Interactive comment on “ Seasonal variations in physical characteristics of aerosol particles at the King Sejong Station , Antarctic Peninsula ”

General remarks The authors report the measurements of aerosol number concentrations, cloud condensation nuclei (CCN), black carbon, and meteorological conditions over a six-year period. The authors claim in the Introduction that “it is necessary to have long-term observations at different regions because aerosol particles vary temporally and spatially.” It seems to me that the authors do not make full use of the data measured for a relatively long time (March 2009 to February 2015). Given extensive data available, interest in the paper would be enhanced if the authors had obtained more significant results. For instance, in addition to a seasonal trend, the authors should check if an annual trend exists for aerosol number concentrations, CCN, and air temperature. I will discuss in detail several points which need to be broadened, analysed and corrected.


Introduction
Aerosol particles in the atmosphere are either emitted directly from various natural and anthropogenic sources (primary aerosol particles) or produced by gas-to-particle conversion processes (secondary aerosol particles).These particles influence the local and global climates directly by scattering and absorbing radiation and indirectly by acting as cloud condensation nuclei (CCN) or ice nuclei (IPCC, 2013).The physical, chemical, and optical properties of atmospheric aerosol particles determine their impact on climate change.Although various studies on the effects of aerosol particles on climate change have been conducted, specific direct and indirect climate effects are still uncertain (IPCC, 2013).Moreover, in order to understand the sources and processes related to atmospheric aerosols, long-term observations for different regions are required because aerosol particles vary temporally and spatially.
The Antarctic region is highly sensitive to climate change due to complex interconnected environmental systems such as snow cover, land ice, sea ice, and ocean circulation (Chen et al., 2009).Previous studies have shown that Antarctica and the Antarctic Peninsula have experienced significant climate changes (Rignot et al., 2004;Steig et al., 2009;Pritchard et al., 2012;Schneider et al., 2012).The Antarctic Peninsula, in particular, is warming at a rate 5 times faster than other regions (Vaughan et al., 2003;IPCC, 2013).The Antarctic climate system can be linked to aerosol particles by complex feedback processes that involve aerosol-cloud interactions.Also, because there are less anthropogenic emission sources in Antarctica, it is a suitable place to study the formation and growth processes of natural aerosol particles.For these reasons, monitoring the physical properties of aerosol parti-cles in Antarctica, such as total particle concentration, size distribution, and concentration of black carbon and activated CCN, is necessary.
The continuously monitored physical properties of aerosol particles at the Korean Antarctic station (King Sejong Station) from March 2009 to February 2015 are used as the basis for this study.Measurements of the aerosol size distribution and concentration of the total aerosol number, black carbon (BC), and CCN were conducted using various instruments.The main aim is to determine the seasonal variation in the physical properties of aerosol particles in the Antarctic Peninsula.Also, the physical characteristics of aerosol particles that originate from the ocean and continent of the Antarctic region are also investigated using air mass backtrajectory analysis.

Sampling site and instrumentation
Continuous observation of the physical properties of aerosol particles was carried out between March 2009 and February 2015 at King Sejong Station (62.22 • S, 58.78 • W) in the Antarctic Peninsula.Detailed information for the sampling site is provided by Choi et al. (2008).In brief, King Sejong Station is located on the Barton Peninsula of King George Island, which has a population density higher than other re- gions in Antarctica due to the research activities conducted on eight permanent on-site stations.The observatory is located approximately 400 m southwest of the main buildings and includes a power generator and crematory.Thus, the northeastern direction (355-55 • ) is designated as a local pollution sector because of emissions from the power generator and crematory.Data collected from this sector as well as BC data with concentrations higher than 100 ng m −3 are discarded to improve the data quality of this study.Moreover, to ensure the reliability of measurements, only validated data that spanned at least 50 % of one day were used for the analysis.In this study, the physical characteristics of aerosol particles collected between March 2009 and February 2015 are analysed.
The physical characteristics of aerosol particles were continuously observed with various instruments including two condensation particle counters (CPCs), an Aethalometer, a cloud condensation nuclei counter (CCNC), and a scanning mobility particle sizer (SMPS).These observation methods are shown in Fig. 1.
Based on Global Atmosphere Watch (GAW) aerosol measurement guidelines and recommendations, a cylindrical stainless common sample inlet was installed.The common inlet was placed on the roof of the observatory (Fig. 1).The diameter and length of the inlet were 0.1 and 5.2 m, respectively.In order to understand flow condition in the common inlet, the Reynolds number was calculated.We used mean values of air temperature and pressure measured over the period from March 2009 to February 2015.The mean values of temperature and pressure were −2.4 • C and 98.8 kPa, respectively.The total flow rate of the sampled air was maintained as 150 L min −1 .The Reynolds number in the common inlet was 2388, which represents a laminar-turbulent transition flow (2000 < Re < 4000).For sampling, a short L-bend stainless steel tube was placed at the centre of the common inlet.Sampling instruments were connected to the inlet using conductive tubing to minimize particle losses.The diameter and length of the conductive tubing connecting the stack with the sampling devices are 3/8 in.and 0.6 m, respectively.To maintain ambient conditions, drying was not conducted during sampling.The total counting efficiency of the sampling equipment was estimated to be 0.92 for 2.5 nm particles, and 1 for particles larger than 10 nm, respectively, at the common inlet (Hinds, 1999;Baron and Willeke, 2011).All sampling lines, expect for the common inlet, were connected with conductive tubing to minimize particle losses.
The total particle number concentration was examined with two CPCs: a TSI model 3776 that measures particles greater than 2.5 nm in diameter, and a TSI model 3772 that measures particles larger than 10 nm.Sample aerosol flow rates for CPC 3776 and CPC 3772 were 1.5 and 1.0 L min −1 , respectively.
An Aethalometer (Magee Scientific, AE16) was used to measure the concentration of light-absorbing particles for 880 nm wavelengths.The flow rate of the sample was constant at 5.0 L min −1 .The main purpose of measuring the BC concentration was to investigate the long-range transport of aerosol particles and assess the impact of local pollution.
To measure the CCN concentration, a CCNC (DMT CCN-100) was used for five different supersaturation (SS) ratios (0.2, 0.4, 0.6, 0.8, and 1.0 %) and a total flow rate of 0.5 L min −1 .The CCN concentration was determined by exposing aerosol particles at supersaturated conditions and then counting the number of activated droplets with a detector.The sampling duration was set at 5 min for each SS value (except for 0.2 % SS) before it was changed to the next value.For 0.2 % SS, the CCN concentration was measured for 10 min because the additional time was required to achieve stability after conducting measurements at 1 % SS.
Aerosol size distributions were continuously measured with a SMPS, which comprises a differential mobility analyser (DMA; HCT Inc., LDMA 4210), a CPC (TSI 3772), a control unit, an aerosol neutralizer (soft X-ray), and a data logging system.The length, inner diameter, and outer diameter of the DMA were 44.42, 0.953, and 1.905 cm, respectively.The resolution of the scanning time was set to 120 s for mobility particle diameters between 0.01 and 0.30 µm.A closed sheath-air loop with a diaphragm pump in the control unit was used to maintain the sheath flow of the DMA.The flow rate of sheath air of the DMA was 10 L min −1 .The ratio of aerosol flow to sheath flow of the DMA was 1 : 10.Also, meteorological parameters including temperature, relative humidity (RH), wind speed (WS), wind direction (WD), pressure, and UV and solar radiation were also continuously monitored over the entire observation period.

Back-trajectory analysis
To relate the physical properties of aerosol particles to their source areas for the sampling period, an air mass back- trajectory analysis was conducted using a Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model (Stein et al., 2015) (http://www.arl.noaa.gov/HYSPLIT.php).For every 6 h period, 120 h air mass back trajectories were analysed, ending at heights of 100, 500, and 1500 m above ground level at the sampling site.Results for the origin and pathway of the air mass that were similar for at least 12 h for the three different heights were used in the present analysis.The air masses were then classified into four groups according to their origin and pathway, including two continental regions (South America and Antarctica) and two oceanic areas (South Atlantic and South Pacific Ocean), as shown in Fig. 2.

Meteorological conditions
Figure 3 depicts the monthly variation in meteorological parameters measured from an automatic weather system (AWS) over the entire observation period.The temperature varied between −19.5 and +5.8 • C, with a mean of −2.4 ± 2.1 • C and relative humidity (RH) was between 60 and 100 %, with a mean of 87.9 ± 3.3 %.As mentioned previous studies (Kwon and Lee, 2002;Mishra et al., 2004), the observation site was relatively humid and warm compared to inland Antarctic stations due to marine environment.The solar radiation varied from 2.3 to 375.4 W m −2 , with a mean of 81.2 ± 38.9 W m −2 .No clear annual trends for temperature were observed during the 6-year period due to a relatively short observation period.In this paper, an analysis of the correlation between temperature (or solar radiation) and the CN concentration is the focus.3.2 Seasonality in the physical characteristics of aerosol particles As can be seen in Fig. 4, there is a clear seasonal cycle for CN concentration, which is highest in the summer (December to February, DJF) and lowest in the winter (June to August, JJA).The maximum concentration of particles larger than 2.5 nm (CN 2.5 ) and larger than 10 nm (CN 10 ) was approximately 2000 cm −3 in De-cember 2012 and about 800 cm −3 in December 2009.The minimum values of CN 2.5 and CN 10 concentration were approximately 110 and 90 cm −3 , both in August 2013, respectively.These results are in agreement with previous studies from other Antarctic stations (Jaenicke et al., 1992;Gras, 1993;Virkkula et al., 2009;Weller et al., 2011).For instance, Virkkula et al. (2009)   mer season (DJF) should be related primarily to non-sea-salt sulfate and methanesulfonate (MSA) derived from the oxidation of dimethyl sulfide (DMS) produced by phytoplankton (Fattori et al., 2005;Weller et al., 2011).The DMS concentration increases sharply when biological activity was enhanced due to increasing temperatures and solar radiation (Virkkula et al., 2009).Since the sampling site for this study was in the Antarctic Peninsula, the biological activity of the ocean was considered as an important factor for the formation and growth of the aerosol particles.Higher differences in concentration occur in summer rather than winter for CN 2.5 and CN 10 concentration correlated with a higher biological activity in summer than winter, as well as solar radiation and temperature.Therefore, it can be supposed that trends of differences are related to secondary aerosol formation caused by biological activity.To better understand the effect of temperature and solar radiation intensity on the CN 2.5 concentration, the relationship between the monthly mean CN 2.5 concentration and solar radiation intensity and the monthly mean CN 2.5 concentration and temperature was compared.

Total particle number concentrations
The correlation coefficient between CN 2.5 and the solar radiation intensity (open circle; R 2 = 0.621) was slightly higher than for the mean CN 2.5 concentration and temperature (open triangle; R 2 = 0.419), as shown in Fig. 5.These results suggest that the CN 2.5 concentration may be more closely coupled with solar radiation intensity than temperature.
Unique results for the CN 2.5 concentration were observed as shown in Fig. 4. The CN 2.5 concentrations in the summer of 2013-2014 were much lower than other years.Unfortunately, the reason for the lower CN 2.5 concentration cannot be explained by the solar radiation intensity or temperature as there was not any distinctive variation in these values compared with other years.A possible reason is the type of air mass reached the sampling site.Although the air mass that originated in the South Pacific Ocean (Case IV: of the descriptions in Sect.3.3) was dominant in summer, based on the air mass back-trajectory analysis explained in Sect.2.2, the frequency of the air mass from the South Atlantic Ocean (Case II) in the summer of 2013-2014 was higher than other years and the frequency of air mass originating from Case IV was lower than other years.Unfortunately, neither CN 10 nor SMPS data are available for the austral summer season of 2013-2014 because of mechanical failures.Thus, it is not possible to directly explain the low concentration for CN 2.5 in this season regarding the potential effects on the air mass characteristics and the concentration of particles sized between 2.5 and 10 nm.Nevertheless, it is likely that an increased frequency of air mass originating from the South Atlantic Ocean (Case II) might have resulted in a lower CN 2.5 concentration in the austral summer season of 2013-2014.These CN concentration is comparable with results from Aboa Station, which is located in the coastal area of Antarctica and is mainly affected by South Atlantic air masses (Virkkula et al., 2009).These results showed a daily CN concentration from December 2003 to January 2007.Although there was a variation in the CN concentration from year to year, the daily CN concentration during the austral summer was ∼ 600 cm −3 .
A more detailed comparison of the monthly trends for the CN 2.5 and CN 10 concentration is presented in Fig. 6.The monthly mean CN concentration increased from September to February during the austral spring and summer periods (Bigg et al., 1984;Jaenicke et al., 1992;Gras, 1993).The CN concentration sharply decreased from March and remained stable from April to August.In particular, the CN 2.5 concentrations during the summer period increased sharply com- pared to the CN 10 concentration.This increase was probably due to new particle formation.High solar radiation and temperature, and low RH values during the summer are conducive to new particle formation (Hamed et al., 2007).

Cloud condensation nuclei (CCN) concentrations
Figure 7a shows the monthly mean CCN concentration at a SS value of 0.4 % over the period from March 2009 to February 2015.There is a large gap in data from July 2011 to December 2013 due to a faulty CCN counter.Anttila et al. (2012) measured cloud droplet number concentration (CDNC) and CCN concentration for five SS values (0.2, 0.4, 0.6, 0.8, and 1.0 %) during the third Palls Cloud Experiment (PaCE-3).These results showed a correlation between the CDNC and CCN concentration for each supersaturation.The relationship between the CDNC and CCN concentration at a SS value of 0.4 % was approximately linear.Furthermore, CCN concentration were lower than the CDNC, when the SS value was lower than 0.4 %, and higher than the CDNC, when the CCN concentration was above 0.4 %.Based on these results, supersaturation of 0.4 % was chosen to investigate seasonal variation in the CCN.Monthly variation in the CCN concentration was found, with a maximum value observed during the summer period (DJF) and a minimum concentration observed during winter periods (JJA).The monthly mean CCN concentration ranged from 20.63 cm −3 in July 2009 to 227.52 cm −3 in January 2014, with a mean of 112.80 ± 39.05 cm −3 .Figure 7b also shows seasonality for the CCN concentration at the SS value of 0.4 %.The CCN concentration gradually decreased from February and remained consistently low during the winter, while the CCN concentration from September increased sharply, as shown in Fig. 7b.The maximum CCN concentration in January was 199.89 ± 37.07 cm −3 and the minimum CCN concentration in August was 42.13 ± 14.51 cm −3 .This clear seasonality for the CCN concentration follows the seasonal variation in the CN concentration.As shown in Fig. 6, the CN 10 concentration as well as CN 2.5 concentration increased during summer.In addition, the aerosol size distribution measured by the SMPS showed that the concentration for accumulation-mode particles in the range of 100 and 300 nm, as well as Aitkenmode particles during summer, increased significantly, as can be seen in Fig. 8. Accumulation-mode particles can easily act as CCN (Dusek et al., 2006); hence CCN concentrations increase during summer and decrease during winter.
In order to investigate the seasonal variation in fractions of the CCN concentration for each SS value for the CCN concentration at a SS of 1.0 %, the CCN data were examined in more detail.An analysis of the cumulative CCN concentrations given in terms of fractions of the CCN concentration measured at a SS of 1.0 % was carried out.The results are shown in Fig. 9. Here, fractions of the CCN concentrations were estimated by dividing the CCN concentration at each SS value by the total CCN concentration at a SS of 1.0 %.Although a clear seasonal trend for CCN concentration with a maximum during the summer and a minimum during the winter was presented, the fraction of CCN concentration at a SS value of 0.2 % for the activated CCN concentration showed a different pattern with a maximum value in July and a minimum value in December, as shown in Fig. 9.The numbers at the top of Fig. 9 represent the mean CCN concentration at a SS values of 1.0 %.The fraction of particles activated to CCN at a SS value of 0.2 % during summer and winter was 0.49 ± 0.07 and 0.62 ± 0.06, respectively.The frac- tion at a SS value of 0.2 % during winter (JJA) was similar to that (∼ 0.52) measured at Mace Head which is a representative site of a marine environment (Paramonov et al., 2015).
Seasonal variation in the mean activation ratio of the CCN concentration at a SS of 0.4 % to the CN concentration measured by the two CPCs (TSI 3776 and 3772) are shown in Fig. 10.The mean value of the activation ratios of CCN / CN 2.5 and CCN / CN 10 were 0.33 ± 0.10 and 0.40 ± 0.08, respectively.Although clear changes were observed in the monthly variation for the CN and CCN concentrations as shown in Figs. 6 and 7b, it was seen that the activation ratio (CCN / CN 10 ) was similar regardless of seasonality.The reason that no clear change is observed in the activation ratio at King Sejong Station might be the variation in the concentration of accumulation-mode particles, as can be seen in Fig. 8.The lower activation ratio in September and November is because of the size and chemical properties of the aerosol particles.Both the size and chemical    composition of the aerosol particles can have a large impact on the activation ratio (Dusek et al., 2006;Leena et al., 2016).The concentration of Aitken-mode aerosol particles increased sharply compared to the concentration in August.Thus, the activation ratio decreased dramatically.Unfortunately, the aerosol size distribution could not be confirmed because the aerosol size distribution data for November were not reliable due to malfunctioning instruments.
The CCN concentrations for different SS values can be represented by the power-law function, defined by Twomey (1959): where N CCN is the concentration of CCN at given a supersaturation value (SS) and C and k T are coefficient constants estimated from CCN spectra.The correlation coefficient between the measured CCN concentration at a SS of 0.4 % and the fitted values of the power-law function was 0.978.The values of C varied from 6.35 to 837.24 cm −3 , with a mean of 171.48 ± 62.00 cm −3 .The daily mean k T values vary between 0.07 and 2.19, with a mean of 0.41 ± 0.10.The monthly variations of k T and C values are summarized in Fig. 11.A comparison with CCN concentrations indicated that the values for k T during the austral winter (June) were also the lowest (0.29 ± 0.06), while during summer (December) they were highest (0.55 ± 0.13).Based on this result, aerosol particles activated to CCN during summer are expected to be more sensitive to SS changes than aerosol particles activated during winter.
It is useful to infer the hygroscopic properties of aerosol particles with the hygroscopicity parameter, κ. κ values varied from 0 for insoluble particles to larger than 1 for watersoluble salt particles (Petters and Kreidenweis, 2007).κ values are defined as where σ w is surface tension of water, M w is the molecular weight of water, R is the universal gas constant, T is the temperature, and ρ w is the density of water.SS is the supersaturation applied in the CCNC.The critical diameter, D act , was estimated following the method of Furutani et al. ( 2008) where N tot is total number concentrations of aerosol particles measured by SMPS.D is the electric mobility diameter observed by SMPS.In this calculation, D 0 was assumed to be 10 nm, at the beginning of the SMPS scan.The CCN / CN ratio indicates the fraction of CCN-active aerosols of the total particle concentration.In this study, the κ values were estimated using a monthly mean CCN concentration at a SS of 0.4 %, the monthly mean CN concentrations measured by the CPC, and the monthly mean size number distribution results obtained from the SMPS data.The annual mean κ value was calculated to be 0.15 ± 0.05.This value is comparable to the previous studies from the Arctic and subarctic areas.For example, Lathem et al. (2013), who measured the CCN activity in the Arctic using aircraft measurements, reported a κ value of 0.32 ± 0.21.Martin et al. (2011) inferred a total κ value of 0.33 ± 0.13 during a cruise observation in Longyear- byen, Svalbard.Kammermann et al. (2010) reported κ values that varied between 0.07 and 0.21 over an 18-day period within the Arctic Circle in Sweden.Jaatinen et al. ( 2014) also showed a κ value of 0.13 ± 0.07 using a 13-day dataset from the subarctic area of Finland (Pallas-Sodankylä station).

Black carbon (BC) concentrations
Figure 12 shows the variation in monthly mean BC concentrations over the whole sampling period.To eliminate the effect of local pollution on the observations, in the present study, data of BC concentration higher than 100 ng m −3 were discarded.The daily mean BC concentration varied between 1.07 and 75.97 ng m −3 , with a mean of 27.43 ± 4.98 ng m −3 .The BC concentration observed at King Sejong Station was higher than those at other stations in Antarctica (Bodhaine, 1995;Wolff and Cachier, 1998;Pereira et al., 2006;Weller et al., 2013).For instance, the annual mean BC concentrations at the South Pole, Halley, Neumayer, and Ferraz stations were 0.65, 1.0, 2.6, and 8.3 ng m −3 , respectively.The reason for the higher BC concentration might be related to location of sampling site.There are nine permanent on-site stations in the Baton Peninsula of King George Island.In particular, six stations are located within a 10 km radius of King Sejong Station.The presence of these stations should affect the measured BC concentration, causing a positive bias.Also, no clear seasonal patterns were observed in this study throughout the entire observation period.However, clear seasonal patterns were observed in previous studies at other stations in Antarctica (Wolff and Cachier, 1998;Weller et al., 2013).Wolff and Cachier (1998)   although BC concentration during summer increases slightly, no clear seasonal trends were observed unlike the results measured by Wolff and Cachier (1998).This suggests that the BC concentrations are dependent on the sampling site and the long-range transport of air mass.

Effect of air mass trajectory on the physical properties of aerosol particles
In this section, the effect of the origin and pathway of the air mass on the physical characteristics of the aerosol particles is presented.As mentioned earlier in Sect.2.2, air masses are classified into four groups based on air mass back-trajectory analysis.The wind data and aerosol characteristics of the four types of air masses over the entire observation period are listed in Table 1.The very few cases of air masses originating from South America (Case I) showed the highest BC and CCN concentration.This may be due to anthropogenic influences at the source and the aging of aerosol particles.The CN 10 concentration was similar regardless of the origin and pathway of air mass, whereas an enhancement of the CN 2.5 concentration was observed when the air mass originated from the ocean (Case II and IV).This is probably due to the high biological activity in the South Atlantic and South Pacific oceans during summer (DJF).A more detailed comparison, excluding the results of Case I from the CN concentration based on an air mass analysis is shown in Fig. 13.It should be noted that the mean values of CN concentration in this analysis fall within the range of standard deviation for the mean values of each case.This is probably because of new particle formation, causing a bias, as new particle formation results in a sudden increase in aerosol concentration (Kulmala et al., 2004;Pierce and Adams, 2007).The minimum concentration of aerosol particles for CN 2.5 and CN 10 originate from the ocean (Case II and IV).These are observed from April to September, whereas the concentration of aerosol particles (CN 2.5 ) originating from the South Atlantic (Case II) and the South Pacific (Case IV) oceans are highest in November and February, respectively.It was found here that the peak month for the CN 2.5 concentration had discrepancies with the air mass history.This is probably due to a difference in the chemical compounds that contribute to aerosol formation processes and/or variations in biogenic activity according to the origin and transport pathway of the air mass.To verify this, further studies on the chemical compositions of aerosol particles are necessary.For air masses  transported from the South Pacific Ocean to King Sejong Station (Case IV), the seasonality of the aerosol size distribution was also investigated.The aerosol size distribution parameters were fitted to a log-normal distribution derived for each season.The seasonally averaged log-normal aerosol size distribution is shown in Fig. 14.The computed modal diameters along with the standard deviation and number concentrations are summarized in Table 2.It is clear that the modal diameter during summer is larger than during winter for both Aitken (0.034 compared to 0.023 µm, respectively) and accumulation modes (0.109 compared to 0.086 µm, respectively).The number concentrations for summer are also higher than those for the winter for the (304.36 ± 20.10 compared to 49.16 ± 3.88 cm −3 , respectively) and accumulation (140.25 ± 10.64 compared to 44.78 ± 14.24 cm −3 , respectively) modes.The enhancement of the number concentrations for the Aitken mode during summer should be linked to new particle formation over the ocean as a product of biological activity.The spring and autumn season shows intermediate values.These results are similar to those of previous laboratory and field experiments (Sellegri et al., 2006;Yoon et al., 2007).O 'Dowd et al. (2004) suggested that primary forma-tion processes play a significant role in marine aerosol production in the North Atlantic Ocean.In addition, the contribution of biological organic compounds to the marine aerosol distribution might be dominant (Kim et al., 2015).

Summary and conclusions
Seasonal variations in the physical characteristics of aerosol particles at King Sejong Station (62.22A clear seasonal variation in particle number concentration (CN) exists, with maximum concentration in the austral summer (DJF) and minimum concentration in the austral winter (JJA).The maximum CN concentration of particles larger than 2.5 nm (CN 2.5 ) and 10 nm (CN 10 ) was approximately 2000 cm −3 in December 2012 and about 800 cm −3 in December 2009 and February 2015, respectively.In particular, the CN 2.5 concentration increased sharply during the summer compared to CN 10 concentration, suggesting that particle formation processes are likely driven by high biological activity.
In addition, clear seasonal trends of CCN concentration at a supersaturation (SS) of 0.4 % were presented.The maximum mean CCN concentration of 199.89 ± 37.07 cm −3 was measured in January and the minimum mean CCN concentration was 42.13 ± 14.51 cm −3 in August.The activation ratio (CCN / CN 10 ) of aerosol particles at King Sejong Station (0.40 ± 0.08) in the Antarctic Peninsula was lower than those at the Arctic area (0.52) (Lathem et al., 2013).The measured CCN spectra were approximated by fitting a power-law function relating the number of CCN at a given SS to the SS, with fitting coefficients, C and k T (Twomey, 1959).The values of C varied between 6.35 and 837.24 cm −3 , with a mean of 171.48 ± 62.00 cm −3 .The value of k T ranged from 0.07 to 2.19, with a mean of 0.41 ± 0.10.The k T values during the austral summer period (DJF) were higher than those during the winter period (JJA).Furthermore, the annual mean hygroscopicity parameter, κ, was estimated as 0.15 ± 0.05, for a SS of 0.4 %.
Based on a backward trajectory analysis, air masses were classified into four groups according to their origin and pathway including two continental regions (South America and Antarctica) and two oceanic areas (South Atlantic and South Pacific Ocean).It was found that most air masses originated from the oceanic areas.The very few cases of air masses originating from the South American continent (Case I) showed the highest BC and CCN concentrations.The CN 10 concentrations were analogous regardless of origin, whereas CN 2.5 concentrations showed differing values.The CN 2.5 concentrations that originated from oceanic areas (Case II and IV) were higher than those from continental regions (Case III).In particular, the CN 2.5 concentration showed clear seasonal variations, with minimum concentrations from April to September and maximum concentrations in November from the South Atlantic Ocean (Case II) and in February from the South Pacific Ocean (Case IV).Furthermore, regarding Case IV, an analysis of aerosol size distributions in the 0.01 to 0.3 µm range was performed.The modal diameters also showed seasonal variations of 0.023 µm in the winter and 0.034 µm in the summer for the Aitken mode and 0.086 µm in the winter and 0.109 µm in the summer for the accumulation mode.
Overall, this study is the first of its kind to analyse seasonal variations in physical characteristics of aerosol particles in the Antarctic Peninsula.The aerosol particle formation process is still not fully understood, and thus more studies should be necessary to determine seasonal variations in the chemical characteristics of atmospheric aerosols.

Figure 1 .
Figure 1.A schematic diagram for the observation methods used in this study.

Figure 2 .
Figure 2. Map of the sampling site (62.22 • S, 58.78 • W; black circle) and classification of the four cases according to the origin and pathway of the air masses.Dot lines represent example of back trajectories according to cases.

Figure 3 .
Figure 3. Monthly mean variation in (a) temperature, (b) relative humidity, (c) wind speed, (d) wind direction, (e) air pressure, and (f) solar radiation over the period from March 2009 to February 2015.The shaded area in panel (d) represents the wind direction for the local pollution sector.

Figure 4
Figure4shows the monthly mean particle number (CN) concentration measured with two types of instruments (TSI CPC 3776 and 3772) over the period between March 2009 and February 2015.All seasons mentioned in this study are austral seasons.As can be seen in Fig.4, there is a clear seasonal cycle for CN concentration, which is highest in the summer (December to February, DJF) and lowest in the winter (June to August, JJA).The maximum concentration of particles larger than 2.5 nm (CN 2.5 ) and larger than 10 nm (CN 10 ) was approximately 2000 cm −3 in De-

Figure 4 .
Figure 4. Monthly variations of mean CN 2.5 (black open circle) and CN 10 (red open circle) concentrations with a standard deviation from March 2009 to February 2015.The error bars represent the standard deviation of the measurements from the mean value.

Figure 5 .
Figure 5. Scatterplot diagram of monthly mean CN 2.5 concentrations and monthly mean temperature (blue open triangle) or monthly mean solar radiation intensity (red open circle).Blue and red solid lines are regression lines.

Figure 6 .
Figure 6.Box plots of seasonality of (a) CN 2.5 and (b) CN 10 concentrations.Lines in the middle of the boxes indicate sample medians (mean value is closed circle), lower and upper lines of the boxes are the 25th and 75th percentiles, and whiskers indicate the 5th and 95th percentiles.

Figure 7 .
Figure 7. (a) Monthly mean CCN concentrations at the SS of 0.4 % with a standard deviation from March 2009 to February 2015.(b) Seasonal variation in mean CCN concentrations at the SS of 0.4 % with a standard deviation.

Figure 8 .
Figure 8. Seasonal mean aerosol size distribution measured by the SMPS at the King Sejong research station over the period from March 2009 to February 2015.The error bars represent the standard deviation of the measurements from the mean value.

Figure 9 .
Figure 9. Monthly mean cumulative CCN concentrations shown as fractions of the CCN concentration at the SS of 1.0 %.Colours indicate the SS bins.The number at the top of figure represents monthly mean CCN concentrations at the SS values of 1.0 %.

Figure 10 .
Figure 10.Comparison of the seasonal mean variation in the activation ratio between measurements (CPC 3776 and CPC 3772) by two CPCs.The error bars represent the standard deviation of the measurements from the mean value.

Figure 11 .
Figure 11.Seasonality of monthly mean values of (a) C and (b) k T over the whole observation periods.The error bar represents the standard deviation of the measurements from the mean value.

-Figure 12 .
Figure 12.Monthly mean concentrations of black carbon over the period from March 2009 to February 2015.The error bars represent the standard deviation of the measurements from the mean value.

Figure 13 .
Figure 13.Seasonal variation in mean (a) CN 2.5 and (b) CN 10 concentrations with a standard deviation depending on the air mass origin.The error bars represent the standard deviation of the measurements from the mean value.

Figure 14 .
Figure 14.Seasonal log-normal fitted size distribution of aerosol particles originating from the South Pacific Ocean, ranging from 0.01 to 0.3 µm (Case IV).

Table 1 .
Pereira et al. (2006)s of BC measured at the Halley Station and the South Pole with an Aethalometer.They found that although BC concentrations varied depending on the sampling site, BC concentration decreased during the austral winter (JJA) and increased during the austral summer (DJF).Contrary toPereira et al. (2006), Summary of meteorology and aerosol data according to the origin and transport pathway of aerosol particles.Case I, Case II, Case III, and Case IV refer to the origin and pathway of the air masses from South America, South Atlantic Ocean, Antarctica and South Pacific Ocean, respectively.

Table 2 .
Seasonal size distribution lognormal fitting parameters for the Aitken and accumulation mode of aerosol particles originating from a Case IV scenario.N , σ , and D g refer to the number concentrations, a standard deviation, and the geometric mean diameter, respectively.
• S, 58.78 • W) in the Antarctic Peninsula were investigated based on in situ measured aerosol data between March 2009 and February 2015.