We measured aerosol size distributions and conducted bulk and
size-segregated aerosol sampling during two summer campaigns in January
2015 and January 2016 at the continental Antarctic station Kohnen (Dronning
Maud Land). Physical and chemical aerosol properties differ conspicuously
during the episodic impact of a distinctive low-pressure system in 2015
(LPS15) compared to the prevailing clear sky conditions. The approximately 3-day LPS15 located in the eastern Weddell Sea
was associated
with the following: marine boundary layer air mass intrusion; enhanced condensation
particle concentrations (1400
The impact of aerosols on global climate, which is in particular mediated by
governing cloud droplet concentrations and hence cloud optical properties
(Rosenfeld et al., 2014; Seinfeld et al., 2016), is of crucial importance
but likewise notoriously charged with the largest uncertainties (Boucher et
al., 2013; Seinfeld et al., 2016). In a seminal review, Carslaw et al. (2013)
concluded that uncertainties in cloud radiative forcing are inter alia
caused by uncertainties in natural emissions of dimethylsulfide (DMS)
producing biogenic sulfur aerosol (i.e., MS
Biogenic sulfur aerosol, i.e., secondary aerosol produced by the photooxidation of DMS and primary sea salt aerosol, by far dominate the aerosol mass over the Southern Ocean around Antarctica (Raes et al., 2000; Quinn and Bates, 2011). This original marine aerosol is partly transported to continental Antarctica and eventually deposited on the ice shield. Ideally, deposited aerosol constituents are archived in chronological order in firn (densified snow) and ice (Legrand and Mayewski, 1997). Therefore ice core records of biogenic sulfur and sea salt tracers potentially provide invaluable information on their (strictly speaking local) atmospheric budget, which is intrinsically tied to the Southern Ocean climate in the past (Wolff et al., 2006; Kaufmann et al., 2009; Mayewski et al., 2009; Abram et al., 2013). More specifically, investigations of sea salt and biogenic sulfur records from the EPICA (European Project for Ice Coring in Antarctica) ice core retrieved at Kohnen station in Dronning Maud Land revealed the relationship of these archived aerosol tracer profiles with climate indices such as the Antarctic Circumpolar Wave or the Antarctic Dipole (Fischer et al., 2004; Fundel et al., 2006). In view of the poor knowledge of aerosol source strength and atmospheric concentrations regarding the Southern Ocean realm, retrieving representative historic data from ice core archives is consequently desirable. Certainly, any meaningful interpretations of ice core records rely on the knowledge of the source region and major transport processes as well as transport efficiency to continental Antarctica which is connected with the instant general weather situation. These crucial points can only be elucidated by dedicated aerosol measurements on-site.
Location of Kohnen station (Dronning Maud Land, DML) and site plan of the station surroundings.
In this way, previous aerosol investigations (bulk and sometimes
size-segregated composition) revealed a striking difference in the
seasonality of sulfur aerosol composition between coastal and inland
Antarctica, with MS
Up to now, all investigations at Kohnen were purely based on bulk aerosol sampling. The aim of our present study is focused on the variability of aerosol number concentration, aerosol size distribution, and size-segregated aerosol sampling to investigate the impact of different general weather situations on the physical and chemical properties of the aerosol for a site located on the Antarctic Plateau (Weller et al., 2017). This extended synoptic documentation of the physics and chemistry of Antarctic aerosol primarily concentrates on biogenic sulfur aerosol due to its distinct seasonal summer concentration peak caused by the seasonality of marine biogenic activity in the surrounding Southern Ocean (Weller and Wagenbach, 2007). Hence deposition in Dronning Maud Land should be virtually entirely governed by the atmospheric concentration maximum on-site, which was sufficiently covered by our dedicated observations.
During two summer seasons between 16 January and 3 February 2015 and 12 and
29 January 2016, we conducted aerosol size distribution measurements and size-segregated aerosol samplings at the continental Kohnen station
(75
For convenience we will use throughout this work the notion “day of the year” (doy) instead of the calendar date and time is indicated in UTC, which is virtually identical to the solar time. All trace compound concentrations refer to standard volume at 273 K and 1013 hPa (STP).
An overview of the experimental setup during both campaigns is given in
Table 1, comprising the respective measuring periods of the different
configurations and the relevant specifications of the deployed
instruments. The size distribution of the sub-micrometer aerosol at Kohnen
was determined by a scanning mobility particle sizer (SMPS; TSI classifier
model 3080; Wang and Flagan, 1990). Setup and respective data evaluation
methods were virtually identical to our installation run at Neumayer
Station, as already described in detail elsewhere (Weller et al., 2015). Hence
we simply highlight here the most relevant points. The SMPS was generally
run in combination with the same condensation particle counter (TSI,
model WCPC 3788; 50 % cutoff diameter
Survey of the experimental setup during both summer campaigns at Kohnen station.
Particle size distributions were complemented with continuous condensation
particle (CP) concentration measurements (TSI, model CPC 3775;
The low-volume sampler (ISAP® 1050; Schulze
Automatisierungstechnik, Germany) and the five-stage low-pressure Berner-type
impactor (GIV, model 80/0.05/2.88) were installed outdoors ca. 15 m
northeast of the bivouac hut with ambient air inlets about 1.70 m above
the ground. Low-volume sampling (1.66 m
According to Piel et al. (2006) sample extracts were analyzed by using ion
chromatography (IC) for methane sulfonate (CH
Relative uncertainty
Time series of the measured meteorological parameters in 2015 (60 s temporal resolution). The period of LPS15 is shaded in yellow. The cloud amount in oktas is denoted on the top of the second figure based on casual visual inspection.
The original concentration data from the different impactor stages were inverted according a procedure described in Winklmayr et al. (1990). In order to assess the validity of the inversion algorithm, we first compared the mass balance of the original with the inverted data. The discrepancy was typically within 1 to 2 % by mass, but never exceeds 5 %. Next we had to consider the influence of the analytical uncertainty of the ion concentrations on the inversion procedure. To this end we ran a Monte Carlo simulation, i.e., 1000 realizations of the inversion for a given compound were calculated, independently varying the concentrations of each impactor stage within 2 SD of the experimental error (we used individual concentration-dependent errors derived from the exponential fit in Fig. 2). From these 1000 realizations, the mean size distribution and the confidence intervals were determined. The result for the two most critical cases is shown in the Supplement (Fig. S1). For clarity, we refrain below from displaying confidence intervals in the presented impactor data.
Comparing ionic concentrations derived from low-volume and impactor sampling
(including results from both campaigns covering the same sampling period,
i.e., 11 out of 12 impactor samples), we found that concentration ratios
from the impactor versus low-volume sampling were 0.64
The basic meteorological parameters wind speed (
Time series of the measured meteorological parameters in 2016 (60 s temporal resolution). The stormy period LPS16 is shaded in yellow. Again, the cloud amount is denoted on the top of the second figure.
Figures 3 and 4 give a detailed synopsis of the measured meteorological
parameters. Regarding local meteorology, bright weather conditions, i.e.,
clear sky conditions with a cloud amount typically well below 2 oktas,
prevailed, except the period between 17 and 20 January 2015 (doy 17 at 12:00
to doy 20 at 18:00) when the impact of a cyclone (LPS15) with a center in the
southeastern Weddell Sea reached the site (a corresponding weather chart is
shown in Fig. S2). Note that the cloud amounts given
in Figs. 3 and 4 were assessed by everyday casual visual inspection between
04:00 and 22:00 and should be rated as subjective but reasonable estimates.
In the following we will especially focus on this episodic and prominent
weather situation, which is preluded by a sharp pressure drop of 15 hPa and associated
with increasing wind velocities to around 10 m s
During the continuous clear sky conditions, katabatic winds from the
northeast (following the terrain slope) around 5 m s
Due to our limited meteorological database, a thorough characterization of
the atmospheric boundary layer was not feasible. Though to this end it seems
reasonable to access a detailed investigation of the summertime
atmospheric boundary layer at Kohnen by Van As
et al. (2006). Concerning
clear sky conditions, our meteorological conditions (
In Figs. 5 and 6, CP concentrations measured with both CPCs (CP
CP (blue diamonds), UCP
CP (blue diamonds), UCP
Particle number concentrations and ionic composition (mean
Expectedly, the time series of the particle number size distributions (PNSDs)
exhibited a corresponding distinct feature (Fig. 7): during LPS15, we observed according to
Dal Maso et al. (2005) a
pronounced class 1 or so-called “banana-type” NPF event. Particle growth
started at doy 19 (02:50) at a modal maximum of 12 nm, reaching 43 nm at doy
20 (24:00; Fig. 7). While CP and UCP concentrations exhibited distinct
breaks around noon at doy 19 (Fig. 5), steady particle growth was observed
throughout. The initial nucleation particle formation rate (doy 19 between
02:50 and 17:20) in the size range 3 to 25 nm (
Time series of the particle size distribution
d
In addition, we observed enhanced UCP concentrations between 10 and 25 nm
during doy 18 in the morning hours, just before the actual NPF, and in the
evening on doy 19 (numbered 2, 4, and 5 in Fig. 7). Apart from that,
discernible natural nucleation bursts occurred around noon on doy 17 and 18
(numbered 1 and 3 in Fig. 7). All these transient UCP maxima did not show
any detectable particle growth. The nucleation bursts were characterized by
increasing particle concentrations from slightly above 5 nm downward towards
the lower instrumental size limit (“open” distribution), indicating local
nucleation. Table 3 provides a summary of the respective particle formation
rates
List of distinct enhanced UCP concentrations and natural nucleation
burst events apart from the main NPF (event numbers refer to Fig. 7). The
initial nucleation rate (mean
For the rest of the measuring period in 2015, PNSDs were prima facie monotonous with a mode maximum around 60 nm (Fig. 8). Notably, an almost persistent Aitken mode around 34 nm was present, which is also obvious in the mean PNSD derived from DMA 3081 data between doy 28 and doy 33 (Supplement, Fig. S5). In contrast, merely an accumulation mode could be identified in the corresponding mean PNSD covering the period from doy 23 to the end of the campaign in 2016 (Fig. 8). In the first part of the latter campaign (doy 12 through doy 22, measured with the nano-DMA 3085), an additional modal maxima between 10 and 30 nm sporadically appeared (Fig. S6).
Mean PNSD during clear sky conditions measured with DMA 3081 during the campaign in 2015 (blue circles) and 2016 (turquoise circles). The red lines are lognormal fits with geometric mean diameters of 34 and 58 nm for the bimodal distribution observed in 2015 and 63 nm for 2016.
Though super-micrometer particles were not captured by the SMPS data, PNSD
appeared clearly governed by sub-micrometer aerosol
(Fig. 8), indicating a crucial role of nss–SO
Generally ion concentrations (except NH
Time series of the measured MS
Figures 11 and 12 show the size-segregated composition of the aerosol
derived from impactor samples taken 2015 and 2016. Note that
in case of 2016 we could not assign particular impactor results to LPS16 due
to the short duration of this event compared to the sampling period. In Fig. 11,
percentage entries denote the portion of the corresponding ion mass in
the size range > 1
Time series of the measured MS
In Dronning Maud Land cyclone-driven marine air intrusions are infrequent,
sporadic events that are often associated with high precipitation rates
(Birnbaum et al., 2006; Schlosser et al., 2010; Welker et al., 2014; Kurita
et al., 2016). Such a cyclone-induced advection of marine boundary layer air
masses towards the Antarctic Plateau had essentially coined the physical and
chemical properties of the aerosol on-site, most noticeably by a maximum of
biogenic sulfur concentrations and the occurrence of an NPF event. During a
previous similar general weather situation at Kohnen (10 and 11 January
2002; LPS02), Piel et al. (2006) reported even strikingly higher
nss–SO
Results from size-segregated (Berner impactor) sampling during
the campaign 2015. Bold reddish lines are the mass size distributions during
LPS15. Percentage entries in the legend denote the portion of the
corresponding ion mass in the size range > 1
Results from size-segregated (Berner impactor) sampling during
the campaign 2016 with the following mass portion of the respective ions in
the super-micrometer range (> 1
The 5-day back trajectories confirmed these conclusions (Fig. 14; starting
point 100 m above Kohnen): air masses during LPS15 were generally marine 2
to 3 days before arrival at Kohnen. Trajectories representing doy 19
(NPF event; reddish and yellow traces in Fig. 14a) spent several hours
within the marine boundary layer close to the East Antarctic coast before
arrival at Kohnen, largely following the contour lines of the local
topography. All trajectories started under cyclonic curvature, finally
approaching Kohnen in an anticyclonic bow from northerly directions (Fig. 14).
In order to estimate the reliability of this trajectory-based finding,
we repeated the calculations with an initial height of 10 m above Kohnen,
calculated trajectory ensembles, and made an extra attempt using the
isentropic approach instead of the 3-D wind field from GDAS data
(Figs. S8–S10). Though appreciable differences
regarding the geographic location of the corresponding source regions and
trajectory course were obvious, in the end the basic aforementioned
implications appeared consistent. In contrast to this finding, air masses
originated and stayed within continental Antarctica during the biogenic
sulfur peaks observed during 14 and 15 January 2002 (Piel et al., 2006),
which could be confirmed by a reanalysis with HYSPLIT trajectories
based on NCEP meteorological data (Fig. S11). Similar
air mass trajectories (not shown) were observed during and in the aftermath
of a short stormy period LPS16, which was again characterized by enhanced
Na
Regarding the chemical composition of the aerosol during the final stage of
LPS15, subsequent increasing nss–SO
Mean size-segregated results for the
MS
Another conspicuous point was the comparatively large part of MS
The 5-day backward trajectories during the NPF event calculated
with a trajectory starting height of 100 m above Kohnen at the points in
time given in the legend
In conclusion it is worthwhile to consider the results from Neumayer. This
coastal site was governed by the same low-pressure system (LPS15) provoking
a blizzard there around 19 January 2015. From there, daily low-volume
sampling (Teflon–nylon filter combination), CP concentration (measured with
a CPC 3022A, TSI;
Obviously the most striking feature during LPS15 was a distinctive NPF
event. The closed shape of PNSD, starting with a modal maximum not less
than 12 nm, implied that the actual particle nucleation event should have
occurred upwind of Kohnen. Assuming a constant growth rate of 0.6 nm
h
In this context it is interesting to compare our results with previous PNSD
measurements conducted at the South Pole during two summer campaigns (Park et
al., 2004) and year-round observations at Concordia (Järvinen et al.,
2013). Results from the South Pole showed similar PNSDs with mean a
Pant et al. (2010) presented a detailed analysis about the impact of passing
cyclones on particle size distributions at coastal Maitri. The authors
observed bimodal PNSD with a coarse-mode maxima around 2
Daily 10-day backward trajectories (3-D approach, starting height
100 m) during clear sky conditions in 2016 (doy 12 to doy 31,
Daily 20-day backward trajectories (3-D approach, starting height
100 m) during clear sky conditions in 2016 (doy 12 to doy 31,
On the one hand, LPS15 was attended with an expeditious advection of marine
boundary layer air, leading to an enhanced entry of marine aerosol and
secondary aerosol precursors, highly variable PNSD, and an NPF event. On the
other hand, clear sky conditions largely prevailed during our recent campaign, but
also during previous summer campaigns (Piel et al., 2006). The present
observations at Kohnen showed that throughout this characteristic synoptic
situation, appreciably lower particle number concentrations restricted
within the accumulation mode were typical. Concerning the ionic composition,
MS
Turning towards biogenic sulfur aerosol,
The air mass history during clear sky conditions was assessed by composite backward trajectory calculations and is summarized in Fig. 15. Even 10 days before arrival at Kohnen with a characteristic anticyclonic curvature, trajectory origins remained principally inside the Antarctic continent and thus remote from marine source regions. Varying initial start height did not essentially change this general feature, but employing the isentropic instead of the 3-D approach showed an increased relevance of marine source regions (see Fig. S13). Upon extending the trajectory travel time to 20 days, the origin of the air masses eventually became marine (Fig. 16), covering a large part of the Southern Ocean, except the western part of the Weddell Sea and the Bellingshausen Sea. The relevance of air mass transport via the free troposphere was difficult to assess, mainly due to the generally highly variable and poorly characterized depth of the marine boundary layer and especially the vertical extent of the atmospheric boundary layer over Antarctica. According to Russell et al. (1998), the border between the marine boundary layer and the free troposphere typically varies between 1400 and 1900 m, while on the Antarctic Plateau only a shallow atmospheric boundary layer of no more than a few hundred meters is typical (Van As et al., 2006). Inspection of the calculated trajectories revealed that they typically originated within the marine boundary layer (i.e., mainly below 1500 m) and essentially stayed below 500 m above the ground across continental Antarctica.
Surprisingly, while the amount of sea salt aerosol was highly variable, the
super-micrometer-mode fraction of sea salt aerosol remained constant at
around 50 % during rapid and efficient marine boundary layer air mass
advection under LPS15 and during long-range transport under clear sky
conditions (Figs. 11 and 12). Since our sea salt mass size distribution
also appeared similar to that typically observed at Concordia (Jourdain et
al., 2008; Legrand et al., 2017b), we infer that the transport of coarse-mode
sea salt particles to continental Antarctica was generally inefficient,
regardless of the general weather situation and transport time. Finally, the
fact that observed
We measured aerosol size distributions and conducted bulk and size-segregated aerosol sampling during two summer campaigns at the continental Antarctic station Kohnen. This extended approach allowed for a detailed synopsis of the physical and chemical properties of summer aerosol in this region. For the first time, the impact of passing cyclones on aerosol advection into the Antarctic Plateau region was examined. Based on these admittedly still limited investigations, we may conclude that during austral summer, the transport of marine aerosol to Kohnen in particular and to continental DML in general was mediated by two different synoptic situations: (i) the impact of low-pressure systems in the western part of the South Atlantic associated with temporarily exceptional marine aerosol concentrations and (ii) persistent long-range transport providing a background aerosol level during clear sky conditions over DML. In the present study, a distinct low-pressure event (LPS15) was additionally associated with NPF. Under prevailing clear sky conditions, on the other hand, aged aerosol and less aerosol (by mass and number concentration) entered DML in air masses that were typically continental for about 10 days before. We tentatively infer that our recent observation, i.e., NPF and peaking marine aerosol concentrations during LPS15, could be of fortuitous occurrence since unexpectedly just the trace compounds seemed to hardly be depleted by precipitation. In contrast, during the mentioned blizzard in 2002 (LPS02) reported by Piel et al. (2006), biogenic sulfur concentrations stayed first quite low but peaked at about 48 h in the aftermath of the storm. Though we refer to two recent and three previous studies, just three LPS and one pronounced NPF event occurred that could be analyzed in detail, emphasizing their sporadic nature. Hence, a worthwhile confirmation of our conclusions would clearly require similar investigations at this site. Such effort is important to better understand the role of biogenic aerosol in general and in particular the impact of NPF events on regional climate forcing.
Though the efficient transport of biogenic sulfur (and also sea salt) aerosol to continental DML may be associated with cyclonic activity in the South Atlantic, in the long run the crucial transport pathway of marine aerosol during austral summer should be long-range transport under typical clear sky conditions. In particular for biogenic sulfur, showing a pronounced summer maximum (Weller and Wagenbach, 2007), we suppose that transport to DML, deposition, and final storage in firn and glacial ice will be dominated by prevailing clear sky conditions. Thus dry deposition, but to an only minor extent wet deposition (partly associated with clear sky precipitation), would be decisive. Consequently future research activities should also envisage assessing dry deposition velocities at this site, e.g., by gradient and/or eddy correlation studies (Grönlund et al., 2002; Contini et al., 2010). On the other hand, retrieving meaningful historic aerosol concentrations from ice core archives also needs a thorough consideration of snow accumulation since snow accumulation co-determines trace compound concentrations in firn and ice (Fischer et al., 1998), which is evidently governed by the infrequent impact of low-pressure systems (Birnbaum et al., 2006; Schlosser et al., 2010; Welker et al., 2014; Kurita et al., 2016). Finally, trajectory analyses indicated that a large part of the Southern Ocean should be considered as a potential source region representative of aerosol deposition in continental DML in contrast to coastal Neumayer where the dominance of the South Atlantic was evident (Minikin et al., 1998).
Data from both campaigns reported here are available at
The supplement related to this article is available online at:
The authors declare that they have no conflict of interest.
The authors would especially like to thank all technicians present at Kohnen Station, namely Holger Schubert, Torsten Langenkämper, and last but not least, Jens Köhler, whose outstanding engagement actually enabled both air chemistry campaigns at this site. We are thankful to the NOAA Air Resources Laboratory for having made available the HYSPLIT trajectory calculation program and all the input data files used. We thank Kevin Manning for providing us with weather charts based on the Antarctic Mesoscale Prediction System (AMPS). Finally, we appreciate the two anonymous reviewers for their helpful comments. The article processing charges for this open-access publication were covered by a Research Centre of the Helmholtz Association. Edited by: Yves Balkanski Reviewed by: two anonymous referees