Accessing the Impact of Sea-Salt Emissions on Aerosol Chemical Formation and Deposition over Pearl River Delta , China

Sea-salt aerosol (SSA) emissions have a significant impact on aerosol pollution and haze formation in the coastal areas. In this study, the Models-3/CMAQ modeling system was utilized to access the impact of SSA emissions on aerosol chemical formation and deposition over Pearl River Delta (PRD), China in July 2006. More SSAs were transported inland from the open-ocean under the southeast wind in summertime. Two experiments (with and without SSA emissions in the CMAQ model) were set up to compare the modeling results with each other. The results showed that the increase of sulfate concentrations was more attributable to the primary emissions of coarse SO4 particles in SSA, while the increase of nitrate concentrations were more attributable to secondary chemical formations, known as the mechanisms of chloride depletion in SSA. In the coastal areas, 17.6 % of SO4, 26.6% of NO3 and 38.2% of PM10 concentrations were attributed to SSA emissions, while those portions were less than 1% in the inland areas. The increases of PM10 and its components due to SSA emissions resulted in higher deposition fluxes over PRD, particularly in the coastal areas, except for the wet deposition of nitrate. Nitrate was more sensitive to SSA emissions in chemical formations than sulfate and dry deposition of aerosol was also more sensitive than that for wet deposition. Process analysis of sulfate and nitrate was applied to find out the difference of physical and chemical mechanisms between Guangzhou (the inland areas) and Zhuhai (the coastal areas). The negative contributions of dry deposition process to both sulfate and nitrate concentrations increased if SSA emissions were taken into account in the model, especially in Zhuhai. The negative contributions of cloud process also increased due to cloud scavenging and wet deposition process. In the coastal area, the gas-to-particle conversions became more active with high contributions of aerosol process to nitrate concentrations.


INTRODUCTION
Haze has received increasing attention with the rapid economic development and urbanization due to its important impact on human's life (Thurston et al., 1994;Liu et al., 2013).High concentrations of aerosol are to blame for the formation of haze.Long-term exposure to airborne particulate matter (PM) is strongly associated with adverse health effects (Yang et al., 2013), especially doing harm to people's respiratory system.Besides, aerosol can affect climate by scattering or absorbing solar radiation (direct effect) (Yu et al., 2006), altering cloud property and lifetime (indirect effect) (Lohmann and Feichter, 2005) or heating the atmosphere (semi-direct effect) (Johnson et al., 2006).With regional scale, air pollution problems such as haze and acid rain are strongly associated with aerosol.Major sources of ambient PM include not only the direct emission of primary particles, but also the formation of secondary aerosol by physical and chemical processes (Murphy et al., 2007).
Sea-salt emissions are important natural sources of aerosols produced by bubbles bursting during whitecap formation in open-oceans and wave breaking in the surfzone near the coastline.Sea-salt aerosols (SSA) are wholly coarse particles and could provide a surface for heterogeneous reactions and gas-to-particle conversion.Due to the broad ocean area and long coastline, the emission fluxes of SSA are tremendous.It is an important fraction of the PM and increases the PM levels, especially in the coastal areas.The mechanism of chloride depletion (McInnes et al., 1994) is well known in SSA.Zhuang et al. (1999) found that nitrate accounted for 65% of chloride depletion while during a heavily polluted period sulfate could contribute up to 29% of chloride depletion in Hong Kong.Thus, it is necessary to consider the SSA emission in the air quality modeling system in order to get more reasonable simulation results.Im (2013) found that 10-20% improvements were calculated in terms of model discrepancies of PM 10 mass in the Eastern Mediterranean coastal regions when considering SSA emissions.
Pearl River Delta (PRD) is one of the largest economic circles in China.The rapid economic development in PRD results in serious problems of air pollution, in which PM is the premier pollutant.The PRD is a coastal area located in the south of China with complex terrain.Influenced by the Asiatic monsoon, the characteristics of PM 2.5 vary from season to season in PRD (Louie et al., 2005).In summer, the wind is southerly due to summer monsoon.Also, under the influence of southerly winds associated with typhoons, more SSA is transported from open-ocean and surf-zone to coastal areas, even in an inland area, making it significant to take the SSA emissions into account for the purpose of accessing the mechanisms of aerosol pollution.
To date, aerosol study using air quality modeling becomes more popular.Air quality modeling systems such as Weather Research and Forecasting (WRF) model coupled with Chemistry (WRF/Chem) (Grell et al., 2005) and Comprehensive Air Quality Model with Extensions (CAMx) (ENVIRON, 2009) was widely used to study the mechanism of regional air pollution.In this study, WRF coupled with the U.S. EPA Community Mutiscale Air Quality (CMAQ) model (Models-3/CMAQ) (Byun and Schere, 2006) was utilized to access the effect of sea-salt emissions on aerosol chemical formation and deposition over PRD on July 2006.Two experiments with and without SSA emissions are designed to get the difference of modeling results.Modeling setting and data description are presented in Sect. 2. The modeling results are shown in Sect. 3 and Sect. 4 includes the conclusions.

Model Description
The Models-3/CMAQ modeling system was designed to reproduce the physical and chemical processes of multipollutants under the concept of "one atmosphere".In this study, the meteorological inputs are from the WRF model (Skamarock et al., 2008) and the emission inputs are from Sparse Matrix Operator Kernel Emissions (SMOKE) (Houyoux et al., 2003) model and Model of Emissions of Gases and Aerosols from Nature (MEGAN) (Guenther et al., 2006).CMAQ chemical transported model (CCTM), the main component in the CMAQ model, is used to calculate the physical and chemical processes with inputs of meteorological field and emissions.The simulation was performed using two-nested domain (horizontal resolution: 36/12 km) with 35 vertical layers.Fig. 1(a) showed the modeling domains for WRF and CMAQ.In this study, the results of the inner domain for CMAQ were analyzed.It was shown that this domain covered the whole area of Guangdong province, with the PRD centered in it.The physical and chemical options in the modeling system were shown in Table 1.Four dimensional data assimilation was also conducted in WRF, helping to improve the performance of simulation (Fan et al., 2013a).The CMAQ simulation time started at 0:00 (UTC) on June 25 th and ended at 0:00 (UTC) on August 1 st , 2006.The first 5-day was spin-up time and was not considered in this study.
Anthropogenic and biogenic emissions were included in the simulation.The emission inventories of point, area and mobile sources were from the work of Feng (2006).It was proved to be suitable for air quality simulations over PRD (Chen et al., 2009;Fan et al., 2013b).These emission inventories were then processed by the SMOKE model.Biogenic emissions were from the work of Guenther et al. (2006) and processed by the MEGAN model.Finally, the merge step in the SMOKE model was performed to combine the anthropogenic and biogenic emissions into one file and provided the inputs for the CMAQ model.

SSA Emission
SSA emissions are divided into open-ocean SSA emissions and surf-zone SSA emissions according to the source region.The open-ocean SSA emissions are calculated online by the CMAQ model using the Gong (2003) method, while surf-zone emission are computed using the open-ocean source function of Gong (2003) with fixed whitecap coverage of 100% and a 50-m-wide surf-zone (De Leeuw et al., 2000).Open-ocean and surf-zone fractions of each grid in the modeling domain are required for calculating the SSA emissions.In this research, these data are got from the coastline information using geographic information system software (ArcGIS).
Fig. 2 presented the distribution of SSA emission fluxes in July 2006 calculated by the CMAQ model.It was shown that SSA emission fluxes were higher in the surf-zone than in open-ocean.Longer coastline, larger areas of surf-zone and higher SSA emission fluxes.Highest emission fluxes in the modeling domain were found along the coastline of Hong Kong.For SSA emissions from open-ocean, emission fluxes were also high in the center of the ocean but decreased getting closer to the continent.In the modeling system, SSA emissions are comprised of Cl -, SO 4 2-, Na + , Mg 2+ , K + and Ca 2+ in coarse and fine mode.As shown in Table 2, coarse Cl -is the main component of SSA.Also, the emission fluxes of coarse particles are much higher than that of fine particles.

Experiments Setting
There are three kinds of emissions considered in the modeling system: anthropogenic emissions, biogenic emissions and SSA emissions.For accessing the impact of SSA emission, simulations with and without SSA emissions were performed.The control experiment, called BASE, was run with all these three emissions.The sensitivity experiment, called CASE, was run the same as the BASE experiment but not considering SSA emissions.The impact of SSA emissions can be achieved by comparing the modeling results of these two experiments.

Data Description
To study the meteorological and chemical characteristics of the PRD, the "Program of Regional Integrated Experiments of Air Quality over the Pearl River Delta" (PRIDE-PRD) campaign of 1-30 July 2006 was performed.Much research has focused on these experimental results (Verma et al., 2010;Fan et al., 2011).In this study, the observations at Guangzhou site (Fig. 1(b)) from this campaign was used to reveal the meteorological condition and air quality and to compare with the simulation results in this month.Guangzhou site (23.13°N,113.26°E) is located in the center of the PRD.The observations at this site included hourly temperature, relative humidity, wind speed and direction, concentrations of SO 4 2-, NO 3 -, PM 2.5 and PM 10 .The meteorological parameter data were collected at an automatic meteorological station during 1-31 July 2006.PM 2.5 and PM 10 concentrations were measured by a dust monitor and counter (model 265, Grimm Aerosol Technik, Germany) during 1-31 July 2006, while SO 4 2-and NO 3 -concentrations were measured by an in situ system with an aerosol collecting device during 6-30 July 2006.Daily precipitations at Shaoguan, Guangzhou, Heyuan, Shantou, Shenzhen and Zhanjiang stations were obtained from the meteorological bureau.The locations of these stations were shown in Fig. 1(b).In the present study, daily precipitations at these sites were summed up monthly to compare with the modeling results.
Besides, acid rain was collected at the measurement stations in Guangdong province.The locations of these stations were also painted in Fig. 1

RESULT AND DISCUSSION
In this section, firstly weather conditions and aerosol pollution episodes were discussed from the observations.Also, the modeling results were compared with the observations to show reasonable performance.And then the impact of SSA emissions on spatial distribution of PM concentrations and deposition fluxes was shown by comparison of modeling results between the BASE and CASE experiments.Also, Guangzhou site and Zhuhai site were selected to reveal the difference of SSA emissions impact on inland areas and coastal areas.Finally, process analysis equipped in the CMAQ model was conducted to find out how the contributions of individual atmospheric processes change due to SSA emissions.

Observation Analysis and Model Evaluation
Fig. 3 presented the temporal variation of meteorological conditions.There were three typhoons that affected PRD in July of 2006: Ewiniar, Bilis and Kaemi.Ewiniar formed on June 30 th in the Northwest Pacific and passed the east of Taiwan on July 8 th .The tracks of Bilis and Kaemi were similar and got closer to PRD on July 14 th and July 25 th , respectively.Influenced by typhoons, temperature increased, relative humidity decreased and wind direction switched from south east to north.Compared with the effects by Typhoon Bilis and Kaemi, the meteorological conditions (including temperature, relative humidity and wind direction) were less affected by Typhoon Ewiniar.It was shown that the WRF model could capture well the hourly evolution of temperature, relative humidity, wind direction and speed.humidity simulated by the WRF model.Wind speed was higher and wind direction was more southerly over the ocean, while wind speed was lower and wind direction was more southeasterly over land, probably due to the difference of surface roughness and thermodynamic properties between ocean and land.The simulated average wind direction at Guangzhou site was southeasterly, which matched with the observation in Fig. 4(b).In the center of the PRD, relative humidity was low due to heat island effect in urban areas.Fig. 4(c) showed the comparison of observed and simulated precipitation at the observed sites.The precipitation was higher in the east of Guangdong province (Heyuan and Shantou station).The WRF model also has a reasonable performance in the simulation of precipitation.
In addition, the CMAQ model did well in capturing the hourly evolution of PM and its components (Fig. 3).Before the typhoon reached PRD, the sinking flows around the periphery of the typhoon were not favorable for the diffusion of pollutants, leading to high PM levels on July 13 th and 24 th , respectively for Bilis and Kaemi.However, when typhoon got closer to PRD, precipitation began and the airborne PM was scavenged to the ground surface, leading to low PM levels since July 14 th and 25 th , respectively.

Impact of SSA Emissions on Aerosol Chemical Formation
For the purpose of revealing the impact of SSA emissions, the modeling results of the BASE experiment were compared with that of the CASE experiment.The difference of results for these two experiments (BASE minus CASE) could show the impact on aerosol air concentrations and deposition fluxes due to the inclusion of SSA emissions.In Fig. 5, it was shown that pollutant concentrations were higher in the center of PRD for the BASE experiments.In the east and west part of Guangdong province where altitudes are high (Fig. 1 2-was seen to be fewer over inland areas such as in Guangzhou.This increase was mostly attributed to the coarse and fine SO 4 2-particles in SSA direct emissions since large differences just appeared in the source region.For NO 3 -, the concentration changes due to SSA emissions rose by up to 1 µg m -3 .The most sensitive area also covered the coastal areas and the areas about several kilometers from the coastline.Unlike SO 4 2-, NO 3 -concentrations also increased in inland areas.Since there are no direct emissions of nitrate from SSA, these increases were attributed to the secondary chemical formation from SSA.Although the increases of nitrate concentrations were less than that of sulfate due to SSA emissions, nitrate concentrations were much less than sulfate concentrations, which showed that nitrate was more sensitive to SSA emissions than sulfate over the PRD region.As for PM 10 , since SSA was mostly comprised of coarse particles such as SO 4 2-, Cl -and Na + , concentration increases due to SSA emissions was significant around the source region, which reached up to 30 µg m -3 tremendously.
The chloride depletion mechanism is an important theory in gas-to-particle transfer for SSA.For the formation of sulfate, SO 2 can be oxidized into SO 3 and SO 4 in the atmosphere, and then reacts into H 2 SO 4 in the water.If there is SSA, fine sulfate particle and HCl will be generated.In this reaction, Cl -particle in SSA is lost.The chemical equation is: For the formation of nitrate, NO is oxidized into NO 2 , while NO 2 is decomposed into HNO 3 and NO in the water.If there is SSA in the atmosphere, the fine nitrate particle will be generated, also HCl.The chemical equation is: (3) Eq. ( 2) and Eq. ( 3) has been implemented in the CMAQ modeling system.As a result, besides the direct SSA emissions, the increases of PM levels were also attributed to the secondary chemical formations described in Eq. ( 2) and Eq. ( 3), while the increase of NO 3 -were all attributed to secondary chemical formation in Eq. ( 3).As shown in Fig. 5, high HCl concentrations were found in coastal areas for the BASE experiments.Looking into the difference between the BASE and CASE experiments, HCl concentrations increased due to SSA emissions, particularly in the coastal areas, where the concentrations have increased by up to 0.4 ppbv.The pattern of HCl concentration increases was similar to that of NO 3 -, which was attributed to the same chemical reactions in Eq. (3).However, chemical reaction in Eq. ( 2) also helped to explain the increases of HCl concentrations, but it was not significant.As a reactant in Eq. ( 3), the concentrations of HNO 3 decreased by up to 0.5 ppbv in the coastal areas.In this study, the partitioning of nitrate between gas and particle phases [NO 3 -/(NO 3 -+ HNO 3 )] (Kelly et al., 2010) in the modeling domain for the BASE and CASE experiments were calculated in Fig. 6.Higher nitrate partitioning meant that more HNO 3 was transferred into NO 3 -.For the CASE experiment, there were hardly no gas-to-particle conversions in coastal areas and open-ocean.But for the BASE experiments, this conversion took place in the ocean and were more active in inland areas.Nitrate partitioning has increased by up to 0.5 due to SSA emissions.

Impact of SSA Emissions on Aerosol Deposition
SSA emissions not only affect aerosol chemical formation, but also alter its dry and wet deposition.Since PM levels have increased due to SSA emissions, dry and wet deposition fluxes have also increased in July 2006.Fig. 7 showed that the maximum of dry deposition fluxed reached 1.2 kg ha -1 for SO 4 2-, 1 kg ha -1 for NO 3 -and 30 kg ha -1 for PM 10 for the BASE experiment over PRD.The distributions for SO 4 2-, NO 3 -and PM 10 were similar and higher dry deposition fluxed were found in coastal areas, where SSA emission fluxes were also higher, and urban areas such as Guangzhou and Shenzhen.Dry deposition fluxes increased by up to 1.2 kg ha -1 for SO 4 2-, 0.8 kg ha -1 for NO 3 -and 15 kg ha for PM 10 due to SSA emissions.Higher increases were found in coastal areas for SO 4 2-and PM 10 .However, as the major sources of NO 3 -increases were secondary chemical formations, regions with higher increases for NO 3 -was different from that of SO 4 2-and PM 10 and located in urban areas. 2-and NO 3 were higher than that of dry deposition fluxes, showing that wet deposition due to rainfall was significant in summertime.Unlike SO 4 2-and NO 3 -most of which were fine particles, PM 10 were almost coarse particles and dry deposition was significant.Thus, the amounts of dry deposition fluxes and wet deposition fluxes for PM 10 were similar.The increases of wet deposition fluxes for SO 4 2-due to SSA emissions were attributed to higher air concentrations in the coastal areas.However, wet deposition fluxes for NO 3 -slightly decreased with consideration of SSA emissions in the model.NO 3 -dissolved in the water can react into HNO 3 in the gas mode and evaporate from the water, leading to lower wet deposition fluxes for NO 3 -.It should be noted that this reaction was not significant and SSA emissions just accounted for about 1% of this decrease.Compared with the increases due to SSA emissions between dry and wet deposition, it was seen that dry deposition was more sensitive to SSA emission than wet deposition.

Comparison between an Inland Site and a Coastal Site
For revealing the impact of SSA emission in different regions, an inland site and a coastal site were selected to make a comparison.As shown in Fig. 1(b), Guangzhou (23.13°N,113.26°E) is an urban site located about 80 km from the coastline and the impact of SSA emissions is considered to be little.However, the anthropogenic emission fluxes here were relatively high.Zhuhai (22.27°N,113.57°E) is a suburb site located just about 10 km from the coastline, and it is thought to be greatly impacted by SSA emissions.Table 3 showed the air concentrations, dry and wet deposition fluxes of SO 4 2-, NO 3 -and PM 10 at these two sites for the BASE and CASE experiments.Air concentrations at Guangzhou site was about 2-5 times higher than that at the  Zhuhai site for the BASE experiment, since much emission sources was attributed to power plants, vehicles, cooking and so on in urban areas, Guangzhou.Air concentrations increases due to SSA emissions were minor and much less than 1%, indicating less impact of SSA emissions on aerosol concentrations in inland areas.However, the impact at Zhuhai site was tremendous.Air concentrations of SO 4 2-, NO 3 and PM 10 increased by 0.75 µg m -3 , 0.29 µg m -3 and 9.74 µg m -3 , respectively, due to SSA emissions.That was to say, 17.6 % of SO 4 2-, 26.6% of NO 3 -and 38.2% of PM 10 were attributed to SSA emissions at Zhuhai site.SSA emissions played an important role in aerosol concentrations in coastal areas, particularly in NO 3 -concentrations.Besides, for dry deposition, the impact of SSA emissions was also significant.Dry deposition fluxes of sulfate, nitrate were much less than 0.1 kg ha -1 for the CASE experiment at Zhuhai site, and that of PM 10 had been just 1.14 kg ha -1 .After considering SSA emissions in the BASE experiment, the dry deposition fluxes of SO 4 2-, NO 3 -and PM 10 have increased to 0.22 kg ha -1 , 0.17 kg ha -1 and 3.84 kg ha -1 , showing that 90.9%, 98.8% and 70.3% of dry depositions were from SSA emissions, respectively.At Guangzhou site, 44.4% and 12.3% of dry deposition fluxes for sulfate and PM 10 were caused by SSA emissions, respectively, less than that at Zhuhai site.However, the portion of nitrate dry deposition fluxes due to SSA emissions was over 97.3%, just a little less than that at Zhuhai site.Unlike SO 4 2-and PM 10 which were affected by SSA through direct emissions, NO 3 -affected by SSA was through secondary chemical formation.While the increases of nitrate from the CASE to BASE experiments at Zhuhai site was attributed to higher NaCl concentrations in Eq. ( 3), that at Guangzhou site was attributed to higher HNO 3 and NO 2 concentrations due to higher NO x emissions from vehicles.For wet deposition fluxes, the portion for SO 4 2-and PM 10 at Zhuhai site were 9.5% and 36.6%, higher than that at Guangzhou site, 0.2% and 1.3%, respectively.As wet deposition was affected by rainfall and aerosol concentrations, the impact of SSA emissions on it was complicated.

Result of Process Analysis
Process analysis is a diagnostic tool equipped in the CMAQ model.It provides quantitative contributions of individual physical and chemical processes to model prediction.There are two types of process analysis methods during a CMAQ simulation: integrated process rates (IPRs) and integrated reaction rates (IRRs).IPRs provide the contributions of individual physical and the net effect of chemical reactions to the concentration, while IRRs provide the contributions of individual chemical reactions to the net effects of chemical reaction on species concentrations.In this study, IPRs were conducted to reveal the physical and chemical processes of PM due to the inclusion of SSA emissions.The schematic equation of IPRs is as follows (Byun and Ching, 1999): IPRs isolate the changes of pollutant concentration into contributions from seven different types of physical and chemical processes as well as a mass conservation adjustment.These processes include horizontal transport (HTRA; sum of horizontal advection and diffusion), vertical transport (VTRA; sum of vertical advection and diffusion), emission process (EMIS), dry deposition (DDEP), aerosol process (AERO), cloud processes and aqueous chemistry (CLDS) and gas phase chemical reaction process (CHEM).In the CMAQ model, aerosol process refers to the net effect of aerosol thermodynamics, new particle formation, condensation of sulfuric acid and organic carbon on preexisting particles, and coagulation within and between Aitken and accumulation modes of PM.The impact of aerosol gaseous precursors on the formation of aerosol particles, such as H 2 SO 4 and HNO 3 , generated by the gasphase chemistry, are also included in the aerosol process.Cloud process refers to the net effect of cloud attenuation of photolytic rates, aqueous-phase chemistry, below-and in-cloud mixing with chemical species, cloud scavenging, and wet deposition.Process analysis tool has been widely used to reveal the mechanism of ozone formation and aerosol pollution (Xu et al., 2008;Liu and Zhang, 2013).In this study, as we focus only on aerosol, there are no contributions of gas phase chemical reaction to PM and it is not discussed in the following section.
Outputs of IPRs in the CMAQ model are just hourly contributions of individual atmospheric processes.Positive contributions to PM indicate that this process helps to increase PM concentrations, while negative contributions indicated that this process help to decrease PM concentration.
To study the results of process analysis in the whole month, monthly sum of hourly contributions from individual atmospheric processes was presented for sulfate (Table 4) and nitrate (Table 5) at Guangzhou and Zhuhai sites, 744 h in total, and positive and negative contributions were summed up separately.For sulfate, emission process was the main sources at Guangzhou and Zhuhai site.The contributions of the emission process at Guangzhou site was much higher than that at the Zhuhai site for the CASE experiment without considering SSA emissions, since there were much more emissions around Guangzhou site.The contributions of emission process made no changes due to SSA emissions.At Guangzhou site, except for the increases of negative contributions of dry deposition process from 124.9 µg m -3 month -1 to 207.0 µg m -3 month -1 , the differences of other contributions due to SSA emissions were minor.With more SSA transported from the coastal areas, positive contributions of horizontal and vertical transport processes slightly increased and negative contributions slightly decreased, which eventually helped to decline the sinks from transportation at Guangzhou site.As for Zhuhai site, the negative contributions of dry deposition process increased tremendously from 36.3 µg m -3 month -1 to 503.2 µg m -3 month -1 due to SSA emissions, showing the significant impact of SSA emissions in sulfate dry deposition.In addition, the negative contributions of cloud process at Zhuhai site increased from 89.0 µg m -3 month -1 to 169.0 µg m -3 month -1 due to SSA emissions by cloud scavenging and wet deposition processes.The contributions of aerosol process changed little due to SSA emissions at both two sites.
For nitrate, the contributions of the emission process at Guangzhou site were also higher than that at Zhuhai site.The negative contributions of dry deposition process increased from 29.0 µg m -3 month -1 to 864.2 µg m -3 month -1 due to SSA emissions at Guangzhou site.While at Zhuhai site, the negative contributions of dry deposition process were just 4.8 µg m -3 month -1 for the CASE experiment, and it increased up to 393.9 µg m -3 month -1 for the BASE experiment, which showed the important role of SSA emissions in coastal areas.For aerosol process, the negative contributions at Guangzhou site decreased due to gas-to-particles conversions in Eq. ( 3), as a portion of aerosol process.It was seen that SSA emissions had fewer impacts on the contributions of cloud process to nitrate concentrations at Guangzhou site, while that at Zhuhai site increase from 99.8 µg m -3 month -1 to 108.9 µg m -3 month -1 .

CONCLUSIONS
The Models-3/CMAQ modeling system was utilized to access the impact of sea-salt emissions on aerosol chemical formation and deposition over Pearl River Delta, China in July 2006.The sea-salt emission was calculated in the CMAQ model and the coarse particle was the dominant components.Under the influence of summertime monsoon, sea-salt was transported shoreward by the southerly wind system, leading to the difference of aerosol mechanism in the coastal area and the inland area.
The results of two experiments (with and without SSA emissions) were compared to access the importance of SSA.The difference of air concentration, dry deposition and wet deposition between BASE and CASE experiment showed that the impact of the SSA was significant in the coastal area and it decreased inland.In the coastal area, the PM 10 concentration increased by up to 30 µg m -3 , dry deposition fluxes increased by up to 15 kg ha -1 and wet deposition fluxes increased by up to 8 kg ha -1 .The increase of sulfate concentrations was more attributed to the primary emissions of coarse SO 4 2-particles in SSA, while the increase of nitrate concentrations was more attributed to secondary chemical formations, known as the mechanisms of chloride depletion in SSA.In the coastal areas, Zhuhai, 17.6 % of SO 4 2-, 26.6% of NO 3 -and 38.2% of PM 10 were attributed to SSA emissions, while those portions were less than 1% in the inland areas, Guangzhou.The ambient HNO 3 could react with Cl -from sea-salt emission and produced NO 3 -and HCl, leading to the decrease of HNO 3 and the increase of NO 3 -and HCl.The high nitrate partitioning ([NO 3 -/(NO 3 -+ HNO 3 )]) also showed that more HNO 3 was converted into NO 3 -.Gas-to-particle conversion was also taking place in sulfate, but it was less significant than nitrate.The increases of PM 10 and its components due to SSA emissions resulted in higher deposition fluxes over PRD, particularly in the coastal areas, except for the wet deposition of nitrate, which can react into gas HNO 3 in the water.Dry deposition of PM was more sensitive to SSA emissions than wet deposition.
Process analysis of sulfate and nitrate was applied to find out the difference of physical and chemical mechanism between Guangzhou (the inland city) and Zhuhai (the coastal city).The negative contributions of dry deposition process for both sulfate and nitrate increased if SSA emissions were included, especially for Zhuhai.Besides, the negative contributions of cloud process also increased for sulfate and nitrate.In the coastal area, the gas-to-particle conversion became more active and leading to high contributions of aerosol process to nitrate.In short, the coastal area was more sensitive than the inland area, and nitrate was more sensitive than sulfate to SSA emissions.
It should be noted that this study only focused on PRD and this type of study on a larger scale will be considered in the future research.Also, we need to assess the combined impacts of seasonality in meteorological conditions and SSA emissions with longer term simulations such as one year.Besides, direct productions of nitrate from N 2 O 5 and NO 2 were not implemented in the CMAQ modeling system in this study.For accessing the impact of the SSA, it is meaningful to make further researches in the following study.

Fig. 1 .
Fig. 1.(a) Two-nested modeling domains for WRF (black frame) and CMAQ (red frame).(b) The inner CMAQ domain with shaded terrain height.The blue square points denote observed sites for wet deposition.The red circle points denote observed sites for precipitation.Guangzhou site represents the inland areas and Zhuhai site represents the coastal areas.
(b).The observations included precipitations and ion concentrations such as SO 42-and NO 3 -for each rainfall in July of 2006.To get the monthly wet deposition fluxes, the equation was as follows: WET denotes monthly wet deposition fluxes (kg ha -1 ); I denontes the number of precipitation in the whole month; ION denotes ion concentrations (mg L -1 ) and PREP denotes

Fig. 3 .
Fig. 3. Hourly evolution of meteorological elements and air pollutant concentrations at Guangzhou site for the BASE experiment.

Fig. 4 .
Fig. 4. (a) Simulated monthly-mean wind field and relative humidity in the model domain in July 2006.(b) Observed wind direction frequency at the Guangzhou site in July 2006.(c) Observed and simulated precipitation at various sites (shown in Fig. 1 (b)) in July 2006.

Fig. 8 .
Fig. 8. Simulated wet deposition fluxes of SO 4 2-, NO 3 -and PM 10 for the BASE experiment and the difference between the BASE and CASE experiments in July 2006.The dots with color denote the observed values.

Table 4 .
Process analysis results for sulfate at Guangzhou and Zhuhai sites.Pos.denotes the sum of hourly positive contributions in July, and Neg.denotes the sum of hourly negative contributions."-" means this contribution is non-existent.Unit in the table is µg m -3 month -1 .

Table 5 .
Process analysis results for nitrate at Guangzhou and Zhuhai sites.Pos.denotes the sum of hourly positive contributions in July, and Neg.denotes the sum of hourly negative contributions."-" means this contribution is non-existent.Unit in the table is µg m -3 month -1 .