2.1 Aerosol sampling

Daily TSP samples were collected at Mount Hehuan site, Taiwan (24.16^∘ N, 121.29^∘ E, 3001 m a.s.l.; see Fig. 1) from September 2011 to September 2012. The sampling station is located in a pristine environment and its vicinity is generally higher than 2900 m, thereby the monitoring site can be considered representative of the free troposphere over the subtropical Pacific region (Y. C. Lin et al., 2014). A high-volume TSP sampler (TISCH, Model TE-5170D), operated at a flow rate of approximately 1.13 m³ min⁻¹, was used to collect aerosol samples. Whatman^® 41 cellulose filters (${\mathrm{8}}^{\prime \prime }\times {\mathrm{10}}^{\prime \prime }$) were used as filtration substrates. After sampling, each filter was folded and stored in a separate plastic bag that was then stored in a polypropylene container, frozen immediately, and returned to the laboratory for further chemical analysis. Carbon monoxide, a tracer for tracking anthropogenic plumes, was monitored by a nondispersive infrared spectrometer (Horiba model APMA-370). The details of the instrument and QA/QC procedure for CO monitoring are described elsewhere (Lin et al., 2013).

Figure 1Clusters of backward trajectories at Mount Hehuan from September 2011 to September 2012.

Download

2.2 Chemical analysis

For the purpose of chemical analysis, the sampled filter was subdivided into eight equal pieces after sampling. One piece was subjected to acidic digestion for elemental determination and another was extracted by Milli-Q water for analyzing water-soluble ions. For acidic digestion, each filter sample was put into an acid-cleaned vessel and digested in a mixed acidic solution (4 mL 60 % HNO₃ + 2 mL 48 % HF) by an ultrahigh throughput microwave digestion system (MARSXpress, CEM Corporation, Matthews, NC, USA). The digestion process was performed in three steps; the sample was (1) heated to 170 ^∘C for 8 min and maintained at this temperature for 7 min at 1440 W, (2) heated to 200 ^∘C for 7 min and maintained at this temperature for 15 min at 1600 W, and (3) cooled for 60 min. Subsequently, the vessel was transferred to XpressVap^™ accessory sets (CEM Corporation) for the evaporation of the remaining acids until nearly dry. Approximately 2 mL of concentrated HNO₃ was added into the vessel and reheated. The resulting solution was then diluted with Milli-Q water to a final volume of 50 mL. After acidic digestion, 31 target elements in TSP samples were analyzed through inductively coupled plasma mass spectrometry (ICP-MS; Elan 6100; Perkin ElmerTM, USA). A multi-element standard prepared from stock (Merk) and composed of 2 % HNO₃ solution was used for calibration. An internal standard containing indium (10 ng mL⁻¹) was used to correct instrumental drift. To minimize the isobaric interference, the nebulizer gas flow rate was adjusted to 0.7–0.9 L min⁻¹. To reduce the formation of doubly charged ions and oxides, ${\mathrm{Ba}}^{++}/\mathrm{Ba}$ and CeO / Ce must be lower than the recommended values of 0.01 and 0.02, respectively. Accuracy and precision were assessed by replicate measurements (N=7) of the standard reference material NIST SRM 1648, following the total digestion process. The results showed that the recoveries for most elements fell within 90 %–110 %, and the precisions were less than 5 %. Arsenic, a target element, exhibited a recovery of 106 % and a precision of 2 %. For each run, a blank regent and three filter membrane blanks were subjected to the same procedure as that of the aerosol samples. The method detection limits (MDLs) were 0.01 ng m⁻³ for both As and Pb.

Another half of the filter sample was extracted with 20 mL Milli-Q water (18.2 Ω) by using ultra-sonic apparatus for 1h. The extracted solution was subsequently analyzed for water-soluble ions, including Na⁺, ${\mathrm{NH}}_{\mathrm{4}}^{+}$, K⁺, Mg²⁺, Ca²⁺, Cl⁻, ${\mathrm{NO}}_{\mathrm{3}}^{-}$, and ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ by ion chromatography (Dionex ICS-90 for cations and ICS-1500 for anions), equipped with a conductivity detector (ASRS-ULTRA). A QA/QC program including calibration, recovery, and precision tests along with MDLs for all ions was conducted during the analyzed processes. A multi-ion solution (Merck) was used for the calibration of the Ion Chromatography (IC) instrument and seven-point calibration curves were made for each batch of samples. One laboratory blank was taken for each batch analysis and the MDL was calculated as 3 times the standard deviation of the values of seven blanks. The average recoveries for all species were in the range of 91 %–105 %; the precisions for all species were less than 5 %.

2.3 Principal component analysis

Principal component analysis (PCA), a technique which attempts to explain the statistical variance in a given dataset in terms of a minimum number of significant components, has been widely employed to identify potential sources for airborne particulate matter observed at a receptor site (Viana et al., 2006; Venter et al., 2017). To compute the PCA model, the first step was to transform the chemical data into normalized form, represented as the following;

where Z_ij is the normalized value of the species j in i sample. C_ij is the concentration of species j in sample i and μ_j, and σ_j are the mean concentration and standard deviation for species j. The PCA model was then expressed as the following;

where $k=\mathrm{1},\mathrm{\dots },n$ represents the different sources, and g_ik and h_kj are the factor loadings and factor scores, respectively. Associated with each component is an eigenvalue; only principal components with eigenvalue greater than 1.0 were selected to identify aerosol sources. The software of SPSS (IBM statistics 19) was performed for the PCA analysis.

2.4 Backward trajectory analysis

To identify potential sources of airborne arsenic at Mount Hehuan, 5-day backward trajectories were computed by the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model developed by the US NOAA Air Resources Laboratory (Draxler and Hess, 1998). The meteorological data for the trajectory model were from the GDAS (Global Data Assimilation System), which were processed by the National Centers for Environmental Prediction (NCEP) with a 6 h time resolution, an approximate 190 km horizontal resolution, and 23 vertical levels. In this work, 5-day backward trajectories arriving at 3000 m a.s.l. were computed at 12:00 LT (local time) once every day, with a time step of 6 hours. Four additional trajectories were generated, of which starting locations were changed ±0.5^∘ from the actual sampling site to reduce the uncertainty of the trajectory analysis. During the sampling period, a total of 1865 backward trajectories were computed. According the originated regions of air parcels, we divided the trajectories into five groups, namely, northern China (NC), the Pacific Ocean (PO), the South China Sea (SS), southeastern Asia (SEA), and southern Asia (SA).

Figure 1 shows the pathways of five different air clusters at Mount Hehuan. The frequency of SA was 33 %, which was the most predominant air clusters, followed by PO (24 %), SS (18 %), SEA (18 %), and NC (7 %). In the NC group, the air mass originated mainly from northern China, where heavily polluted air is contaminated by industrial emissions, moved to the southern areas slowly, and arrived at the receptor site. The NC air cluster was predominately found in March, August, and September with a frequency of >16 % (shown in Fig. 2). In the case of PO, the air parcel generally came from the western Pacific Ocean, spending much time in the marine atmosphere before arriving in Taiwan. This air cluster was most predominately found from July to September with a frequency of >48 %. High frequencies (>20 %) in the PO cluster were also surprisingly found in October and November. For SS air clusters, the air parcel was regularly from South China Sea, crossing the marine areas or the Luzon Islands and then arriving at Mount Hehuan. This air group accounted for 18 % with a high frequency in June, July, and November. For the SEA group, the air mass typically came from the mainland Southeast Asian peninsula, occasionally passing across parts of polluted southern China like the Sichuan Basin and the Pearl River Delta (PRD) region before reaching Taiwan. The SEA air group profoundly occurred from March to June, with the frequency exceeding 30 %. Finally, the air parcel of the SA cluster was mainly from Middle East and the Indian subcontinent, passing over the northern parts of Myanmar, Thailand, Laos, and Vietnam along with the PRD region and then descending to Mount Hehuan. The SA group was frequently found during the sampling periods, except for July to September. The air masses of NC, SEA, and SA groups were associated with continental origins, as they spent much time on the Asian continent before arriving at Mount Hehuan. The continental air masses mostly prevailed from mid-autumn to late spring (see Fig. 2). On the contrary, PO and SS air clusters were grouped into marine air parcels and were profoundly found from June to September. Nevertheless, the air parcels from NC, SEA and SA groups are anticipated to pick up polluted air and transport it to Mount Hehuan, compared with the PO and SS air clusters that spent much time in the marine atmosphere.

Figure 2Monthly distributions of the fractions for various air clusters at Mount Hehuan during the sampling period.

Download

2.5 WRF-Chem model

To simulate the long-range BB plumes transported to Mount Hehuan, the coupled with chemistry module of the WRF model (WRF-Chem; Ver. 3.2.1) was employed. Previously, this model has successfully simulated and identified the BB- plume transportation from SE and S Asia (Chi et al., 2010; C. Y. Lin et al., 2009, 2014). In this study, a tracer module in WRF-Chem developed by Lin et al. (2009) was employed to identify the transport. The tracers were assigned to the fire locations derived from MODIS satellite data over the study domain. They were placed at the first level above the surface at each fire location with a concentration of 1 unit per day. The dry and wet deposition functions are also considered in the model. The meteorological initial and boundary conditions for WRF-Chem were acquired from the NCEP-FNL Global Forecast System (GFS) $\mathrm{0.5}{}^{\circ }\times \mathrm{0.5}{}^{\circ }$ analysis datasets (35 vertical levels). The Mellor–Yamada–Janijc (MYJ) planetary boundary-layer scheme was selected in this study. The horizontal resolution for our BB simulations was 27 km. To assure the meteorological fields were well simulated, the four-dimensional data assimilation (FDDA) scheme was activated based on the NCEP-GFS analysis data.

Figure 3Average concentrations (ng m⁻³) of chemical compositions in TSP samples collected at Mount Hehuan site from September 2011 to September 2012.

Download

2.6 MODIS fire spots

The fire spots from BB activities were extracted by the Moderate Resolution Image Spectroradiometers (MODIS) on board NASA's polar-orbit Aqua and Terra satellites. The MODIS fire detection algorithm employs infrared spectrum channels of 4 and 11 µm (Kaufman et al., 1998). The 1 km Level-2 active fire products, abbreviated as MOD14 and MYD14, provide the detection time, coordinates, confidence, and bright temperature for each fire pixel (Giglio, 2013). Details of the fire detection algorithm was discussed by Giglio et al. (2003). In this work, the MODIS Level-2 fire spots observed within a domain (65–135^∘ E and 5–40^∘ N) from September 2011 to September 2012 were obtained. We used the datasets to analyze the monthly distributions of BB activities over the SE and S Asian continent.

3.1 Overview of airborne particulate As

A total of 302 daily TSP samples were collected at Mount Hehuan during the sampling period. Each TSP sample has been determined as the concentration of water-soluble ions and elements by IC and ICP-MS, respectively. Because the net mass of each collected aerosol sample was not measured, the abundance of each species relevant to the TSP mass can not be obtained. Figure 3 displays the average concentrations of ionic species together with metallic elements in TSP samples. All the concentrations are presented under prevailing conditions. Without the determination of carbon contents, sulfate was the most predominant species in airborne TSP samples with a mean concentration of 4.1 µg m⁻³, followed by nitrate (2.0 µg m⁻³), ammonium (1.7 µg m⁻³), and chloride (0.24 µg m⁻³). Aluminum (Al), a typical geological material, exhibited a mean concentration of 184 ng m⁻³ and was the predominant element. In addition to K, Ca, and Fe (higher than 100 ng m⁻³) were also major metals, followed by Na, Mg, Cu, Ti, Zn, and P (10 to 100 ng m⁻³), and then followed by Pb, Mn, Ba, and Sr (1 to 10 ng m⁻³). The rest metals had concentrations of <1 ng m⁻³ over the free troposphere. As expected, high concentrations for all species were found for the continental air clusters, including NC, SEA and SA air groups (see in Table S1 in the Supplement). In particular, the SA air parcel picked up heavily polluted air to the receptor site since the concentrations of secondary inorganic aerosols (SIA, including ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$, ${\mathrm{NO}}_{\mathrm{3}}^{-}$ and ${\mathrm{NH}}_{\mathrm{4}}^{+}$) and crustal materials (Al, Fe, Ca, K, Mg, and Sr) were significantly higher than those of other continental air clusters. Although the reason was not well-understood, it might be attributed to the different emission sources passed through the air and the atmospheric processes during their long-range transport.

Figure 4Monthly distributions of 95th, 75th, 50th, 25th, and 5th percentile values of (a) As, (b) Al, (c) Pb, and (d) K⁺ concentrations observed at Mount Hehuan from 2011 September to 2012 September.

Download

Arsenic, a target element in this study, exhibited a daily concentration ranging from 0.02 to 5.9 ng m⁻³, with a mean value of 0.5±1.0 ng m⁻³ (see Supplement Fig. S1). As expected, arsenic concentrations in the continental air groups, such as SA, NC, and SEA, were much higher than those in the marine air categories (Fig. 1 and Table S1). The As concentrations (∼0.1 ng m⁻³) in PO and SS air groups were in agreement with those of Mauna Loa, Hawaii (Zieman et al., 1995), indicating that the low As value can be considered a background concentration in the subtropical free troposphere (Zieman et al., 1995). A large standard deviation suggested that the As concentration at this mountainous site had a large day-to-day variation. Some As peaks were found with enhancements of both CO and potassium ion (K⁺) especially between January and May, indicating BB origins.

Figure 4a shows monthly variations of 95th, 75th, 50th, 25th, and 5th percentile values of arsenic concentrations observed at Mount Hehuan. As can be seen, the median concentration of arsenic increased from January (0.18 ng m⁻³), maximizing in May (0.81 ng m⁻³) and then decreasing abruptly from June to December (from 0.03 ng m⁻³ in November to 0.11 ng m⁻³ in August). The seasonality of As was different from that of Al (a tracer of dust) and K⁺ (a marker of BB) as shown in Fig. 4b and d, but it was very similar to that of Pb (Fig. 4c), suggesting that As and Pb might originate from similar sources. The seasonal distributions of As at this mountainous site were associated with emission sources, regional circulations, and local meteorological conditions. Marine air prevailed from July to November, except in October, resulting in lower As concentrations over the subtropical free troposphere. On the contrary, continental air prevailed in the wintertime and springtime, picking up polluted air and transporting it to the receptor site; as a result, an increase of As concentrations was expected. Besides this, favorable local meteorological conditions for the dispersion of air pollution might be another reason for the lower As concentrations in the summertime (Lin et al., 2011, 2013).

Table 1Summary of principal component analysis for aerosol species and carbon monoxide observed at Mount Hehuan. Factor loadings lower than ±0.4 are not given. Factor loadings lower than ±0.7 are marked in bold.

Download Print Version | Download XLSX

The 95th percentile value for As concentrations is better for understanding the distributions of high As events over the free troposphere. Higher 95th percentile values of arsenic were found between February and May compared to those of other seasons, reflecting that more high-As plumes crossed over Mount Hehuan from late winter to late spring. Over the subtropical free troposphere, two distinct haze plumes were usually observed from late winter to spring: one is dust storm that originated from eastern Asian and non-eastern Asian continents (Lin et al., 2001; Hsu et al., 2012), while another one is the BB plume that mainly comes from SE and S Asia (Lin et al., 2009, 2010). As shown in Fig. 4b and 4d, substantially elevated Al and K⁺ concentrations were observed in the springtime, especially for 75th percentile values, suggesting that Mount Hehuan was influenced by both dust and BB aerosols. Both specific plumes would impact the atmospheric compositions, of course, including airborne As in the Pacific region.

Figure 5Scattered plots of As against (a) K⁺, (b) Al, and (c) Pb in different As concentration bins observed at Mount Hehuan.

Download

3.2 Potential source of As in the BB seasons

Table 1 lists the PCA results of the TSP observed at Mount Hehuan during the S and SE Asian BB seasons (from January to May) and non-BB seasons (from June to December). The results showed that there were three factors of TSP during the BB seasons. PC1 was associated with a mixed source of crustal materials (high loadings of Al, Fe, Mg, Ca, Sr, Ti, Mn, and Rb along with La, Ce, and Nd), SIA (high loadings of ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$, ${\mathrm{NO}}_{\mathrm{3}}^{-}$, and ${\mathrm{NH}}_{\mathrm{4}}^{+}$) and industrial emissions (high loadings of Ni, Mo, Tl, V, and Se). In PC2, a high loading was found for K⁺ and a moderate loading for CO, indicating BB sources. Interestingly, moderate loadings were also found for As and Pb, suggesting that both species might originate from BB activities. The PC3 was regarded as sea-salt aerosols, since the high loadings were found for Na⁺ and Cl⁻. During the non-BB periods, three factors were also assessed. The PC1 was assigned as a source related to crustal materials, with high loadings of Al, Fe, Mg, Ca, Sr, Ti, Mn, Rb, La, Ce, and Nd. In terms of PC2, high loadings of Tl, As, and Se were found, suggesting industrial emissions. In particular, highly positive loadings were found for As and Se, indicating that the As at Mount Hehuan during the non-BB seasons was mainly attributed to coal-fired power plants, which was very different from the results between January and May. The PC3 exhibited high loadings of ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$, ${\mathrm{NO}}_{\mathrm{3}}^{-}$, and ${\mathrm{NH}}_{\mathrm{4}}^{+}$ together with Na⁺ and Cl⁻, reflecting SIA and sea-salt aerosols.

Figure 5 shows the scattered plots of As against K⁺, Al, and Pb in different arsenic concentration bins. We found that As correlated well with K⁺ (r=0.78, p<0.05 for the 95th percentile value of As) when severely high As events occurred, suggesting BB origins. Oppositely, arsenic correlated weakly with Al (r ranged from 0.05 to 0.42) in all As concentration bins, indicating that wind-erosion soil was not a major source of airborne As at the sampling site. However, significantly positive correlations were observed between As and Pb within the 75th percentile of As values, reflecting that airborne As and Pb were from similar sources in the high arsenic events.

As discussed above, BB activities may be an important regional source of high As concentrations over the subtropical free troposphere, especially during the S and SE Asian BB seasons. Consequently, we tried to prove the hypothesis using backward trajectory analyses and MODIS fires observations along with the WRF-Chem-model simulated results. Figure S2 shows the seasonality of fire spots over SE and S Asia observed by MODIS from 2011 September to 2012 September. In SE Asia, the BB activities showed strong seasonal variations with a gradual increase from January to March, a time when they reached a peak. They then decreased substantially from late spring to a minimum in summer. In southern Asia, the total annual counts of fire spots were approximately 20 % of those in SE Asia. A similar seasonality was found in intensive fire spots in the springtime, with a maximum in May. The fire spots then decreased between summer and mid-winter, with a minimum in July. However, the total fire spots (SE Asia plus S Asia) reached a maximum in March. This might explain why particulate K⁺ and CO concentrations at Mount Hehuan were highest in March.

For convenience, prior to further analysis, we arbitrarily chose a K⁺ concentration of >109 ng m⁻³ (the 75th percentile value of potassium ion) as a criterion value for identifying the suspected BB events. A second criterion (CO concentration higher than 160 ppb) was also added for selection of the BB plume. Ultimately, a total of 49 suspected BB TSP samples were identified during the entire sampling period. Figure 6 shows time series of daily concentrations of As, K⁺, and CO observed at Mount Hehuan from January to May 2012, when intensive BB activities occurred over SE and S Asia. The air clusters are also shown in this figure for helping to identify the air origins. As can be seen, several As spikes coincided with increasing CO and K⁺ (e.g., 19 February, 30 and 31 March, 3 April, 5 and 7 May, etc.), implying BB origins. Backward trajectory showed that the air parcels for the high arsenic events originated mainly from S Asia (see in Fig. S3). A high arsenic plume passed over Mount Hehuan with As concentration increasing from 1.2 ng m⁻³ on 25 March to 5.3 ng m⁻³ on 3 April though low As concentration was found on 2 April. Figure S4a illustrates the 5-day backward trajectories starting at Mount Hehuan during this period. The results showed that the air parcels mainly passed over northern India, Nepal, Bangladesh and southeastern China before arriving in Taiwan. Figure 7a plots the distributions of MODIS fires from 25 March to 3 April and the WRF-Chem model results at an altitude of 700 hPa on 3 April, when the high daily As concentration (5.3 ng m⁻³) was observed. As seen in the figure, extensive fire spots were observed over the Indian subcontinent from 25 March to 2 April. In this case, the tracers were assigned to the fire locations derived from MODIS satellite data over the Indian subcontinent, ranging from 5 to 38^∘ N and 65 to 90^∘ E. They were placed at the first level above the surface at each fire location with a concentration of 1 unit per day. The WRF-Chem model results showed that significant BB plumes originated over the burned areas, were transported to the east, and passed over Mount Hehuan, resulting in increased concentrations not only of K⁺ and CO, but also of arsenic. As shown in Fig. 8a, during the BB events over the S Asian continent, arsenic correlated well with K⁺ (r=0.73, p<0.05). On the contrary, the correlation coefficient between As and K⁺ in the non-BB events was 0.53 (p>0.05). This supported our argument that airborne arsenic at Mount Hehuan is attributed to BB activities over S Asia.

Figure 6Time series of daily airborne particulate As, Pb, K⁺, and CO concentrations and clusters of backward trajectories observed from January to May in 2012. In the bottom panel, the green and grey crosses denote the BB and non-BB samples identified in the text.

Download

Figure 7MODIS fires on surface of ground and WRF-Chem modeled results of BB plumes at the altitude of 700 hPa on (a) 3 April and (b) 15 March. The blue arrows and lines denote the wind directions and wind speeds, respectively. The grey shadows represent tracer concentrations.

Download

Some BB plumes were observed at Mount Hehuan, but the As concentrations did not increase. For example, a suspected BB plume was found from 8 to 15 March, since K⁺ and CO concentrations increased concurrently. Based on backward trajectory analysis, the air parcels during this BB event were mainly from SE Asia, passing over southeast China and then arriving at Mount Hehuan (Fig. S4b). Because backward trajectories were mainly from the mainland Southeast Asian peninsula, the tracers were then placed at the first level above the surface at each fire location in the mainland Southeast Asian peninsula ranging from 5 to 30^∘ N and 90 to 110^∘ E. The WRF-Chem model showed that the significant tracer concentration which was located in the northeast–southwest belt and covered Taiwan on 15 March (as shown in Fig. 7b). Nonetheless, the As concentration did not rise but maintained a low level of 0.2 ng m⁻³. Another similar case was also found in the end of February (25 to 28 February). The backward trajectories also showed that the air masses were mainly from the mainland Southeast Asian peninsula (see in Fig. S3). Unlike BB events over S Asia, arsenic correlated weakly with K⁺ (r=0.4, p>0.05, Fig. 8b) in the BB events from SE Asia as well as those in the marine air groups (Fig. 8c). These findings suggested that some specific sources might release numerous arsenic into atmosphere during BB activities over S Asia but not over the mainland Southeast Asian peninsula.

Figure 8Scattered plots of As against K⁺ observed at Mount Hehuan in (a) SA, (b) SEA, and (c) other air groups during the S and SE Asian biomass burning seasons.

Download

Wind-erosion soil particles are an important source for airborne arsenic. According to the investigation by Nriagu (1989), arsenic derived from wind-erosion dust was 2.1 Gg yr⁻¹, accounting for 18 % of natural As emissions. Figure S5a–c show the scattered plots of As against Al in all air groups during the S and SE Asian BB periods. Weak correlations were found between As and Al in the various air groups, except for the SS air category (r=0.88, p<0.05), indicating that wind-erosion soil was not a major source of As over the free troposphere. Interestingly, a strong correlation of As and Al was found in the SS air group. The marine air parcels, which spent a long time in the clean marine atmosphere, are subjected to dilution, which can affect the air pollution (Lin et al., 2011), probably resulting in similar behaviors of As and Al.

Recently, numerous studies pointed out that S Asia, especially in West Bengal in India and Bangladesh, is an area extremely contaminated with As (Burgess et al., 2010; Neumann et al., 2010; Roberts et al., 2010). In these regions, highly As-contaminated groundwater, typically caused by geological process, is not only used for drinking water but is also used for the irrigation of crops. The accumulation of arsenic has been found in rice roots and rice plants along with crop soils (Norra et al., 2005). After burning, the As might be released from these crops into the atmosphere and transported easterly to Pacific regions with BB plumes. On the other hand, uses of the pesticide as an insecticide for cotton, paddy, and wheat in India and Bangladesh might be another reason for As contamination in crops (Aktar et al., 2009). Lead arsenate (LA, [Pb₅OH(AsO₄)₃]; As ∕ Pb ∼ 0.22) was the most extensively used arsenical insecticide in the world. It was used as an insecticide for gypsy moths invading hardwood forests in 1892. LA can be adhered to the surfaces of plants. Although LA was officially banned as insecticide in the 1990s in many developed countries, it has not been banned completely in India. Figure 9a and b show the scattered plots of As against Pb in TSP samples for various air groups during the S and SE Asian BB season. The higher As concentrations were generally found in the SA air category. In case of SA air group, the average As concentration in the BB events were 1.6±1.4 ng m⁻³, exceeding that (0.6±0.7 ng m⁻³) of non-BB events by a factor of 2.7 (p<0.05) and suggesting a special arsenic emission source over S Asian continent during the BB season. In some cases, low As concentrations were also found when the BB plumes were transported from S Asia. The reason has not been clearly understood but might be explained by a mixed source of the BB plume with other emissions during the air transportation. In terms of the SEA group, no substantial discrepancy of As concentrations was found during BB and non-BB periods, indicating that the BB activities over the mainland Southeast Asian peninsula were unable to enhance As concentrations over the subtropical free troposphere.

Figure 9Scattered plots of As against Pb observed at Mount Hehuan in (a) SA, (b) SEA, and (c) other air groups during the S and SE Asian biomass burning seasons.

Download

During the S and SE Asian BB period, strong correlations between As and Pb (ranging from 0.84 for SA-BB to 0.96 for NC, see in Fig. 9) were found in various air groups; hence, a ratio of As ∕ Pb might give us an insight in how to trace the specifically regional arsenic sources in the SA air group when a BB activity occurred. During the SA-BB plumes, the average As ∕ Pb ratio was 0.18 (see in Fig. 9a), which was much higher than the average value (0.11) of non-BB (SA-non-BB) events along with those (ranging from 0.08 to 0.1) of other air categories (see in Fig. 9b and c), implying a special source of As during the BB events over S Asia. Some datasets of SA-BB groups showed low As ∕ Pb ratios, probably reflecting the air mixing of BB plumes and other emission sources transported to the subtropical free troposphere. Wind-erosion soil particles and metal smelting (lead smelting) along with coal combustion industries are major natural and anthropogenic sources of airborne As, respectively. In northern India, the As ∕ Pb ratio in natural soil, paved road and unpaved road dust varied from 0.02 to 0.13, while low As ∕ Pb ratios were found in lead smelting (0.002), coal combustion in stoves (0.0016), and coal fire power plants (0.0026) (Patil et al., 2013). Our As ∕ Pb ratios in the SA-BB events were much higher, suggesting that wind-erosion dust, lead smelting and coal combustion seemed not to be major As sources. In particular, the As ∕ Pb ratio was normally higher than 0.20 when severely high As concentrations were observed. This ratio was in line with that of LA (∼0.22), suggesting that burning crops contaminated by LA in S Asia could be a crucial candidate for extremely high As concentrations at Mount Hehuan during the BB periods.

3.3 Impact of Biomass Burning

The differences in As concentrations between the BB and non-BB days could be considered as the net influence of BB activities on the airborne As concentrations over the subtropical free troposphere (Lin et al., 2010, 2013). Table 2 lists the differences of As, Pb, K⁺, and CO concentrations of BB and non-BB samples for SA and SEA air groups in the S and SE Asian BB seasons. For the SA air cluster, all species apparently increased in the BB events. On average, the As concentrations in the BB and non-BB events were 1.6 and 0.6 ng m⁻³, respectively. The difference (1.0 ng m⁻³) accounted 63 % for the average As concentration on the BB days. This indicated that S Asian BB activities played an important source of high As concentrations at Mount Hehuan. On the contrary, the differences of concentrations in K⁺ and CO were observed in the BB and non-BB events for SEA air clusters but were not found for As and Pb. Again, this suggested that BB activities from SE Asia does not release enormous amounts of arsenic into atmosphere and transport it to the subtropical free troposphere by the westerly belt.

Table 2The max, min, mean, and standard deviation values of As, Pb, K⁺, and CO concentrations on the BB and Non-BB days in SA and SEA air clusters during the S and SE Asian BB periods (from January to May).

^∗ Difference for each species are calculated by the mean values in BB and non-BB events.

Download Print Version | Download XLSX

From Table 2, we identified ΔK⁺, ΔCO, and ΔAs as the differences of concentrations in K⁺, As, and CO between the BB and non-BB days in SA and SEA air clusters during the BB seasons. Here, we also converted the units of daily K⁺ and As concentrations from ng m⁻³ to ppb based on the ambient temperature and molecular weight of K⁺ and As. Then, we can obtain the ratios of $\mathrm{\Delta }{\mathrm{K}}^{+}/\mathrm{\Delta }$CO and ΔAs ∕ ΔCO without units. The parameters are useful in estimating the K⁺ or As emissions from BB activities over S and SE Asia. The BB air masses emitted from the S and SE Asian continents contained $\mathrm{\Delta }{\mathrm{K}}^{+}/\mathrm{\Delta }$CO ratios of 0.0043 and 0.0018, respectively. Each value was within the same order of magnitude of that estimated by Tang et al. (2003), who claimed that the BB events emitted from SE Asia had a $\mathrm{\Delta }{\mathrm{K}}^{+}/\mathrm{\Delta }$CO ratio of 0.0038. Besides, a ratio of ΔAs ∕ ΔCO in the S Asian BB events was estimated to be 0.00001, which was one order of magnitude higher than that (ΔAs ∕ ΔCO ∼0.000001) in SE Asian BB events, indicating that much more As released into atmosphere from the S Asian continent. According to the emission inventory, the annual CO emission rate from biomass burning over S Asia was nearly 17 Gg yr⁻¹ (Stress et al., 2003), we then estimated that approximately 0.17 tons yr⁻¹ of arsenic was released into atmosphere due to S Asian BB activities, resulting in enhancements of As concentrations over the subtropical free troposphere.