Influences of the 2010 Eyjafjallajökull volcanic plume on air quality in the northern Alpine region

Introduction Conclusions References

hanced PM 10 and SO 2 concentrations were detected on 17 April at mountain stations (Zugspitze/Schneefernerhaus and Schauinsland) as well as in Innsbruck by in situ measurement devices. On 19 April intensive vertical mixing and advection along with clear sky-conditions facilitated the entrainment of volcanic material down to the ground. The subsequent formation of a stably stratified lower atmosphere with limited 10 mixing near the ground during the evening of 19 April led to an additional enhancement of near-surface particle concentrations. Consequently, on 19 April and 20 April exceedances of the daily threshold value for particulate matter (PM 10 ) were reported at nearly all monitoring stations of the North Alpine foothills as well as at mountain and valley stations in the northern Alps. The chemical analyses of ambient PM 10 at moni-

Introduction
In the past, the Laki fissure eruptions of 1783-1784 in Iceland had a major impact on air quality in Central Europe. It has been estimated that during these eruptions 5 approximately 122 megatons (Mt) of sulphur dioxide (SO 2 ), 15 Mt of hydrogen fluoride (HF) and 7 Mt of hydrochloric acid (HCl) were released into the atmosphere Self, 1993, 2003). These emissions extended vertically up to 13 km. SO 2 was rapidly removed from the relatively moist troposphere by forming about 200 Mt of sulphuric acid (H 2 SO 4 ) and finally sulphate aerosol particles which were present over Introduction Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | mountain stations (Zugspitze/Schneefernerhaus and Schauinsland) detected the volcanic cloud in Southern Germany (Emeis et al., 2010;Flentje et al., 2010;Gasteiger et al., 2010). Mountain hikers, e.g. at Arlberg at about 2500 m above sea level (a.s.l.), reported an odour of sulphur (personal communication C. Plass-Duelmer, DWD/HPB, 2010) on 17 April. During 19 April and 20 April, high concentrations of particulate 5 matter (particle sizes smaller than 10 µm -PM 10 ) were reported at nearly all monitoring stations in the North Alpine forelands as well as at stations in the northern Alps. SO 2 concentrations were also enhanced at many monitoring stations. All over Europe, monitoring stations observed enhanced air pollutant concentrations.
To consider the strength of the influences of the 2010 Eyjafjallajökull volcanic plume 10 on ambient air composition the legally relevant criterion of the European Union air quality directive 2008/50/EC (for Germany, see BImSchV, 2010) is the basis. This directive sets a 24-hour PM 10 mass concentration threshold value of 50 µg m −3 which must not be exceeded more than 35 times a calendar year at any air quality monitoring station. The annual threshold value of PM 10 is 40 µg m −3 and the annual target value of 15 particles with sizes smaller than 2.5 µm (PM 2.5 ) is 25 µg m −3 . Correspondingly, a 24hour threshold value of 125 µg m −3 for SO 2 and an 8-hour target value of 120 µg m −3 for ozone apply, where only 3 and 25 exceedances are allowed per year, respectively. Further threshold values currently exist for nitrogen dioxide (NO 2 ), carbon monoxide (CO), benzene and lead (Pb) and from 2013 on, for arsenic (As), cadmium (Cd), nickel 20 (Ni), and benzo(a)pyrene. According to 2008/50/EC, days with threshold value exceedances which are attributable to natural sources e.g., volcanic eruptions or Saharan dust are reported to the European Commission. These days are then not counted for the 2008/50/EC regulation. Evidence for the contribution of natural sources can for example be provided by elemental analysis of PM 10 samples. Therefore, environmen-25 tal agencies analysed the elemental composition of particle samples and studied the PM 10 and SO 2 concentrations in order to quantify the influence of the Eyjafjallajökull volcano on air quality (see e.g. Krabbe et al., 2010 The study of this volcano eruption event on the basis of existing monitoring networks allows us to determine the potential impacts of volcanic plumes on air quality and the corresponding health risks in the northern Alpine region.

Objectives and study area
This paper provides, in a first step, a documentation of the horizontal and vertical 5 distribution of suspended particles of a volcanic origin in the northern Alpine region following the eruption of Eyjafjallajökull and the associated meteorological transport. The analysis is based on a physical and chemical characterisation of particles and gases at multiple ground-based observation sites as well as remote-sensing data: optical observations from the ground and space; 10 time series of PM 10 and SO 2 concentrations; particle number size distributions (PSD) and particle number concentrations (PNC); multi-elemental composition of PM 10 filter samples; chemical analysis of wet deposition;

15
air-borne measured parameters like spectral optical absorption and ozone concentrations.
This study focuses on the primary event lasting from 17 April to 20 April 2010. In particular, we focus on the development of the near-surface concentrations of SO 2 and PM 10  The study area is delineated by the Black and Bavarian Forests to the West and East and the regions around Nürnberg and Innsbruck to the North and South, respectively (see Fig. 1). This region was characterized by the highest PM 10 concentrations at ground level within Germany and Austria. The following measurement sites have been selected: Schauinsland (SSL), Augsburg (AUHS), Hohenpeissenberg 5 (HPB), Zugspitze/Schneefernerhaus (ZSF) and Innsbruck (IBK). Measured parameters at these sites which are not influenced during the volcanic plume events are not considered.
Finally, the potential influence of the volcanic material with respect to human health aspects is discussed.

Data, instruments and methods
This work is based on analyses of ground-based and air-borne in situ measurements, remote sensing derivatives and synoptic maps. All data were considered as hourly means for continuous monitoring instruments on the time basis Central European Time (CET).

Ground-based in situ measurements
An overview of the selected key monitoring sites, including their characterization and instrumentation, is given in Table 1 At AUHS, a twin differential mobility particle sizer (TDMPS), IfT, Leipzig, Germany (detectable particle size range from 3 to 800 nm) is deployed in conjunction with an aerodynamic particle sizer (APS), TSI Inc., model 3321, Shoreview, USA (detectable particle size range from 0.8 to 10 µm). At HPB, SSL, and ZSF scanning mobility particle sizers (SMPS), TSI Inc., model 3936, Shoreview, USA are operated with a detectable size 5 range from 10 to 800 nm (HPB, SSL) and from 10 to 600 nm (ZSF), respectively. At HPB, SSL and ZSF, a thermodenuder with 300 • C (TD), TSI Inc., Shoreview, USA is running upstream of the electromobility spectrometers during each second sample to evaporate volatile PM components. Quality assurance procedures at AUHS are documented in Pitz et al. (2008a, b) In addition, the in situ aerosol forward and backscattering coefficient and Angstroem exponent were measured at SSL and HPB by a three wavelength integrating nephelometer, TSI Inc., model TSI 3563, Shoreview, USA. The total PNC was measured at ZSF with both a butanol-based and water-based condensation particle counter (CPC), TSI Inc., model 3025a, Shoreview, USA.
imagery based on red-green-blue (RGB) colour composites (Schmetz et al., 2002) from the SEVIRI satellite radiometer. Due to the high spatial and temporal resolution of SEVIRI, dust outbreaks, fire plumes and volcanic plumes can often be tracked and monitored for hours and even days. The European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT) provides the so-called "ash" 10 and "dust" products at hourly resolution as RGB-images in a 5 day rolling archive (http://oiswww.eumetsat.int/IPPS/html/MSG/RGB/). The RGB "ash product" is generated using the following SEVIRI channels (Ch) and channel combinations: red = radiance difference (Ch 12.0 µm -Ch 10.8 µm); this is the prime detection method for ash and dust in the atmosphere (also used for cloud detection); 15 green = radiance difference (Ch 10.8 µm -Ch 8.7 µm), partially sensitive to SO 2 ; blue = Ch 10.8 µm, cloud detection.
Samples were collected during 24 h on quartz fibre filters (PALL TISSUQUARTZ 2500 QAO-UP, Lot 56276) by low volume samplers Leckel, model SEQ47/50, Berlin, Germany. The elemental composition of the PM 10 samples from the five sites was analysed by ICP-MS after microwave decomposition of the samples as described in DIN EN 14902 (2005).
Rain water was collected at HPB using a wet only sampler, Eigenbrodt, model NSA 181/KE, Königsmoor, Germany. The daily samples are changed at 07:30. They are analysed for ions using suppressed ion chromatography with a Dionex AS14 column for anions and a Dionex CS12 column for cations. The pH value is measured with a 15 WTW pMX3000/pH meter and a Mettler Toledo InLab Science electrode, conductivity with a WTW LF3000 instrument and a WTW LR01/T electrode, both electrodes suited for low conductivities. The data are part of HPB's routine contribution to the Global Atmospheric Watch (GAW) programme as a Global Station and are quality controlled according to the GAW procedures.

Synoptic situation and transport of volcanic emissions
On 14 April and 15 April, a North Atlantic high pressure system extended over the northern British Isles towards Southern Scandinavia and induced north-easterly winds over central Europe (for more details and affects in Iceland, see Petersen, 2010 and the Netherlands, as evidenced, e.g., by ceilometers that progressively detected the volcanic plume during its southbound motion at about 6-7 km a.s.l. (see Emeis et al., 2010). By 17 April, the centre of this high pressure system was located over Southern England and the main air flow continued from North-West to South-East. Since large parts of central Europe were cloud-free during 17 April and partially also 5 on 18 April, the volcanic plume could be tracked by hourly METEOSAT/SEVIRI satellite imagery. However, the Alpine area and large parts of south-western Germany and Switzerland were affected by a low pressure system over the northern Adriatic Sea which brought clouds and rain showers to the area on 18 April and the following days, thus hampering observation of the volcanic plume by passive satellite instruments. This cyclonic system was mainly responsible for further dispersal of volcanic material in the area of interest. A first weak sign of the eruption became visible in SEVIRI data, the so-called "ash product" as described in Chap. 3.2, from 14 April, 08:00 UTC while clear signs were apparent by 15 April,15:00 UTC. Observational conditions were not optimal for passive 15 satellite instruments since high clouds hampered the view of lower atmospheric levels. However, on 16 April, 00:00 UTC, the ash front was clearly identified over the North Sea (Fig. 2, upper panel) in a distance of already about 1500 km from Iceland. The southern front of the volcanic plume was almost parallel to circles of latitude. Only nine hours later, the front line already passed the coastal zones of Germany and Poland 20 and quickly moved across Germany. It further passed large parts of Poland, where the most southerly section of the front reached the Czech Republic and Slovakia in the late afternoon of 16 April (Fig. 2, lower panel). At this point it is evident that the curvature of the ash front had grown, indicating a differential propagation by different sections of the volcanic plume. The area of highest speed towards South-East is indicated by the Introduction layer across Germany and parts of Austria, including south-east of the Alps (Fig. 3, upper left panel, faint coloured layer). Surface winds were from north-easterly directions at the Alps (Fig. 3, middle panels) and the low pressure system over Northern Italy is apparent in the satellite image as well as in the relative humidity fields (Fig. 3, middle panels). The more rapid motion of parts of the plume towards South-East led to a 5 more elongated front line across Europe (Fig. 3, upper right panel). The most eastern part of the volcanic plume initially passed the eastern Alpine area, while the western patches were to later slow down over south-western Germany and in Switzerland on 18 April. As a consequence of this dilution, particle concentrations in the plume decreased and the ash signal in the satellite image became weaker. On 17 April, the ash plume covered all the stations in Southern Germany and Austria but became invisible for SEVIRI over large parts of this area due to reduced concentration levels, with clouds hampering the detection. On 17 April, 12:00 UTC, remnants of the ash front are not only still visible over south-eastern Austria and Slovenia, but also over France (see arrows in Fig. 3, upper right panel). During 17 April, the low pressure system over 15 northern Italy became more and more important for the spatial and temporal evolution of the volcanic plume north of the Alps. As seen in Fig. 3 (middle panels), the wind direction over the northern Alpine area turned from north-east to easterly directions. The southern Alpine area experienced south-easterly flow inducing cloudiness and some precipitation. Signs of subsiding air masses (blue areas in Fig. 3, lower right panel) 20 are noted along the northern fringes of the Alps on 17 April, 12:00 UTC (Fig. 3, lower panels) which supports the conclusions drawn from the ceilometer data as discussed in Chap. 4.4.1.
Signatures of the volcanic plume are hardly visible in the satellite images from 18 April (Fig. 4, upper panels). This is due to the increasing influence of the low 25 pressure system south of the Alps hampering the direct observation of the volcanic plume over Tyrol and the northern Alpine area. The further approach of the cyclonic system induced southerly winds across the Alps contrasting the north-easterly winds the day before (Fig. 4,  18 April (subsidence in the North of the Alps, blue areas in Fig. 4) is suggested by the banded structure of vertical velocity and short-term southerly winds at ZSH during noon. However, the foehn did not break through to e.g. the Inn valley bottom. However, the associated cyclonic system rapidly crossed the Alps and lost control over the air flow in the area of interest. The remaining volcanic material over the southern 19 April, the ash load of the atmosphere was no longer visible from space. This was mainly due to stronger downward mixing down to the ground and thus dilution, as it will be discussed later. Westerly winds dominated in the following days when ash remnants over Europe were no longer detected by passive satellite instruments. The station ZSF detected the first enhanced SO 2 concentrations on 17 April while  Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | correlated with correspondingly higher PM 10 levels observed at the monitoring sites. Enhanced formation of H 2 SO 4 by atmospheric oxidation of SO 2 to sulphur trioxide (SO 3 ) and the subsequent reaction with water vapour may have led to increasing concentrations of secondary particles, i.e. ultrafine particles (UFP -particle size smaller than 100 nm and shown in Fig. 7) during the period of interest.

5
Time series of ultrafine PNC in the size range 10-30 nm and 30-100 nm observed at the four core measurement sites AUHS, SSL, HPB and ZSF are shown in Fig. 7. On 17 April, ultrafine PNC levels rose from background concentrations to elevated levels at ZSF shortly before 11:00 and two hours later at SSL. As discussed by Emeis et al. (2010), the delay at SSL is probably due to the much lower height of 10 this monitoring site (1200 m a.s.l. compared to the 2670 m a.s.l. of ZSF). The signal peaks between 15:00 and 16:00 at both sites with a PNC of more than 16 000 and 12 000 particles cm −3 at SSL and ZSF, respectively (size range 10-100 nm). During the following hours, the measured ultrafine PNC decreased to a relatively low level of less than 4000 particles cm −3 at SSL at 20:00, and later on also at ZSF. The latter PNC 15 is rather typical for ZSF if the station is within the PBL. After reaching their first maximum, the ultrafine PNC decreased on 18 April, but showed a second and stronger maximum in the afternoon of 19 April. This peak occurred between 12:00 and 13:00 at SSL, and four hours later (between 16:00 and 18:00) at the HPB and ZSF sites. The ultrafine PNC at AUHS showed a maximum 20 between 13:00 and 14:00, but the association of this peak with volcanic material is partially masked by the elevated regional and urban background level in the urban air of Augsburg. The ultrafine PNC were higher during the second episode on 19 April than during the first episode on 17 April at all four locations AUHS, SSL, HPB and ZSF.
The PNC for particles in the accumulation range i.e. particle sizes from 100 to 25 800 nm showed no significant increase (data not shown). Further in situ measurements at HPB are reported in detail by Flentje et al. (2010) and are briefly summarized here. There is no significant increase of PNC for particle sizes from 300 nm to 500 nm measured with an optical particle counter (OPC), Grimm, Comparison of particle samples recorded with and without TDs at SSL and ZSF between 14:00 and 16:00 on 17 April showed a considerable fraction of particles which consisted of evaporating material. This indicates the presence of evaporable secondary aerosol which was initially not of volcanic origin but may have been formed in the volcanic plume from SO 2 . At the same time, elevated PNC were also found at the urban background site AUHS. Because of the much higher anthropogenic regional and urban background at this site, this maximum cannot be attributed to the volcanic plume without additional measurements (see Chap. 4.4.1). Note also that the increase of ultrafine PNC at HPB was much less pronounced in comparison with SSL or ZSF (see Chap. 4.4.3 too). 15 Apart from urban anthropogenic emissions, UFP are regularly formed in the atmosphere during mid-day secondary particle formation bursts (Kulmala et al., 2004). In Central Europe, such bursts seem to be triggered by the photochemical formation of H 2 SO 4 , originating mainly from the oxidation of SO 2 (Birmili et al., 2003;Paasonen et al., 2010). 20 Volcanoes are a major source of SO 2 in the atmosphere (see also Thomas and Prata, 2011). The Institute of Earth Sciences of the University of Iceland publishes current SO 2 gas fluxes from volcanoes in Iceland (http://www.earthice.hi.is/page/ IES-EY-CEMCOM). It is obvious from measured SO 2 values (Fig. 6,  is supported by prevailing clear-sky conditions over Central Europe on 17 April and 18 April which facilitate hydroxyl (OH) radical production and their subsequent reaction with SO 2 (for more information about these processes, see Hamed et al., 2010;Kazil et al., 2010). To estimate the contribution of volcanic plume material to PM 10 , it is basically possible to use elements with a significant increase in PM 10 and a high enrichment factor found in the ash samples. Some elements like Fe and Mn show however rather high 5 background values due to the impact of anthropogenic sources (e.g. abrasion of braking pads, industry etc.) and are therefore less suitable for this purpose. The elemental composition of Eyjafjallajökull ashes have been published on the website of the Institute of Earth Sciences of the University of Iceland (http://www.earthice.hi.is/page/ IES-EY-CEMCOM). Compared to the average composition of the Earth crustal mate-10 rial (Binder, 1999), the elements Mn, phosphorus (P), Sc, strontium (Sr), Ti, Y and Zr are enriched by a factor higher than 2 (see also Table 2). The exception is P, which was not quantified in our samples. Therefore, apart from P these highly enriched elements were used to estimate the impact of the volcanic plume on PM 10 concentrations. Figure 9 shows plume to ambient PM 10 samples has been estimated by multiplying the surplus in concentration ( Fig. 9) with corresponding conversion factors (see Table 2). On this basis, the impact of the volcanic plume to PM 10  for ADRO are 8 ± 3 µg m −3 on 19 April and 10 ± 4 µg m −3 on 20 April (26 ± 10% on both days). The high variability of these values is reflecting the uncertainty in the basic data (composition of volcanic ash and PM 10 sample analysis), and the unquantifiable impact of sources other than the volcanic plume on elevated element concentrations.

Identification of volcanic
On the other hand, the estimation of the volcanic plume contribution is supported by 5 the fact that the fraction of both ammonium and nitrate in the PM 10 samples, which are also indicators for secondary aerosol, were substantially reduced during the volcanic plume events. Primary aerosol dominated those PM 10 samples which are usually dominated by secondary aerosol. Further, only small changes of sulphate are observed, because part of it has probably been of volcanic origin.

Evidence of volcanic plume influences in Augsburg
Within the PM 10 record at Augsburg (Fig. 6), peak concentrations were at least partially masked by the enhanced anthropogenic background. However, consideration of additional volcanic fingerprint data as described in Chap. 4.3 allows for an enhanced 15 interpretation of the observed PM 10 time series and reveals a volcanic influence, too. Also HYSPLIT backward trajectories (Draxler and Rolph, 2011) for the monitoring site AUHS were used for the days from 17 April to 20 April. These analyses confirm the former findings (Chap. 4.1) regarding the motion of the volcanic plume across Germany. At all levels, the trajectories consistently show an air flow from north-westerly 20 directions on 17 April which caused effective transport of the volcanic plume to Southern Germany during this time. On 17 April and 18 April, the transport to Augsburg was from easterly directions but on 19 April and 20 April it turned to westerly directions, which is in line with the satellite observations and the surface wind analysis (see Figs. 3 and 4).
Ceilometer backscatter intensities at AUHS showed a layer of strongly enhanced backscatter above the PBL only on 17 April until 13:00 (see Fig. 10 until 13:00, the volcanic plume subsided and was finally mixed into the PBL where its clear signature finally disappeared. From 17:00 until mid-night, a structured layer in the upper part of the PBL became visible and is interpreted as a remnant of the formerly confined plume layer above a stable lower atmosphere. The ceilometer observations further indicate that there was a defined upper boundary of the PBL and the lower at-5 mosphere was well mixed on 17 April from 13:00 until 17:00 up to about 1500 m a.g.l., i.e. the vertical extension of the PBL was relatively large. Entrainment of volcanic material into the PBL must be assumed in this phase, but a corresponding signal on the near-surface air composition at Augsburg was not detectable due to strong dilution and high background concentrations. On 19 April, the situation became different: Due to 10 convection, the distinct separation of the PBL and the free troposphere above disappeared. Flentje et al. (2010) reported thin plume layers all day on 19 April which could partly reach Augsburg due to good vertical mixing and advection. During the following night, a stable near-surface layer and a residual second layer (Fig. 10) were formed. Such meteorological conditions favoured the enrichment of air pollutants near surface 15 level, as seen during the night from 19 April to 20 April in Figs. 5 and 6. Time series of further parameters of PM (size ranges 0.1-1 µm, 1-2.5 µm and 2.5-10 µm) and sulphate concentrations measured in PM 2.5 at AUHS from 17 April to 20 April are shown in Figs. 11a and b. Beginning on 19 April, 18:00, an increased PNC at the ground was observed. This could be seen as a consequence of the meteo-20 rological conditions described above and associated downward mixing and horizontal replacement, i.e. the exchange of air mass which is loaded with volcanic material. PNC is essentially pronounced for particle sizes larger than 1 µm (Fig. 11a). The mass concentration of smaller particles (0.1-1 µm) was not increased compared to other days, i.e. on 17 April. The small increase of these accumulation mode particles is consistent 25 with the increase of sulphate concentration in PM 2.5 (Fig. 11b). It is also in line with the non-significant PNC increase of PM with sizes from 100 to 800 nm.
It is striking that an increased concentration of sulphate in PM 2.5 occurred without a major SO 2 peak during the volcanic plume episode in Augsburg was observed (see Chap. 4.2). This supports our hypothesis (Chap. 4.2) that the formation of H 2 SO 4 by atmospheric oxidation of SO 2 in the volcanic plume leads to the increased concentration of secondary particles. This is also corroborated by a small increase of accumulation mode particles. There can also be coating to the surface of existing fine and coarse particles with UFP (H 2 SO 4 ). Furthermore, we assume the formation of H 2 SO 4 (i.e. ox-5 idation of SO 2 ) has already taken place in the higher layers of the atmosphere during advection of the volcanic plume from the volcano. As the travel time is about 3 days and H 2 SO 4 particles grow at about 6-7 nm h −1 (Hamad et al., 2010) it seems possible that H 2 SO 4 is a compound of particles with sizes between 1 and 2.5 µm as is typical for sulphate (Fig. 11). Further dilution during transport down to the surface inhibited significant increases of SO 2 concentration in the urban air in Augsburg, whereas such an increase was observed at core monitoring sites at higher elevations. It can be concluded that H 2 SO 4 in UFP is formed much faster than the sulphate in accumulation mode particles. The increased content of sulphate in PM 2.5 correlates with the first maximum of 15 particle mass concentration (PM 0.1−1 , PM 1−2.5 and PM 2.5−10 ) on 19 April, 23:00. This indicates that this increase of particle mass is associated with the influence of the volcanic plume. It should be added here that a source apportionment on the basis of PSD analyses will be published soon which is used to separate volcanic ash and dust from other sources. Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | and the dewpoint declined from about 1500 m down to 800 m a.g.l. In both profiles, the dewpoint above this local maximum steadily decreased with height, a feature which in stagnant high pressure conditions is normally associated with cumulus clouds with a by far lower cloud base close to the maximum dewpoint. However, clouds were present at a greater height only. The change in the profiles is thus more likely due to a re-5 placement of the air mass in the area beginning at about 3000 m a.g.l. and gradually progressing towards the ground where the new air mass was detected about one hour after the landing of the aircraft. Fine particle numbers and size distributions did not change significantly between the aged air mass during the ascent and the volcanic plume influenced air mass of the descent but the optical properties were different. The 10 spectral absorption in the UV was nearly completely suppressed while the absorption at 880 nm was comparable high in both profiles (see also Emeis et al., 2010). Concurrent with the reduction in UV absorption, a strong increase in the extinction in the near infrared was observed, a clear indication of a different chemical composition of the air mass.

Ultrafine particles at Zugspitze/Schneefernerhaus
To further characterize the ZSF time series, the total PNC measured by a CPC was analyzed. Especially at the beginning of the event with high PNC of very small aerosols (10-20 nm size range, see Fig. 7) we observed a considerably higher signal of the water-based CPC. It is therefore likely that particles measured at that time were mainly 20 water soluble, being in line with the aerosol measurements using the TD channel. A possible source of such water-soluble aerosols would be the high SO 2 content which, after chemical transformation, leads to sulphate-rich, water-soluble particles (Hamed et al., 2010 Diverging from this finding, decreased aerosol scattering was already observed from 16 April, 12:00 onwards while the SO 2 concentration did not increase until 17 April, 11:00. Precipitation starting on 18 April, 15:00 and on 20 April,17:00 and 21:00, was likely responsible for increasingÅngström exponents, presumably due to scavenging of coarse aerosols. The SO 2 concentration levels however were not reduced by pre-5 cipitation. The entrainment of the volcanic plume into the PBL proceeded from East to West due to predominant south-easterly winds for 24 hours (see Figs. 3 and 4). After 18 April, 03:00, the wind changed first to western, then to northern direction and again to westerly directions. From 18 April, 08:00 onwards, easterly winds again dominate.

Composition of wet deposition at Hohenpeissenberg
On 18 April from about 12:00 to 14:00 and on 20 April from about 13:50 to 15:10 and with some breaks from 19:10 onwards, it was slightly raining at HPB. These periods are characterised by reduced PM and SO 2 concentrations on the ground in comparison to the concentrations of those compounds during the time periods some hours before 15 (see Fig. 6). Samples of these precipitation events were analysed for water soluble ions (nitrate (NO 3 -N), sulphate (SO 4 -S), ammonium (NH 4 -N) and hydrogen ion (H + )), elements (chlorine (Cl), Na, K, Mg and Ca), acidity or asicity (pH) and conductivity. Table 3 shows the results and is complemented with two days in May, when additional volcanic plumes were detected through high SO 2 concentrations at ZSF (see also Chap. 5.1).

20
For reference, the table also includes the highest, mean and lowest concentrations and depositions observed at HPB in April and May, during the ten preceding years.
Due to the small precipitation amounts, concentrations were generally high. Conductivity and H + concentration on 18 April were the second highest of all April precipitation events from 2000 to 2009, the sulphate concentration on that day ranked in third place.

25
It is therefore more reasonable to look into the deposition data which are calculated with the concentration data from the precipitation sample analyses. They do not exceed the maximum values measured in April  18 April, deposition of H + and sulphate was unusually high, but both components were similarly high during the week before the eruption. These high concentrations can therefore largely be attributed to the advection of polluted continental air in the PBL rather than to the volcanic plume on that day. On 20 April, the precipitation event was at the end of the volcanic influence upon the PBL. In summary, the wet deposition 5 analysis results from HPB do not provide information about the volcanic influence on the air composition.

Volcanic influences on air quality
Entrainment of volcanic plume material into the lower atmosphere was evident be- mixing near the ground during the evening of 19 April led to the additional accumulation of locally emitted particles. Advection was also important. Between the second and third event the synoptic situation became different and the air flow changed from an easterly up to a westerly direction. That means that in the afternoon of 19 April, an exchange of air masses 5 occurred in the northern Alpine region.
The enhanced PM 10 values on 19/20 April could just as well have been caused by anthropogenic sources. However, chemical analyses of PM 10 samples revealed a very unusual elemental pattern with an increase in the concentration of elements like Ti, Zr and Y by a factor of more than 10 at five LÜB monitoring stations. Elevated PM 10 10 concentrations were recorded at all monitoring stations in the northern Alpine area almost simultaneously and the elemental composition data support the view that the high PM 10 concentrations were associated with the presence of the volcanic plume. Also, the compounds Mn, P, Sc, Sr, Ti, Y and Zr are enriched in PM 10 in comparison to the Earth crust composition. This is different to Saharan dust events which are composed 15 of the Earth crust elements too (see Table 2). During such events, enriched concentrations of the primarily oxides of silicon (Si), Fe and Al from 24 March to 25 March 2007(Vanderstraeten et al., 2008 or Fe, Al, Ca, Ni and As from 27 May to 01 June 2008 (Bruckmann et al., 2008) were found. But enhanced Ti concentrations as observed during the Eyjafjallajökull volcano event were never seen in the long-term PM 10 elemental 20 analyses. A study for France by Colette et al. (2010) found the similar conclusions especially over 18 April and 19 April in Mulhouse. There is a difference between the observations at ZSF and SSL as well as HPB and AUHS during the volcanic plume event from 17 April to 20 April. Increased ultrafine PNC were measured at the first two stations but no comparable increase of these 25 parameters existed at the other two stations. The high amount of SO 2 measured at the same time at ZSF and SSL enabled the new formation of aerosols from SO 2 and resulted in these high values of ultrafine PNC at the beginning of the volcanic plume event. However, enhanced concentrations of sulphate are measured at AUHS and of  April (see Flentje et al., 2010) so that SO 2 is transformed into these compounds during the transport of the volcanic plume. Following the European Union air quality directive 2008/50/EC, short-term threshold exceedances in the northern Alpine region were found for the 24-hour PM 10 mass concentration, but not for SO 2 , O 3 , PM 2.5 and further regulated compounds. Con-5 sequently, the volcanic plume influenced the near-surface atmosphere and thus the ambient air quality. However, the air pollutant emissions caused by the Eyjafjallajökull volcano eruption were much lower than those estimated for the Laki eruption of 1783-1784. The PM 10 and PM 2.5 concentrations observed during the Saharan dust event from 27 May to 01 June 2008 at the monitoring stations used here were also higher 10 (see Bruckmann et al., 2008) than during the Eyjafjallajökull volcano event but with different particle composition during both events.
For a second time after the April 2010 event, volcanic plume influences have been observed with a significant signal at ZSF from 18 May, 09:00 to 19 May, 24:00. In contrary to April, in May enhanced SO 2 concentrations were observed which exceeded 15 14 µg cm −3 on 18 May and 8 µg cm −3 on 19 May. At the same time, precipitation at HPB was relatively clean and did not show any influence from the volcano. The observed SO 2 concentration levels were clearly beyond the concentrations in April 2010. In contrast to the April event, frequent precipitation events and high relative humidity occurred during the May episode, which considerably reduced PM 10 concentrations to 20 maximum levels at about 20 µg cm −3 . During the May episode, various airports in Italy and again the airport at München were nevertheless again closed for safety reasons. A discussion of further observational results together with modelling results can be found in Emeis et al. (2010

Possible health damages caused by air pollution originating from the volcanic plume
Ambient particulate matter has been a long standing concern to induce short-term as well as long-term health effects (Brunekreef and Holgate, 2002;Craig et al., 2008;Dokery, 2009). The size, shape and density of the particles determine their behaviour 5 in the lung. As the human airways are the major surfaces of interaction, particles of sizes less than 10 µm entering the airways and of sizes less than 2.5 µm entering the lungs are of primary concern. This indicates that in principle, volcanic plume particles have the potential to affect human health. Indeed, reports from volcano outbreaks have demonstrated respiratory effects in study subject with underlying pulmonary disease (Longo et al., 2010;Gudmundsson, 2011;Shimizi et al., 2007;Naumova et al., 2007). These observations seem highly plausible as high concentrations of particles have been indicated to irritate upper airways (Bascom et al., 1996;Peters et al., 1997;Tong et al., 2010). However, these respiratory disease exacerbations in association with acute volcanic particle exposure occurred at concentrations one to two orders of 15 magnitude higher than in the investigated Eyjafjallajökull event. Analyses of medical surveillance data from the UK did not indicate an access of respiratory disease in association with regionally transported particles from the Eyjafjallajökull outbreak (Elliot et al., 2010). The data presented in this paper indicates that the transported mass of particles in 20 populated areas was largely in the size range between 2.5 and 10 µm. The control of anthropogenic particles focuses frequently on PM 2.5 (Craig et al., 2008). Consistent associations with cardiovascular diseases have been demonstrated in this lower size range (Brook et al., 2010). The strength of adverse health effects of the coarse mode particles are still being debated (Brunekreef and Forsberg, 2005;Zanobetti and Schwartz, 2009;Tong et al., 2010). Regionally transported volcanic particles will mix with other particles from anthropogenic sources as well as may absorb components from the gas phase. The transported volcanic plume particles may have potentially high acidity due to interaction with sulphur species. For example, H 2 S has been implicated as a biologically active species in relation to volcano outbreaks (Hansell and Oppenheimer, 2004). These compounds are expected as gases and / or within UFPs which are reaching the lungs. Thus the health risk of sulphur species during volcanic plume emissions can 5 also be high.

ACPD
Toxicological investigations are underway to further elucidate the biological activity of particles collected during days with volcanic plume exposure. In conclusion, there seems to be little evidence that the particles associated with the Eyjafjallajökull outbreak have a dramatically increased toxicity. At the same time, it seems not to be 10 warranted to assign to the transported volcanic particles the relatively low toxicity of non-anthropogenic influenced re-suspended dust particles (e.g. mineral dust).

Conclusions
On 17 April, enhanced PM 10 and SO 2 concentrations were detected at mountain stations (ZSF and SSL) as well in Innsbruck by in situ measurement devices. The air 15 quality monitoring networks in Southern Germany (LÜB and LUBW) and ground-based in situ measurement systems at low elevations observed only a moderate increase of PM 10 and SO 2 during that time. However, peak concentrations of both PM 10  From the beginning of 17 April, the ground-based in situ measurement systems in Augsburg as well as of LÜB detected increased PM mass concentrations resulting in a peak in the early morning hours on 20 April which could be assigned to an increase of particles with diameters larger than 1 µm. Introduction

Tables Figures
Back Close

Full Screen / Esc
Printer-friendly Version

Interactive Discussion
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | event was supported by a specific development in the regional weather situation which otherwise prevented an influence of the volcanic plume on air quality at monitoring stations on 17/18 April. The particles attributed to the volcanic plume appeared either as UFP with diameters smaller than 100 nm, or primary particles larger than 1 µm. It is concluded here that the 5 high numbers of UFP were caused by photochemical processes acting in the SO 2 -rich regime of the volcanic plume. This chemical composition is a feature that distinguishes the volcanic plume from Saharan dust. The high acidity of volcanic plume particles is a risk for human health in addition to the PM 10 threshold exceedance.
Without further evidence from toxicological studies, we assume that the volcanic 10 plume particles contribute to the overall exposure of the population on 19/20 April and therefore in principal may lead to exacerbation of respiratory and cardiovascular symptoms. However, for Germany the effect, if any, is expected to be minor.