Observational study of influence of aerosol hygroscopic growth on scattering coefficient over rural area near Beijing mega-city

We investigated aerosol hygroscopic growth property and its influence on scattering coefficient using M9003 nephelometers in coupling with humidity controlled inlet system at a rural site near Beijing mega-city from 24 April to 15 May 2006. Inlet relative humidity was controlled in an increasing range of 40%–90% while aerosol hygroscopic growth factor of scattering coefficient, f (RH=80%) as ratio of scattering coefficient at RH=80% to “dry” scattering coefficient (RH<40%) varied in a range of 1.07–2.35 during the measurement. Further analysis indicated that under dust episode, measured f (RH=80%) is 1.2±0.02, and estimated periodic mean value of f (RH=80%) was 1.31 ±0.03 under clean periods; during urban pollution periods, the aerosol displayed relative strong water absorbing properties with f (RH=80%) of about 1.57 ±0.02. An examination of chemical composition of daily filter samples highlighted that aerosol hygroscopicity was generally depressed with the increasing ratio of organic matter (OMC)/ammonium sulfate (AS) in particle mass, similar with the results of many previous studies. However, a special case with high value of f (RH=80%)=2.21 and high OMC/AS ratio was also observed, this exception reflected physico-chemical particularities of organic matter and its complex interaction with other compounds during this episode.


Introduction
Aerosol hygroscopic growth with increasing relative humidity (RH) may lead to dramatic changes in its mass concentration, size distribution and corresponding optical properties (scattering coefficient, single scattering albedo, asym-Correspondence to: J. Tang (tangj@cams.cma.gov.cn)metry factor, etc.), which could enhance the cooling effect of aerosols in the atmosphere by directly scattering more light radiation (Carrico et al., 1998(Carrico et al., , 2000;;Kotchenruther et al., 1998;Randles et al., 2004), or change cloud microphysical properties (Crumeyrolle et al., 2008) by serving as cloud condensation nuclei (CCN) (Houghton et al., 2001).To precisely evaluate the direct radiative forcing, light scattering coefficient (σ sp ) of aerosol particles and its dependency on relative humidity (RH), defined as f (RH)=σ sp (scanning RH)/σ sp (dry), have been investigated for decades through Integrating Nephelometers equipped with humidity control devices in clean background regions, marine boundary layers, rural and urban areas (Kotchenruther et al., 1999;Malm et al., 2001Malm et al., , 2005;;Carrico et al., 2003;Magi et al., 2003;Kim et al., 2006).Previous studies reported that different types of aerosol particles usually have distinct hygroscopic growth properties (Tang, 1996;Cruz et al., 2000;Kim et al., 2006).Hand and Malm (2006) indicated that the scattering coefficients of (NH 4 ) 2 SO 4 and (NH 4 )HSO 4 aerosols could be enhanced by a factor of three when relative humidity is over 85%.Dust particles, dominant in coarse mode, are mostly insoluble with f (RH=80%) smaller than 1.1 (Li-Jones et al., 1998), but they could also be hygroscopic when coated by sulfate or other soluble inorganic aerosols during transportation (Perry et al., 2004;Shi et al., 2007).During the period of Aerosol Characterization Experiment (ACE-Asia), dust particle's hygroscopic factor of f (RH=80%)=1.25 was observed in Ron Brown cruise (Carrico et al., 2003), and even as high as 2.0 in Korea (Kim et al., 2006).Combustionemitted smoke organic matter and photochemical formed organic compounds, accounting for a significant proportion of particulate aerosols, could be hygroscopic or hydrophobic depending on the types of organics and their oxidation status in the atmosphere.Studies (Carrico et al., 2005) conducted in the Yosemite National Park indicated an inverse relationship between the ratio of organic carbon mass to Published by Copernicus Publications on behalf of the European Geosciences Union.X. L. Pan et al.: Observational study of aerosol hygroscopic growth factors (NH 4 ) 2 SO 4 (OMC/AS) and the hygroscopicity of aerosols.Researches carried out by Malm et al. (2005) in the same location showed that f (85<RH<90) decreased from 2.0 to <1.2 as OMC/AS increased from 0.57 to 11.15.Nevertheless, inorganic aerosols' hygroscopicity enhanced by organic acids were also observed in some studies (Cruz et al., 2000;Choi et al., 2002).Therefore, following the great temporal and spatial variations of aerosols' hygroscopic growth properties, in situ and laboratory investigations are needed.
In the past two decades, rapid economy growth triggered China's annual GDP reaching over 9%, growing number of mega-cities (population over 10 millions), and increasing number of private cars (http://www.stats.gov.cn/), and consequently, the country become an important source of pollutants and aerosols production.Aerosols mass concentrations and their optical properties must have changed to some extent.In such condition, providing in sights into aerosols optical properties and its dependence on relative humidity in some Chinese key regions seems to be an important scientific issue to better understand the relationships between aerosols and climate.Previously, studies of Yan et al. (2008) at Shangdianzi (SDZ), a baseline air pollution monitoring station of North China, showed that the f (RH=80%) was 1.2 in clean air conditions and 1.48 during pollution episodes.Kim et al. (2006) obtained higher value of 2.75 during the ACE-Asia period.Experiment carried out in the Pearl River Delta (PRD) region (Liu et al., 2007) also indicated the urban pollutants aerosols hygroscopic growth factor f (RH=80%) of about 2.04 for Guangzhou.
The present analysis of the aerosol scattering hygroscopic factors f(RH) over the rural area near Beijing mega-city is based on an observational experiment performed from 24 April to 15 May 2006.The main purpose of this study is to investigate the characteristics of aerosol scattering properties and its dependency on the relative humidity f(RH).The effect of aerosol chemical composition, meanwhile influence of origin and mixing process during transportation with air mass back trajectory analysis will be also discussed.

Site description and meteorology
Measurements of aerosol scattering coefficient, f (RH), and aerosol chemical compositions were conducted at Xin'An weather operational station in Baodi county (referred simply as Xin'An site, 39.73 • N, 117.51 • E, altitude: 6 m) located in the rural areas of Jing-Jin-Tang region (Fig. 1).Baodi is a major agricultural district (region) of Northern China where rice and corn are cultivated.The principal human economical activities are farming and tree growing.This site which is about 85 km, 70 km and 105 km far from Beijing, Tianjin and Tangshan respectively, seems to be strongly under mega-cities air pollution influence.During the observation, meteorological condition was characterized by clear and sunshiny days, except the floating dust event at approximately 16:00 LST (Local Standard Time) 30 April.Air temperature average of experimental period was 16.1 • , and relative humidity about 52.9%, and surface wind direction at the Xin'An site was dominated by NW and E with wind speed less than 5 m/s.The observation site was particularly frequently likely influenced by anthropogenic perturbed air masses from industrial areas of Beijing and Tangshan City (Fig. 1).

Integrating nephelometer
Aerosol scattering coefficient (σ sp ) was measured with M9003 integrating nephelometer at 525 nm wavelength with light integrating angle 10 • -170 • .The measurement range of this instrument is 0-2000 Mm −1 (1 Mm −1 =10 −6 m −1 ) with its lower detection limit of about 0.3 Mm −1 (lower than one tenth of Rayleigh scattering of air molecules), detailed information about this instrumentation have been described in many previous studies (Liu et al., 2007;Yan et al., 2008).During the experiment, a water-proof wire strainer was mounted in the head of inlet tube, and the ambient air was inhaled into nephelometer measuring chamber through 1.5 m 1/4 anti-static tube, air temperature and relative humidity were measured by sensor in the front of the measuring cavity.

Humidity control device
The dependency of increase of aerosol scattering coefficient on relative humidity, f (RH), is represented by the ratio of aerosol scattering coefficient at conditioned RH (from <40% to 90%) and that under the reference "dry" conditions (usually with RH<40%).This method was initially introduced by Covert et al. (1972).Relative humidity of Inhaled air was regulated by a connected heating tube and humidifier.Schematics of humidifier is similar with studies at Earth System Research Laboratory, National Ocean & Atmospheric Administration, USA(NOAA/GMD) (http://www.esrl.noaa.gov/gmd/aero/instrumentation/humid.html).The Humidifier consists of a deionized water bath in sheath plexiglass pipe and a water-vapor penetrating membrane tube immerged through it.The water Vapor permeated to mix with sample air was modulated by controlling of the water temperature with power of heating equipment.Conditioned relative humidity and temperature of inflow air were measured with a build-in sensor in the nephelometer.

Experimental procedure and quality control
During the observation period, the nephelometers were calibrated with the zero gas (dry filtered air) every 24 h, and with the standard calibration gas (HFC-134a) every 7 days.The data were automatically recorded every 5 min.Additional test on humidity sensor and dry scattering measurements were performed.The comparison of humidity sensors of two nephelometers with HMP41/45 Visala humidity indicator showed in good agreement with the difference less than 3%.Measurement of f (RH) has been obtained using two nephelometers in conjunction with two inlet in parallel; one nephelometer was kept inflow air mass in dry condition (RH<40%) through its built-in heating tube, while another regulated the sampled air RH from <40% to 90% using humidity control device.In spite of well maintenance of the nephelometers, there still occasionally subsisted some discrepancy between the measured signals of two nephelometers when RH<40%, that probably due to drifting electronic signal of Photomultiplier or disturbance of inlet tube.We firstly removed the periodic data of bias more than 20%, and then normalized the signals to make sure f (RH<40%) was set to one (Day et al., 2000).Comparison results of two integrating nephelometers are showed in the Fig. 2.After these processes, hygroscopic growth factor of aerosol scattering coefficient could be determined, as showed in Eq. (1), The changing ratio of scattering coefficient obtained by two nephelometers with RH are then fit into the following empirical equation. (2)

Aerosol sampling and chemical analysis
Ambient aerosol particles were collected by filters using two sets of Andersen size-segregated impactors (KA200), the cut-off size ranges are: <0.43 µm, 0.65-0.43µm, 1.1-0.65 µm, 2.1-1.1 µm, 3.3-2.1 µm, 4.7-3.3µm, 7-4.7 µm, 11-7 µm and >11 µm, respectively.Teflon membranes were used as impaction substrates (collecting water soluble inorganic samplings) for the top eight stages, and Zeflour membranes (Pall Corporation) were for the backup filter (the bottom stage).Organic carbon matter was sampled with Quartz for analysis.Collection period of organic samplings were on 3, 4, and 10 to 15 May, and aerosol mass concentration and inorganic samplings were performed from 24 April to 15 May (Table 1).Aerosol mass concentrations were determined by means of weighting filters with Satorius microbalance (precision: 10 µg) before and after sample collection in the glove box.The filters were balanced in the glove box at a stable relative humidity (40±2%) and temperature (20±1 • ) environment for about 72 h prior to weighting process.Each membrane was weighted three times in 2 days according to the standard procedure, and the difference of gross mass of the filter weighted each time was less than 50 ng.

Classification of different observation periods
During spring over Northern China, aerosol chemical properties are rather complex because of the mixture of different sources including transportation of heavy air pollution from urbanization and industrial activities, desert dust particles from frequent dust events, and local emitted plant organic compounds during vegetal "greening".We categorized observation periods into four different groups according to records of weather phenomena, aerosol mass concentration and 72-h air masses pathways.Air masses back trajectories were calculated by means of the Hybrid Single-particle Lagrangian Integrated Trajectory (HYSPLIT) model (Draxler et al., 1998).Meteorological fields with 3-h time interval (formatted as GDAS) were obtained from US National Centers for Environmental Prediction (NECP).The back trajectory starting geographical position was located at Xin'An site with starting altitude of 100 m.The procedure of dissociation of different periodicity can be described as follows: 1. Detect period influenced by pronounced dust aerosols in accordance with observational records and 3-h meteorological information (Provided by China Meteorological Administration, CMA).The floating dust episode occurred in the afternoon of 30 April (approximately 17:00 LST) while the local visibility was less than 1 km.72-h back trajectories shows that air masses are originated from central Mongolia and past through Hunshandake desert before arriving at Xin'An site.
2. Select the periods where mass concentration of fine particles (PM 2.1 , particle diameter less than 2.1 µm) are less than 50 µg•m −3 as "clean periods", and use this value as delimitation of clean-to-pollution on the ground where the value is approximately equivalent to Air Pollution Index (http://www.mep.gov.cn/) of about 50 (indication of "good" air quality), and back trajectories shows incoming air masses have not passing through anthropogenic sources with probable pollutant emission for 72 h.
3. Concerning the geographic location of Xin'An, observation site located right in the center of three pollution source areas, Beijing, Tianjin and Tangshan (shown in Fig. 1), and urban emitted anthropogenic pollutants might be easily transported to the site, here we choose the periods when PM 2.1 mass concentrations over 50 µg•m −3 as influenced by regional urban pollutions.Back trajectories analysis also indicates that most of air masses came from the pollutant area of nearby regions in prevail of southern wind, and there are also some periods when air masses were stationary over Beijing and Hebei provinces under synoptic high pressure weather system.et al., 2001;Duan et al., 2004) showed an approximately same OC concentration level at the near study site (suburban of Beijing); the high OC concentration was mostly in connection with urban residential activities (Yang et al., 2005) and biomass burning (Duan et al., 2004).In PM 2.1 category, mass concentration of water soluble SO 2− 4 ion was 11.5(7.6)µg•m −3 , followed by NO − 3 and NH + 4 ions with value of 6.05(4.52)µg•m −3 and 4.02(3.17)µg•m −3 .This indicated anthropogenic urban pollution influence during the observation period.In the PM 11 , soluble Ca 2+ ion makes up the second largest proportion of aerosol masses that may be Atmos.Chem.Phys., 9, [7519][7520][7521][7522][7523][7524][7525][7526][7527][7528][7529][7530]2009 www.atmos-chem-phys.net/9/7519/2009/related to flowing soil dirt during the springtime generally featured by low ambient relative humidity and strong wind.

Size distribution
Aerosol mass size-distribution displays bi-modal feature during the periods of pollutant condition (Fig. 3a), and the lower-diameter peak appears in the range of 0.33∼0.47µm with mass concentration of 27 µg•m −3 averagely, maximum percentage of ammonium sulfate in all detected water soluble inorganic matter accounts for 52.2% in particle size less than 0.43 µm and that of ammonium nitrate is 25.1% in diameter range of 0.65∼1.1 µm, indicating the influence of urban emission pollutant and the production of gas-to-particle formation.In the coarse mode, high concentration (averagely 33.41 µg•m −3 ) of aerosol particle occurs in the diameter of 4.7∼7.0µm, and Ca 2+ ion takes up about 47.8%, it is more likely resulted from the floating dirt and crustal substances for strong wind and low ambient relative humidity during spring time (Fig. 3b).
During clean and dust influencing periods, aerosol mass concentration display mono-peak (appears in diameter of 3.3∼7.0µm) distribution, with maximum value of 28.5 µg•m −3 and 179.3 µg•m −3 , respectively (Fig. 3a).In clean condition (Fig. 3d), the average percentage of sulfate and nitrate in fine mode (<PM 2.1 ) are 24.6% and 9.04%, less than half values in pollutant periods.Ca 2+ is most abundant materials with average percentage of 45.2% in fine mode (<PM 2.1 ) and 60.3% in coarse mode (>PM 2.1 ).During dust episode, the proportion of Ca 2+ mass concentration occurs in the size range of 7.0∼11.0µm with value of 74.4%, and gradually decreases to averagely 17.2% in the fine mode.Percentage of sum of sulfate and nitrate arise from 16.5% (7.0∼11.0µm) to 46.7% (<0.43 µm) (Fig. 3c).
Lorenz-Mie theory, analytical solution of Maxwell's equations for the scattering of electromagnetic radiation by spherical particles, provides insights into chemo-optical relationships of aerosol community in the atmosphere.During calculations, we assumed that all aerosol particles were externalmixed and in shape of perfect sphere (McMurry et al., 1996).The chemical species involved EC, OC, (NH 4 ) 2 SO 4 , NH 4 NO 3 , and water soluble ion (F − , Cl − , Na − , Mg 2+ ) and rested crustal substances; complex refractive indices and densities of species were taken from reference (Larson et al., 1988).Aerosol light scattering coefficient was integrated with aerosol number distribution and Mie scattering efficiency (derived directly from Lorenz-Mie calculation) for each particle species.
Reconstructed total ambient light scattering by particles in the Interagency Monitoring of Protected Visual Environment (IMPROVE) protocol are given by Eq. (3) in the premise of an externally mixed aerosol community, as follows: where f (RH) is hygroscopic factor of aerosol scattering coefficient as mentioned above (it is equal to one in the dry condition), R OC is multiplied factor to convert organic carbon to total particulate organic matter, Turpin and Lim (2001) (recommended a factor of 1.6 for urban organic aerosols, and even 2.1 for non-urban organic aerosols).According to closure result in Yosemite Nation Park, R OC value of 1.8 allowed for fine mass closure in both biomass burning and regional haze conditions (Malm et al., 2005), and we took value of 1.8 for calculation as IMPROVE suggested (http: //vista.cira.colostate.edu/improve/Publications/GrayLit/016 IMPROVEeqReview/IMPROVEeqReview.htm) hereafter.The mass concentration of chemical compositions in PM 2.1 was used in closure calculation under the assumption that fine particles contribute large portion of scattering, and this hypothesis has been supported by the Lorenz-Mie theory results (estimated contribution of PM 2.1 scattering effect to total scattering coefficient averagely 80.6% in this study) and previous studies (Meszaros et al., 1998).
Calculation results of two closure methods mentioned above are displayed in Fig. 4. The scattering coefficients closures show good agreements to the measurements, with R 2 of 0.89 and 0.82 for IMPROVE method and Lorenz-Mie theory method.Based on analysis of mass chemical information, relatively large discrepancy between the calculations and the measurements occurs on these days in particular under coarse particles high concentrations condition.Local perturbed aerosols arise from agriculture activities may be a plausible explanation.In fact, some soil tillage occurred near the site within 100 m in distance according to meteorological and aerosol observation records.

Distinction of different scattering hygroscopicity
Measured aerosol scattering growth feature with relative humidity is showed in Fig. 5.All the f (RH) displayed a similar appearance in the general shape of the hygroscopic growth (monotonic increasing) in spite of significant scattering of data points, and temporal variation of f (RH=80%).Different hygroscopic growth curves are categorized according to weather condition (Sect.3.1) and high variations of chemophysical features (listed in    dominant aerosols types at Xin'An site will be discussed afterwards.Comparisons with other experimental results for aerosol's hygroscopic properties in different weather conditions are listed in Table 3.

Clean period
During clean days, aerosol scattering growth features with relative humidity, showed large temporal variations (Fig. 5) with averaged measured of f (RH=80%)=1.31(Table 2), corresponding to about 10% higher than (f (RH=80%)=1.2) indicated by Yan et al. (2009) for another rural site (100 km northeast far from Xin'An site) of north Beijing, and about 8.1% and 24.1% lower than the results (f (RH=80%)=1.42) reported by Koloutsou-Vakakis et al. (2001) for the Northern Hemisphere continental site (Bondville, Illinois, US) and (f (RH=80%)=1.67)measured by Carrico et al. (2000) in the clean air conditions in Sagres (Portugal, ACE-2) respectively (Table 3).In clean days, site prevails northerly wind and mean mass concentration of PM 2.1 and ammonium sulfate are 43.5 µg•m −3 and 6.4 µg•m −3 , respectively (Fig. 3d), dry scattering coefficient is averagely 89.2 Mm −1 .In Bondville site, aerosol scattering coefficient is averagely only 33.4 Mm −1 during spring time, comparably lower than the result in surrounding area of Beijing, however local high proportion of sulfate burden in PM and SO 2 concentration, reported by Charlson et al. (1991) and Harris et al. (1994) might be the possible contributions to relatively higher f (RH=80%).According to the results in Sagres, lower dry (RH<30% ) scattering coefficient of aerosol (only 19.4±8.9Mm −1 for PM 10 during clean period) related with different aerosol physical chemical properties could partially account for that discrepancies.Studies of Yan et al. (2009) were performed in the baseline air monitoring station of Northern China, where geographical environment within 30 km radius distance was characterized by rolling hills farmland, orchard and forests with lower pollutants concentrations and scattering coefficient (averagely 30 Mm −1 ).

Dust period
Measured f (RH=80%) was averagely 1.20 in the dust episode.This value seemed to be close to the results obtained by Carrico et al. (2003), of which f (RH=82.5%)were 1.18 and 1.39 for particles size less than 10 µ and 1 µm respectively in East Asia (ACE-Asia) during dust dominant period.Observational experiment performed by Kim et al. (2006) in Kosan (South Korea) indicated aerosol f (RH=85%)=1.73-2.20 during dust episode (Table 3).They reported the aerosol dry scattering coefficient of 238.6 Mm −1 , only 33.7% of value in Beijing.These differences might be resulted from the mixing effect of dust particles with other inorganic materials during transportation.This is supported by many previous studies on dust particles over marine (Yamato et al., 1994;Zhang et al., 2003) or urban areas (Zhang et al., 1999; 2) for different aerosol types.Black, red, blue and green refer to the results of dust, clean, urban pollution and special case for urban pollution episodes, respectively.Wang et al., 2007) of which the authors demonstrated that the mixing of sulfate, nitrate, and coarse calcium in dust particles could substantially enhance the hygroscopicity.coarse calcium dust particles with nitrate-containing could appears in aqueous phase even at 15% RH (Shi et al., 2007).

Pollution episode
Under urban anthropogenic pollutants dominance condition, humidograms showed little evidence of deliquescence but monotonic changes in scatting coefficient with increasing control RH (Fig. 5).Measured aerosol hygroscopic growth factor f (RH=80%) is 1.57, about 6% greater than the results (f (RH=80%)=1.48)obtained by Yan et al. (2008) for periods under the downwind of urban plum from Beijing city.
In this episode, measured dry aerosol scattering coefficient is 311.2Mm −1 (three times higher than that of clean period), mean mass concentration of PM 2.1 is 87.9 µg•m −3 and ratio of PM 2.1 to PM 11 increase from 30% (clean periods) to 42.8%, percentage of sulfate and nitrate in the water soluble inorganic enhance to 42.8% and 20.4% in PM 2.1 and they accounts for 63.2% of total ions (Fig. 3b), which all indicates the influence of urban emitted pollutants.Air mass back trajectory analysis show that air pathway has passed through polluted regions in Tianjin, Hebei and Shandong provinces.Kim et al. (2006) reported approximately the same f (RH=80%)=1.55-1.77for local urban pollution (from Korea industrial area) episode in Gosan regional background site, and f (RH=80%)=2.0∼2.43 of pollutants from east region of China (about 26%∼43% higher than the results).We speculate possible alteration of physico-chemical properties of aerosol particles during the transport through over the East Chinese Sea in terms of the measured aerosol single scattering albedo (0.89±0.02).The latter is higher than results reported by Bergin et al. (2001) in urban area of Beijing (0.81) and those of Yan et al. (2008) and Qiu et al. (2004) in background monitor station in North China (0.85) and in www.atmos-chem-phys.net/9/7519/2009/Atmos.Chem.Phys., 9, 7519-7530, 2009 urban area of North China (0.83) respectively.ACE-2 experiments conducted by Carrico et al. (2000) in Sagres showed that f (RH=80%) value of 1.40 for anthropogenic pollutant (PM 10 ) from Europe, about 6.6 % lower than measured results in the present study.
Many other studies (listed in Table 3) indicated more hygroscopic properties of urban aerosols, 2.04±0.28 in PRD region (Liu et al., 2007), 1.7-2.0 in Yangtze delta region (Xu et al., 2002) and 1.81±0.37∼2.30±0.24 in the flight over east coast of the United States (Kotchenruther et al., 1999).Among these studies, Liu et al. (2007) determined the urban pollution periods following air mass back trajectories, and found that aerosol hygroscopicity was due to higher water soluble mass fraction (averaged 44.8%) and lower mass fraction of total Carbon (34.7% in PM 2.5 ).Kotchenruther et al. (1999) classified the periods by airflow patterns, result indicates that airflows consistently influenced by anthropogenic sources show less scattered data at high RH, the mean total carbon as percentage of total aerosol mass and that of sulfate mass are 49.8% and 38.6%, with f (80%, 550 nm)=2.1±0.4 averagely.

Special pollution episodes
Under significant influence of urban pollution, we distinguish two main episodes (in the morning of 11 May and on 15 May, respectively) of which aerosols hydrophilicity and its effects on scattering coefficient are relatively higher with measured mean hygroscopic growth factor f (RH=80%) of 2.33-2.48(green circles in Fig. 5).Detailed descriptions of experimental processes during those two days are as follows.
There are three hygroscopic growth experiments on 11 May.During the first experiment (began at 07:45 LST, and ended at 12:30 LST), aerosol scattering hygroscopic growth factor f (RH) gradually enhances with RH increase from 38.0% to 91%, and f (RH=80%)value of 2.51.The second experiment (began at 15:40 LST and ended at 1820 LST) and the third experiment (from 19:05 LST to 22:30 LST) result in approximately same values of hygroscopic growth factor, f (RH=80%)=1.31 and f (RH=80%)=1.33 respectively.During the first experiment, measured dry aerosol scattering coefficients indicates mean dry scattering coefficient of about 211.2 Mm −1 with standard deviation of 96.8 Mm −1 ; the observation site prevails southerly wind (averagely 1.85 m•s −1 ) with gradually change of wind direction to northerly strong blow (averagely 3.12 m•s −1 ) in the afternoon, and the dry scattering coefficient decreases to 102.5 Mm −1 .This suggests that aerosol types over observation site along air movement are physically and chemical distinct.Therefore, 24-h averaged chemical composition representing daily average condition seems to be insufficient to provide accurate explanation on the relationship between f (RH) and chemical compounds.
Experiments (from 10:35 LST to 22:30 LST) performed on 15 May all show high f (RH=80%) of 2.56 in daytime and relatively lower value of 1.96 at night.Corresponding scattering coefficients are 379.3Mm −1 and 310.0 Mm −1 , respectively.Meteorologically, the site is characterized by weak southerly and relative high ambient temperature (26.5 • averagely), and low relative humidity (31.4% averagely), so that there is no evidence change of synoptic of weather condition with respect to 72-h back trajectories.Scattering closure calculations performed by the two methods indicate good consistency with measured scattering coefficient.High mass concentration of organic matter (32.1 µg•m −3 in PM 2.1 ) may play certain positive role in aerosol particle absorbing waters as discussed in Sect.3.5.

Scattering hygroscopic growth function
Large variations of aerosols water absorbing ability appear among different dominant aerosols types, and fitting aerosols hygroscopic growth trends in dust, clean, urban pollution and special cases periods are shown in Fig. 5. Aerosols hygroscopic growth functions are obtained from Eq. ( 2), and used for monotonic growth illustration (Kotchenruther et al., 1998(Kotchenruther et al., , 1999;;Carrico et al., 2003;Magi et al., 2003;Liu et al., 2007).The parameters of (a) and (b) are described in Table 4.Both (a) and (b) seem to be undergone a sharply increase from dust particles (a=0.64,b=5.17) to pollutant period (a=2.30,b=6.27).

Influence of organic compounds on aerosol hygroscopicity
Water soluble inorganic ions and organic carbon matter are key factors directly correlated to aerosols water absorbing abilities as reported by (Virkkula et al., 1999;Malm et al., 2005).To characterize the influence of chemical compounds on hygroscopic properties of aerosols, ratio of OMC to AS is used, and Fig. 6 displays relationship between OMC/AS with f (H=80%) for the measurements.As many previous studies indicates, organic matter generally serve as hydrophobic compounds in water absorbing process of aerosol particles, however, large variations still exist for different weather conditions.During the urban pollutant episodes, aerosol hygroscopic capacity is gradually inhibited with increasing proportion of hydrophobic organic carbon matter, which might be originated from or mixed by urban anthropogenic emitted pollutant such as industrial and traffic activities.While in clean periods, plants released active carbon organics might sever as positive factor (such as surfactant) to promote inorganic salt to absorb water, as shown (Fig. 6), and laboratory studies carried by Choi et al. (2002), which show that malonic acid and citric acid, behaving as non-deliquescent species in single particle studies, caused Ammonium Sulfate particles to absorb a significant amount of water before deliquescence.
On 15 May, daily averaged OC mass concentration in PM 2.1 is 20.54 µg•m −3 , increased of about 24.9% comparable to mean value of other periods, and OMC/AS ratio is 2.89, while f (RH=80%) oppositely enhance to 2.21 averagely, this result has great disparity with many previous studies.The "closure" analysis by IMPROVE method indicated that the data of aerosol chemical compositions and measured scattering coefficients for the day are in quite good consistency, with the measurement and calculated aerosol scattering coefficient for dry aerosols are 396.5 Mm −1 , 341.0 Mm   Relationship between f (RH=80%) and value of OMC/(NH 4 ) 2 SO 4 .might havea complex interaction with inorganic salt which enhanced the hygroscopic growth of aerosols as suggested by Saxena et al. (1995) for the non-urban area.Cruz et al. (2000) also pointed out that the organic portion was observed to enhance the water uptake of the (NH 4 ) 2 SO 4 -organic aerosol systems by as much as a factor of 2-3 for particles consisting of 80% organic acids.Another result reported by Garland et al. (2007) on relative humidity dependence of light extinction in Organic-ammonium sulfate mixing aerosols, indicated that f RH ext (80% RH, dry) (the ratio of measured extinction coefficient at RH=80% to at RH<10%) was about 2.15 while wt% of organic species reached 50% for dry mean optical diameter of 0.35 µm.The mixing state and interactions between sulfate and nitrate with coarse particle (mass concentration of 131 µg•m −3 ) might also play a crucial role in aerosol hygroscopicity enhancement process.Nevertheless, present chemical composition data could not provide more satisfactory explanations, and further field experiments and information on organic compounds are needed to provide insights into schematics of organic and inorganic interaction.

Conclusions
Aerosols scattering hygroscopic growth experimental study over rural area near Beijing mega-city through analysis of key meteorological parameters and aerosols chemical composition, results in smooth and monotonic growth features associated with increasing RH in most of observation periods.Experimental data allow categorizing pollution episodes in three different cases (clean, dust, urban pollution) in accordance with aerosols mass concentrations and chemical compositions variations.(1) In the "clean" period, aerosols hygroscopic growth factor f (RH=80%) is 1.31±0.03,higher than the results report by Yan et al. (2008) in rural area of North China (1.2), and lower than that of Koloutsou-Vakakis et al. (2001) measured in Northern Hemisphere continental site (1.4-1.5) and Carrico et al. (2000) in Sagres Portugal (1.69±0.16).( 2) During the dust episode, f (RH=80%)=1.2±0.02,more lower than observational experiment result (1.73-2.20)obtained by Kim et al. (2006) The mixing effect of dust particles with sulfate, nitrate and sea-salt aerosols may be a plausible explanation of results divergence.(3) During urban pollutant phase, measured aerosol hygroscopic growth factor f (RH=80%) is 1.57±0.02,lower than values of 2.04±0.28 in southern coast area of China, 1.7-2.0 in Yangtze delta region and 1.81±0.37-2.30±0.24 in East coast of the United States.
Generally, the f (RH=80%) is declining with increase of OMC/AS during urban pollutant episode, as many previous works showed; however, an interesting phenomena of high ratio of OMC/AS (2.89) in PM 2.1 and high f (RH=80%)=2.21 was observed on 15 May.Well consistency of scattering closure results and running condition of two nephelometers suggested existence of positive effect of organic particles on aerosol absorbing water.Through present experimental study, organic matter seems to importantly contribute to uncertainties of aerosols hygroscopic properties as well as the variations of hygroscopic growth.Facing this complexity, further experimental long-term study associated with specific modeling analysis on the relationship between organic compounds and aerosols hygroscopicity will be done in near future to thoroughly understand such physico-chemical process.

Fig. 3 .
Fig. 3. Aerosol mass size distribution (a) and percentage of chemical compounds in detected water soluble inorganic during clean (b), dust (c) and pollutant (d) periods.The label diameter in the plot is arithmetic average of aerodynamic diameters.Mass concentration of (NH 4 ) 2 SO 4 and NH 4 NO 3 are derived from equation in reference(Malm et al., 2001).

Fig. 4 .
Fig. 4. Results of scattering closure calculation using both IMPROVE method (a) and Lorenz-Mie theory method (b).

Fig. 5 .
Fig. 5.Fitting curves of different dominant aerosol types.Among them, dot data represent the measurement results and lines stand for the fitting output of Eq. (2) for different aerosol types.Black, red, blue and green refer to the results of dust, clean, urban pollution and special case for urban pollution episodes, respectively.

Table 1 .
Instrumentation of measurements.

Table 2 .
Designation of different dominant types of aerosols.

Table 3 .
Other observation results of hygroscopic growth factor of scattering coefficient for different categories of aerosols.

Table 4 .
Curve-fitting parameters in different aerosol dominant episodes in terms of Eq. (2).