Influence of Seasonal Variation and Long-Range Transport of Carbonaceous Aerosols on Haze Formation at a Seaside Background Site , China

The Yellow River Delta is a crucial background site that is located in a heavily polluted area in China. Carbonaceous aerosol concentrations were measured using 2.5-μm-diameter particle (PM2.5) samples collected from the Yellow River Delta, Shandong Province, China, from January 2011 to November 2011 by using the thermal/optical reflectance method. In the Yellow River Delta, the organic carbon (OC) concentration ranged from 0.74 to 27.51 μg/m; the annual average concentration was 7.61 μg/m. Moreover, the elemental carbon (EC) concentration ranged from 0.16 to 15.00 μg/m; the annual average concentration was 2.98 μg/m. In addition, the carbonaceous aerosols concentrations were the highest in winter and lowest in summer. The EC tracer method showed that the secondary OC (SOC) contribution to the total carbonaceous concentration tended to be higher in winter than in other seasons. An analysis of carbonaceous showed that haze was derived from different matter in different seasons, particularly haze in winter was dominated by OC, EC, and SOC. The Yellow River Delta can be considered a background site because of the strong correlation between OC and EC (R = 0.83–0.97). Furthermore, the OC/EC ratios for cold seasons (winter and spring) were higher than those for warm seasons (summer and autumn), suggesting that the OC originated from biomass burning in nearby villages in cold seasons. Back trajectories indicated that short-distance air mass from region area contribute most to the sample site. However, the highest carbon concentrations during haze days were related to the air mass travelled through the Bohai Rim except in summer haze episodes. Based on the entire sampling period, the air mass travelled through the polluted areas of Beijing and Hebei Province toward the Yellow River Delta may contribute most to carbonaceous species due to long-range transport.


INTRODUCTION
Atmospheric carbonaceous aerosols have been the focus of many recent studies because of their effects on climate and human health.Carbonaceous aerosols are conventionally classified into three categories: organic carbon (OC), elemental carbon (EC), and carbonate carbon (CC).The aerosols in these categories account for 20%-80% of the total fine particle concentration in the atmosphere.Compared with OC and EC, CC can typically be neglected because of its small mass contribution (Chow et al., 2002).Particulate OC can be emitted directly into the air (primary organic aerosol, POA) or formed in the atmosphere through chemical reactions of reactive organic gases and subsequent gas-to-particle partitioning processes (secondary organic aerosol, SOA) (Chu, 2005).
Currently, there is no direct experimental method for distinguishing secondary OC (SOC) from primary OC (POC).However, methods such as the EC tracer method have been used to predict SOA concentrations.A previous study suggested that an OC/EC ratio of 2.9 represents the upper value of the range of primary OC/EC ratios (Strader et al., 1999).Some studies have used low-ratio (5%-10%) samples as a subset.In this data subset, the median sample is typically determined and used to represent the minimum OC/EC ratio (Pio et al., 2011).The second method is the multiple regression method, which involves using both tracers of primary emission and secondary activity to determine POC and SOC concentrations, respectively (Wang et al., 2012).The third approach entails using receptor models, such as chemical mass balance and positive matrix factorization methods (Pachon et al., 2010) and organic molecular markers (distribution of carbonaceous aerosol in the southeastern United States by using molecular markers and carbon isotope data, Zheng et al., 2006), to estimate SOC concentrations.
In China, atmospheric emissions have markedly increased because of the rapidly growing economy.In 2006, the Intercontinental Chemical Transport Experiment-Phase B (INTEX-B), conducted by the National Aeronautics and Space Administration (NASA), estimated China's black carbon (BC, a product of incomplete combustion of vegetation and fossil fuels) and OC emissions to be 1.8 Tg and 3.2 Tg, respectively, accounting for 49% and 62% of all emissions in Asia.Because the main difference between EC and BC is the analytical methods used for their detection (OC is detected using thermal/optical reflectance and BC is detected using light absorption), the BC concentration can represent the EC concentration to an extent.Also, this campaign showed a 14% increase in 2.5-µm-diameter particles (PM 2.5 ), OC, and BC during 2001-2006(Cao et al., 2006).Another study showed that more carbonaceous aerosols are emitted from eastern China than from western China (Zhang et al., 2009).By using a top-down model, a previous study showed that secondary formation accounts for 21% of Chinese annual mean surface OC concentrations (Fu et al., 2012).Carbonaceous aerosol is a crucial factor for haze formation because it reduces visibility and exhibits high concentrations during haze episodes (Wang et al., 2014).In China, several studies have investigated the effects of carbonaceous aerosols on megacities and mountains (Cao et al., 2009;Duan et al., 2005;Hou et al., 2011).In addition, some studies have focused on river deltas (Cao et al., 2003).However, research in offshore areas, particularly regarding background pollution characteristics, remains limited.
The Yellow River Delta (N 37°35'-38°12', E 118°33'-119°20') is the largest river delta in China and is located in Dongying, Shandong province.In winter, the delta is affected by a continental air mass (Xu et al., 2004); however, in summer, the area is affected by the monsoon from the Pacific Ocean.The area has an annual average temperature of 12.3°C, an annual sunshine rate of 62%, and an annual average sunshine duration of 2682 h (Liu et al., 2010).The Yellow River Delta is a nature reserve, where human activities are limited.Therefore, the delta is a background site.However, The Yellow River Delta is located in China's most heavily polluted area (i.e., the Circum-Bohai Sea industrial region) and its atmosphere may be influenced by regional transport (Yuan et al., 2014).
A previous study suggested that pollution in Asia can be transported over the Pacific Ocean to North America or even further (Tanaka et al., 2005).Therefore, studying the Yellow River Delta, a typical background site in East Asia, is of global importance.This study aimed to determine the carbon content and how carbonaceous aerosols are involved in haze formation in background sites.This study provides the first known information regarding carbonaceous aerosols from the offshore area of the Bohai Sea.

Sample Site
The sample site for this study was the Yellow River Delta, Shandong province.The Yellow River Delta is the largest delta in China and is located adjacent to the Bohai Sea.Because the area is a natural reserve, no industrial facilities are present.The study site in the Yellow River Delta was on the roof of a four-story building.The sampling period in 2011 was divided into four seasons: winter (Jan 14-Feb 14), spring (April 1-May 3), summer (Jul 4-Jul 29), and autumn (Oct 11-Nov 7).

Sampling
PM 2.5 samples were collected on a 47-mm quartz filter (prefired at 600°C for 2 h) by using a reference ambient air sampler (RAAS, Model: RAAS2.5-400,Thermo Anderson).The typical sampling period was from 8:30 a.m. to 8:00 a.m. the following day; samples were not collected during rain and snow events.Furthermore, meteorological data were determined using an automatic weather station set together with our sampler, while visibility was determined by a visibility sensor (Vaisala, PWD22), which used the proven forward scatter measurement principle to measure.Four filters, a Teflon filter and three quartz filters, were placed in the RAAS simultaneously.
One of four channels which contained quartz files was used to analyze carbonaceous aerosols, and OC and EC concentrations were determined using a thermal/optical analyzer from the Desert Research Institute; this analyzer is based on the thermal/optical reflectance (TOR) method.For quality control, one of every 10 samples was measured twice.Data obtained from the 10 samples were considered to be valid only when the difference between the two results of the selected sample was < 5%.If the difference was > 5%, then the 10 samples were reanalyzed until the results of the selected sample met the criteria.In total, 106 valid samples with four blank samples were acquired (1 blank sample was obtained in each season; 28, 23, 23, and 32 valid samples were obtained for spring, summer, autumn, and winter, respectively).We placed the filter in the RAAS; however, the vacuum pump did not function.Therefore, the only difference between blank and filter samples was that air flow from the ambient atmosphere was zero for blank samples.The Teflon filter samples were weighed using a Sartorius ME-5F balance (readability: 1 µg) before and after sampling at a constant temperature (20 ± 1°C) and relative humidity (RH, 50% ± 2%) to determine their mass concentrations (Zhou et al., 2012a).The OC,EC and mass concentrations of all valid samples were determined after those for blank samples were deducted.

Estimation of SOC Concentration
We calculated the SOC concentration in the Yellow River Delta by using the EC tracer method.The SOC concentration can be obtained using the following formula: (1) OCsec = OCtot -OCpri (Stadler et al., 1999;Hou et al., 2011;Pio et al., 2011).
In these equations, OCpri is the primary OC, N is the noncombustible OC, (OC/EC) pri is the ratio of primary OC to primary EC, OCsec is the secondary OC, and OCtot is the total OC.
In the EC tracer method, N is typically neglected or estimated using an OC-EC regression equation.According to the results of our study (Section 3.4), noncombustible OC concentrations were < 1 µg/m 3 in all seasons; therefore, N can approximately be 0 µg/m 3 .A crucial step in the EC tracer method is determining (OC/EC) pri.As discussed above, there are two primary methods for calculating SOC concentrations.In this study, we used data for which OC/EC was within the lowest 10% in our sample time to construct a subset (Zhou et al., 2012a), then used median of the sample data to represent (OC/EC) pri.

Seasonal Variation in Carbonaceous Species
Table 1 lists OC, EC, SOC and PM 2.5 mass concentrations and their ratios in the four studied seasons.The OC mass concentration varied from 0.74 µg/m 3 in summer to 27.51 µg/m 3 in winter; the average concentration was 7.61 µg/m 3 .Moreover, the EC concentration exhibited a similar trend; it ranged from 0.16 µg/m 3 in spring to 15.00 µg/m 3 in autumn; the average concentration was 2.98 µg/m 3 .The SOC concentration was highest in winter (14.30µg/m 3 ) and lowest in summer (0 µg/m 3 ); the annual average concentration was 3.64 µg/m 3 .Statistical data showed that carbonaceous aerosol concentrations in winter were approximately two times higher than those in summer.Moreover, the PM 2.5 proportion in winter was higher than that in summer, implying that carbonaceous aerosol pollution increased in winter.Table 2 shows a comparison of carbonaceous aerosol concentrations in the Yellow River Delta site with those in other urban and rural sites.The OC and EC concentrations in the Yellow River Delta were higher than those recorded at some of the remote sites in China and some sites in developed countries; however, the OC and EC concentrations in the Yellow River Delta were lower than those obtained in another river delta, the Pearl River Delta in southeastern China.Fig. 2 illustrates the daily variations in the carbonaceous aerosol and visibility in the Yellow River Delta.The seasonal periods are defined in Section 2. Both the OC and EC exhibited similar temporal variations, suggesting that in the Yellow River Delta, these species had similar sources and atmospheric processes.In addition, the OC and EC mass concentrations exhibited a pronounced decrease in summer; this observation was similar to an observation of Jinan, a heavy polluted city in North China.This decrease was related to typical meteorological conditions in summer, such as frequent rain.There were nine rain episodes in July (i.e., July 2,7,10,18,19,24,25,27,and 30).By contrast, there were only two and four rain episodes in spring and autumn, respectively.There were three snow episodes in winter.Moreover, summer exhibited deeper mixing heights that favor pollutant dispersion.A cold stagnant atmosphere and transportation from heavily polluted areas to the Yellow River Delta (Section 3.5) were responsible for the high OC and EC concentrations observed in winter.

Analyzing Haze According to Meteorological Conditions; OC, EC and SOC Meteorological Conditions
Haze is defined as a weather phenomenon in which the atmospheric visibility decreases to < 10 km and the RH is < 90% because of suspended particles and/or smoke.During the sampling period, seven haze episodes occurred in the Yellow River Delta: January 22 and 23, January 30-February 7, April 14 and 15, July 10 and 11, July 22-24, October 22  Na et al., 2004) and 23, and October 26-November 1 (Fig. 2).Fig. 3 shows OC, EC, and SOC concentrations and their ratio to PM 2.5 during both haze days (HDs) and clear days (CDs).In the presence of haze, visibility decreased rapidly; however, the haze events exhibited different characteristics depending on the season.The highest OC and EC concentrations were observed when the wind speed was 2-4 m/s.High wind speeds favor pollutant diffusion; however, because of a lack of local sources, regional pollution cannot be transported to the Yellow River Delta when the wind speed is < 2 m/s (Fig. 2).

Concentrations of OC, EC, SOC
The characteristics of haze events varied according to the season.Table 3 summarizes the average OC, EC, SOC and PM 2.5 concentrations for both HDs and CDs in the four studied seasons, and HD/CD ratio of these species.Moreover, Fig. 3 shows the OC, EC, and SOC concentrations and the contribution of OC, EC, and SOC to PM 2.5 on both HDs and CDs.The results demonstrate the following crucial points.(1) The PM 2.5 concentration observed in winter was lower than that observed in both summer and autumn.This is different from the trend observed in urban areas (PM 2.5 concentration is higher in winter than in summer and autumn because of heating, Yang et al., 2013) and shows that transported pollution may control winter haze events.In winter, the HD/CD for OC, EC and SOC were 2.9, 3.4 and 2.6, respectively.In addition, the highest OC+ EC concentrations were observed in winter haze days, indicating that carbonaceous aerosols was responsible for haze formation.(2) In spring, the HD/CD values for carbonaceous aerosols were low (1.3, 1.3 and 1.0, for OC, EC and SOC, respectively), implying that "light haze" events occurred in spring because frequent strong gales in spring may favor pollutant advection.Furthermore, the HD/CD values for OC and EC were approximately 1.0 in summer; high RH favors high polar OC formation (Cheng et al., 2012), which is readily eliminated by rain.In consideration of the high PM 2.5 concentrations, high RH, and multiple rain events in summer, summer haze is possibly a mixture of haze and fog.In autumn, HDs exhibited features similar those in winter (i.e., carbonaceous aerosol concentrations were higher on HDs).However, the OC + EC contributed less to PM 2.5 in autumn, indicating that carbonaceous aerosols contribute more to winter haze than autumn haze.

Relationship between OC and EC in the Yellow River Delta
EC is often stable in the atmosphere and is a suitable tracer of primary OC because primary OC and EC are mostly emitted from the same combustion sources (Strader et al., 1999;Chu, 2005).The relationship between OC and EC can affect their origins.Fig. 4 shows strong correlations between OC and EC (R 2 = 0.83-0.97),suggesting that they have similar emissions and transport processes.In consideration of the lack of industrial and vehicle emission sources in the Yellow River Delta, the transportation of carbonaceous aerosol contributes a portion of the total concentration and local rural emissions (e.g., plants and biomass burning in nearby rural areas) contribute the other portion.The strong correlations suggest that the local Fig. 3. Comparison of the concentrations of OC, EC, and SOC (μg/m3 ) and their contribution to PM 2.5 (%) between haze and clear days for the four studied seasons in 2011 within the Yellow River Delta.emissions and transport processes of carbonaceous aerosol have similar OC/EC ratios.In northern China, fossil fuel emissions and biomass burning can weaken the relationship between OC and EC.The slopes for cold seasons (winter and spring) were lower than those for warm seasons (summer and autumn).Moreover, OC/EC of biomass burning are often much higher than those of vehicle exhaust (9.0) and residential coal combustion (3.0) (Han et al., 2008), implying that the OC originated from biomass burning in nearby rural areas or transported pollution from other areas in cold seasons.All four seasons had intercepts close to zero, implying the presence of low background OC concentrations from noncombustible sources in the Yellow River Delta.Therefore, studying the OC/EC ratio may reveal the energy source structure of a region.

Transport Pattern of Carbonaceous Aerosols
To determine the transport patterns of air masses during a haze episode, 72-h back trajectories for the Yellow River Delta for HDs (2h interval) and 120-h back trajectories for all sample episode (6h interval) were calculated using the hybrid single-particle Lagrangian integrated trajectory (HYSPLIT Model, Version 4.9) (Xue et al., 2011).Our meteorological data was obtained from ftp://arlftp.arlhq.noaa.gov/pub/archives/gdas1, start height was at 50.0 m and the meteorological model's vertical velocity fields were used.A cluster analysis was performed to segregate calculated trajectories into distinct groups.Fig. 5 shows the effects of air masses transport on the haze process in different seasons.
The air mass over Central and North China, Clusters 1 and 3, which travelled through heavily polluted areas in the free atmosphere, contained 73% of the total air mass and may have primarily contributed to winter haze.The lower boundary layer caused this air mass to be affected mainly by pollutants from the surface.In spring, 70% of the air mass originated from eastern China, and heating and cooking in rural areas may influence carbonaceous aerosols at the Yellow River Delta through regional transport.Summer and autumn exhibited similar characteristics; that is, most of the air masses originated from nearby regions (100% from eastern China in summer and 86% in autumn).All air mass trajectories were below the boundary layer.Although a local source that travelled a short distance contributed most of the air mass to our sample site, the highest OCEC concentration was found in an air mass originating from Mongolia.The air mass then travelled through the Bohai Rim (including Beijing, Tianjin, and Hebei Province), except in summer, suggesting that the carbonaceous aerosol concentration in the air mass was affected by the heavily polluted area (the Bohai Rim) when haze occurred.According to all of the samples depicted in Fig. 6 (the Asian emission data in 2006 were obtained from the NASA INTEX-B mission), high BC and OC emissions implied that regional pollution sources influenced the background site.The frequent occurrence of local air masses (42%) suggested that pollution from local areas exerted a substantial influence.The highest OC and EC concentrations occurred in Cluster 2, which also originated from Mongolia and then travelled through polluted areas, including Beijing and Hebei Province before reaching the sampling site, suggesting the importance of the effects of pollution sources in the Bohai Rim on carbonaceous species through longrange transport.By comparison, the North China Gobi and northeastern China air masses, which originated in relatively clean regions and in the free troposphere, were related to relatively lower carbonaceous concentrations in CD, particularly during cold seasons.

CONCLUSIONS
Carbonaceous aerosols in PM 2.5 samples in the Yellow River Delta were observed during winter (Jan 14-Feb 14), spring (April 1-May 3), summer (Jul 4-Jul 29), and autumn (Oct 11-Nov 7).The mean OC and EC concentrations were 7.61 µg/m 3 and 2.98 µg/m 3 , respectively.The seasonal variations in OC exhibited the following order: winter > autumn > spring > summer.The variations in EC exhibited the following order: autumn > winter > summer > spring.
An analysis of haze events showed that carbonaceous aerosols were more responsible for haze formation in winter than other seasons.Strong correlations between OC  and EC (R 2 = 0.83-0.97)suggest that these species have similar emissions and transport processes.Moreover, high OC/EC ratios in cold seasons imply that the OC originated from biomass burning in nearby rural areas or transport from other areas.Analysis of back trajectories showed that the Yellow River Delta was considerably affected by local haze events, and an air mass that travelled through the Bohai Rim contributed the highest carbonaceous aerosol concentration, except in summer haze episodes.According to all samples, the air masses travelled through polluted areas, such as Beijing and Hebei Province, which typically have the highest average OC and EC concentrations in China.

Fig. 1 .
Fig. 1.Sample site in map of China.

Fig. 4 .
Fig. 4. Correlation analysis of OC and EC in the Yellow River Delta for 2011.

Fig. 5 .
Fig. 5. 72-h air mass trajectory clusters and OCEC concentrations for the Yellow River Delta during typical haze episodes in the four studied seasons in 2011.

Fig. 6 .
Fig. 6. 120-h air mass trajectory clusters and concentrations of OC and EC in the Yellow River Delta based on all samples episode in 2011.

Table 1 .
Statistics of EC, OC, SOC and PM 2.5 (µg/m³) in the Yellow River Delta in 2011.

Table 2 .
Comparisons of EC and OC concentrations (µg/m³) and OC/EC ratios from rural sites around the world.

Table 3 .
Concentration of OC, EC and SOC from the four studied seasons during haze and clear episodes in the Yellow River Delta (2011).