River Delta

Abstract. In order to characterize the features of particulate pollution in the Pearl River Delta (PRD) in the summer, continuous measurements of particle number size distributions and chemical compositions were simultaneously performed at Guangzhou urban site (GZ) and Back-garden downwind regional site (BG) in July 2006. Particle number concentration from 20 nm to 10 μm at BG was (1.7±0.8)×104 cm−3, about 40% lower than that at GZ, (2.9±1.1)×104 cm−3. The total particle volume concentration at BG was 94±34 μm3 cm−3, similar to that at GZ, 96±43 μm3 cm−3. More 20–100 nm particles, significantly affected by the traffic emissions, were observed at GZ, while 100–660 nm particle number concentrations were similar at both sites as they are more regional. PM2.5 values were similar at GZ (69±43 μg m−3) and BG (69±58 μg m−3) with R2 of 0.71 for the daily average PM2.5 at these two sites, indicating the fine particulate pollution in the PRD region to be regional. Two kinds of pollution episodes, the accumulation pollution episode and the regional transport pollution episode, were observed. Fine particles over 100 nm dominated both number and volume concentrations of total particles during the late periods of these pollution episodes. Accumulation and secondary transformation are the main reasons for the nighttime accumulation pollution episode. SO42−, NO3− accounted for about 60% in 100–660 nm particle mass and PM2.5 increase. When south or southeast wind prevailed in the PRD region, regional transport of pollutants took place. Regional transport contributed about 30% to fine particulate pollution at BG during a regional transport case. Secondary transformation played an important role during regional transport, causing higher increase rates of secondary ions in PM1.0 than other species and shifting the peaks of sulfate and ammonium mass size distributions to larger sizes. SO42−, NO3−, and NH4+ accounted for about 70% and 40% of PM1.0 and PM2.5, respectively.


Introduction
Atmospheric aerosols have attached more and more attention in recent years because they influence the global climate change and human health (Dockery et al., 1994) and degrade visibility (Sokolik and Toon, 1996;Jung and Kim, 2006).In order to understand these effects, accurate knowledge on physical and chemical properties of aerosol is required.A large number of studies showed that the size resolved properties of the atmospheric aerosols are more powerful to explain their atmospheric behavior than their bulk properties (Dusek et al., 2006;See et al., 2006).
On one hand, the absorbing and scattering effect of aerosols on the incoming radiation is dependent on the particle size and composition (Nishita et al., 2007) and the D. L. Yue et al.: Variation of particle number size distributions and chemical compositions accumulation mode particle number concentrations could explain the visibility degradation on hazy days (See et al., 2006).On the other hand, only particles within a certain size range have cloud-nucleating ability and affect the microphysical and optical properties of cloud condensation nuclei (CCN, Iorga and Stefan, 2005).In addition, whether the adverse health effects of aerosols are number-or massconcentration-dependent is still a debating issue.Recently, studies have proved that ultrafine particles with very small sizes can be uptaken directly by cells as well as be translocated to other sensitive target organs such as the heart and central nervous system (Oberdörster et al., 2005).Compared with larger particles of similar composition ultrafine particles are more toxic and induce more intense oxidative stress in cells (Nel, 2005;Nel et al., 2006).The chemical compositions are also key elements deciding the health effect as well as the influence on climate change.Therefore, characterizing number size distributions and chemical compositions of atmospheric aerosols is very important to understand their effects on climate change, human health, and air quality.
The Pearl River Delta (PRD) is one of the most economically invigorating and densely populated regions and one of the biggest city clusters in the world.Rapid urbanization and economic development have deteriorated the air quality and changed the properties of the air pollution: the primary pollutants, such as SO 2 and inhalable particulate matter (PM 10 ) have been reduced by abatement measures.However, the secondary products such as ozone and fine particles of high concentrations become two of the most formidable air quality and public health issues facing the PRD region.Moreover, the scale of the pollution problems in the PRD region has also expanded (Zhang et al., 2008).The occurrence of haze remains very high on about 150 days per year on average in Guangzhou from 1980 to 2006 (Deng et al., 2008).Haze characterized of very low visibility and high mass concentrations of fine particles has been reported in summer as well as in winter (Tan et al., 2009).The particle pollution in the PRD region have been reported regarding to the chemical compositions in size resolved particles or in PM 2.5 and PM 10 concentration at one or more sites and particle number size distributions at a coastal rural site Xinken (Cao et al., 2004;Hagler et al., 2006;Liu et al., 2008a, b;Zhang et al., 2008).However, simultaneous measurements of particle number size distributions and chemical compositions at over one site in the PRD region have not been reported.Resulted from the intense photochemical activity in summer, particulate pollution in the PRD region will be characterized with regional and secondary properties, which should be different from that in the winter or in other cities with less intense solar radiation.The average ratios of PM 2.5 to PM 10 in Guangzhou were larger than three other big cities in China, i.e.Wuhan, Chongqing, and Lanzhou (Wei et al., 1999).High concentrations of secondary products in fine particles, mainly oxidized organics and sulfates were observed during PRIDE-PRD2004 and 2006(Andreae et al., 2008;Jung et al., 2009).
The worse correlation between organic carbon (OC) to elemental carbon (EC) in the summer (R = 0.6) than in the winter (R = 0.8) and the higher OC but lower EC concentrations in Guangzhou than in Beijing during summertime (Cao et al., 2004) indicated the significance of the secondary transformation in the PRD region in summer.Therefore, it is of scientific significance to investigate the properties of particles especially of fine particles in the PRD region during summertime.
Within the "Program of Regional Integrated Experiments of Air Quality over the Pearl River Delta" intensive campaign in July 2006 (PRIDE-PRD2006) focusing on gas phase photochemistry and the aerosol formation and properties during summertime, the particle number size distributions were measured simultaneously at both Guangzhou urban site (GZ) and Back-garden downwind regional site (BG), as well as the concentrations of mass and chemical composition of fine particles.Previous papers in the same special issue already show that the conditions are mainly characterized by strong particulate pollution at ground level (Li et al., 2010) and size matters more than chemistry for the CCN activity of aerosol particles at the BG site in the summer of 2006 (Rose et al., 2010).Hence, the purpose of this study is to characterize the particulate pollution in the PRD region on the basis of comparison of particle number size distributions and chemical compositions between GZ and BG sites and to explore secondary formation and regional transport with the discussion of pollution episodes.

Experimental methods
The intensive field campaign was performed simultaneously at both GZ and BG sites in July 2006 (Zhang et al., 2010).At the GZ site the instruments were set up on the top floor of Guangdong Provincial Environmental Monitoring Center (about 50 m above the ground level), which is located in the western urban area of Guangzhou city.At the BG site the instruments were installed on the roof of a hotel building (about 15 m above the ground level), which is located in the north of Huadu district, about 50 km north from the GZ site.
At the GZ site dry particle number size distributions between 15 nm and 10 µm were measured with a system consisting of a Scanning Mobility Particle Sizer (SMPS, TSI model 3080, TSI Inc., St. Paul, MN, USA) and an Aerodynamic Particle Sizer (APS, TSI model 3321).The SMPS (a long differential mobility analyzer (TSI model 3081) with a Condensational Particle Sizer (TSI model 3025A)) was used to measure particle number size distributions from 15 to 660 nm with a time resolution of 5 min.The system was kept dry by silica gel tube within the inlet line.
At the BG site the particle number size distributions from 3 nm to 10 µm were measured with a system consisting of a Twin Differential Mobility Particle Sizer (TDMPS) and an APS (TSI model 3321, USA).The TDMPS is composed of two Hauke-type differential mobility analyzers and two CPCs (TSI model 3010 and 3025, respectively, USA), deployed to measure the particle number size distributions from 3 to 900 nm every 10 min.The relative humidity within the whole system was kept below 30% by silica gel tubes within the inlet line and both sheath air cycles.The size range of particle number size distributions observed by APSs was 500 nm −10 µm.The time resolution of APS was set as 5 or 10 min according to SMPS's or TDMPS's to keep consistent.APS data of particle number size distributions between 660 or 900 nm and 10 µm were transformed from aerodynamic diameter to Stokes diameter with a supposed particle density of 1.7 g cm −3 (Yue et al., 2009).
Size-dependent losses due to diffusion and sedimentation within the inlet lines were corrected with empirical particle loss corrections for both two systems (Willeke and Baron, 1993).The information on these instruments and the time periods of valid data is listed in Table 1.
Other data including PM 2.5 and mass concentrations of water soluble ions (SO 2− 4 , NO − 3 , and NH + 4 ) and organic matter (OM) in PM 1.0 or PM 2.5 , meteorological factors (temperature, relatively humidity, wind speed, and wind direction (T , RH, WS, and WD, respectively)), and gaseous pollutants (CO, SO 2 , O 3 ) at both sites are also involved in this paper.PM 2.5 was measured by a Tapered Element Oscillating Microbalances (TEOM), ions in PM 2.5 by two coupled Wet Annular Denuder sampling/Ion Chromatograph analysis systems (WAD/IC), size-resolved chemical composition mass concentrations by Micro Orifice Uniform Deposit Impactor (MOUDI), and CO, SO 2 , and O 3 by CO Analyzer, SO 2 Analyzer, and O 3 Analyzer (model 9830A, 9850A, and 9810A, ECOTECH, Australia), respectively.Meteorological stations (Met.Station) were also set up at both sites.In addition, ions and OM in PM 1.0 were detected by an Aerodyne Mass Spectrometer (AMS) at the BG site.Relevant information is also listed in Table 1.

Overview of particle number size distributions and mass concentrations
Weather system during summertime in the PRD region is controlled by tropical cyclones and subtropical high pressure alternately.The former brings frequent precipitation and scavenge the pollutants, while the latter leads to high atmospheric stability with high temperature and high RH, causing regional pollution.The temperatures during PRIDE-PRD2006 at both sites were similar, 31 ± 3 • C at GZ and 30 ± 3 • C at BG. RH were nearly the same at GZ and BG, 76 ± 14%.Low wind speeds (below 2 m s −1 ) were observed during about 60% of the measurement time in the PRD region.Over 50% of the time during PRIDE-PRD2006 at GZ and BG the wind came from south or southeast.
The mean particle number size and volume distributions at both sites during the whole campaign are shown in Fig. 1.The ultrafine particle number concentration at the GZ site was significantly higher than that at the BG site.During the measurement period, the particle number concentration (20 nm −10 µm) at GZ site ((2.9 ± 1.1)×10 4 cm −3 ) is 70% higher than that at BG site, (1.7 ± 0.8)×10 4 cm −3 (Table 2).The explanation is there are more intensive traffic emission sources in the Guangzhou urban area than those in the Back-garden suburban area.The number concentrations at GZ were also significantly higher than the total particle number concentrations (3 nm −10 µm) at Xinken rural coastal site in the PRD region during PRIDE-PRD2004, (1.6 ± 0.8)×10 4 cm −3 (Liu et al., 2008), which are comparable to the total particle number concentrations at BG, (1.8 ± 0.8)×10 4 cm −3 .
At both sites fine particles with diameter below 1000 nm were the main contributor to the total particle volume concentrations, as shown in the lower panel of   b in the summer of 2002 (Cao et al., 2004).were similar and no significant difference within the ultrafine sizes (below 100 nm) was observed at GZ and BG sites.In coarse mode, the peak of average particle volume size distribution at the BG site shows at about 2 µm, smaller than that at the GZ site at about 3 µm.This indicates that the major sources for the coarse particles are different at the BG and GZ sites.Construction and road dust are probably major sources for coarse particles in the Guangzhou city, while coarse particles at the BG site are more affected by the biological sources and biomass burning.The total particle volume concentration at the GZ site of 96 ± 43 µm 3 cm −3 is similar to that at the BG site (94 ± 34 µm 3 cm −3 ).In addition, the measured mean particle PM 2.5 mass concentrations are also similar at both sites (69 ± 43 µgm −3 at GZ and 69 ± 58 µgm −3 at BG) with R 2 of 0.71 for the daily average PM 2.5 at these two sites.These findings suggest that the fine particulate pollution in the PRD region is a regional problem.The average PM 2.5 in the summer of 2006 is lower than that in Guangzhou city in the summer of 2002, i.e. 78 ± 30 µgm −3 (Cao et al., 2004).The higher fine particle mass concentrations and total parti-cle volume concentrations at the GZ and BG sites than those at Xinken, 60 km southeast of the GZ site with a rural/coastal background character, is probably caused by the influence of the sea breeze at Xinken (Zhang et al., 2008).

Characteristics of pollution episodes
In the summer of PRD region, the mass concentrations of particles can increase quickly from very low level to very high level such as with PM 2.5 exceeding 100 µg m −3 resulted from accumulation, secondary transformation, and/or regional transport.During such days, the daily average PM 10 does not violate the national standard of the second grade, although heavy particulate pollution occurs with high hourly average particle mass concentrations and low visibility.The daily average particle mass concentrations conceal the pollution conditions and do not reflect them in detail.Therefore, an hourly criterion will capture the properties of the particulate pollution better.According to the frequency distribution of hourly average PM 2.5 , conditions with PM 2.5 exceeding 100 µg m −3 for more than two hours (excluding those caused by short time local emissions) were classified as pollution episodes in this paper.100 µg m −3 is set with the 90% percentile of the hourly average PM 2.5 concentrations during this measurement and it is almost the same as the PM 2.5 value of 103 µg m −3 calculated from the ambient air quality standard of PM 10 of 150 µg m −3 with the average ratio of PM 2.5 to PM 10 over the PRD region in summer of 68.7% (Cao et al., 2004).Totally, pollution episodes were observed on five days (12, 14, 19, 21, and 23 July) simultaneously at both sites from 6 to 23 July.Mainly two different kinds of pollution episodes were identified, accumulation pollution episode (cases on 12, 14, and 23 July) and regional transport pollution episode (cases on 19 and 21 July).

Accumulation pollution episode
Pollution episodes with gradual increase of PM 2.5 mass concentrations were observed at both sites simultaneously.Such pollution episodes took place under stagnant meteorological conditions with wind speed below 1 m s −1 , RH over 80%, and low boundary layer at night.One accumulation pollution episode occurred from about 18:00 LT on 11 July to about 06:00 LT on 12 July is illustrated in Fig. 2 and Fig. 3 (Accumulation pollution episodes on 14 and 23 July will not be discussed in detail in this paper as they were not observed completely.).During this episode, a clear particle growth process was observed: The number peak diameter at about 80 nm in the beginning grew gradually to at about 120 nm in 12 h (Fig. 4).The evident increase in particle number concentration from 100 to 660 nm was observed.Conversely, the number concentrations for particles from 3 to 20 nm and from 20 to 100 nm decreased during the episode.In the early morning of 12 July, the lowest number concentration of the 3-20 nm particles occurred (around 10 cm −3 ).This can be ascribed to the strong coagulation scavenging produced by the high concentration of the accumulation mode particles (Mönkkönen et al., 2004).
The obvious increases of PM 2.5 and secondary ions in PM 2.5 including SO 2− 4 and NO − 3 were also observed at both sites, as shown in Fig. 3. Two main reasons for this increase can be postulated: (1) the dispersion of primary emissions was weak under stable weather conditions.(2) Secondary transformation processes played a key role in the particle growth.Evident growth in the mass concentrations of secondary water soluble ions was observed.In addition, contribution of unknown sources with significant emission of EC to this pollution episode might be important.

Fig. 1 .
Fig. 1.Average particle number and volume size distributions at GZ and BG during the whole campaign.

Fig. 3 .
Fig. 3. Variations of trace gases at BG, PM 2.5 , and mass concentrations of chemical compositions in PM 2.5 at both sites from 00:00 LT, 11 July to 12:00 LT, 12 July.

Table 1 .
Measurement of particle number size distributions and other parameters at GZ and BG.