3, the 210Pb records from the coastal Aveiro and from the mountainous Schauinsland sites show significant weekly to biweekly 210Pb variations superimposed on the seasonal cycle (if present). The general temporal patterns shown in Figure 3 are calculated using a Fast Fourier Transformation (FFT) filter with a cutoff frequency of 1/15 weeks and a parabolic transfer function (this method was used throughout the paper). Relative deviations of around ±50% from these patterns appear to be common to all sites (see Table 2), though slightly decreasing with increasing continentality. These variations in the 210Pb concentration are expected to reflect changing circulation patterns with alternating influence of maritime and (aged) continental air masses, which can partially come along with variations of the mixing height and precipitation probability at inland sites.

Figure 3.

Weekly 210Pb variability at the coastal site Aveiro and the low mountain range site Schauinsland. The general (seasonal) pattern is highlighted by the FFT filtered dashed line.

4.1.1. Marine Island Situation

[29] In the case of the Azores, the air mass residence time over the island is much shorter than the 222Rn life time. That is why no substantial local contribution to the observed 210Pb changes is expected here (to be seen in Figure A1 in Appendix A). Same holds true for secondary aerosol species, which are produced by similar gas-to-particle transfer rates. However, episodic advection of continental aerosol plumes; diluted en route by dispersion (mixing with marine background air) or partly depleted by rain scavenging and dry deposition; should be mirrored by the short-term 210Pb variability. Especially continental submicron aerosol may be traced by the changes in the 210Pb concentration.

[30] Inspection of the residual concentrations of 210Pb, OC and anthropogenic sulphate, derived by subtracting the FFT-filtered general long-term trend, revealed that at the Azores, 210Pb catches the synoptic changes of anthropogenic sulphate surprisingly well (correlation coefficient r2 = 0.4), whereas the covariance to OC is found to be relatively weak (r2 = 0.16) and to non-sea-salt Ca (i.e., mineral dust) even absent. However, the large uncertainty in the calculation of non-sea-salt Ca (even leading to a negative value for ∼40% of all values) may already explain the missing relationship to the continental 210Pb tracer. No straightforward explanation for the weak correlation can be given in the case of OC, though again large analytical uncertainties and/or a substantial marine OC contribution might be responsible. For this low-altitude island site we conclude that 210Pb and anthropogenic sulphate are both clearly above true marine background levels, at 0.26 mBq/m3 and 770 ng/m3, respectively. Both show significant covariation on the synoptic timescale, which points to a substantial influence of long-range transported continental aerosol plumes. The covariance between 210Pb and anthropogenic sulphate and the 210Pb to sulphate ratio, which is quite similar to that seen at high-elevation continental CARBOSOL sites (as most representative for the continental outflow), might questions the significance of SO2 ship emissions proposed by Pio et al. [2007] and addressed by Corbett and Köhler [2003], and Derwent et al. [2005], at least at the Azores site.

4.1.2. Low-Altitude Sites

[31] At the two low-altitude sites on the continent (Aveiro and K-puszta), distinct synoptic changes of 210Pb, anthropogenic sulphate and OC are found, which may be due to circulation patterns governing the air mass origin and boundary layer stability. However, the changes in the boundary layer stability are expected to be less significant here, since 210Pb and secondary aerosol species of comparable precursor life times cannot be enhanced much during reduced vertical mixing episodes persisting over only a few days (outlined in Appendix A). While it would be a matter of roughly one day to replace the boundary layer aerosol by a depleted maritime or an enriched continental air mass, it would take a week or more to reestablish a typical secondary aerosol level solely by local ground-level precursor emissions (see equation (A4) in Appendix A).

[32] The residual concentrations displayed in Figure 4b for the costal site Aveiro suggest a simultaneous variation of the three species. The residuals are calculated subtracting the FFT-derived long-term changes (i.e., smoothed seasonal cycle) and are subsequently normalized to the overall mean of the displayed data. Most likely this covariant behavior is driven by the alternating influences of inland and marine air masses (sea-breeze influence), although strong seasonal contrasts exist especially for OC (see Figure 4a), but not for 210Pb (see section 4.2.1). The same picture is seen for K-Puszta. Correlations of the seasonal cycle adjusted residuals of OC to those of 210Pb gave r2 values of 0.42 at Aveiro and 0.49 at the continental K-puszta site (with 210Pb explaining very similar variance fractions to those of OC for anthropogenic sulphate at both sites). Thus a major part of the short-term anthropogenic sulphate and OC variability at these sites appears to be independent of emission and atmospheric chemistry changes, but driven by air mass transport processes.

Figure 4.

The 210Pb, anthropogenic sulphate and OC time series at the costal site Aveiro. (a) Raw data and (b) residuals with respect to the general (seasonal) FFT smoothed changes, normalized to the overall means of the displayed data.

4.1.3. High-Altitude Sites

[33] At the high-altitude CARBOSOL stations, synoptic circulation patterns result not only in air mass replacements, but may also determine whether specifically the low mountain range sites are transiently lying within or outside the seasonally varying mixing height. Contrasting the short-term variability of 210Pb and OC, at SIL and PDD rather coherent intersite and intrasite concentration variability is observed. Following the example displayed in Figure 5, we note that this observation seems to follow the regional-scale circulation patterns, which are frequently common to both sites. Therefore we conclude that the almost uniform short-term 210Pb and OC changes seen at SIL and PDD with respect to timing and extent, would imply a spatial OC source distribution comparable to the definitely homogenous one of 210Pb. This suggests that a substantial contribution of the relatively uniform distributed sources is related to biosphere OC It should be noted that at all mountain sites (including SBO) the coherence of anthropogenic sulphate with 210Pb residuals is not as clear as the one found for OC. This might be due to a less uniform spatial source distribution of SO2 compared to OC.

Figure 5.

Example of synoptic intersite and intrasite variations of 210Pb and OC at Puy de Dome (PDD) and Schauinsland (SIL) in 2003. Data refer to the deviation from the FFT smoothed seasonal cycles and are normalized to the overall mean of the displayed period. Vertical marks in the bottom plot denote consecutive periods of different European weather situations, according to German Weather Service (DWD, 2005, http://www.dwd.de/de/wir/Geschaeftsfelder/Medien/Leistungen/GWL/index.htm) (see below). Shaded areas refer to selected weekly samples, which cover almost single meteorological periods characterized by a high-pressure system located over western Europe supporting mainly bright weather conditions (indicated by a), a northerly cyclonal pattern over central Europe leading to showers and gale force winds (indicated by b), a westerly cyclonal pattern with daily rainfalls (indicated by c), and anticyclonal conditions with eventually raising temperature displaying the forerunner of the 2003 heat wave (indicated by d).

4.2. Seasonal Changes

[34] The vertical mixing intensity is tied closely to the seasonal insolation changes. It is expected to control the annual cycle of the overall pollutants buildup within the planetary boundary layer, as well as their handover into the free troposphere. Therefore continental aerosol concentrations like 210Pb are generally expected to be relatively enhanced at low-altitude sites and depleted at mountain sites during winter, with the vice versa feature showing up during summer. That is why, at continental sites, the seasonal cycle of 210Pb will be mainly driven by this transport process and may substantially differ from that of those chemical aerosol species, that are controlled by seasonally varying source or sink properties. In this context we investigated the general annual changes of 210Pb by contrasting it to the correspondent OC cycles. Figure 6 illustrates the FFT-derived general concentration patterns normalized to zero means for 210Pb and OC at all sites. For quantitative comparison of the seasonal cycles we refer to the core winter and core summer means given in Table 2.

Figure 6.

FFT-derived seasonal cycles of 210Pb (solid line) and OC (dashed line) normalized to the overall mean for each site and scaled to zero mean: (left) low-altitude and (right) high-altitude CARBOSOL sites.

4.2.1. Low-Altitude Sites

[35] Although local emissions and the seasonal development of surface inversions are not expected to be very relevant at the Azores site, we observed an annual 210Pb cycle with a mean relative amplitude (scaled to overall mean) of 32%. Its extent is quite comparable to the OC and anthropogenic sulphate amplitudes (not shown), but no coherent phasing exists (especially obvious for OC). The reason for the summer maximum of 210Pb, and the roughly similar late summer maxima of continental OC and anthropogenic sulphate at the Azores is still unknown. Various factors might be responsible for this observation: (1) seasonal changes inherent to circulation patterns, (2) deep convective mixing over the continents, and (3) subsidence of long-range transported aerosol from the free troposphere into the marine boundary layer. At the coastal Aveiro site, no regular seasonal 210Pb cycle could be found (though a cross check in the 222Rn record shows a clear peak around late fall, since 222Rn is more sensitive to changes in the vertical mixing intensity). The 210Pb signal is almost unrelated to the strong, seasonal changes of OC and anthropogenic sulphate, which themselves seem to be in antiphase with each other. This confirms that at Aveiro the inversion episodes during the winter half-year do not persist long enough to allow for a substantial 210Pb pileup (see equation (A4) of Appendix A). Also, dilution of boundary layer air by strong vertical mixing in the summer half-year appears not to be important. The episodic dilution occurring during advection episodes of clean maritime air masses seems to be responsible for the absence of a regular seasonal 210Pb pattern. We note that the rather strong winter-summer contrasts of OC and of opposite-signed anthropogenic sulphate observed near the Portuguese coast are without any 210Pb pendant. They seem not to be much influenced by the seasonal-varying vertical mixing, but instead mainly driven by changes of their local source (and maybe sink) strength.

[36] At the continental low-altitude site KPZ we observed an unexpectedly weak seasonal 210Pb cycle, however, different to Aveiro, with regular winter maxima roughly in phase with OC and sulphate. The higher frequency of the persisting surface inversion during the winter seasons (in addition to the generally lower mixing height) should lead at this site to a more pronounced relative 210Pb enhancement than the ∼5% observed at KPZ. Indeed, at roughly comparable continental low-altitude sites like, e.g., Braunschweig and Munich (Germany) or Belgrade (Serbia), 210Pb winter maxima exceed the summer level by 20% to 30% [Winkler, 1997; Wershofen and Arnold, 2004; Todorovic et al., 2000], whereas from the record of Merešová et al. [2004] at Bratislavia (Slovakia), a winter enhancement of roughly a factor of 2 can be deduced. The substantial leaks of data from KPZ sampling could be the reason for not finding a clear seasonal cycle. The 222Rn record of K-Puszta clearly shows the expected winter inversion episodes, but the 222Rn record has an inadequate temporal coverage. Nevertheless, we may state that the relative seasonal 210Pb amplitude at KPZ reaches at most one third of the seasonal anthropogenic sulphate cycle and only ∼10% of the OC cycle. Consequently we expect that the observed seasonal changes of these chemical aerosol species were mostly related to the seasonal cycle of their emissions and associated gas-to-particle transfer rates.

4.2.2. High-Altitude Sites

[37] As expected we observe prominent seasonal 210Pb cycles at all three mountain sites. Dominated by the regional-scale convection change, these cycles stand out at all sites by summer maxima and winter minima and match the OC and anthropogenic sulphate cycles quite well. Referring to core winter and summer periods, the relative amplitude of the 210Pb seasonality systematically increases with station altitude. This reflects the extent of the decoupling from the planetary boundary layer. Comparing the 210Pb amplitudes to the respective relative changes of sulphate and OC, we obtain a rather narrow range in which the seasonality of the OC is explained by the 210Pb cycle: i.e., 60%, 57% and 52% for PDD, SIL and SBO, respectively. For anthropogenic sulphate, the ratios are 83%, 62% and 64%. These estimates suggest that roughly 60% of the seasonal OC and anthropogenic sulphate variations seen at the high-elevation sites may be attributed to vertical mixing. At least for OC, the remaining fraction may be partly associated with the higher (regional-scale) biogenic emissions and oxidation efficiencies during summer. At the high Alpine Sonnblick site, where no local aerosol sources exist, we may conclude vice versa that the obvious seasonal cycle of (partly secondary) aerosol species, commonly observed in Alpine ice cores [Preunkert et al., 2001] is mainly, but not entirely, driven by the winter/summer contrast of the regional-scale vertical mixing.

4.3. Continental Pileup Evidence

[38] Although a clear overall 210Pb pileup shows up at the CARBOSOL sites, its general quantification is not feasible because of the different station altitudes. However, the PDD and SIL high-altitude sites are at fairly comparable elevations. Since they are frequently sharing westerly air mass trajectories (EMEP trajectory database available at http://www.emep.int/Traj_data/traj2D.html), this may offer a possibility for a detailed investigation of the continental pileup of 210Pb and related aerosol components. Stimulated by the rather concurrent synoptic intersite changes of 210Pb, anthropogenic sulphate and OC (see Figure 5), we investigated if the concentrations of these species are systematically enhanced at SIL against the ones at PDD due to a common pileup effect.

[39] Referring to equation (A3) of Appendix A, giving the 210Pb increase versus the continental air mass residence time, we get for not too large Δ210Pb (i.e., relatively small continental fetch differences) the simple relationship

equation image

[40] Here, to and Δt denote the air mass residences time over the catchment area at the upwind site (corresponding to the continental fetch) and over the intersite area, respectively, with the 222Rn emission rate PRn and the mixing height H kept constant.

[41] In the intersite regression of 210Pb and OC displayed in Figure 7, a relevant intercept shows up for both species (which also holds true for sulphate). It is worth mentioning that the so-called geometric mean regression is not solely driven by the obvious seasonal cycle, since similar positive intercepts (however, being undistinguishable because of their relatively large, formal errors) are obtained for regressions calculated separately for to the winter and summer half-years.

Figure 7.

Puy de Dome/Schauinsland intrasite variability for 210Pb and OC, respectively, based on Geometric Mean Regressions analysis.

[42] Provided that any secondary aerosol system (e.g., sulphate) is characterized by a spatial- temporal precursor source distribution roughly similar to that of 210Pb (i.e., homogenous properties), equation (1) may be applied here as well. This allows to express the observed intersite difference ΔC by the known 222Rn emission rate, scaled to the life time toward particle formation:

equation image

[43] Note that only in the simplified approach given by equation (1), the mixing height H and the continental air mass residence times to and Δt may be canceled, as they are estimated to be the same for both species.

[44] We find a regression-based mean annual sulphate pileup between SIL and PDD of (520 ± 220) ng SO4/m3, along with a corresponding 210Pb enhancement of (0.1 ± 0.04) mBq/m3. This translates into a SOx emission rate of (17 ± 10) ng S/m2 s, taking 1 atom/(cm2 s) for the 222Rn exhalation rate and crudely assuming λRn = λSOx.

[45] Somewhat fortuitously our inferred SOx emission rate appears to be virtually the same as the 14 ng S/m2 s reported by EMEP [Vestreng et al., 2005] for France during 2003 and 2004. On the one hand, the 210Pb approach should systematically underestimate the SO2 emission, since SO2 is partially lost by dry deposition (unlike 222Rn) and since the typical overall SO2 sulphate transfer rate is probably higher than the 18% per day of 222Rn [Benkovitz et al., 2004].

[46] Extending the 210Pb pileup calculation to OC appears to be even more critical, because only a unknown fraction of OC may be addressed as secondary produced, including a tremendous range of transfer rates involved in the secondary organic aerosol production [Millet et al., 2004; Kanakidou et al., 2004]. Nevertheless, according to equation (2) and assuming λRn = λOC, the observed OC pileup between PDD and SIL of (0.3 ± 0.2) μgC/m3 would correspond to a total organic aerosol emission rate of (30 ± 20) ngC/m2/s, which is on the same order as various model-based compilations of OC-related emission rates. According to estimated emissions of monoterpenes and total BioVOCs for France and Italy by Lenz [2001] and Simpson et al. [1999], we infer 20 ngC/m2/s and 10 ngC/m2/s, respectively, for these precursor species. If referring to T. C. Bond (cited by Kanakidou et al. [2004]) and the OSCE-Europe surface area, the figure is virtually the same with mean flux densities for primary organic aerosol, terpenes and other-reactive-VOC of 10 ngC/m2/s, 20 ngC/m2/s, and 30 ngC/m2/s, respectively. On the one hand, the OC offsets obtained from the SIL-PDD regression are uncertain regarding the limited number of samples (though still statistically significant). On the other, calculating the OC and sulphate flux densities simply via the SIL-PDD differences of the annual mean concentrations does not change the results much (i.e., 17 ngS/m2/s versus 14 ngS/m2/s and 51 ngC/m2/s versus 30 ngC/m2/s, with the latter numbers obtained by intersite regressions), indicating that the estimated values are numerically robust.

[47] The 210Pb-based flux densities are scaled to the gas-particle transfer rate ratio λRn/λC, which means we would get accordingly lower values for sulphate and OC if their transfer rates were assumed to be larger than the fixed 18% per day of 222Rn.

[48] Note that conversely to 210Pb, OC and anthropogenic sulphate concentrations, no significant correlation between SIL and PDD shows up for elemental carbon (EC, not shown) and hence no pileup related EC offset is obtained. This observation suggests that the combustion sources contributing to the observed variability of the purely primary EC-aerosol are more local or at least more patchily distributed than those related to 210Pb, OC and anthropogenic sulphate. Accordingly, the relatively small 20% enhancement in the mean EC concentration level between SIL and PDD may not be traced back to a continental pileup effect, which is underlined as well by the marginal EC intersite correlation of only r2 = 0.09.

[49] We conclude that the well-constrained 210Pb continental pileup at two comparably clean air sites appears to be reflected by OC as well as by anthropogenic sulphate, eventually implying that the regional source distribution of these species is comparably uniform like that of 210Pb.

5. Final Remarks

[50] The 210Pb observations within the CARBOSOL sampling network proved to be a relatively simple approach in assessing to which extent the spatiotemporal variability of the simultaneously recorded secondary chemical aerosol species may be attributed to transport-related variations of 210Pb. In addition to changes in the source or sink strengths, a large part of, e.g., the organic aerosol fraction was found to be governed by the continental imprint of the sampled air mass, generally implying an OC source distribution comparably uniform like that associated with 210Pb. However, because of the secondary production mode, low-altitude observations of 210Pb are found to be relatively insensitive to short-term variations of the boundary layer height.

[51] Spatiotemporal 210Pb observations may provide a productive test field for CTM simulations of secondary aerosol species [Simpson et al., 2007]. Those simulations would have to capture the mostly transport-driven 210Pb variability. First attempts to model 210Pb at the CARBOSOL mountain sites with the regional-scale MOCAGE model showed reasonable agreement with observations [Dombrowski-Etchevers et al., 2005].

[52] Particularly at clean air sites outside of the intrinsic aerosol sources region, 210Pb records may be a simple tool to assist in the selection of concurrently observed aerosol data for dedicated evaluations or for the subsample collection for intricate, expensive or filter consuming analyses. Without this tracer, such tasks would be complex if, e.g., based on back trajectory analysis (at least for weekly samples). We used this approach successfully in determining the fossil fraction of the CARBOSOL filters by 14C analyses [Gelencsér et al., 2007; B. May et al., manuscript in preparation, 2007], where samples were selected according to comparable 210Pb activity as to obtain composite filter batches distinguished by their relative imprint of continental sources.

[53] Even in nondestructive analysis, the 210Pb detection limit can be easily improved by approximately a factor of 6 through deploying a low-level well-detector system. This advancement would allow a time resolution of days for atmospheric 210Pb records and/or to analyze low volume sample aliquots available from various aerosol filter archives. Therefore meteorology-driven interannual effects in the relation of secondary aerosol species to 210Pb (like the European 2003 heat wave, not significantly resolvable in the two year CARBOSOL observations) or decadal trends may be approached. As an example, inspection of long-term 210Pb records revealed the 2003 summer half-year to stand out at Schauinsland mountain by a broad maximum (BfS monitoring), but show a relatively persistent depletion at the Braunschweig low-elevation site [Wershofen and Arnold, 2004].

Appendix A

[54] In the most simple conceptual model, 222Rn is emanating at the fixed flux rate P into a well mixed ground-level layer of constant mixing height H to produce 210Pb at a rate λ (i.e., the 222Rn decay constant). 210Pb is steadily removed at the constant rate 1/τ, with τ denoting the residence time of the 210Pb carrying aerosol.

[55] Taking the initial 222Rn(t) and 210Pb(t) box concentrations as zero the temporal 222Rn concentration increase would be

equation image

[56] Coupling 222Rn(t) with 210Pb(t) via

equation image

leads to the temporal 210Pb increase

equation image

which may be reduced for the special, but realistic case τ = 1/λ to

equation image

[57] A structurally identical picture is obtained for the continental 210Pb pileup. Following a Lagrangian approach, an isolated air mass with initial zero concentration is drifting at constant velocity inland, where it is steadily uploading 222Rn, while becoming well mixed within the planetary boundary layer. Changing the meaning of the variable t in equations (1) to (4) into a continental residence time tres (i.e., replacing t by l/v, with l denoting the trajectory length over land and v the associated drift velocity) gives the most simple formulation of the continental pileup effect.

[58] As displayed in Figure A1, the initial increase of 210Pb remains relatively flat, thereby lagging behind the linear increase of 222Rn. In the case of equation (4), the maximum 210Pb increase rate is reached after 5.5 d at t = 1/λ and accounts only to 1/e of the maximum 222Rn increase rate occurring at t = 0.

Figure A1.

Comparison of 222Rn and 210Pb increase within a stagnant boundary layer of initial zero concentrations, according to equations (A2) and (A4), respectively. Ordinates are given relative to the respective steady state concentrations. The vertical dashed line marks identical increase rate at t = 1/λ. A similar relationship can be applied for the concentration decrease of an isolated air mass transported over 222Rn-source-free areas with initially given concentrations.

[59] Hence it becomes obvious that reduced vertical mixing persisting over some days only (i.e., transient inversion layers) would be less clearly seen in the 210Pb enhancement compared to its progeny. This is equivalent to pileup effects corresponding to short continental fetches. Note, however, that the first-order temporal depletion of a continental 210Pb-222Rn plume drifting over the almost source-free ocean would show the opposite behavior: i.e., an initial linear 222Rn decrease, accompanied by a delayed 210Pb decrease. Therefore 210Pb is the more sensitive radioisotope for the identification of continental air masses at remote islands.

Acknowledgments

[60] We acknowledge the contribution of many members of the CARBOSOL team for aerosol sampling, sample handling and transfer. The authors are specifically grateful to Ingeborg Levin for invaluable support with respect to accompanying 222Rn monitoring and to Michel Legrand for cooperative steering activities throughout this work. The project has been funded by DG XII of the European Commission, contract EVK2 CT2001-00113.

Ancillary