Chemical Composition of Sea Fog Water Along the South China Sea

Transcription

1 Pure Appl. Geophys. 169 (212), Ó 212 Springer Basel AG DOI 1.17/s Pure and Applied Geophysics Chemical Composition of Sea Fog Water Along the South China Sea YANYU YUE, 1 SHENGJIE NIU, 1 LIJUAN ZHAO, 1 YU ZHANG, 2 and FENG XU 3 Abstract The chemical and microphysical properties of sea fog were measured during a field experiment on Donghai Island, Zhanjiang of China from March 15 to April 18, 21. The average ph and electrical conductivity (EC) value of the six sea fog cases during the experiment was 5.2 and 1,884 ls/cm. The observed total ion concentration of sea fog was four orders of magnitude higher than those in the North Pacific and other sea areas of China. The dominant anion and cation in all sea fog water samples were Cl - and Na?, respectively. From backward trajectory analysis and ion loading computation, it can be concluded that the ions in the samples were transported either from pollutants in distant industrial cities or from local ion deposition processes. The concentration of Ca 2? in the sea fog water samples in Case 2 suggested that a dust storm in the Inner Mongolia, a northern region of China several thousand kilometers away, could reach the South China Sea. The data also showed that the sea fog droplet spectrum over the South China Sea is unimodal. Through relationship analysis, it is illustrated that the evolution of microphysics (such as droplet concentration, diameter, and liquid water content) during fog process could affect the chemical properties of sea fog. Key words: Fog chemistry, ion concentration, fog microphysics. 1. Introduction Sea fog is a condensation phenomenon occurring in the lower layer of the atmosphere over the sea or coastal areas. Sea fog typically occurs as a result of relatively warmer marine air moving over a region of cold ocean surface (GULTEPE et al. 27). It can affect 1 Key Laboratory of Meteorological Disaster of Ministry of Education, School of Atmospheric Physics, Nanjing University of Information Science and Technology (NUIST), 2144 Nanjing, People s Republic of China. yueyanyu123@163.com; niusj@nuist.edu.cn; ; ; 2 Zhanjiang Meteorological Bureau, 5241 Zhanjiang, People s Republic of China. zy22264@163.com 3 Guangdong Ocean University, 5249 Zhanjiang, People s Republic of China. gdouxufeng@126.com transportation and fishing at the sea. The fog frequency increasing may result from increased air pollution of various sources (MOHAN and PAYRA 29), so fog can be a hazard for human health if there are high concentrations of chemical compositions in fog water. How these high concentrations are generated in fog water is still not clear, particularly for fogs in the southern coast of China, although there have been great efforts dedicated to this study in the other countries (e.g., MILLET et al. 1996; MOORE et al. 24; ZAPLETAL et al. 27; SHIMADERA et al. 29). Monitoring cloud and fog chemistry is a good way for comprehensive interpretation and identification of long-distance transport of air pollutants (BLAŚ et al. 21). RAJA et al. (28) noted a smog fog smog cycle: When fog dissipates, it leaves behind a portion of the scavenged pollutants on residual aerosol particles, enhancing fog formation by high aerosol concentration. Ion species that form coarse particles were more easily and effectively scavenged by fog water than those that form fine particles (AIKAWA et al. 27b). On the basis of field research over the North Pacific, chemical composition of sea fog and acidification processes were analyzed (SASAKAWA and UEMATSU 22, 25). Some studies on fog chemistry took into account the connection between fog microphysics characteristic quantity and chemical compositions (WALDMAN et al. 1982; FUZZI et al. 1984; LU et al. 21). The sea fog research in China began in the 196s. WANG (1983) discussed sea fog distribution, synoptic system and meteorological conditions for sea fog occurrence, and sea fog forecasting. Recently, numerous field experiments on fog water have been conducted in various regions in China (NIU et al. 21b), including in the urban areas, such as Nanjing (FENG et al. 29) and Shanghai (BAO et al. 1995), mountainous regions (WU et al. 24) and coastal

2 2232 Y. Yue et al. Pure Appl. Geophys. areas (HUANG et al. 29). In contrast, sea fog observations were limited in the southern coast of China. Through field experiments in the coastal areas of the Yellow Sea (Qingdao) and East China Sea (Zhoushan), researchers concluded that the physical and chemical characteristics of sea fog were different from those of the fogs appeared on the continent. The most important anion and cation of sea fog were Cl - and Na? (MO et al. 1989; SONG et al. 1992), and the spectrum width of sea fog in Zhoushan Island was larger than that on land (YANG et al. 1989). Sea fog over coastal areas in south China has significant impacts on navigation and aviation in spring seasons. During the early fog studies in the South China Sea area, the research was concentrated on synoptic systems, microphysics and boundary layer. Studies on sea fog chemistry had not begun. To gain an understanding of sea fog water chemistry, a field experiment was carried out in early 21 with the following objectives: to investigate the chemical characteristics of fog in the area, to study the factors that influence chemical composition, to reveal the relationship between fog chemical components and fog microphysics, and to find differences in chemical composition and microphysics properties between sea fog and continental fog. The paper is organized as follows. Section 2 introduces the experiment site, the principle of instruments used in the experiment, and the data collected. Section 3 illustrates the methods of analysis and calculation. Section 4 presents the results, including sea fog water chemical compositions, variation of ions for the six fog cases, ph, electrical conductivity (EC), and ions loading. Section 5 discusses the sources of air mass, ion concentration and fog microphysical variables in different regions. Moreover, the relationship between sea fog microphysics and ion concentration is explained. The concluding remarks are presented in Sect. 6. The observation site is on the east coast of Donghai Island in Zhanjiang, at 11.5 E and 21 N, which is in the east part of the Leizhou Peninsula (Fig. 1). The urban district of Zhanjiang is to the northwest, about 3 km away from the island. The instruments were installed on the fourth floor (about 15 m above the sea level) on the east side of a building 2 m away from the sea shore, with no obstacles toward the sea. Therefore, the observational data can reflect the characteristics of sea fog. The instruments are listed in Table 1. The main data used in the analysis are visibility from the VPF73 Visibility and Present Weather Observing System, droplet spectra from FM-1 Droplet Measurement Systems, and fog water from a fog water collector. The surface meteorological conditions including temperature, wind, humidity and pressure near the surface were observed by an automatic weather station. We also conducted a visual inspection hourly to record cloud cover, weather phenomena, and instrument operational status. Next, we give some introductions on these instruments. Bulk fog water was sampled with a selffabricated active fog water collector. The design mainly followed a compact version of the original Caltech Active Strand Cloud-Water Collector (CASCC2) (DEMOZ et al. 1996). A pump draws the foggy air through two rows of Teflon strands. After impaction and coagulation of droplets on the strands, they run down the strands through a Teflon tube, and 2. Experiment and Observations The field observation was conducted from 15 March to 18 April 21 on Donghai Island, which is in the South China Sea. It is one of the five high sea fog occurence areas in China (ZHANG and BAO 28). Figure 1 Map of the observation site

3 Vol. 169, (212) Chemical Composition of Sea Fog Water Along the South China Sea 2233 Table 1 Instruments used in the field experiment on Donghai Island Instrument Model Measurement Accuracy Fog measuring device (DMT, USA) FM-1 Fog droplet spectrum Range 2 5 lm Fog water collector (self-fabricated) Fog water Visibility meter (Biral, England) VPF-73 Visibility and present weather ±1 %, Vis \ 16 km; ±2 %, 16 km \ Vis \ 3 km Automatic Weather Station (Guangdong Meteorological Institute of Computer Application, China) WP313 Temperature, wind direction, wind speed, RH, pressure T, ±.3 C, wind direction, ±1, wind speed,.5 ±.3 m s -1 ; humidity, ±4 %, pressure, ±.3 hpa Conductivity meter (Leici company, China) DDSJ-38A Electrical conductivity ±.5 % (FS) ± 1 dgt ph meter (Shenhua laboratory instrument PHS-25, PHS-29 ph value.3 ph company, China) Ion chromatograph (Dionex, USA) ICS-2 Ion concentration Flow velocity ±.25 % eventually settle into a polyethylene sample bottle. When the volume of each sample reaches 1 15 ml, the sampling bottle is replaced by a new one. Before sampling, the collector is washed by high purity deionized water. The sampling bottle is washed by the industrial alcohol and deionized water. The instrument used in measuring the size and number concentrations of fog droplets was Droplet Measurement Systems (DMT) Fog measuring device (model FM-1, Boulder, Colorado, USA; GULTEPE and MILBRANDT 27). It is based on light scattering by small particles. Particles scatter light from a laser diode of approximately 5 mw, and collecting optics guides the light from 5 to 14 into forward and masked (qualifier) detectors. The vacuum source pulls fog particles through a sample area at a known velocity, allowing particle concentrations to be calculated. According to the intensity of scattered light attributed to the size of droplets, we can classify the droplets and count them to calculate the number of fog droplets in each bin. It measures fog droplets from 2 to 5 lm in diameter, with 2 size classes in the experiment. The maximum concentration of fog droplets is up to 1 4 cm -3. The near-surface meteorological elements (temperature, wind, humidity, pressure) were measured by the automatic meteorological station on Donghai Island, which is located at E, 21.2 N. It is 16 km away from the fog field observation site. The resolution of the sensors is.1 C (temperature), 3 (wind direction),.1 m s -1 (wind speed), 1 % (humidity), and.1 hpa (pressure). The VPF-73 Visibility and Present Weather Observing System (Biral, Bristol, UK) was used to measure the visibility and present weather automatically every 3 s. The sensor measures visibility, precipitation identity, rain rate, and snowfall rate. The instrument has an optical transmitter, a forward scatter receiver, and a back scatter receiver. The scattering angle coverage is between 39 and 51, and the sample volume is 4 cm 3. When the transmitter emits a near-infrared pulse in a modulation frequency of 2, Hz with a bandwidth of.8 mm, the intensity of scattered light can be measured when small particulates are suspended in the air or large particles are passing through the sample volume. Given the reasonable assumptions that the visibility is restricted to values of less than 1 km and that absorptions by fog, aerosols, and precipitation are negligible, the total scattering coefficient can be equated to the total extinction coefficient. The visual range is from 1 m to 75 km. When the visibility is less than 16 km, the measurement error of the instrument is less than 1 %. When the visibility is between 16 and 3 km, the measurement error is up to ±2 %. Backward trajectory analysis is employed to examine the transport path of air mass to Donghai Island. The analysis is based on the HYSPLIT4 Model (Hybrid Single-Particle Lagrangian Integrated Trajectory). It is a complete system from computing simple air parcel trajectories to performing complex dispersion and deposition simulations, and can be applied in synoptic meteorology, climatology and

4 2234 Y. Yue et al. Pure Appl. Geophys. environmental sciences, among others. More information on the model can be found at the website of NOAA s Air Resources Laboratory in Silver Spring, Maryland USA (DRAXLER and HESS 1997, 1998). The model takes vertical velocity as the vertical motion mode, and uses the Global Data Assimilation System (GDAS) data available from the National Centers for Environmental Prediction (NCEP). Backward trajectories are calculated for three days arriving at the altitudes of 1, 5, and 1, m above the ground level. 3. Analysis 3.1. Chemical Composition Analysis Methods A total of 12 fog cases were observed during this field experiment; however, only six consecutive sea fog cases are analyzed and presented here. Fog water was not collected in Cases 7 12, mainly due to the short duration of fog. Overall, 19 samples collected in Cases 1 6 were analyzed for nine ionic species plus ph and EC. The EC was measured on the spot with a conductivity meter (DDSJ-38A). Because of malfunction of the ph meter, the ph of fog could not be measured on site. The samples were brought back immediately to the Nanjing University of Information Science and Technology (NUIST) for chemical analysis after the field experiment. The ph values were obtained with a digital ph meter (PHS-25/PHS- 29) at the NUIST. Dionex model 2 ion chromatographic system was used in the analysis of ion concentration. The fog water samples were filtered through a cellulose acetate membrane filter (.45 lm). Because high concentrations of Na? and Cl - would damage the column, we diluted the samples first. Before detection, the standard solution was made, injected, and analyzed. Then, the fog water samples were injected into the ion chromatograph twice for measuring the main anions or cations, and their mean values represented the ion concentration. The system setup for anion analysis included separation column (AS11- HC, mm), micro-membrane suppressor (ASRS 4 mm), and eluent (3 mmol/lkoh) with a flow rate of 1 ml min -1. The setup for cation included separation column (CS16, mm), auto-suppressor (CSRS 4 mm), and eluent (32 mmol L -1 methane-sulfonic acid) with a flow rate of 1 ml/min. Each sample was injected twice, and the analysis times for anions and cations were 1 and 2 min, respectively Calculation Method Data quality control was performed by evaluating percentage difference of the ion balance (PDI). PDI was calculated by the following formula (FUZZI et al. 1992; BLAŚ et al. 21): Conc:anions Conc:cations PDI ¼ 1 %; Conc:anions þ Conc:cations ð1þ A good ion balance would be between and?1.4 with regard to the perfect ion balance. In order to ascertain the ph value, we use the pai proposed by HARA et al. (1995). It is the hypothetical ph of atmospheric water if no neutralization takes place for both sulfuric and nitric acids and can be calculated as follows: pai ¼ log nssso 2 4 þ NO 3 : ð2þ The levels of non-sea-salt sulphate (nssso 4 ) and non-sea-salt Ca 2? (nssca 2? ) were estimated using the following equations: nssso 2 4 nssca 2þ ¼ SO 2 4 SO 2 4 =Na þ seawater ½ Naþ Š; ¼ Ca 2þ Ca 2þ =Na þ ð3þ seawater ½ Naþ Š: ð4þ The ratios of (SO 4 /Na? ) seawater and (Ca 2? / Na? ) seawater are.12 and.44, respectively (KEENE et al. 1986; WATANABE et al. 26). The acidifying potential (AP) and neutralizing potential (NP) (TSU- RUTA 1989) are calculated via the following equation: AP ¼ nssso 2 4 þ NO 3 ; ð5þ NP ¼ nssca 2þ þ NH þ 4 : ð6þ According to the Fog Monitor Operator Manual provided by DMT, the sampling volume per second (V) of the units of cm 3 s -1 is calculated by the following expression:

5 Vol. 169, (212) Chemical Composition of Sea Fog Water Along the South China Sea 2235 V ¼ TAS A ¼ 2:6 M Ta :5 A; ð7þ where M is the Mach number derived from the dynamic and static pressure (in units of mb), T a is the actual ambient temperature (in units of K), and A is the sampling area equal to.264 mm 2. The electric signals of instruments could be calculated to get droplet count, and then the fog counts divided by V provides n(r). The number concentration (N; in units of cm -3 ), liquid water content (LWC; in units of gm -3 ) and surface area, which represents the surface area of fog droplets per unit volume of air (S; in units of m 2 m -3 ) of the whole spectra can be calculated as follows: N ¼ X nðrþdr; ð8þ LWC ¼ q X 4p 3 r3 nðrþdr; ð9þ S ¼ 4p 1 6 X nðrþr 2 dr; ð1þ where n(r) is the number concentration of particles corresponding to radius r, and the units of n(r) and r are cm -3 and lm, respectively. q = 1gcm -3 is the density of water. The ion loading C air (in units of nmol m -3 ) can be calculated by the following equation (IGAWA et al. 1991), C air ¼ C fog LWC=q: 4. Results 4.1. ph and Electrical Conductivity ð11þ Table 2 shows the ph value, EC, and some meteorological elements for the 12 cases. The fogs on Donghai Island can be classified primarily as advection fog. The typical weather systems for sea fog occurrence on Donghai Island are the rear part of a high pressure system that is moving eastward to enter the sea, and in front of a depression. The wind brings the warm and moist air to the cool sea surface of the northern South China Sea coastal area, and then the air approaches saturation to form the advection fog. In all but Case 1, the wind direction during fog was from the east. Warm, moist, and foggy air was advected to the coast. Such a process is typical for this region (ZHANG 1999). The average wind speed in the six fog cases was 2 3 m s -1, similarly to the result of the South China Sea fog by LIN and SONG (199). The ph range was between 4.8 and 6.5, with a mean of 5.2, which was narrower than for the range on Zhoushan Island near the East China Sea (MO et al. 1989). The ph value of advection fog originated from the stratus cloud deck over the eastern South Pacific varied between 2.9 and 3.5 (STRÄTER et al. 21). The ph value in the Sea of Okhotsk and the Sea of Japan was below 3 (SASAKAWA and UEMATSU 22). Compared to these areas, the acidification of fog water on Donghai Island was not significant. However, the pai values shown in Table 3 are much lower than the ph. This index focuses only on the acidic components, while the ph is determined by the balance between acidic and neutralization components (SASAKAWA and UEMATSU 22). The EC in the samples ranged from 229 to 6,185 ls cm -1 with a mean of 1,884 ls cm -1, which was much higher than those in other sites (WRZESINSKY and KLEMM 2;BEIDERWIEDEN et al. 25). Figure 2 illustrates the relationship between AP and NP in sea fog water. In this graph, the pdi is a theoretical line that equals -log(ap NP) (FUJITA et al. 2). Almost all our fog water data is below the pdi7 line of the theoretical neutrality, indicating that the sea fog in this region was acidic due to the low concentration of neutralizing agents. The AP is slightly higher than the NP. The neutralizing and acidifying components from continental sources affect the fog water in this region. Both acidifying and neutralizing agents have strong influences. However, they have similar influences on fog chemistry and counteract against each other pretty well. So, if the acidifying component was more than the neutralizing one, the sea fog water would be acidic. One of the other sources of acidifying component is the ocean. The stable end-products of dimethylsulfide (DMS) oxidation in the atmosphere from the ocean may have contributed to the sea fog acidification (SASAKAWA and UEMATSU 25) Ion Concentration The average ion balance ratio (PDI) for the six cases is -6.2 %, with a standard deviation of 7.2. The

6 2236 Y. Yue et al. Pure Appl. Geophys. Table 2 Mean ph, electrical conductivity (EC), and meteorological elements during the 12 fog events Time Case ph EC (ls cm -1 ) Visibility (m) T ( C) Wind speed (m s -1 ) Wind direction : 6: ( ) 2.23 ( ) 47 (3 82) : : ( ) 2,81 (87 3,81) ( ) 2.96 ( ) 99.2 (67 129) : : ( ) (229 1,319) ( ) 2.15 (.9 4.2) 112 (62 149) : : ( ) 2,56.58 ( ) ( ) 2.67 (.7 5.3) 18 (3 158) : : ( ) 2,348 (1,196 3,5) ( ) 2.56 ( ) (43 151) : 7: ( ) 3, (1, ,14) ( ) 2.64 (1.4 4.) 114 (86 129) :51 4: ( ) 1.1 (.5 1.9) 86.7 (58 121) :17 8: ( ) 1.27 (.6 1.9) (93 15) :35 7: ( ) 2.14 (1.2 4.) 85 (68 16) :58 4: ( ) 3.22 ( ) 8.9 (62 97) :51-7: ( ) 2.22 ( ) 85 (71 112) :33 5: ( ) 3.43 ( ) 8.9 (63 99) The values in parenthesis represent the minimum and maximum values small anion deficit was probably due to some nonmeasured inorganic and organic acids, such as carbonate, nitrite, acetate, formate and so on. A similar phenomenon was found in the Los Angeles Basin (MUNGER et al. 1983), Strasbourg (MILLET et al. 1996) and Nanjing (LU et al. 21). Considering the anion deficit and analytic errors, the ion balance is thought to be well established with the experimental data (MILLET et al. 1996). As shown in Fig. 3, the dominant anion and cation in all of sea fog water samples are Cl - and Na?, with mean concentrations of 11,79 and 11,666 leq L -1, respectively, which are higher than the values obtained in other sea fog observations (MO et al. 1989). The second most important ions are NO 3 - and Mg 2?.In contrast, SO 4,NH 4?, and Ca 2? were the predominant ions in Chongqing (LUO and XIAN 25) and Nanjing (LI et al. 28); both locations are far away from the coast. Since Donghai Island is along the South China Sea, the ocean is a major source of aerosols and gases for the overlying atmospheric boundary layer (KEENE et al. 1986). Over the sea, bubble breaking, photochemistry, and biological processes could produce sea salt aerosol and gases released by marine organisms. A more important source of aerosol particles results from the bursting of bubbles produced by the entrainment of air at wave crests. Subsequent to its formation, a sea-salt particle may change its composition further as a result of both chemical reactions with atmospheric trace gases and coagulation with other aerosol particles in the atmosphere (PRUPPACHER and KLETT 1997). During the transport over the ocean, there are two aerosol removal processes from the atmosphere: one is dry deposition, and the other is wet deposition. We also need to take into account for an additional removal process by sea fog. Sea fog will enhance the deposition of atmospheric components to the sea surface by scavenging aerosols and acidic gases (SASAKAWA and UEMATSU 25). Based on the investigation on Amsterdam Island, the composition of precipitation is in agreement with that of sea water (KEENE et al. 1986). On the other hand, in northern Chile, which is a site with strong air pollution, only about 5 % of the total ionic concentration was due to sea salt (STRÄTER et al. 21). Therefore, it is reasonable to assume that at least 5 % of the ions in the sea fog originate from sea water.

7 Vol. 169, (212) Chemical Composition of Sea Fog Water Along the South China Sea 2237 Table 3 Chemical coefficients and indices of fog water (leq L -1 ) from Case 2 to Case 6 on Donghai Island Case 2 Case 3 Case 4 Case 5 Case 6 Max pai Min Mean Max Cl - /Na? Min Mean Max SO 4 /Na? Min Mean Max K? /Na? Min Mean Max Ca 2? /Na? Min Mean Max Mg 2? /Na? Min Mean Max 5, ,464 1,192 1,96 nssso 4 Min 1, mean 3, ,272 max 16,363 2,63 4,388 2,916 2,42 nssca 2? min 7, ,58 1,13 mean 1, ,517 2,248 1,761 NP:[nssCa 2+ ]+[NH 4 + ](µeq/l) AP:[nssSO 4 ]+[NO 3 - ](µeq/l) NP= AP R=.95 P<.1 Figure 2 Relationship between AP and NP in sea fog water samples of the six cases. Dotted line represents pdi7 The sea salt particles responsible for the emissions of Cl - and Na? are in the form of coarse particle mode (TANG et al. 26). Because there are almost no factories that emitted large amount of pollutants on the island, and the wind is almost always the Concentration(µeq/L) Na + NH 4 K + Mg 2+ Ca 2+ - F - Cl - NO 3 SO 4 Cations and Anions Figure 3 Box plots for sea fog water ions in the six cases. The large boxes represent the interquartile range from the 25th to the 75th percentile. The small squares and the lines inside the large boxes indicate the mean and median values, respectively. The whiskers extend upward to the 95th percentile and downward to the fifth percentile easterlies, the Cl - associated with coal burning is only a small portion (MILLET et al. 1996; MCCULLOCH et al. 1999; WATSON et al. 21). Na? is usually used as a reference element of sea source due to its stable chemical properties in atmospheric aqueous solutions. The ratio of [SO 4 ]/[NO 3 - ] in sea fog water on Donghai Island was.67, indicating that the pollution was dominated by nitrate. The SO 2 and nitrogen oxides emitted by the plants in the local areas are assumed to contribute insignificantly. A major source of NO 3 - in the sea fog water samples is industrial pollutants, such as nitrate aerosol and nitric acid gas (COLLETT et al. 22) emitted and transported from those distant industrial regions. As these pollutants were carried to the fog region, the fog deposition can scavenge and collect the suspending ions in the air. On the other hand, due to the poor road conditions, the traffic had small contribution to nitrogen oxide emission (BLAŚ et al. 21). As for sulphate in the water sample, oxidation of SO 2 and non-sea-salt SO 4 from other anthropogenic emissions away from the island were considered to be the most important sources. The SO 4 in sea fog water was considerably lower than that in Nanjing (LU et al. 21). In the sea, marine organisms may be one of the sulfur gas sources. DMS generated by phytoplankton is thought to be the most important of these possible sources, such as H 2 S, CS 2, and DMS

8 2238 Y. Yue et al. Pure Appl. Geophys. (KEENE et al. 1986). Mg 2? was the second important cation in the sea fog water, and was mainly derived from oceanic aerosols. The radar observation station near the observation site on Donghai Island was under construction at the time of our field experiment, which would produce lots of dust and sand. Ca 2? concentration can be attributed to soil (MILLET et al. 1996; ALI et al. 24). The mean ratio of Cl - and Na? was about one, suggesting that no deficit of Cl - was observed due to the proximity of the observation site to the sea and the apparent lack of free acidity (high phs). However, the ratio range of SO 4 /Na? was.14.42, which is clearly larger than that in the sea water (.12). This implies that there were other sources for SO 4 besides the sea water. As we have mentioned earlier, SO 2 was from anthropogenic emissions that came from the large cities to the northeast and northwest of Donghai Island with dense populations and industries. Another source might be marine biogenic. The mean equivalent ratio of Mg 2? and Na? is.23 in the sea water (KEENE et al. 1986), which is a little lower than the ratio in the fog water. However, the ratio of Ca 2? /Na? is much higher than that in the sea water (.44). This indicates that there was a strong influence from continental sources, such as wind erosion of soil and rock materials on the chemical composition of sea fog water. As shown in Fig. 4, various ions during the first fog case were relatively low. Because the fog water sample in Case 1 was not enough to conduct many additional analyses, so NH 4? was not obtained. In Case 2, Ca 2? and Mg 2? increased significantly. The average percentage of Ca 2? to all ions was nearly the same as that of Na?. The dust storm originated largely from Inner Mongolia, a region several thousand kilometers to the north, on 19 March 21, which was the most serious sandstorm since January 29. The dust aerosols were transported to central and east China on March 2. The fog water samples obtained in the second fog case had dust sediments, while there were almost no sediments in the other samples, clearly showing the impact of the sandstorm from north China on the South China Sea. The dust particles were made up of mineral particles and carbon grain aggregation (SHI et al. 28). When the dust storm occurred, the concentration of crust Concentration(µeq/L) Concentration(µeq/L) Na + NH 4 + K + Mg 2+ Ca Fog case F - Cl - NO 3 - SO Fog case Figure 4 Variation of average concentrations of cations and anions in the six cases elements of high enrichment factor would increase, such as Mg, Zn, Ca, Fe, and Na (NIU and ZHANG 2). The dust particles carrying numerous aerosols were transported to the South China Sea area, which led to the concentrations of aerosols containing Ca or Mg increased in the atmosphere. During the fog, the dissolved material entered fog droplets, and CaCO 3 could react with H? in fog water; as a result, Ca 2? and Mg 2? increased in fog water. By the time the third fog case happened on 23 March, the dust process had ceased. The ion concentration in Case 3 was very low. During this case, the wind direction and wind speed did not change; the wind was still the easterlies. It suggests that the deposition of dust aerosol scavenging the particles suspended in the atmosphere was a crucial process to reduce the ions. The ions concentration values, such as Na? and Cl -, were almost the same in Cases 4 and 5, which were

9 Vol. 169, (212) Chemical Composition of Sea Fog Water Along the South China Sea 2239 lower than those in Case 6. In Case 6, the ions derived from ocean sources were significantly higher. The reasons may be that on one hand the synoptic system of this case belonged to a uniform pressure field, which had low wind speed and high temperature, favoring conditions of intense evaporation at the sea surface. The sea salt particles were produced by the bubble-bursting mechanism, some of the drops fell back to the ocean surface, and the remainder evaporated, causing small sea salt particles to escape because they were light enough to be carried aloft by air flows (PRUPPACHER and KLETT 1997). High temperature favored drops evaporation and made more sea-salt particles suspended in the air. On the other hand, the microphysics properties of sea fog had an influence on the ion concentration. Their relationship will be further discussed in Sect Figure 5 shows the variation of ions with time during Cases 3 and 4. The concentrations of ions in Case 3 were much lower than those in Case 4. The temporal trends of various ions were similar to each other in the two cases, indicating that the LWC was the key player in the variation of ions concentrations. F - and NH 4? did not change obviously, probably due to their low concentrations. At the beginning of these events, the ion concentrations were high. From the initial stage (3.23 during BST 23: 3: and 3.31 BST during 19:2 22:) to the maintenance stage (3.24 during BST 3: 7: and 3.31 during BST 22: 4.1 4:), the ions concentrations declined greatly. Because of gravitational settling, wet precipitation, and turbulent transport of fog droplets in the initial stage, gases and aerosol particles near the surface layer could be scavenged. Consequently, the ions concentrations in fog water were lower at the maintenance stage. Before the fog dissipation in Case 3 and Case 4, namely, at 5: 7: on 24 March and 2: 4: on 1 April, the ions concentrations of fog water had risen again. Combined with Fig. 8 (Case 3), we can learn that the visibility changed a little compared to that during 3: 5: in Case 3, but the LWC presented a downward trend. For a given concentration of pollutants per unit volume of air, if the LWC decreased, the concentration in the fog water samples would increase. For a given droplet size, the concentration of the liquid phase is governed by the ratio between the Concentration(µeq/L) Concentration(µeq/L) concentration of dissolved pollutants per volume of air and the number of droplets per unit volume of air (MILLET et al. 1996). During dissipation, the ions concentrations were a little higher than those in the maintenance stage, which can be mainly attributed to evaporation (WU et al. 24). It was obvious that the ions concentrations increased slightly during the dissipation of fog, as in HOGAN and HICKMAN (1965) Ion Loading F - Cl - NO 3 - F - Cl - NO 3 - SO 4 Anions and Cations SO :-3: :-4: :-5: :-7: :-1: Na + Ca 2+ K + Mg 2+ NH 4 + Ca :2-22: :-24: 3.31 :-1: 4.1 1:-2: 4.1 2:-4: 4.1 4:-9: Na + K + Mg 2+ NH 4 + Anions and Cations Figure 5 Variation of ions with time during the third and fourth fog cases. The time is in BST format It is important to estimate the deposition of air pollutants via fog because of the high concentrations of ions in fog water (TAGO et al. 26; KLEMM and WRZESINSKY 27). The ion loading is often used to estimate the efficiency of nucleation and gas scavenging in cloud or fog, and it is defined as the amount

10 224 Y. Yue et al. Pure Appl. Geophys. Fog case F - SO 4 NO 3 - Cl - NH 4 + Ca 2+ Mg 2+ K + Na Ion loading(nmol/m 3 ) Figure 6 Variation of ion loadings in the six cases of a chemical species dissolved in the liquid phase within 1 m 3 of air (ELBERT et al. 2). As shown in Fig. 6, the ion loadings of Na? and Cl - were obvious, and the peak values were observed on 6 April 21. The ion loading values of Na?,Mg 2?, and Cl - were much higher than their average values of the six cases, which were 224, 34, and 245 nmol m -3, respectively. The LWC was.23 g m -3. The relatively low LWC and high ion loading led to the high ions concentrations in fog water in Case 6. As mentioned by MINAMI and ISHIZAKA (1996), LWC, gas scavenging, and chemical reactions affect the solution concentration in fog water; so LWC alone could not always explain the change of ions concentrations. The Ca 2? ion loading had a peak on 22 March. The rise of Ca 2? was probably caused by dust particles. The ions concentrations in Case 4 (from March 31 to April 1) and in Case 5 (during April 1 2) were almost the same, but their ion loadings were significantly different from each other. The average ion loadings of Na? and Cl - in Case 4 were nearly two times larger than those in Case 5. However, the dilution effect in Case 4 was more significant than in Case 5. The LWC played an important role in changing the ions concentrations in the two cases; therefore, their ions concentrations were almost the same Ionic Correlations Correlation coefficients between different chemical compositions of fog samples can indicate whether the components have the same sources and originate from the same chemical compound or not (Table 4). The ph values had no significant correlation with any kinds of ion compositions. A strong correlation existed between EC and ions of high concentrations such as Cl -, Mg 2? and Na?. The correlation coefficients among Cl -,Mg 2? and Na? were more than.98. Cl -,Mg 2?, and Na? originated from oceanic aerosols primarily, which existed in the forms of MgCl 2 and NaCl. For the fog of marine origin, the Cl - /Na? ratio was 1.4, which reflects the composition of sea water. This is consistent with the results in Szrenica and Borucino (BLAŚ et al. 21). The correlation coefficients between SO 4 and other four cations, except for NH 4?, were all about.8, while it was much lower between SO 4 and NH 4? ; Table 4 Correlation coefficients between different chemical species of sea fog samples for the six cases ph EC F - Cl - NO 3 - SO 4 NH 4? Ca 2? K? Mg 2? Na? ph 1.42* ** *.46 * **.46 **.48 ** EC 1.75 **.91 **.81 **.84 **.69 **.55 **.89 **.94 **.92 ** F **.79 **.78 **.57 *.59 **.76 **.8 **.77 ** Cl **.78 **.82 **.39 *.86 **.98 **.99 ** - NO **.4 *.9 **.8 **.82 **.74 ** SO **.87 **.84 **.87 **.8 **? NH **.76 **.8 ** Ca 2? 1.61 **.55 **.43 ** K? 1.9 **.89 ** Mg 2? 1.99 ** Na? 1 ** Significant at.1 level, * significant at.5 level

11 Vol. 169, (212) Chemical Composition of Sea Fog Water Along the South China Sea 2241 this is different from fog water composition correlation over the continent (LU et al. 21). A good relationship between NO - 3 and Ca 2? was observed. Nitric acid could react with calcium carbonates and form calcium nitrates (MAMANE and GOTTLIEB 1992). There was a strong correlation between NO - 3 and SO 4, reaching.98, which illustrates that the two elements could be influenced by the same source, such as anthropogenic pollutions. 5. Discussions 5.1. Microphysical Factors that Affected Chemical Composition of Sea Fog To examine sea fog characteristics on Donghai Island, we first focus on the average spectra, represented by fog droplet size distribution. As shown in Fig. 7, the average spectra of Cases 1 6 show a unimodal distribution. The peak diameter was 2.8 lm, with an average maximum of number density at 2 cm -3 lm -1. Complex relationships were observed between fog ion composition and microphysical properties. Variation in the chemical composition of fog drops depended on the drop size. SO 4, NO 3 - and NH 4? were observed to be enriched in small droplets; and concentrations of Na? and Mg 2? were enriched in larger drops (RAJA et al. 28; LAJ et al. 1998; MOORE et al. 24; HOAG et al. 1999). The concentration of the constituents in fog water was also related to LWC as shown in detail Number density(/cm 3 µm) 1 Case1 Case2 Case3 1 Case4 Case5.1 Case6.1 1E-3 1E-4 1E Diameter(µm) Figure 7 Average spectra of Cases 1 6 by many scholars (Aikawa et al. 27a; FUZZI et al. 1984; ELBERT et al. 2). The negative correlations are shown between LWC and chemical compounds, such as Cl -,NO 3 -,SO 4,NH 4?,Mg 2? and Ca 2?, which indicates that an important part of the dissolved material entered the solution from aerosol particles (MILLET et al. 1996), while changes in fog solute concentrations were not a simple function of LWC (LI et al. 211). A new parameter S was used to describe gas-liquid interface more quantitatively by LU et al. (21), the scavenging potential calculated by S/LWC can be used as an indicator for explaining the relationship between ions concentrations of fog water and microphysics of fog. The higher surface/ volume ratio of small droplets could promote greater gas/liquid exchange and chemical reactions (SCHWARTZ 1988). S/LWC ratio is calculated to comprehensively reflect fog absorbing potential and dilution effect. Taking Cases 2, 3, and 6 as examples (Fig. 8), the variations of fog droplet number concentration, LWC, average diameter (D), and S/LWC ratio during these cases are discussed. Fog droplet number concentration (N) strongly depends on the phase and chemical compositions of particles that serve as the nuclei. It is also largely determined by how many aerosol particles are activated as nuclei. LWC could also be determined by N and droplets sizes (MA et al. 23). At the beginning, the characteristic quantities of LWC, N, and D were low; as a result the value of S/LWC ratio was relatively high with low S and LWC. The higher S/LWC ratio represented large scavenging potential of fog body with respect to the dilution effect of per unit volume of LWC. The concentration of pollutants was high, and the dissolved material and gas suspended in the atmosphere could enter the liquid water and be scavenged by the fog water. Consequently, the ions concentrations of fog water were high. During the maintenance process, Case 2 could be regarded as two separated cases with two main oscillations in LWC, N, and D, according to visibility. Cases 3 and 6 each had one main oscillation, but in Case 3 there were declines of LWC and N when the visibility was still below 5 m. When the fog droplets were mature and LWC was high, S/LWC ration was low. The pollutants settled down to the ground, and the ions concentrations decreased.

12 2242 Y. Yue et al. Pure Appl. Geophys. Visibility(m) Vis LWC N Case N(/cm 3 ) LWC(g/m 3 ) S/LWC(m 2 /g) S/LWC D D(µm)..2 22T2: 22T22:3 23T1: 23T3:3 Time(BST) 23T6: 23T8:3 23T11: Visibility(m) Vis N LWC Case N(/cm 3 ) LWC(g/m 3 ) S/LWC(m 2 /g) S/LWC D D(µm).. 23T2: 23T22:3 24T1: 24T3:3 Time(BST) 24T6: 24T8:3 Visibility(m) Vis N LWC Case N(/cm 3 ) LWC(g/m 3 ) S/LWC(m 2 /g) S/LWC D D(µm).. 6T: 6T1:1 6T2:2 Time(BST) 6T3:3 6T4:4 6T5:5 Figure 8 Temporal evolution of visibility, number concentration, average diameter, liquid water content, and S/LWC in Cases 2, 3, and 6

13 Vol. 169, (212) Chemical Composition of Sea Fog Water Along the South China Sea 2243 From the formation stage to the maintenance stage, the ions concentrations had a conspicuous decline. Moreover, they kept their fluctuations in a small range with a low value, in accord with the variation of all kinds of microphysics quantities. In the dissipation process, the three quantities mentioned above, except for S/LWC ratio, were decreasing obviously, along with the evaporation of fog droplets. Sometimes, the ions concentrations for fog water Table 5 Maximum and average values of N, LWC, and D in Cases 1 6 N (cm -3 ) LWC (g m -3 ) D (lm) Case 1 67 (16).33 (.13) 9.1 (4.8) Case 2 86 (3).6 (.1) 7. (4.4) Case 3 19 (6).97 (.26) 8. (4.4) Case 4 16 (7).89 (.27) 8.7 (4.6) Case 5 12 (26).7 (.13) 9.6 (5.2) Case (74).49 (.23) 4.5 (4.1) The figures in the parenthesis represent the average value of N, LWC and D might increase slightly compared to the previous stage. As shown in Table 5, the average N in Case 6 was maximum. Nevertheless, the LWC and diameter were low. For the fog droplets that consisted of the same condensation nuclei, the smaller the radius that made the surface enlargement, the more substances were dissolved (FENG et al. 29). What is more, when the number density was almost the same, the less the LWC, the more substances were dissolved, which is similar to the results of OGREN et al. (1992) and SCHELL et al. (1997). The high N, low D, and low LWC together contributed to the high total ions concentrations in Case Air Mass Origin In order to estimate the effects of transported pollutants on chemical compositions of sea fog water, the 7h backward trajectories for the six cases are described (Fig. 9). In all these cases, the air masses at low levels moved from the South China Sea and the Figure 9 Results of backward trajectory analysis at three levels (1, 5, and 1, m) for the six fog cases. a f Cases 1 6, respectively, and the starting time is in UTC format

14 2244 Y. Yue et al. Pure Appl. Geophys. Philippine Sea areas. Advection from different directions brought different kinds of pollutants to the area with little local pollutants, such as high ridges, to influence fog water compositions (KIM et al. 26; WATANABE et al. 26). As air mass passes over sea areas, it can transport the aerosols suspended in the atmosphere, including both oceanic and anthropogenic aerosols, over the sea to the coastal area. In Cases 2, 3, 5, and 6, the air mass at an altitude of 5 1, m transported over Hainan Island. The ion loadings in Case 4 were higher than those in Case 5. When the air mass passed over Hainan Island, which is full of human activities, the ion loadings or concentrations did not increase. One reason is that the island is a relatively clean place without heavy industries. In addition, the properties of oceanic air masses (that was generated over the ocean) and fog microphysics also had impacts on the ions concentrations. In addition, the air parcels from different regions affected by wind directions may have impacts on fog water and its microphysical structure (GOODMAN 1977; ZANNETTI et al. 1977). Different air mass would have different fog condensation nuclei (FCN) sources. The aerosol particles acting as cloud or fog condensation nuclei (CCN) determine the initial composition of fog droplets, which can be further altered by uptaking of soluble gases and by aqueous phase chemical reactions (JOHNSON et al. 1987; RAJA et al. 28) Comparisons with Other Studies Microphysical Properties Based on the studies of fog in the coastal areas (Table 6), it can be observed that the LWC of sea fog along the southern coast of China is lower than that in the continent regions. This result is similar to the studies by Fitzgerald (FITZGERALD 1978) in Canada and Goodman in US (GOODMAN 1977). Maoming and Donghai Island were located along the South China Sea with LWC less than.1 g m -3 (Table 6). LWC distribution is a consequential balance among advection, cooling, droplet gravitation settling, and turbulence in the liquid water budget of fog (ROACH et al. 1976; WELCH and WIELICKI 1986; ZHOU and FERRIER 28; WU et al. 29). Because the underlying surface has different roughness at the interface of the sea and the land, turbulence would be enhanced. Therefore, a large number of fog droplets would deposit. Another reason might be that the fog intensity was lower on Donghai Island than that in southern Ontario, Canada (GULTEPE and MILBRANDT 27) and Nanjing (LU et al. 21). The minimum visibility was higher than 1 m on Donghai Island. According to the new visibility parameterization (GULTEPE et al. 26), the higher visibility and lower number concentration would lead to low LWC. The fog droplet concentration of sea fog was also much smaller than that over the continent. In Chongqing, the value reached 1,746 cm -3, while the number concentration was no more than 2 cm -3 during sea fog processes. In some Chinese megacities, for example, in Shanghai, a considerable number of gases and particles are released into the air due to human activities (LI et al. 211). When the vapor is enough, the aerosol particles could act as FCN to form numerous fog droplets. The average maximum diameter and peak diameter of sea fog were similar to the continental fog. In some sea fog events, the maximum diameter was very large (YANG et al. 1989). Table 6 Mean values of key microphysics properties on Donghai Island and at other sites Observation site and period N (cm -3 ) LWC (g m -3 ) D (lm) D 1 (lm) D 2 (lm) 1. This work, 15 Mar 18 Apr Maoming, Guangdong, 16 Mar 29 Apr Nova Scotia Coastal, 2 Aug California coastal, 2 21 July Shapingba, Chongqing, Dec 23 1, Nanjing, Jiangsu, 25 Dec 27 Dec D, D 1 and D 2 represented average diameter, average maximum diameter and peak diameter, respectively. Sources: HUANG et al. (29), FITZGERALD (1978), GOODMAN (1977), LIU et al. (21), and NIU et al. (21a)

15 Vol. 169, (212) Chemical Composition of Sea Fog Water Along the South China Sea 2245 The reason might be the larger droplets are drizzle-sized drops; unfortunately, no direct measurements of drizzlesized drops were made. GULTEPE et al. (27, 29) proposed that a separation at about 15 lm sizewas likely due to the presence of drizzle-size droplets with diameter great than 15 lm Chemical Composition Properties Fog process could promote the deposition of atmospheric components to surface. Fog water composition is significantly influenced by scavenging aerosol particles. Obvious differences of chemical composition are present in different areas. SO 4,NH 4?, and Ca 2? are the predominant anions and cations in fog water over the continent. As shown in Table 7 (3), the concentration of SO 4 was 17,3 leq/l, which was 6.8 times and 36 times higher than the concentrations of NO 3 - and Cl -, respectively. Ca 2? was the most important cation in Shanghai (LI et al. 1999), whereas NH 4? was the main cation in the other four cities (4, 6 8 in Table 7). In the coastal area, the dominant anion and cation were Cl - and Na?, different from the continental fog. The maximum values among the sea area were 11,79 and 11,666 leq/l for Cl - and Na?, respectively. The different aerosols suspended in the atmosphere led to significant differences in fog water compositions. Along the coast, the wind from the ocean could transport a large number of oceanic aerosols, such as sea salt particles and so on. Therefore, the values of Cl - and Na? were the highest in sea fog water. While over the continent, the influence of industries and urbanization on ambient atmosphere is outstanding, which leads to high concentrations of SO 4,NH 4?, and Ca 2?. The sample-collection efficiency on Donghai Island (collecting 19 samples in 54 h) was higher than that in Nanjing, although the LWC on the island was low. The collection times of fog water during the experiment in Nanjing in 26 and in Zhoushan in 1987 were 99.8 and h, respectively, corresponding to the samples of 17 and 31 (MO et al. 1989; LU et al. 21). Along the coastal area, there were some large droplets that exceeded the measuring range and were not measured. These drizzle-size drops occurred during the fog events may contribute to fog water collection. 6. Conclusions During the sea fog field observation from 15 March to 18 April 21 on Donghai Island, China, Table 7 Mean values of ph, electrical conductivity (EC), and ion concentration on Donghai Island and at other observation sites Location Year ph EC F - (leq/l) Cl - - NO 3 1. This work (Donghai island) , ,79 4, Zhoushan, China Shanghai, China , ,53 4. Nanjing, China , Sea of Okhostsk El Tiro (continent samples) Houston South of Korea Location SO 4 Mg 2? K? Na?? NH 4 Ca 2? 1. This work (Donghai island) 1,532 2, , , Zhoushan, China Shanghai, China 17,3 6,1 51 1,76 1,63 13,22 4. Nanjing, China 6, ,282 6,654 3, Sea of Okhostsk 1, , El Tiro (continent samples) Houston South of Korea Sources: MO et al. (1989), LI et al.(1999), LU et al.(21), SASAKAWA and UEMATSU (25), BEIDERWIEDEN et al.(25), RAJA et al.(28), and KIM et al. (26)