Atmospheric inorganic nitrogen input via dry , wet , and sea fog deposition to the subarctic western North Pacific Ocean

Aerosol, rainwater, and sea fog water samples were collected during the cruise conducted over the subarctic western North Pacific Ocean in the summer of 2008, in order to estimate dry, wet, and sea fog deposition fluxes of atmospheric inorganic nitrogen (N). During sea fog events, mean number densities of particles with diameters larger than 0.5 μm decreased by 12–78 %, suggesting that particles with diameters larger than 0.5 μm could act preferentially as condensation nuclei (CN) for sea fog droplets. Mean concentrations of nitrate (NO −3 ), methanesulfonic acid (MSA), and non sea-salt sulfate (nss-SO 2− 4 ) in sea fog water were higher than those in rainwater, whereas those of ammonium (NH+4 ) in both sea fog water and rainwater were similar. These results reveal that sea fog scavenged NO −3 and biogenic sulfur species more efficiently than rain. Mean dry, wet, and sea fog deposition fluxes for atmospheric total inorganic N (TIN; i.e. NH+4 + NO − 3 ) over the subarctic western North Pacific Ocean were estimated to be 4.9 μmol m −2 d−1, 33 μmol m−2 d−1, and 7.8 μmol m−2 d−1, respectively. While NO−3 was the dominant inorganic N species in dry and sea fog deposition, inorganic N supplied to surface waters by wet deposition was predominantly by NH +4 . The contribution of dry, wet, and sea fog deposition to total deposition flux for TIN (46 μmol m−2 d−1) were 11 %, 72 %, and 17 %, respectively, suggesting that ignoring sea fog deposition would lead to underestimate of the total influx of atmospheric inorganic N into the subarctic western North Pacific Ocean, especially in summer periods.


Introduction
Atmospheric transport of particulate matter from the continents to the oceans is well recognized as a major pathway for supply of natural and anthropogenic materials to open ocean surface waters, including nutrients (Duce et al., 1991;Jickells, 1995).Previous studies have highlighted the significance of the atmosphere as a pathway for transport of essential nutrients for biological growth such as nitrogen (N), from continents to marine surface waters and its critical role in oceanic biogeochemical cycling (e.g.Duce et al., 1991Duce et al., , 2008;;Prospero et al., 1996;Paerl, 1997;Galloway et al., 2004;Dentener et al., 2006;Baker et al., 2010;Kim et al., 2011;Jung et al., 2011).
While numerous studies have dealt with the input of nutrients via atmospheric deposition, especially dry and wet deposition, relatively little is known about the deposition flux of atmospheric constituents by fog (Lange et al., 2003).Scavenging processes of water-soluble gases (e.g.HNO 3 , NH 3 , and SO 2 ) and aerosols in the atmosphere by fog events are determined by the properties of ionic compositions in fog water and by the growth rate of fog droplets during fog events (Aikawa et al., 2007).The chemical compositions of the particles acting as condensation nuclei (CN) determine the initial compositions of the fog droplets, which can be further altered by uptake of water-soluble gases and by aqueous phase chemical reactions (Sasakawa et al., 2003;Raja et al., 2008).In fog, the condensation of water vapor on pre-existent particles in the boundary layer shifts the aerosol size distribution towards larger sizes and accelerates their removal from the atmosphere (Jacob et al., 1984;Sasakawa et al., 2003;Herckes et al., 2007;Li et al., 2011).The deposition of fog can contribute significantly to the hydrologic, pollutant, and nutrient cycles in coastal and mountainous regions, since it is an important transfer process for water and various inorganic and organic substances from the atmosphere to the biosphere (Lovett et al., 1982;Jacob et al., 1984;Collett et al., 2001;Zhang and Anastasio, 2001;Klemm and Wrzesinsky, 2007).However, quantifying fog deposition flux for atmospheric nutrient or pollutant and assessing its impact are still a challenge in atmospheric science and ecosystem research (Klemm and Wrzesinsky, 2007).
Considerable effort has been devoted to investigating the chemical and physical properties of fog in valleys, mountains, and urban areas (e.g.Collett et al., 2001;Burkard et al., 2002;Collett et al., 2002;Moore et al., 2004;Lu et al., 2010;Li et al., 2011).However, sea fog has not been extensively investigated (e.g.Sasakawa andUematsu, 2002, 2005;Sasakawa et al., 2003); although it may stimulate phytoplankton growth over the oceanic regions where sea fog occurs frequently and atmospheric nutrients derived from natural and anthropogenic sources are transported and/or affected (Sasakawa et al., 2003).It is therefore necessary to clarify the scavenging process of atmospheric nutrients by sea fog and to quantify their deposition flux to the sea surface.
Rapid growth in human population and industrial activity have led to increases in the concentrations of reactive N species throughout the environment (Galloway et al., 2008).In particular, the increase in nitrogen oxides (NO x ) emissions in eastern Asia has been dramatic over the last decade (Akimoto, 2003;Uno et al., 2007).The western North Pacific receives a large influx of mineral particles and pollutants from the Asian continent through atmospheric long-range transport (Uematsu et al., 1983(Uematsu et al., , 2010;;Gao et al., 1992;Nakamura et al., 2005).Accordingly, estimating deposition flux of atmospheric N and evaluating its impact on marine biogeochemical cycles over the western North Pacific have become increasingly important.In addition, the subarctic western North Pacific (>40°N) has a high sea fog frequency, with a maximum of ∼50 % during the summertime period from June to August (Wang, 1985).Nevertheless, no study has been carried out over this region to estimate atmospheric N input via dry, wet, and sea fog deposition simultaneously.This study therefore aims to (1) investigate general characteristics of sea fog, (2) estimate the fluxes of atmospheric N via dry, wet, and sea fog deposition, (3) estimate the contribution of each deposition to atmospheric N input, and (4) evaluate the impact of atmospheric N deposition on the ocean marine ecosystem.In this study, we focus on ammonium (NH + 4 ) and nitrate (NO − 3 ) that are dominant components for N supply to the oceans (Krishnamurthy et al., 2010).The results for atmospheric inorganic N deposition from this study should be valuable for filling the data gap, especially for the atmospheric inorganic N input by sea fog deposition to the subarctic western North Pacific Ocean.

Aerosol collection
A high-volume virtual impactor air sampler (AS-9, Kimoto Electric Co., Ltd.) was used to collect marine aerosols on a Teflon filter (PF040, 90 mm in diameter, Advantec) (Nakamura et al., 2005;Jung et al., 2011).The virtual impactor air sampler utilizes an acceleration nozzle to divide the intake air into the major flow carrying the fine particles (D < 2.5 µm) and the minor flow carrying the coarse particles (D > 2.5 µm).The major air flow is diverted laterally after passing through the acceleration nozzle and the inertia of the coarse particles carries them into a collection probe aligned with the acceleration nozzle.Both particle fractions are then collected on the same filter.Details of the virtual impactor are described in Loo and Cork (1988).The aerosol sampler was put on the front of the upper deck (17 m a.s.l.) of the ship.A wind-sector controller was used to avoid contamination from the ship's exhaust during aerosol sampling.The windsector controller system was configured to allow collection of marine aerosol samples only when the relative wind directions were within plus or minus 100°relative to the ship's bow and the relative wind speeds were over 1 m s −1 during the cruise.The flow rate was approximately 13 m 3 h −1 and the total sampling time was 1−3 days, representing a total sampling air volume of 310-930 m 3 .After sampling, the filter was stored frozen at −24 °C prior to chemical analysis.Deployment blanks (n = 3) were obtained by placing Teflon filters in the aerosol sampler for 5 min on idle systems (i.e.no airflow through the filters) and processed as other aerosol samples.
During the KH-08-2 cruise, number densities of ambient particles in four size fractions of D > 0.3, 0.5, 1.0, and 2.0 µm were also measured continuously using an optical particle counter (KC-01D, Rion Co., Inc).Meteorological variables (i.e.wind speed, wind direction, air temperature, sea temperature, dew point, and relative humidity) were continuously monitored by weather monitoring systems equipped on the research vessel.

Rainwater collection
Rainwater sampling was conducted on an event basis during the cruise.Rain samples were collected with a 36.5 cm diameter polyethylene funnel fitted to a 500 ml polyethylene bottle (Jung et al., 2011).Both the funnel and bottle were first soaked in detergent (Contaminon B, Wako Pure Chemical Industries) for 24 h, then soaked in 1 N HCl for 24 h, before being rinsed at least three times with Milli-Q water (>18 M cm −1 ; Millipore Co.) and finally dried in a drying oven (DK 600, Yamato, Japan) prior to deployment.The rain sampler was put on the front of the upper deck of the ship, and was opened manually just before or as soon as possible after precipitation.During rain collection, the relative wind directions were monitored.If rain occurred when the relative wind directions were outside the ranges of the wind-sector controller, the rainwater was not collected.After collection, the rain sampler was washed thoroughly with Milli-Q water and closed.The rainwater samples were immediately separated into three aliquots.Two of the aliquots were used for measurements of pH (Model 290A, Orion) and conductivity (Model 115, Orion), respectively.When the amount of precipitation was less than 10 ml, pH and conductivity were not measured.The pH and conductivity meters were calibrated before each measurement.Standard pH 4.01 and 7.00 buffer (Thermo Scientific) and conductivity/total dissolved solids (TDS) standard (1413 µS cm −1 , Thermo Scientific) solutions were used for calibrations of the pH and conductivity meters, respectively.As the third aliquot, remaining rainwater was sealed in pre-cleaned 100 ml high-density polyethylene (HDPE) bottles, and stored in a freezer at −24 °C prior to chemical analyses.Procedural blanks (n = 3) for rainwater were collected by pouring 100 ml of Milli-Q water through the clean funnel-bottle assembly.The procedural blanks were also treated as other rainwater samples.

Sea fog water collection
Sea fog water sampling was conducted on an event basis during the cruise.A fog water sampler (FWG-400, Usui Co. Inc.) was used to collect sea fog water and put on the front of the upper deck of the ship.The fog water sampler is composed of a net of Teflon strings (0.5 mm in diameter), a net holder and a 500 ml low-density polyethylene (LDPE) bottle.Prior to deployment, both the net and the bottle were first soaked in the detergent for 24 h, then soaked in a 1N HCl for 24 h, then rinsed at least three times with Milli-Q water and finally dried.The net of Teflon strings and the LDPE bottle were set only during the sea fog occurrence and removed once sea fog event ceased.When the ship sails the sea fog occurrence area, sea fog droplets collide with the strings and drop along the strings into the 500 ml LDPE bottle beneath the strings (Sasakawa et al., 2003).Likewise rain sampling, the relative wind directions were monitored during sea fog water collection, and sea fog water samples were collected only when the relative wind directions were within the ranges of the wind-sector controller.If the relative wind directions changed from within to outside the ranges during sea fog water sampling, the sea fog water was considered contaminated and discarded.After collection, sea fog water samples were immediately separated into three aliquots.Two of the aliquots were used for measurements of pH and conductivity, respectively.The third aliquot was sealed in pre-cleaned 100 ml HDPE bottles, and stored in a freezer at −24 °C prior to chemical analyses.Procedural blanks (n = 5) for sea fog water samples were collected by pouring 100 ml of Milli-Q water through the clean net-bottle assembly.The procedural blanks were also treated as sea fog water samples.
Size distributions of sea fog droplets were measured with a fog monitor (FM-100, Droplet Measurement Technologies).The fog monitor detects the number and size of individual fog droplets with a diameter from roughly 2 µm up to 50 µm by the forward scattering principle, and can classify droplets in up to 40 size classes (Klemm et al., 2005).Liquid water content (LWC) for each of the 40 droplet size classes was computed based on an idealized mean volume of spherical droplets with aerodynamic diameter.Total LWC was obtained from the sum of LWC for all size.Until now, the uncertainties of droplet measurements for the fog monitor have not been considered in most studies.Recently, Spiegel et al. (2012) evaluated the influence of Mie scattering on the droplet size spectra collected with the fog monitor and the droplet losses during sampling with the fog monitor.They also recommended choosing the 40 channel thresholds to reduce the error from Mie scattering, and doing loss calculations for the droplet measurements using an ultrasonic anemometer and other instrument (e.g.Particulate Volume Monitor) for a reference.In this study, the 40 channel thresholds were used for the measurement of sea fog droplets, indicating that it is sufficient for the determination of the total droplet number concentration or the total liquid water content (Spiegel et al., 2012).However, the droplet losses during sampling with the fog monitor were not calculated since an ultrasonic anemometer and other instrument for a reference were not used in this study.
Non sea-salt (nss-) concentrations of some ionic components were calculated by subtracting the component's seasalt-derived (ss-) concentration from its total concentration.In this study, it was assumed that all Na + in aerosols, rainwater and sea fog water were derived from sea-salt.Contributions from sea-salt were calculated from the Na + concentration in aerosols, rainwater or sea fog water using the mole ratio of the component of interest to Na + in seawater (Keene et al., 1986).

Backward trajectory analysis
Air mass backward trajectories (AMBTs) provide a better understanding of air flow and long-range transport of aerosols.In particular, AMBTs have been used to identify the origin of primary aerosols collected far away from their source region (Chiapello et al., 1997).In this study, 7-day AMBTs were calculated from the National Oceanic and Atmospheric Administration (NOAA) GDAS (Global Data Assimilation System) database using the Hybrid Single-Particle Lagrangian Integrated Trajectories (HY-SPLIT) model (NOAA Air Resources Laboratory, http:// www.arl.noaa.gov/ready/hysplit4.html).AMBTs were performed at 500, 1000, and 1500 m above ground level.Chen and Siefert (2004) reported that atmospheric aerosols may not follow the resulting trajectories because of scavenging processes and gravitational settling; however, the AMBTs provide useful background data on airstreams and the potential origins of the source of the sampled air mass.

Dry deposition
Dry deposition fluxes (F d ) were calculated from aerosol concentrations (C a ) in the coarse (c) and fine (f) modes and dry deposition velocities (V d ) for each size mode (Duce et al., 1991;Baker et al., 2007): Here, dry deposition velocities of 2 cm s −1 for coarse mode and 0.1 cm s −1 for fine mode were used since these two values are known to be best estimates based on experimental and model studies (e.g.Duce et al., 1991;Baker et al., 2003;Nakamura et al., 2005).This estimate results in an uncertainty of a factor of 2-3 in the calculated flux, since deposition velocity includes terms for gravitational settling, impaction and diffusion of particles, all of which vary in complex functions of particle size and meteorological conditions (e.g.wind speed and relative humidity) (Duce et al., 1991).

Wet deposition
Wet deposition fluxes (F w ) were estimated from the concentration of the species of interest in rainwater (C r ) and the precipitation rate (P ) (Baker et al., 2010): The precipitation rate was calculated from the monthly averaged precipitation rate (mm d −1 ) using the CMAP model output (http://www.cdc.noaa.gov/cdc/data.cmap.html)(Xie and Arkin 1997).Similar to estimates for dry deposition flux, the choice of precipitation rates based on limited data causes the greatest uncertainty in wet deposition flux estimates, particularly in the open ocean (Spokes et al., 2000).However, Baker et al. (2010 and references therein), who used the same method for estimating wet deposition flux, argued that the uncertainty arising from selection of precipitation rate is minor since the precipitation rate data agreed relatively well with other studies in terms of total rainfall amount.

Sea fog deposition
Sea fog deposition fluxes (F f ) were estimated by multiplying sea fog water flux (F fw ) of each event by the concentration of the species of interest in each sea fog water sample (C f ) (Thalmann et al., 2002): The F fw was estimated by Eq. ( 4), where LWC(D p ) represents the mean LWC for each of the 40 droplet size classes during each sea fog event and V (D p ) indicates the deposition velocity for sea fog droplets with an aerodynamic diameter (Klemm et al., 2005).The sum of sea fog water flux for all size classes then yielded the F fw .
For V(D p ), the modeled values reported by Matsumoto et al. (2011), who calculated the deposition velocities for the particles with diameters larger than 3 µm, were and rain events, respectively.5 6 Fig. 2. Temporal variations of meteorological variables (i.e.wind speed, wind direction, air temperature, sea surface temperature, dew point, and relative humidity) during the cruise.Sea fog (red circle) and rain (violet triangle) events indicate the occurrences of observed sea fog and rain events, respectively.used (i.e.4.69 cm s −1 for 3 < D < 5 µm, 10.1 cm s −1 for 5 < D < 10 µm, 13.5 cm s −1 for 10 < D < 20 µm, 15.6 cm s −1 for 20 < D < 30 µm and 19.0 cm s −1 for 30 < D < 50 µm).This ambiguity in values used for sea fog deposition velocities leads to the greatest uncertainty in sea fog flux estimates.
In this study, it was assumed that all sea fog droplets measured with the fog monitor were deposited to the sea surface without changes in size distribution of LWC; however, due to evaporation and coalescence of sea fog droplets, their size and deposition velocities are subject to change, suggesting that the deposition velocities used here leads to the uncertainty in sea fog deposition flux estimates.The fog water sampler used in this study has a 50 % efficiency collection diameter of 6 µm under flow rate 3 m s −1 (Minami and Ishizaka, 1996).It is difficult, however, to calculate the precise collection efficiency of this fog sampler in this study, because the relative wind directions and the relative wind speeds change extremely with the movements of the ship (Sasakawa and Uematsu, 2005).Hence, the estimates of sea fog deposition fluxes in this study contain the uncertainties that are related to the changes in size distribution of LWC and the collection efficiency.

Meteorological conditions associated with sea fog occurrences
Sea fog typically occurs as a result of cooling of humid air over a cold ocean surface (Lewis et al., 2004;Eugster, 2008).During the sampling period, sea fog occurred pre-dominantly when the dominant wind direction was southerly and/or southwesterly and air temperature dropped to its dew point (Fig. 2).This result indicates that the warm and humid air masses from the low and middle latitudes of the North Pacific Ocean passed over the cold sea surface of the subarctic western North Pacific Ocean and they were cooled down to a saturation temperature.Sea fog events, however, did not occur when sea surface temperature was higher than air temperature, although the dominant wind direction was southerly, suggesting that the difference between the air and sea surface temperatures is a key factor controlling sea fog formation (Cho et al., 2000).
According to previous studies (Cho et al., 2000;Fu et al., 2006;Tokinaga and Xie, 2009), the difference between air and sea surface temperatures is often observed to be positive in frequent sea fog occurrence regions, since the relatively cold sea surface temperature stabilizes the lower atmosphere, making a favorable condition for sea fog formation.Therefore, the meteorological conditions during the sea fog sampling period show that advection of warm and humid air masses from the subtropical North Pacific Ocean and the positive difference between air and sea surface temperatures make favorable conditions for sea fog occurrence over the subarctic western North Pacific Ocean.

Effect of sea fog on particle number density
Temporal variations of total LWC, particle number densities for aerosols in the range of 0.3-2.0µm, and fog droplet size distribution for each of the 40 droplet size classes during the cruise are shown in Fig. 3.The total LWC varied from < 0.2-140 mg m −3 .Mean particle number densities during non sea fog events were 25 cm −3   and rain events occurred simultaneously, only rainwater sample was collected.Then that 6 sample was considered as a rainwater sample (i.e.rainwater sample number 3 and 4).Since 7 rainwater sample number 1 was collected on 30 July, it was not presented here.8 9 Fig. 3. Temporal variations of total LWC, particle number densities for aerosols in four size groups, and LWC for each of the 40 droplet size classes during the cruise.The green and pink shaded areas indicate sea fog water and rainwater sampling durations, respectively.The black and pink numbers indicate sea fog and rainwater sample number, respectively.When sea fog and rain events occurred simultaneously, only rainwater sample was collected.Then that sample was considered as a rainwater sample (i.e.rainwater sample number 3 and 4).Since rainwater sample number 1 was collected on 30 July, it was not presented here.
for aerosols in the range of 0.3 < D < 0.5 µm, 2.6 cm −3 for 0.5 < D < 1.0 µm, 0.53 cm −3 for 1.0 < D < 2.0 µm, and 0.17 cm −3 for D > 2.0 µm.In comparison, the mean particle number densities during sea fog events decreased by 4 % (mean particle number density 24 cm −3 ) for aerosols in the range of 0.3 < D < 0.5 µm, 12 % (2.3 cm −3 ) for 0.5 < D < 1.0 µm, 55 % (0.24 cm −3 ) for 1.0 < D < 2.0 µm, and 78 % (0.038 cm −3 ) for D > 2.0 µm.The differences   of mean particle number densities between two periods were statistically significant (Wilcoxon Signed Rank Test, p = 0.0005), except that for aerosols in the range of 0.3 < D < 0.5 µm.In addition, the sea fog droplet size distributions during sea fog events were shifted towards the larger droplet sizes.These results suggest that particles with diam-eters larger than 0.5 µm could act preferentially as CN for sea fog droplets (Sasakawa et al., 2003), and that the formation of liquid droplets by condensation of water vapor on pre-existed particles acting as CN leads to the acceleration of particle removal from the atmosphere (Pandis and Seinfeld, 1990).

Chemical composition of aerosols, rainwater, and sea fog water
Sea-salts (Na + , Cl − , and ss-ions) were the dominant components in aerosols, rainwater, and sea fog water, representing approximately 72 %, 61 %, and 86 % of total ionic concentration, respectively (Fig. 4 and Table 1).In aerosols and sea fog water, the mole equivalent ratios of Cl − /Na + were slightly lower than that in seawater (Table 2), suggesting that Cl − depletion occurred through the volatilization of hydrogen chloride (HCl) from sea-salt particles that became acidified by the incorporation of nitric (HNO 3 ) and/or sulfuric (H 2 SO 4 ) acids in the marine atmosphere (Graedel and Keene 1995;Andreae and Crutzen 1997), and that the acidified sea-salt particles acted as CN of sea fog droplets (Sasakawa and Uematsu, 2002;Raja et al., 2008).Furthermore, the Mg 2+ /Na + , K + /Na + , and Ca 2+ /Na + ratios in aerosols and sea fog water were similar or slightly higher than those in seawater, suggesting that most of Mg 2+ , K + and Ca 2+ in aerosols and sea fog water were derived from sea-salt particles.
Two volcanoes on the Aleutians erupted during the cruise.The eruption at Okmok volcano (53.40°N, 168.17°W) started on 12 July 2008 and ended in late August 2008 (Larsen et al., 2009;Lu and Dzurisin, 2010), and Kasatochi volcano (52.18°N, 175.51°W) became active on 7 August 2008 (Schmale et al., 2010).During the sampling period, air masses originated from the Asian continent and the Kamchatka Peninsula, indicating that these air masses were likely affected by anthropogenic and crustal sources as well as the eruptions of two volcanoes (Fig. 5a-c).
Unlike aerosol and sea fog water samples, the Cl − /Na + , Mg 2+ /Na + , K + /Na + , and Ca 2+ /Na + ratios in rainwater were much higher than those in seawater (Table 2).Atmospheric HCl are derived from sea-salt particles, volcanoes, and anthropogenic activities (e.g.fossil fuel combustion and incineration) (Graedel and Keene, 1995).Gioda et al. (2011) observed high Cl − /Na + ratios in rainwater (2.2) and cloud water (3.2), collected in Puerto Rico from December 2004 to March 2007, when ash from the Soufriere Hills volcano reached the sampling site.During the collection of rainwater samples, air masses originated from the Asian continent or the subtropical western North Pacific Ocean and thereafter swept over large regions of the Korean Peninsula and/or the Japanese Islands, indicating that these air masses were most likely affected by strong anthropogenic and crustal sources rather than by the influences of two volcanoes (Fig. 5b).The high Cl − /Na + ratio in rainwater thus is likely due to scavenging of HCl derived from sea-salt particles and/or anthropogenic source by rainwater.Sasakawa and Uematsu (2002) reported that NH + 4 /nss-Ca 2+ ratio in rainwater (0.53) collected over the northwestern North Pacific (34°N-48°N, 137°E-155°E) from 15-29 July 1998, was two orders of magnitude lower than that in sea fog water (11), reflecting that mineral particles, such as calcium carbonate (CaCO 3 ), mainly existed over higher altitudes than those where sea fog appeared.While sea fog occurs near the sea surface (Fu et al., 2006) and scavenges only lower atmospheric substances (Ali et al., 2004), precipitation removes the substances existing in the whole air column in and below clouds (Deboudt et al., 2004).The high Mg 2+ /Na + , K + /Na + and Ca 2+ /Na + ratios in rainwater therefore suggest that most of these ionic species in rainwater were derived from other sources, such as crustal materials and biomass burning, and that scavenging processes of aerosols by sea fog are different to those by rain.

The pH of rainwater and sea fog water
The pH values of rainwater and sea fog water varied from 3.5-4.5 and 3.4-5.9,with averages of 4.1 and 4.2, respectively (Table 1).Several compounds, such as H 2 SO 4 , HNO 3 , HCl, NH 3 , and CaCO 3 , contribute to the acid-base balance of rainwater and fog water (Millet et al., 1996).
The mean nss-Cl − concentration in rainwater was two orders of magnitude greater than those of nss-SO 2− 4 and NO −  (Table 1), suggesting that nss-Cl − exerted a larger influence on acidity of rainwater collected over the subarctic western North Pacific Ocean.Assuming the nss-SO 2− 4 , NO − 3 , and nss-Cl − in rainwater existed in the form of free acids, the expected pH of rainwater was 3.4, whereas the mean pH value measured in rainwater was 4.1.This discrepancy indicates that rainwater had experienced some neutralization.From the difference between the sum of nss-SO 2− 4 , NO − 3 , nss-Cl − , and the mean H + concentration obtained from the mean pH value, it was estimated that approximately 61 % of these acidic substances was in neutralized forms.The mean concentrations of nss-Ca 2+ and nss-Mg 2+ in rainwater were 15 times and 3.8 times higher than that of NH + 4 , respectively (Table 1).Moreover, the mole equivalent ratios of NH + 4 /nss-Ca 2+ and NH + 4 /nss-Mg 2+ in rainwater were lower than 1, suggesting that nss-Ca 2+ and nss-Mg 2+ played key roles in neutralization of rainwater acidity.
In sea fog water, nss-SO 2− 4 and NO − 3 were the dominant acidic species (Fig. 4 and Table 1).The mole equivalent ratio of NO − 3 /nss-SO 2− 4 in sea fog water was 0.57 (Table 2).This result suggests that nss-SO 2− 4 was the major component to lower the pH of sea fog, and that the pH of sea fog water was controlled by nss-SO 2− 4 derived mainly from marine biological activity than that of rainwater since mean concentrations of methanesulfonic acid (MSA) and nss-SO 2− 4 in sea fog water were an order of magnitude greater than those in rainwater (Table 1).For sea fog water, it was estimated that approximately 48 % of nss-SO 2− 4 and NO − 3 was in neutralized forms.While nss-Ca 2+ and nss-Mg 2+ were the dominant neutralization substances in rainwater, NH + 4 was the major basic component in sea fog water (Fig. 4 and Table 2), suggesting that neutralization of sea fog water was predominantly caused by NH + 4 , and that nss-SO 2− 4 and NO − 3 in sea fog water were probably in fully or partially neutralized forms, such as ammonium sulfate ((NH 4 ) 2 SO 4 ), ammonium bisulfate (NH 4 HSO 4 ) and ammonium nitrate (NH 4 NO 3 ).
Table 4. Mean dry, wet, sea fog, and total deposition fluxes of NH + 4 and NO − 3 , and the contribution of each deposition to total inorganic N input in the subarctic western North Pacific Ocean during the cruise.

Dry deposition
Wet deposition Sea fog deposition Total deposition a TIN represents total inorganic nitrogen.In this study, total inorganic nitrogen is defined as including NH + estimated to be 1.9 µmol m −2 d −1 and 3.0 µmol m −2 d −1 , respectively.Although the mean concentration of total NH + 4 in aerosols collected over the subarctic western North Pacific Ocean was approximately 2 times higher than that of total NO − 3 (Table 1), inorganic N supplied to surface waters by atmospheric dry deposition was mainly from NO − 3 that was largely associated with coarse mode particles, since fluxes to the ocean are dominated by the coarse mode, resulting in NO − 3 being deposited much more rapidly (Fig. 6a and b).In this study, dry deposition fluxes of gaseous inorganic N species such as NH 3 and HNO 3 were not estimated, resulting in the underestimate of dry deposition fluxes.Due to the lack of observational data for gaseous inorganic N species over the western North Pacific Ocean, we used the results of Zhang et al. (2010) to estimate the contributions of gaseous inorganic N to dry deposition fluxes, who reported that dry deposition fluxes for NH 3 and HNO 3 were estimated to be 2.0 µmol m −2 d −1 and 4.7 µmol m −2 d −1 respectively over the East China Sea in July using the MM5/CMAQ model with the 2004 national emission inventory of China.Assuming that the same amounts of NH 3 and HNO 3 are deposited over the sampling region of this study, the contribution of dry deposition flux to mean total (dry + wet + sea fog) deposition flux for TIN would increase from 11 % to 23 % (Table 4).However, it is worth noting that the dry deposition fluxes for NH 3 and HNO 3 would decrease as the distance from shore increased since land-derived gaseous inorganic N species are probably transformed to aerosol particles through acid-base reaction and/or absorption on sea-salt particles during their transport over the ocean.
Wet deposition of atmospheric inorganic N was highly variable from one event to the next depending on the concentrations of NH + 4 and NO − 3 in the precipitation as well as the frequency and amount of precipitation.Wet deposition fluxes of atmospheric inorganic N species ranged from 3.5 to 98 (mean 25) µmol m −2 d −1 for NH + 4 and from 1.0 to 32 (mean 8.0) µmol m −2 d −1 for NO − 3 , accounting for ∼77 % by NH + 4 and ∼23 % by NO − 3 of TIN from wet deposition flux.While NO − 3 was the dominant inorganic N species in dry deposition, inorganic N supplied to surface waters by atmospheric wet deposition was predominantly by NH + 4 (72-89 % of the wet deposition fluxes for TIN).
Likewise wet deposition, sea fog deposition of atmospheric inorganic N was highly variable depending on the size distribution of LWC, the amount of LWC, and the duration of sea fog event as well as the concentrations of NH + 4 and NO − 3 in the sea fog water.The estimated sea fog deposition fluxes for atmospheric inorganic N species ranged from 0.18-5.2µmol m −2 d −1 for NH + 4 and from 0.13-22 µmol m −2 d −1 for NO − 3 , contributing ∼39 % by NH + 4 and ∼61 % by NO − 3 to the sea fog deposition flux for TIN.Mean sea fog deposition fluxes for NH + 4 and NO − 3 were estimated to be 2.1 µmol m −2 d −1 and 5.7 µmol m −2 d −1 , respectively, indicating that inorganic N supplied to surface waters by sea fog deposition was mainly by NO − 3 , since aerosol NO − 3 and HNO 3 were scavenged more effectively by sea fog.
While dry deposition is a continuous process occurring at all times over all surfaces, wet and sea fog deposition are highly episodic.The relative importance of dry, wet, and sea fog deposition obviously varies greatly on short time scales, and varies spatially on longer time scales with global rainfall patterns (Jickells, 2006) and trends in fog frequency (Gultepe et al., 2007).Mean total (dry + wet + sea fog) deposition flux of atmospheric TIN in the subarctic western North Pacific Ocean was estimated to be 46 µmol m −2 d −1 , with 72 % of this in the form of wet deposition (Table 4).This indicates that wet deposition plays an important role in the supply of atmospheric inorganic N to the subarctic western North Pacific Ocean compared to dry and sea fog deposition, although the relative contributions are highly variable.The estimate of the proportion of atmospheric N input via wet deposition were comparable to previously published values: the Pacific 86 % (Duce et al., 1991), the Atlantic 78-85 % (Baker et al., 2010), and the world oceans 71 % (Duce et al., 1991).
Fog can lead to substantial N deposition if the event persists long enough (several hours) with sufficient LWC (dense fog), particularly for those formed in continental air masses (Jordan and Talbot, 2000).Although the mean contribution of sea fog deposition to total atmospheric TIN input to the subarctic western North Pacific Ocean was ∼17 % (Table 4), in some cases, atmospheric TIN deposition flux via sea fog exceeded the combined dry and wet deposition flux of TIN (Fig. 9), suggesting that sea fog can deposit as much N as a high N deposited by rain event, and that sea fog is an important transfer process for atmospheric inorganic N from the marine atmosphere to the subarctic western North Pacific Ocean.

Potential impact of atmospheric inorganic nitrogen deposition on primary production over the subarctic western North Pacific Ocean
The potential impact of atmospheric deposition on marine ecosystems depends on the nutrient status of the receiving waters, and is related to both the total amount and ratio of atmospherically supplied nutrients and to the limiting nutrient for the existing local water column (Baker et al., 2006).
It is known that NH + 4 and NO − 3 can be readily utilized by a variety of aquatic microorganisms (Gilbert et al., 1991).In order to evaluate the impact of atmospheric N on the marine ecosystem, potential primary production was estimated using the result for total deposition flux of TIN and the Redfield C/N ratio of 6.6.The average elemental C/N requirement for many plankton species and for bulk organic matter in the oceans has been found to be relatively constant at ∼6.6 (mol mol −1 ) (Redfield, 1958).Assuming that phytoplankton can take up all the N coming from atmospheric deposition with no losses, and that there is no co-limitation by other nutrients (i.e.P and Fe), total mean deposition flux of atmospheric TIN over the subarctic western North Pacific Ocean (46 µmol m −2 d −1 ) was found to be maximally responsible for the carbon uptake of 300 µmol C m −2 d −1 .Elskens et al. (2008) reported that the integrated new primary production in the upper part of the euphotic zone (0-50 m) at station K2 (47°N, 161 • E) in the subarctic western North Pacific Ocean from 30 July 2005 to 18 August 2005 ranged from 67-119 mg C m −2 d −1 (5.6-9.9 mmol C m −2 d −1 ).Wong et al. (2002) estimated that the annual new production from the surface (upper 50m) of the subarctic western North Pacific Ocean, which covers the sampling area of this study, to be 32.8-82.8g C m −2 yr −1 .To facilitate evaluation, we have converted g C m −2 yr −1 units for the annual new production reported by Wong et al. (2002) to the µmol C m −2 d −1 units used in this study.Based on these estimates by Elskens et al. (2008) and Wong et al. (2002), the result of this study suggests that inorganic N deposited to the subarctic western North Pacific Ocean from the atmosphere can support 1.6-5.3% of the new primary production.Atmospheric inorganic N deposition, however, could be an important N source in the ocean where sporadic atmospheric N deposition events caused by the transport of the continental dust affect and the supply of deep nutrient-rich water is restricted by the stratification of the surface ocean that is enhanced by global warming.

Conclusions
The subarctic western North Pacific Ocean (> 40°N) has a high frequency of sea fog, with a maximum of ∼50 % during the summertime period (June-August).The fog deposition is an important transfer process for atmospheric substances from the atmosphere to the biosphere.It is therefore suggested that sea fog may play a key role in supplying atmospheric nutrients to this region.Nevertheless, no study has been carried out over the subarctic western North Pacific Ocean to quantify sea fog deposition flux for atmospheric N.This is the first study to estimate atmospheric inorganic N fluxes via dry, wet, and sea fog deposition simultaneously over the subarctic western North Pacific Ocean.The mean dry, wet, and sea fog deposition fluxes for TIN were estimated to be 4.9 µmol m −2 d −1 , 33 µmol m −2 d −1 , and 7.8 µmol m −2 d −1 , respectively.Wet deposition delivered more atmospheric inorganic N to the subarctic western North Pacific Ocean than dry and sea fog deposition, contributing ∼72 % to total deposition flux for TIN (46 µmol m −2 d −1 ), although the relative contributions are highly variable.The mean contribution of sea fog deposition to total deposition flux for TIN was ∼17 %.Despite the relatively lower contribution of sea fog deposition, in some cases, atmospheric inorganic N input via sea fog deposition exceeded the combined dry and wet deposition fluxes.Thus, it is suggested that sea fog can result in substantial N deposition if the event persists long enough and has sufficient LWC (dense sea fog), and that ignoring sea fog deposition would lead to underestimate of the total influx of atmospheric inorganic N into the subarctic western North Pacific Ocean, especially in summer periods.
A schematic diagram of atmospheric inorganic N input to the subarctic western North Pacific Ocean via dry, wet, and sea fog deposition is shown in Fig. 10.In dry deposition, NO − 3 was the dominant inorganic N species, accounting for ∼57 %.This reflects higher deposition velocity of NO − 3 than that of NH + 4 since NO − 3 is largely associated with coarse mode particles in the marine atmosphere.In comparison, inorganic N supplied to surface waters by atmospheric wet deposition was predominantly by NH + 4 (72-89 % of the wet deposition fluxes for TIN), suggesting that NH + 4 is more important inorganic N species supplied by wet deposition over the subarctic western North Pacific Ocean.The contributions of NH + 4 and NO − 3 to the sea fog deposition flux for TIN were ∼39 % and ∼61 %, respectively, indicating that sea fog scavenged more effectively not only coarse mode particles (e.g.sea-salt particles and NaNO 3 ) that acted as CN of sea fog droplets, but also gaseous HNO 3 .
In this study, atmospheric inorganic N input via dry, wet, and sea fog deposition to the subarctic western North Pacific Ocean was estimated by simplified calculation (e.g. using assumption for sea fog water flux) during limited sampling period in the summer, but these results contributed to the understanding of atmospheric N cycle in open ocean environment.Further studies, however, are required to understand the biogeochemical cycles of N more clearly and should focus on long-term monitoring of atmospheric reactive N species, including organic N, in the subarctic western North Pacific Ocean.

Figure 1 .Fig. 1 .
Figure 1.Cruise track of Leg 1 of the KH-08-2.White circles, pink triangles, and blue 2 diamonds indicate aerosol, rain, and sea fog sampling locations during the cruise, respectively.3 Each aerosol sampling start point represents the end of the previous sampling period.Dotted 4 line indicates that no aerosol sampling was conducted.Black, pink, and blue numbers 5 represent the sample numbers of aerosol, rain, and sea fog, respectively.6 7 Fig. 1.Cruise track of Leg 1 of the KH-08-2.White circles, pink triangles, and blue diamonds indicate aerosol, rain, and sea fog sampling locations during the cruise, respectively.Each aerosol sampling start point represents the end of the previous sampling period.Dotted line indicates that no aerosol sampling was conducted.Black, pink, and blue numbers represent the sample numbers of aerosol, rain, and sea fog, respectively.

Figure 2 .
Figure 2. Temporal variations of meteorological variables (i.e.wind speed, wind direction, air 2 temperature, sea surface temperature, dew point, and relative humidity) during the cruise.Sea 3 fog (red circle) and rain (violet triangle) events indicate the occurrences of observed sea fog 4

Figure 3 .
Figure 3. Temporal variations of total LWC, particle number densities for aerosols in four size 2 groups, and LWC for each of the 40 droplet size classes during the cruise.The green and pink 3 shaded areas indicate sea fog water and rainwater sampling durations, respectively.The black 4 and pink numbers indicate sea fog and rainwater sample number, respectively.When sea fog 5

Fig. 5 .
Figure 5.The 168 h (7 days) air mass backward traje 2 (triangle symbols), 1000 (circle symbols) and 1500 m ( 3 (AGL) during the collections of aerosol (a), rain (b), and 4 from the Global Data Assimilation System (GDAS) d 5 Atmospheric Administration (NOAA) and simulated b 6 Langrangian Integrated Trajectory (HY-SP 7 http://www.arl.noaa.gov/ready/hysplit4.html).White c 8 diamonds indicate aerosol, rain and sea fog sampling loc 9 Kasatochi and Okmok volcanoes are black diamonds.10 11 Fig. 5.The 168 h (7 days) air mass backward trajectories for starting altitudes of 500 (triangle symbols), 1000 (circle symbols) and 1500 m (square symbols) above ground level (a.g.l.) during the collections of aerosol (a), rain (b), and sea fog (c) samples were calculated from the Global Data Assimilation System (GDAS) database of the National Ocean and Atmospheric Administration (NOAA) and simulated by using the Hybrid Single-Particle Langrangian Integrated Trajectory (HY-SPLIT) model (web site http://www.arl.noaa.gov/ready/hysplit4.html).White circles, pink triangles and yellow diamonds indicate aerosol, rain and sea fog sampling locations during the cruise, respectively.Kasatochi and Okmok volcanoes are black diamonds.

Figure 8 . 4 Fig. 8 .
Figure 8. Variations of mean LWC for each sea fog event as a function of mean droplet 2 diameter during the cruise.3 4

Figure 9 .Fig. 9 .
Figure 9. Temporal variations of dry, wet, and sea fog deposition fluxes for NH4 + and NO3 - 2 over the subarctic western North Pacific Ocean during the cruise.3 4 Fig. 9. Temporal variations of dry, wet, and sea fog deposition fluxes for NH + 4 and NO − 3 over the subarctic western North Pacific Ocean during the cruise.

Figure 10 .Fig. 10 .
Figure 10.Schematic diagram of atmospheric inorganic N input to the subarctic western 2 North Pacific Ocean during 29 July-19 August 2008.The numbers in circles represent 3 concentrations of NH 4 + and NO 3 -in aerosols (in neq m -3 ), rainwater (in µeq l -1 ) and sea fog 4 water (in µeq l -1 ).The mean percentages of total (fine + coarse) aerosol concentrations in fine 5 (D < 2.5 µm, white) and coarse (D > 2.5 µm, yellow) modes for NH 4 + and NO 3 -are shown in 6 the circles for aerosol.The orange and blue arrows and numbers (in µmol m -2 d -1 ) indicate 7 NH 4 + (orange) and NO 3 -(blue) fluxes via dry, wet, and sea fog deposition.8 Negative values that arise for non sea-salt ionic species as a result of analytical uncertainty and samples where concentration of each ionic component was below the detection limit have been included in the calculation of the average as 0.
a b These values indicate percentiles.c Sample number of each ionic component detected (or calculated) in aerosols, rainwater, and sea fog water.

Table 2 .
Mole equivalent ratios for major ionic species in aerosols, rainwater, and sea fog water, together with seawater ratios for comparison a .