Characterization of aerosol number size distributions and their effect on cloud properties at Syowa Station, Antarctica

We took aerosol measurements at Syowa Station, Antarctica to characterize the aerosol number–size distribution and other aerosol physicochemical properties. Four modal structures (i.e., mono-, bi-, tri-, and quad-modal) were identified in aerosol size distributions during measurements. Particularly, quad-modal structures were associated closely with new particle formation (NPF). To elucidate where NPF proceeds in the Antarctic, we compared the aerosol size distributions and modal structure to air mass origins computed using backward trajectory analysis. Results of this comparison imply that NPF 15 occurred in free troposphere during spring and autumn, and in the free troposphere and boundary layer during summer. Photochemical gaseous products, coupled with UV radiation, play an important role in NPF, even in the Antarctic troposphere. With the appearance of the ozone hole in the Antarctic stratosphere, more UV radiation can enhance atmospheric chemistry, even near the surface in the Antarctic. However, linkage among tropospheric aerosols in the Antarctic, ozone hole, and UV enhancement is unknown. Results demonstrated that NPF started in the Antarctic free 20 troposphere already in the end-August – early September by UV enhancement resulting from the ozone hole. Then, aerosol particles supplied from NPF during spring grow gradually by vapor condensation, suggesting modification of aerosol properties such as number concentrations and size distributions in the Antarctic troposphere during summer. Here, we assess the hypothesis that UV enhancement in the upper troposphere by the Antarctic ozone hole modifies the aerosol population, aerosol size distribution, cloud condensation nuclei capabilities, and cloud properties in Antarctic regions during summer. 25

Abstract. We took aerosol measurements at Syowa Station, Antarctica, to characterize the aerosol number-size distribution and other aerosol physicochemical properties in 2004-2006. Four modal structures (i.e., mono-, bi-, tri-, andquadmodal) were identified in aerosol size distributions during measurements. Particularly, tri-modal and quad-modal structures were associated closely with new particle formation (NPF). To elucidate where NPF proceeds in the Antarctic, we compared the aerosol size distributions and modal structures to air mass origins computed using backward trajectory analysis. Results of this comparison imply that aerosol size distributions involved with fresh NPF (quad-modal distributions) were observed in coastal and continental free troposphere (FT; 12 % of days) areas and marine and coastal boundary layers (1 %) during September-October and March and in coastal and continental FT (3 %) areas and marine and coastal boundary layers (8 %) during December-February. Photochemical gaseous products, coupled with ultraviolet (UV) radiation, play an important role in NPF, even in the Antarctic troposphere. With the existence of the ozone hole in the Antarctic stratosphere, more UV radiation can enhance atmospheric chemistry, even near the surface in the Antarctic. However, linkage among tropospheric aerosols in the Antarctic, ozone hole, and UV enhancement is unknown. Results demonstrated that NPF started in the Antarctic FT already at the end of August-early September by UV enhancement resulting from the ozone hole. Then, aerosol particles supplied from NPF during periods when the ozone hole appeared to grow gradually by vapor condensation, suggesting modifica-tion of aerosol properties such as number concentrations and size distributions in the Antarctic troposphere during summer. Here, we assess the hypothesis that UV enhancement in the upper troposphere by the Antarctic ozone hole modifies the aerosol population, aerosol size distribution, cloud condensation nuclei capabilities, and cloud properties in Antarctic regions during summer.

Introduction
The Antarctic is isolated from human activities occurring in the mid latitudes. In spite of the slight amount of human activity in the Antarctic, such as research activities at each station and tourism mostly in the Antarctic Peninsula during summer, the source strength of anthropogenic species (e.g., black carbon from combustion processes) is negligible in the Antarctic circle at the moment (e.g., Weller et al., 2013;Hara et al., 2019). Consequently, aerosol measurements in the Antarctic have been taken to ascertain aerosol physicochemical properties, atmospheric chemistry, and their effects on climate change under Earth's background conditions (i.e., cleanest and pristine conditions).
Knowledge and discussion of condensable vapors (i.e., aerosol precursors) is fundamentally important to elucidate microphysical processes such as NPF and growth and locations at which NPF occurs. Earlier works have emphasized examination of the following condensable vapors (i.e., aerosol precursors) for NPF and particle growth: H 2 SO 4 , CH 3 SO 3 H, HIO 3 , NH 3 , amines, and other organics with low vapor pressure (Yu et al., 2012;Kulmala et al., 2013;Kyrö et al., 2013;Weller et al., 2015;Sipilä et al., 2016;Jen et al., 2016;Jokinen et al., 2018;Shen et al., 2019;Burrell et al., 2019). Gaseous H 2 SO 4 has been regarded as an important aerosol precursor (e.g., Kulmala et al., 2013;Kerminen et al., 2018). Actually, H 2 SO 4 in the Antarctic is converted dominantly via photochemical oxidation of dimethyl sulfide (DMS) released from biogenic activity in the ocean and SO 2 derived from DMS oxidation (e.g., Minikin et al., 1998;Weller et al., 2015;Enami et al., 2016;Jang et al., 2019). Indeed, high DMS concentrations in the Antarctic were found in the sea-ice margin and marine boundary layer during summer (Koga et al., 2014). Concentrations of H 2 SO 4 and CH 3 SO 3 H showed diurnal change with a maximum in daytime at Palmer Station (Jefferson et al., 1998a, b). Additionally, NPF often occurred in daytime at Neumayer and Aboa during summer (Asmi et al., 2010;Weller et al., 2015;Jokinen et al., 2018). Measurements of aerosol hygroscopicity at Aboa indicated an important role of organic vapors in the growth of freshly nucleated particles (Asmi et al., 2010). Recently, simultaneous measurements of aerosol precursors and aerosol size distributions showed that H 2 SO 4 and NH 3 made important contributions to NPF and growth at Aboa in air masses from the marine boundary layer (Jokinen et al., 2018). Condensable vapors other than H 2 SO 4 and NH 3 , for example, iodine compounds such as HIO 3 and organics, are associated with atmospheric and snowpack chemistry (e.g., Saiz-Lopez et al., 2008;Atkinson et al., 2012;Roscoe et al., 2015;Sipilä et al., 2016;Hara et al., 2020) and oceanic bioactivity (Dall'Osto et al., 2019). Contributions of the condensable vapors to NPF and growth have been discussed, but this topic has remained unclear for the Antarctic.
Some condensable vapors for NPF and growth are formed through photochemical reactions with atmospheric oxidants such as OH, O 3 , and BrO (Read et al., 2008). They relate to UV radiation (Matsumi et al., 2002;Matsumi and Kawasaki, 2003). Appearance of the ozone hole in the Antarctic stratosphere during September-November (Hoppel et al., 2005) is expected to enhance UV radiation and atmospheric oxidation potential in the troposphere (Jones et al., 2003). Nevertheless, effects on aerosols in the Antarctic troposphere have not been elucidated sufficiently. Because of low aerosol number concentrations, direct effects of aerosol radiative forcing are negligible in the Antarctic (Bodhaine, 1995). After aerosol activation to cloud condensation nuclei (CCN), indirect effects might affect the atmospheric radiation budget and cli-mate (Bromwich et al., 2012). With the present study, we are striving to understand the occurrence of NPF in the Antarctic, linkage among NPF, the ozone hole, and cloud properties in the Antarctic coast (around Syowa Station). In OPC measurements, number concentrations were recorded every minute. SMPS, CPC, and OPC were operated at room temperature of ca. 20 • C in the clean air observatory located at the windward side ca. 400 m distant from the main area of the station, where a diesel generator was operated. When winds came from the main area, local contamination might have occurred. Before analysis and discussion, locally contaminated data were screened using condensation nuclei concentrations (measured as aerosol monitoring at Syowa Station) and wind data (provided by the Japan Meteorological Agency). Data screening procedures and criteria were identical to those described in earlier reports (Hara et al., 2011b(Hara et al., , 2019. After screening of locally contaminated data, daily mean aerosol number concentrations and size distributions were calculated for data analysis. Because of lower aerosol number concentrations during winter through early spring, data quality of hourly mean aerosol size distributions was insufficient to identify NPF and to analyze growth rate after NPF through the year. Consequently, daily mean aerosol size distribution was analyzed and discussed in this study. Details of the respective NPF events during the periods with higher aerosol concentrations will be discussed elsewhere. Aerosol sampling for chemical analysis was made using a two-stage mid-volume impactor and a back-up filter. Water-soluble aerosol constituents such as Na + , SO 4 2− , and CH 3 SO 3 − were determined using ion chromatography. Details of aerosol sampling and chemical analysis were presented in reports of our earlier work (Hara et al., 2004(Hara et al., , 2018.

Estimation of coagulation sink and condensation sink
Nano-size aerosol particles are removed rapidly through coagulation. To elucidate the speed of removal by coagulation, coagulation sink (Coag.S) was calculated using the following equation (Kulmala et al., 2001).
Therein, K ij and N j , respectively, represent the coagulation coefficient in the transitional regime and the number concentrations of the size bin of j . Condensation sink (Cond.S) was calculated to elucidate the rate of condensable vapor removal by condensation on aerosol particles using the following equation (Kulmala et al., 2001).
Therein, r and β M , respectively, denote the particle radius and the transitional correction factor given by the following equation.
In that equation, Kn and α, respectively, represent the Knudsen number and the sticking coefficient. The value of α is assumed to be unity. Also, Kn is expressed as In Eq. (4), λ denotes the mean free path. Details of calculation procedures were presented in an earlier report (Kulmala et al., 2001).

Estimation of the nucleation rate of aerosol particles
The nucleation rate of aerosol particles (D p = 5 nm: J 5 ) was estimated as elucidating NPF and particle growth in the Antarctic troposphere. Number concentrations (N) of nanosize aerosol particles in each size bin depend on (1) growth from the smaller size by vapor condensation, (2) coagulation loss, and (3) growth to larger size by condensation. To calculate J 5 , three size bins based on particles were used: D p 1-5, 5-10, and 10-20 nm (Fig. 1). The change in the number concentrations for each size bin can be given by the following equation using Coag.S and the condensation growth rate (Cond) in accordance with Dal Maso et al. (2002).
J 5 = (Coag.S 5-10 + Cond 5-10 ) Cond 5-10 × N 10-20 t + Coag.S 10-20 N 10-20 (8) In this study, growth rate after NPF was not estimated in respective NPF events, as described above. Here, the growth rate (0.01-1 nm h −1 ) and N 10-20 = 0 were assumed to estimate J 5 . Therefore, J 5 is obtainable from Coag.S 5-10 N 5-10 , Cond 5-10 , and Coag.S 10-20 N 10-20 . Details of procedures for estimation of Cond 5-10 were described by Dal Maso et al. (2002). It is noteworthy that primary emissions of ultrafine aerosol particles can engender false estimation of J 5 . Sea-salt particles with less volatility were dominant even among ultrafine aerosol particles with sizes of D p > 10 nm under storm conditions that prevailed during winter-spring at Syowa Station (Hara et al., 2011b). Indeed, a high value of J 5 was identified occasionally in the high contribution of sea-salt particles during winter.

Lognormal fitting of aerosol size distributions
Multi-modal fitting analysis of aerosol size distributions is used commonly to understand and discuss microphysical processes in the atmosphere. In this study, daily mean aerosol size distributions were approximated by a lognormal function, which is given by the following equation.
In Eq. (9), D p , n, D p,i , σ i , and N i , respectively, denote the particle diameter, mode number (n = 1-4), modal size in mode i (i.e., mean diameter of the mode in aerosol number size distributions), modal standard deviation in mode i, and aerosol number concentrations in mode i. Lognormal fitting was performed using the nls.lm function of the minpack.lm library of R (R interface to the Levenberg-Marquardt nonlinear least-squares algorithm found in MINPACK). To avoid unrealistic lognormal fitting, we set the following restrictions: (1) N i > 1 % of total particle concentrations, (2) 1.2 ≤ σ i ≤ 2.2, and (3) 1.3D p,i < D p,i+1 . The daily mean aerosol size distributions (D p = 5-168 nm: SMPS only) were approximated by lognormal functions in 2004. The size distributions (D p = 5-300 nm: SMPS + OPC) were analyzed using lognormal fitting in [2005][2006]. For this study, we defined each mode based on the particle size range as follows: fresh-nucleation mode (D p < 10 nm), aged-nucleation mode (D p = 10-25 nm), first Aitken mode (D p = 25-50 nm), second Aitken mode (D p = 50-100 nm), and accumulation mode (D p > 100 nm).

Backward trajectory analysis
The 120 h (5 d) backward trajectory was computed using the NOAA-HYSPLIT model (Stein et al., 2015; https://www. ready.noaa.gov, last access: 6 August 2021) with the NCEP meteorological dataset (reanalysis) in model vertical velocity mode. The initial point was at 500 m above ground level over Syowa Station, Antarctica, because the trajectory at lower altitudes has high uncertainty as a result of topographic effects and turbulence. According to tethered balloon measurements at Syowa Station (Hara et al., 2011a), the top of the boundary layer over Syowa was identified at 1400 m above ground level and lower (annual mean, 840 m). For this study, we use the following criteria to divide each air mass origin: marine, < 66 • S; coastal, 66-75 • S; Antarctic-continental, > 75 • S; boundary layer (BL), < 1500 m; and FT, > 1500 m. Then, the time passing in each area, such as marine BL (MBL), coastal BL, continental BL, continental FT, coastal FT, and marine FT (MFT), was counted for each backward trajectory. Areas with air masses staying for the longest times in the 5 d backward trajectory were classified into their respective air mass origins.

Results and discussion
3.1 Aerosol size distributions Figure 2 presents examples of number-size distributions of aerosol particles observed at Syowa Station. Our measurements show size distributions of ultrafine aerosol particles with mono-, bi-, tri-, and quad-modal structures. In earlier studies Weller et al., 2015), multimodal aerosol size distributions with mono-modal, bi-modal, and tri-modal structures were identified in the Antarctic. Although quad-modal structures were observed clearly for this study, they were not described in reports of earlier works.
To characterize the aerosol size distributions, we compare the modal size in each mode (Fig. 3). In mono-modal structures, the modal size ranged mostly from 40 to 105 nm. In the bi-modal structure, the modal sizes in the first and second modes were in the ranges of 20-40 nm and 60-135 nm. In the tri-modal structure, the modal sizes in the first to third modes were, respectively, in the ranges of 8-20, 20-63, and 65-135 nm. In the quad-modal structure, the modal sizes in the first to fourth modes were, respectively, in the ranges of 7-13, 14-30, 30-65, and 70-140 nm. The mono-modal structure was observed often under storm and strong wind conditions with blowing snow during winter-early spring (Hara et al., 2011b(Hara et al., , 2020. In these conditions, sea-salt particles in ultrafine to coarse modes were released from snow and seaice surface in polar regions (Hara et al., 2011bFrey et al., 2020). In the bi-modal structure, modal sizes in the modes of the smallest modal size were greater than those in tri-modal and quad-modal structures, so that the bi-modal structure was well aged relative to tri-modal and quad-modal structures. In tri-modal and quad-modal structures, modal sizes in the smallest mode appeared mostly for diameters smaller than 20 nm. As demonstrated by Asmi et al. (2010), Kyrö et al. (2013), Järvinen et al. (2013), Weller et al. (2015), Jokinen et al. (2018), and Kim et al. (2019), aerosol particles were grown to a few tens of nanometers after NPF, even in the Antarctic troposphere during summer. Because the smallest mode appeared with a diameter smaller than 20 nm, occasionally smaller than 10 nm, in tri-modal and quad-modal structures, aerosol size distributions with trimodal and quad-modal structures might be associated with NPF and growth by vapor condensation. In bi-, tri-, and quadmodal structures, the modal sizes with the modes with the largest modal sizes had a similar diameter larger than 50 nm, which corresponded to a critical diameter for CCN activation in the Antarctic (Asmi et al., 2010). Figure 4 shows the seasonal variation of abundance of modal structures at Syowa Station during our measurements. Mono-modal structure was identified mostly during May-August. Abundance of mono-modal structures was found for 16 %-60 % of days (mean, 37 %) during winter. As described above, mono-modal structures during winter-early spring were associated with sea-salt aerosol emissions. Indeed, sea- salt aerosols released from sea-ice areas were dominant during winter-early spring at the Antarctic coasts (e.g., Hara et al., 2012Hara et al., , 2013Hara et al., , 2020Frey et al., 2020). Although bi-modal structures were observed throughout the year, abundance of bi-modal structures occurred as 23 %-76 % (mean, 56 %) in April-September and 9 %-52 % (mean, 26 %) in December-March. Particularly, abundance of mono-modal and bi-modal structures was dominant (more than 90 %) during May-August. High abundance of tri-modal structures (14 %-75 %, mean 45 %) was observed in September-April. Particularly, abundance of tri-modal structures exceeded 50 % in January-March. Surprisingly, tri-modal structures were identified even under dusk and polar night conditions during May-August. Modal sizes in the smallest mode of tri-modal structure were greater in winter than in spring-summer (details are presented in a later section). It is noteworthy that the quad-modal structure was found not only in December-February, but also in August-November and March-April. Considering the modal size in the smallest mode of quadmodal distributions, NPF might proceed in August-April in the Antarctic. Indeed, CN concentrations started to increase in August, with high concentrations in October-February at coastal stations (e.g., Hara et al., 2011a, Weller et al., 2011Asmi et al., 2013).

Relation between modal structures and air mass history
Annual cycles of air mass origins in each modal structure using 120 h backward trajectory analysis are shown in Seasonal features of air mass origins in mono-modal structures (Fig. 5b) resembled the general features (Fig. 5a), although the abundance of air masses from coastal FT was slightly higher in August and October. Considering that mono-modal structures corresponded mostly to storm conditions and strong winds during winter-spring (Hara et al., 2010(Hara et al., , 2011b(Hara et al., , 2020, the appearance of mono-modal structures was associated with primary emissions of sea-salt aerosols from the snow surface on sea ice by strong winds rather than air mass history (i.e., transport pathway), as presented by Hara et al. (2012Hara et al. ( , 2013Hara et al. ( , 2020. Similarly, seasonal features of air mass origins in bi-modal structure (Fig. 5c) resembled the general features. In general, bi-modal structures were recognized as well-aged distributions by condensation growth, coagulation, and cloud processes. Therefore, the appearance of bi-modal structures might be compared only slightly to air mass origins classified by 120 h backward trajectory analysis.
Regarding tri-modal structures (Fig. 5d), the abundance of air masses from continental FT and coastal FT increased in spring, compared to general features (Fig. 5a). Similarly to bi-modal structures, the appearance of some tri-modal structures, particularly with larger modal sizes (e.g., D p > 20 nm), in the smallest mode needed aging processes for a longer time. Consequently, seasonal variations of air mass origins in tri-modal structures were similar to general features, although the sum of abundance of air masses from continental FT and coastal FT exceeded 50 %-60 % in August-October. This abundance was slightly higher than that of general features.
Unlike the features in mono-modal, bi-modal, and trimodal structures, continental FT and coastal FT were the most abundant air mass origins in quad-modal structures during spring and fall (Fig. 5e). In general, features of air mass origins (Fig. 5a), MBL, and coastal BL showed an important contribution during spring and fall. Nevertheless, quadmodal structures in spring and fall were identified only in the air masses from continental FT and coastal FT. This feature implies strongly that NPF proceeded in FT during spring and fall in the Antarctic. In contrast to the high contributions of air masses from continental FT and coastal FT during spring and fall, quad-modal structures were observed also in air masses from MBL and coastal BL during summer. Therefore, NPF might occur also in MBL and coastal BL during summer, as reported from results of earlier works (Weller et al., 2011;Asmi et al., 2013;Lachlan-Cope et al., 2020).
Based on aspects of location where NPFs occur in the Antarctic troposphere, seasonal features of abundance of trimodal and quad-modal structures and their air mass origins are presented for comparison in Fig. 6. The abundance of trimodal and quad-modal structures reflects the frequency of NPF in the Antarctic troposphere, although the appearance of tri-modal and quad-modal structures does not necessarily mean fresh or local NPF events near Syowa in this study. The abundance of tri-modal and quad-modal structures was less than 10 % of days during May-August. In September-January, the abundance of tri-modal and quad-modal structures was 40 %-48 % (mean, 44 %). By contrast, abundance reached 60 % and 84 % in February and March. Seasonal features of NPF occurrence with highs in spring-summer and minima in winter were observed at Concordia  and King Sejong . The monthly occurrence (frequency) of the NPF, however, varied greatly at Syowa, Concordia and King Sejong. Differences in the abundance of NPF occurrence among Syowa, Concordia, and King Sejong might derive from different atmospheric conditions such as the concentrations of aerosols and precursors and different criteria for identification of NPF-growth events.
Abundance of tri-modal and quad-modal distributions with air mass origins of coastal FT and continental FT (Fig. 6) ranged from 14 % to 27 % of days (mean, 22 %) during September-November, from 8 % to 16 % (mean, 11 %) during December-February, and from 11 % to 32 % (mean, 22 %) during March-April. Particularly, fresh NPF (quadmodal structures) was identified in FT (12 % of days) dur-ing September-October and March, in contrast to 1 % in BL. Abundance of quad-modal structures decreased to 3 % in FT during December-February. Considering high abundance of the quad-modal structures in the air masses from FT in September-October (Fig. 5e), spring NPF might occur dominantly in FT. However, abundance of tri-modal and quad-modal structures with air masses of MBL and coastal BL increased (27 %-52 %, mean 41 %) during December-March. Abundance of quad-modal structures in BL increased to 7 % in BL (3 % in FT) during December-February. In addition to NPF in FT, the high abundance in BL during December-March implies that more NPF proceeded in the BL during summer, as reported from earlier works (Koponen et al., 2003;Virkkula et al., 2007;Asmi et al., 2010;Kyrö et al., 2013;Weller et al., 2015;Jokinen et al., 2018;Kim et al., 2019;Jang et al., 2019;Lachlan-Cope et al., 2020). It is noteworthy that the abundance of air masses from continental FT and coastal FT during summer decreased remarkably, even in general features (Fig. 5a). Furthermore, the quad-modal structure was observed in air masses from continental FT and coastal FT in December (summer). CN enhancement by NPF and growth was observed in the lower FT over Syowa Station during summer (Hara et al., 2011a). Therefore, the difference of contributable air mass origins in quad-modal structures between spring-fall and summer might reflect not only the locations of NPF occurrence, but also seasonal features of general air mass origins (Fig. 5a). Consequently, NPF might occur dominantly in FT and partly in BL during spring and fall and in BL and FT during summer. 2− in summer and sea salt (e.g., Na + ) in winter (Fig. 7a). The CH 3 SO 3 − concentrations, which start increasing at the end of August-September, show maximum concentrations in February-March at Syowa Station. Seasonal variation of CH 3 SO 3 − concentrations implies that oceanic bioactivity and atmospheric photochemistry contribute to the maintenance of aerosol systems in September-early April because precursors of CH 3 SO 3 − (e.g., DMS) derive from oceanic bioactivity (e.g., Minikin et al., 1998;Enami et al., 2016).

Seasonal variations of aerosol physicochemical properties
Variations of CN concentrations (Fig. 7b) exhibited clear seasonal features with a minimum in winter and a maximum in summer. During winter, CN concentrations increased occasionally under storm and strong wind conditions. Increase in CN concentration from the winter minimum started at the end of August-early September at Syowa. CN concentrations and seasonal variations were similar to those measured at other coastal stations (e.g., Weller et al., 2011;Fiebig et al., 2014).
Our measurements show size distributions of ultrafine aerosol particles with mono-, bi-, tri-, and quad-modal structures (Fig. 2). The presence of tri-modal and quad-modal structures during spring-fall suggests a frequent occurrence of NPF and growth in the atmosphere. Indeed, freshnucleation and aged-nucleation modes (D p : < 10 nm and 10-25 nm) appeared often in spring-summer (Fig. 7c). Particularly, the fresh-nucleation mode appeared most frequently at the end of August-early October and February-March. Furthermore, the modal sizes increased gradually from freshnucleation mode to the first Aitken mode (D p = 25-50 nm) during September-December. Such a gradual size shift implies that aerosol particles derived from NPF grew by the condensation of condensable vapors such as H 2 SO 4 .
Aerosol number concentrations of D p = 5-10 nm (N 5-10 ) show three maxima in September-October, December, and February-March (Fig. 7d), when tri-modal and quad-modal structures appeared frequently. These periods, respectively, correspond to the existence of the ozone hole (September-October marked by grey-shaded bands), the maximum of solar radiation (summer solstice marked by blue-shaded bands), and the minimum of the sea-ice extent (February-March marked by green-shaded bands). In general, N 5-10 was controlled by (1) NPF and growth from smaller size (ca. 1 nm), (2) coagulation loss, and (3) particle growth to larger sizes by vapor condensation. Additionally, J 5 showed three maxima similar to N 5-10 variation (Fig. 7f). The estimated J 5 values from this study are comparable to the formation rates measured and estimated at Aboa (Kyrö et al., 2013), Neumayer , and Concordia , although a high formation rate was obtained in the Antarctic Peninsula . During this study, high values of J 5 were observed often during May-August. The high J 5 values corresponded to high CN concentrations and storm conditions during winter. Aerosol volatility measurements showed clearly that less volatile particles (i.e., seasalt particles) were distributed in size ranges of D p > 10 nm at Syowa during winter (Hara et al., 2011b). Indeed, sea-salt aerosols in ultrafine mode can be formed through sublimation of saline snow particles released from sea-ice areas (Yang et al., 2019). High values of J 5 during winter might derive from dispersion of sea-salt aerosols from sea-ice areas by strong winds (Hara et al., 2011b). Therefore, high J 5 in the spring and summer are expected to be related to NPF.
The high values of J 5 and N 5-10 in December and February-March are explainable by photochemical reactions that form condensable vapors and by the source strength of aerosol precursors released from oceanic bioactivity near the were calculated using Na + concentrations and molar ratios in seawater (SO 4 2− /Na + = 0.0602; Millero et al., 2008) during November-March and ratios in sea salts (SO 4 2− /Na + = 0.01; Hara et al., 2012Hara et al., , 2018 in April-October because the SO 4 2− /Na + ratio is changed by sea-salt fractionation on sea ice during April-October (Hara et al., 2012(Hara et al., , 2018. The ozone hole extent and sea-ice extent data were provided, respectively, by NASA (https://ozonewatch.gsfc.nasa.gov/, last access: 6 August 2021) and the National Snow and Ice Data Center (https://nsidc.org/data/seaice_index, last access: 6 August 2021). Daily mean UV data at Syowa Station were monitored by the Japan Meteorological Agency (http://www.jma.go.jp/jma/index.html, last access: 6 August 2021) using a Brewer spectrophotometer. sea-ice margin. Indeed, high concentrations of DMS released from oceanic bioactivity were observed near sea-ice margins (Koga et al., 2014). By contrast, the sea-ice extent showed a maximum value in spring (September-November, Fig. 7g). Consequently, other factors affecting the likelihood of high J 5 and N 5-10 in spring should be considered. To elucidate the presence of fresh-nucleation mode in spring, the location at which NPF occurred in the Antarctic region must be discussed. Air masses with quad-modal structures with freshnucleation mode originated mostly from the upper FT over the Antarctic continent (Fig. S1 in the Supplement). Coag.S of D p = 5 nm aerosol particles increased gradually from 2 × 10 −6 to 10 −5 s −1 during September-December (Fig. 7e). From Coag.S, we estimated the e-folding time of the particles (D p = 5 nm) by coagulation loss as 3.9-5.8 d in early September and 1.2-1.4 d in December (Fig. S2 in the Supplement). This finding implies that the particles in the freshnucleation mode observed at Syowa had too short a lifetime to be supplied by long-range transport from the mid latitudes via the FT. Therefore, NPF can be expected to start in the Antarctic FT already by the end of August.
Generally speaking, NPF is more likely to occur under conditions with (1) lower number concentrations of preexisting particles, (2) higher concentrations of condensable vapors, and (3) presence of sufficient photochemical oxidants such as OH. Because of low aerosol number concentrations in the Antarctic FT (Hara et al., 2011a), NPF can proceed preferentially in the FT if condensable vapor is present. Earlier studies presented that the following condensable vapors participate in tropospheric NPF: H 2 SO 4 , CH 3 SO 3 H, NH 3 , iodine species such as HIO 3 , and amines (Yu et al., 2012;Kulmala et al., 2013;Kyrö et al., 2013;Weller et al., 2015;Sipilä et al., 2016;Jen et al., 2016;Jokinen et al., 2018;Shen et al., 2019;Burrell et al., 2019). Actually, these vapors were identified along the Antarctic coasts and the Southern Ocean (Jefferson et al., 1998a, b;Sipilä et al., 2016;Jokinen et al., 2018;Dall'Osto et al., 2019). High contributions of photochemical processes in the atmosphere were required for an important relation among J 5 , aerosol size distributions, and CN concentrations. In addition, condensable vapors can be removed by condensation onto pre-existing particles. Therefore, the condensable vapors must be formed in FT if NPF occurs in FT. Particularly, H 2 SO 4 , CH 3 SO 3 H, and iodine species such as HIO 3 are converted from precursors via photochemical reactions (e.g., Enami et al., 2016;, 2014. Direct measurements of these vapors and their precursors in the Antarctic FT have never been reported. Atmospheric iodine cycles are related closely to snowpack chemistry in the Antarctic (e.g., Atkinson et al., 2012;Roscoe et al., 2015;Saiz-Lopez et al., 2015;Hara et al., 2020). Because of fast reactions of reactive iodine species (I x O y ; Saiz-Lopez et al., 2008;Atkinson et al., 2012;Roscoe et al., 2015), atmospheric chemical processes of I x O y proceed dominantly near the surface (below 100 m). Hara et al. (2014) showed that significant injection of sea-salt aerosols originated from sea-ice areas into FT by cyclone activity. In addition, sea-salt aerosols originated from sea-ice area were distributed widely in ultrafine-coarse particles at Syowa during winter-spring (Hara et al., 2011b(Hara et al., , 2013(Hara et al., , 2020. Considering that sea salts in snow and frost flowers on seasonal sea ice had iodine enrichment by sea-salt fractionation (Hara et al., 2017), the rapid vertical mixing of sea-salt particles as shown by Hara et al. (2014) plays an important role in supplying iodine (mostly iodide) in aerosols to FT. Heterogeneous reactions on sea-salt particles containing iodine during transport can engender I x O y release and formation of the iodine condensable vapors such as HIO 3 . High concentrations of sea-salt particles, however, might prevent NPF by the iodine condensable vapors in FT because of efficient condensation losses of vapors onto sea-salt aerosols.
DMS measurements taken at Concordia suggest that DMS was transported from coastal areas (or oceans) to the inland station via FT (Preunkert et al., 2008). Therefore, photochemical reactions with DMS, H 2 SO 4 , and CH 3 SO 3 H might occur in the FT over the Antarctic continent. Sufficient photochemical oxidants such as OH and UV radiation are necessary for the conversion of condensable vapors such as DMS and SO 2 . Appearance of the O 3 hole in the Antarctic stratosphere can engender UV enhancement with wavelengths shorter than 310 nm, even in the Antarctic troposphere during September-November. Therefore, atmospheric OH in the Antarctic troposphere is producible by UV radiation (λ ≤ 310 nm) (Matsumi et al., 2002;Matsumi and Kawasaki, 2003).
Polar sunrise in the upper FT occurs earlier than it does near the surface. Additionally, the existence of the ozone hole enhances UV radiation, even in the troposphere, during September-November (Fig. 7g-h). More noteworthy is the higher UV of wavelengths shorter than 305 nm in October-November than in December near the surface at Syowa. For example, the monthly mean amounts of UV radiation of 300-305 nm at Syowa were, respectively, 0.080, 0.098, and 0.068 kJ m −2 in October, November, and December, respectively. In general, light intensity including UV radiation is greater at higher altitudes because of atmospheric scattering and absorption. Considering the greater amount of UV radiation in the upper troposphere during September-November, the existence of the ozone hole might enhance the formation of photochemical oxidants (e.g., OH) and condensable vapors, which then engender NPF in the upper troposphere ( Fig. 7c-d).

Vertical variations of aerosol e-folding time by coagulation loss
After NPF, aerosol particles in fresh-nucleation mode are grown mainly by condensation of condensable vapors. They are coagulated efficiently onto pre-existing particles. For a better understanding of the aerosol life cycle including mi-crophysical processes in the Antarctic troposphere, aerosol lifetime must be discussed before discussion of the relation between NPF and CCN ability. Williams et al. (2002) reported that aerosol particles of D p < 60 nm were removed dominantly by coagulation in FT. Therefore, vertical variations of e-folding time were estimated as aerosol lifetime in the Antarctic FT from Coag.S in each size (Figs. 7e and S2). Aerosol size distributions in D p < 100 nm remain unknown in the Antarctic FT through the year. Here, the e-folding time in FT was estimated, assuming that aerosol mixing ratio of aerosol number in respective size bins in FT was equal to that measured at the surface. Aerosol size distributions in the air masses from coastal FT and continental FT and monthly mean vertical profiles of air temperature measured by the Japan Meteorological Agency (URL: http://www.jma.go.jp/ jma/indexe.html, last access: 6 August 2021) were used for this estimation. Figure 8 depicts vertical variations of the e-folding time of aerosol particles with sizes of D p = 1-50 nm in September-December. In general, smaller aerosol particles have a greater diffusion coefficient, so that the e-folding time was shorter, particularly at D p < 10 nm. In other words, aerosol particles larger than D p = 20 nm had a longer e-folding time. Because of lower air temperature at higher altitudes, the estimated efolding time tended to increase at higher altitudes for all sizes and months. The e-folding time at the upper troposphere (ca. 8.3 km, 300 hPa) was 3-4 times longer than that at the surface. When aerosol particles were grown to sizes larger than D p = 20-30 nm, the e-folding time reached > 30 d in the middle and upper FT. In estimation of aerosol lifetime by Williams et al. (2002), the aerosol lifetime was approximately 35 d in the upper FT and approximately 15 d in the middle FT in the INDONEX campaign. It must be noticed that these lifetimes were calculated by Williams et al. (2002) from aerosol concentrations higher than those at the surface of Syowa and that the e-folding time in Fig. 8 was estimated based on a constant mixing ratio and based on aerosol size distributions measured at the surface. Tethered balloon measurements at Syowa Station (Hara et al., 2011) presented a vertical gradient of CN concentrations that was lower in FT (up to ca. 2500 m) except for an aerosol-enhanced layer in the lower FT. Additionally, aerosol number concentrations of particles with D p > 300 nm in the upper FT over Syowa were 1-2 orders lower than those at the surface (Kizu et al., 2010). With a vertical gradient of the number concentrations of preexisting particles, the e-folding time of 1-2 months might be available and realistic for aerosol particles with sizes of D p > 20-30 nm in the Antarctic FT.
In contrast to the longer e-folding time in FT, the e-folding time was shorter in BL (below 850 hPa) because of efficient coagulation loss. Particularly, the e-folding time at the surface was shorter in air masses originating from MBL and coastal BL than that in FT (not shown) because of high concentrations by primary emission of sea-salt aerosols in strong winds and NPF in MBL and coastal BL. In other words, a high growth rate (i.e., high concentrations of condensable vapors) was necessary for particle growth for shorter periods (within the e-folding time) in BL. For example, under conditions with growth rates of 0.05, 0.1, 0.2, and 0.5 nm h −1 , it takes approximately 3.3, 1.7, 0.8, and 0.33 d, respectively, for initial growth from D p = 1 to 5 nm. Under the slow growth conditions, the time for initial growth exceeded the e-folding time of D p = 1 and 5 nm in BL. In FT, however, aerosol particles can continue to exist for a longer time under conditions with slower particle growth (lower concentrations of condensable vapors) than those in BL. Consequently, aerosol particles can be grown gradually through condensation of condensable vapors in the Antarctic troposphere once new particles grow to the aged-nucleated particles (D p > 10 nm). Finally, UV enhancement in the upper troposphere by the ozone hole during the end of August through November might modify the aerosol population and size distributions in the Antarctic troposphere during spring-summer. Therefore, aerosol properties in the Antarctic troposphere during spring-summer are not "pristine" under Anthropocene conditions, although the Antarctic troposphere remains the cleanest on Earth.

CCN potential and cloud amounts during summer
Gradual change in modal sizes in fresh-nucleation mode and first Aitken mode were observed during spring-summer (Fig. 7c). When aerosol particles derived from NPF during periods with the existence of an O 3 hole grow to a critical diameter, the aerosol particles can act as CCN. Here, aerosol particles grown after NPF during periods with O 3 holes were specifically examined to elucidate their relation to CCN ability and CCN properties during summer. As shown in Fig. 7c, modal sizes increased gradually in aged-nucleation mode and first Aitken mode from the end of August through December. This gradual feature of the modal sizes was likely to be associated with particle growth after NPF. Measurements of aerosol hygroscopicity at Aboa showed the critical diameter to be approximately ca. 50 nm during summer (Asmi et al., 2010). Considering gradual particle growth in the agednucleation and first Aitken modes (Fig. 7c), NPF during the periods with O 3 hole and growth are expected to be linked to cloud properties during summer. Therefore, we attempt to estimate the contributions of aerosols derived from NPF during periods with the O 3 hole to act as CCN in November-January based on the following assumptions: aerosol particles supplied from spring NPF were grown to aged-nucleation and first Aitken modes. For this estimation, the contribution to CCN (R CCN ) was defined as the ratio of the aerosol number concentrations of D p > 50 nm in aged-nucleation and first Aitken modes (N D>50 ) to the total aerosol number concentrations in D p > 50 nm (total N D>50 , Fig. 9). The value of R CCN was estimated using the following equation. Figure 10 depicts R CCN in the continental FT and coastal BL in November-January. Results obtained using an ANOVA test indicated a significant difference (p = 0.0125) between the values of R CCN obtained in the continental FT and coastal BL. Differences in aerosol size distributions are regarded as indicating R CCN in the continental FT and coastal BL. As shown in Fig. S3 in the Supplement, the modal size in the smallest mode was smaller in BL than those in FT in December-January, although the modal size was not significantly different in November. Aerosol particles of D p < 30 nm were supplied by NPF and growth, as described above. Therefore, NPF in the BL in December-January might engender a difference in modal sizes in the smallest mode. The modal size in the smallest mode in BL was smaller than the critical diameter. Therefore, aerosol particles supplied from summer NPF in the BL were not grown sufficiently to the critical diameter by condensation. We infer that aerosol enhancement by NPF during periods with the existence of an O 3 hole might modify not only aerosol size distributions, but also CCN ability and cloud properties in summer.
Trends of cloud amounts at Syowa Station were examined to assess the effects of spring NPF enhanced by the existence of the O 3 hole on CCN ability and cloud properties during summer. Figure 11 depicts variations of cloud amounts in December and January that have been recorded since 1969. No increasing trend was found for air temperature or water vapor concentrations related to the cloud amount, but a marked increasing trend with p values lower than 0.01 was found. Nevertheless, it was found only for December and January during 1969-2012. Particularly, cloud amounts at Syowa Station were significantly greater during the existence of an O 3 hole (Figs. 11b and S4 in the Supplement). Consequently, the cloud amount trend during summer might result from spring NPF enhanced by O 3 hole existence. Because of the conformable relation between the cloud amount and radiative fluxes (Yamanouchi et al., 2007), UV enhancement by the O 3 hole can affect atmospheric radiation budgets by aerosol and cloud properties in the Antarctic during summer.
Our hypothesis can be summarized as shown in the schematic figure (Fig. 12). In polar sunrise, solar radiation re-  covers earlier in the stratosphere and the upper troposphere. Then, ozone depletion by catalytic reactions of chlorine cycle starts in the Antarctic stratosphere. The ozone hole appears in the Antarctic stratosphere from the end of August until the end of November. Appearance of the ozone hole engenders UV enhancement and then production of atmospheric oxidants such as OH in the troposphere. Atmospheric oxidants such as OH are likely to be formed in the upper troposphere by UV enhancement. Condensable vapors (i.e., aerosol precursors with lower vapor pressure) are producible by photochemical oxidation. H 2 SO 4 and CH 3 SO 3 H are important condensable vapors for NPF and growth in the Antarctic FT. In addition, NPF occurs already by the end of August in the upper troposphere. As a consequence, NPF is enhanced in FT under UV enhancement by the O 3 hole during spring (end of August-November). Aerosol particles in fresh-nucleation mode grow gradually by the condensation of condensable vapors in the FT. Some aerosol particles in fresh-and aged-nucleation modes are transported to the lower troposphere on the Antarctic coasts. With condensation growth, some aerosol particles originating from spring NPF in the FT can be grown to sizes greater than the critical diameter (ca. 50 nm) in the summer. Results show that NPF proceeds even in coastal BL and MBL from the end of spring-summer. In contrast to FT, H 2 SO 4 , CH 3 SO 3 H, and other condensable vapors such as HIO 3 , NH 3 , and amines can engender NPF and growth in BL. Aerosol particles originating from NPF in BL are removed efficiently by coagulation under conditions with higher aerosol concentrations of pre-existing particles than those of FT, except under high growth rate conditions. Consequently, lesser aerosol particles derived from summer NPF in the boundary layer grow up to the critical diameter than from spring NPF in the FT. Our results demonstrate that spring NPF plays an important role in aerosol populations and cloud properties in the summer under conditions that include the existence of an Antarctic ozone hole. Knowledge of the following issues will be necessary for future investigations to test our hypothesis: (1) aerosol size distributions of particles smaller than 100 nm in the FT, (2) chemical forms of condensable vapors and their concentrations in the FT and BL, and (3) their vertical and seasonal variations in the Antarctic.

Conclusions
Aerosol measurements were taken using SMPS and OPC at Syowa Station, Antarctica, in 2004Antarctica, in -2006. Aerosol size distri-  butions were found to have mono-, bi-, tri-, and quad-modal distributions during our measurement period. The monomodal distribution was dominant under strong wind conditions during May-August. The bi-modal distribution was identified through the year. Tri-modal and quad-modal distributions were observed mostly during September-April. Seasonal features of N 5-10 and J 5 imply that NPF identified at Syowa Station was associated with UV enhancement by ozone hole existence in spring, maximum of solar radiation in summer, and minimum of sea-ice extent in February-March. In addition, NPF occurs in FT during spring and fall and in the FT and boundary layer during summer. Additionally, spring NPF and particle growth are linked to cloud properties during summer.
We obtained direct evidence indicating that spring UV enhancement by the ozone hole engendered spring NPF and growth in the Antarctic FT. With ozone hole recovery (Solomon et al., 2016;Kuttippurath et al., 2017), aerosol properties and populations might be modified for the next several decades. Consequently, indirect effects on atmospheric radiation budgets and climate change in the Antarctic regions during the summer can revert to levels that prevailed before the ozone hole existence. More aerosol measurements must be taken in Antarctic regions to monitor these trends and future effects.
Data availability. Data are available by contacting the corresponding author (Keiichiro Hara: harakei@fukuoka-u.ac.jp).
Author contributions. KH, KO, and TY designed aerosol measurements at Syowa Station. KH, KO, and MY conducted wintering aerosol measurements at Syowa Station in 2004-2006. KH and CNH contributed to data analysis including SMPS/OPC data and meteorological data. KH prepared the manuscript and led data interpretation. All the co-authors contributed to discussions about data interpretation and the manuscript.