Evaluation of Atmospheric PCDD / Fs at Two High-Altitude Stations in Vietnam and Taiwan during Southeast Asia Biomass Burning

Dioxin and dioxin-like compounds such as polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans (PCDD/Fs) are persistent organic pollutants (POPs) generated through human activities. In recent times, extreme weather events such as wild fires have significantly affected the remobilization and successive bioavailability of PCDD/Fs. In Seven South East Asian Studies (7-SEAS), a Southeast Asia biomass burning event that influenced the environmental outcome and transport of PCDD/Fs in Taiwan was investigated on the basis of a climate change situation. During the 7-SEAS campaign on 20–28 March, 2011, significantly high levels of atmospheric PCDD/Fs were observed at Lulin mountain in central Taiwan and in the source region of Northern Vietnam (Son La). Measurements indicated that the patterns of variation of atmospheric PCDD/Fs at both locations were similar, but with a time lag of approximately 2 to 3 days. At Mt. Lulin, there was a significant increase of PCDD/F concentrations from 3.69 to 11.1 fg I-TEQ m and 3.32 to 19.1 fg I-TEQ m to reach their peaks on 23 March and 26 March. In this study, a tracer simulation using the Weather Research and Forecasting model coupled with chemistry was conducted to investigate the effects of the Southeast Asia biomass burning. The combined results of air mass paths simulation and satellite data can be used as evidence supporting the hypothesis that the source of the increasing PCDD/F level is originated from biomass-burning regions in Indochina, particularly Northern Vietnam and Northern Thailand.


INTRODUCTION
Dioxin and dioxin-like compounds such as polychlorinated dibenzodioxins and dibenzofurans (PCDD/Fs) are persistent organic pollutants (POPs) that can be generated through anthropogenic activities (Yamagishi et al., 1981;Kao et al., 2006;Guo et al., 2014).Persistence and lipophilicity enhance PCDD/F accumulation in environmental matrices for years.Therefore, it takes a long time for concentrations of these pollutants to be suppressed (Masunaga et al., 2001).The association between emission, distribution and degradation of POPs with environmental conditions has been characterized, and climate change and increasing climate variability have the potential to influence POP presence (Bogdal et al., 2010;Chi et al., 2014Chi et al., , 2015)).Factors including (1) re-evaporation of secondary sources, (2) wind, (3) precipitation, (4) ocean currents, (5) the melting of ice caps and mountain glaciers at the poles, (6) the frequency of extreme weather events, (7) the degradation and transportation of POPs, (8) environmental partitioning, and (9) biotic transportation, have all been proposed by the United Nations Environment Programme (UNEP/AMAP, 2011) as key factors influencing the environmental fate and transportation of PCDD/Fs.Since the fate of these pollutants is influenced by environmental conditions, an important component influencing measured concentration of POPs is climate-induced (UNEP/AMAP, 2011).
Climate change and greenhouse effects are enhancing global warming (Haldar, 2010).Higher temperatures, in turn, could increase the volatilization rate of PCDD/Fs, especially for lower chlorinated level compounds caused by open sources (UNEP/AMAP, 2011).This factor is mainly related to primary emissions, and may be the dominating effect of climate change on the environmental distribution of PCDD/Fs (UNEP/AMAP, 2011).However, climate change will also affect the environmental fate of PCDD/Fs once they have been emitted into the environment from primary sources (Lamon et al., 2009).Moreover, there are additional factors that can affect atmospheric PCDD/F concentrations.Photolysis and reaction with OH radicals can degrade PCDD/Fs in the troposphere (Lohmann et al., 1999).The half-lives of PCDDs in atmosphere are estimated at 8 days, whereas those of PCDFs are estimated at 4 days (Sinkkonen and Paasivirta, 2000).
According to a report by the U.S. Environmental Protection Agency in 2005, PCDD/F emissions from industrial processes have declined steadily over the past decade, such that forest burning and backyard open-trash burning have become the major source of dioxin emissions in the United States (US EPA, 2005).In Asia, annual emissions of PCDD/Fs from the open burning of crop residues in mainland China were estimated to range from 1.38 × 10 3 to 1.52 × 10 3 g international toxicity equivalency quantity (I-TEQ) between 1997 and 2004, which contributed approximately 10%-20% of total PCDD/F emissions in the country (Zhang et al., 2008).Recent changes in global temperatures have also increased the quantity and severity of forest fires and biomass burning (UNEP/AMAP, 2011).Biomass burning hotspots are mostly located in tropical or subtropical regions, such as South America, Africa, Southeast Asia, and Australia.Pollutants emitted from biomass burning, such as particulate matter or chemical gases like CO, SO x , NO x , and VOCs, can negatively impact human health (Gullett et al., 2006;Feng and Christopher, 2013;Lin et al., 2014).Biomass burning activities in Southeast Asia combined with monsoonal meteorological condition during the regional dry season can also enhance transport of emitted pollutants, including PCDD/Fs to downwind areas through long-range transport (Jian and Fu, 2013).
Previous studies have found that prevailing winds can transport pollutants and tracers (e.g., PCDD/Fs, CO, NO X , NH 3 , SO 2 , K + , PM 2.5 , ozone, and organic and black carbon) emitted from biomass burning in Southeast Asia to Taiwan (Huang et al., 2013;Lin et al., 2013;Reid et al., 2013;Thuan et al., 2013).In this study, atmospheric concentrations of seventeen 2, 3, 7, 8-substituted PCDD/Fs in the vapor phase and suspended particles were monitored in Northern Vietnam (Son La, 21.32°N, 103.91°E; 675 m above mean sea level) and Central Taiwan (Lulin Mountain, 23.51°N, 120.92°E; 2,862 m above mean sea level) during the spring season of 2011 (March).This study evaluates the association of biomass burning in Southeast Asia (particularly Son La, Vietnam) with atmospheric concentration variability of dioxin compounds, and assesses the potential for long-range transport of PCDD/Fs.Moreover, this study speculates on the relative contribution of emission sources by applying the Weather Research and Forecasting model coupled with chemistry (WRF-Chem) and using tracers to apportion PCDD/Fs in the atmospheric simulation.

Sampling Sites
The Seven South East Asian Studies (7-SEAS) project was initiated in 2007 through collaboration of the Taiwan government with those from Southeast Asian countries and the U.S. As a part of this project, a sampling campaign was conducted in Son La, Vietnam in 2011.Until then, measurements of PCDD/F concentrations in ambient air from this area were very limited or incomplete.Due to climatological wind trajectory direction at high altitudes, long-range regional transports of PCDD/Fs (and other pollutants) via the southwest monsoon can be clearly observed (Lin et al., 2013;Wang et al., 2015).During the campaign, using data collected at the two high-altitude stations (Fig. 1), we target the potential relationship between biomass burning events in Son La and PCDD/F concentrations in Central Taiwan.
The Son La sampling site is located at the northeastern regional hydro meteorological observatory, National Hydro-Meteorological Service in Son La Province, Vietnam.Son La is a mountainous, rural area, which is approximately 300 km northwest of the capital Hanoi, approximately 140 km south of the border with China, and approximately 110 km to the east of the Laos border.The area of Son La province is approximately 14,200 km 2 with an estimated population of 1,134,000 people in 2011.The sampling station was placed on a high hill in Son La city, surrounded by vegetation.There is no occurrence of industrial activities or anthropogenic emissions within nearly 20 km of the Son La station.
Located at significant surface altitude in Jade Mountain National Park, conditions near Mt.Lulin are normally not affected by local pollution.Thanks to its unique location (free troposphere and free from boundary layer pollution), Lulin atmospheric background station play a very important part in the Global Atmosphere Watch project, especially for Southeast Asia (Wai et al., 2008).

Sample Collection and Analysis
During the 7-SEAS Son La Campaign (18-31 March, 2011), both 24-hour PCDD/F and total suspended particle (TSP) samples were collected at the Son La and Lulin stations simultaneously.PCDD/F compounds were collected from both vapor and solid phases using high-volume sampling instruments (HV-1000F, Sibata, Japan).Whatman quartz fiber filters (8 × 10 inches) were utilized to keep PCDD/Fs bound within particles, while the vapor phase of PCDD/Fs was captured by polyurethane foam (PUF).The flow rate was set so that a total volume of more than 1,000 m 3 could be sampled in 24 hours.After being conditioned, the net mass of collected TSPs on filter were measured using a microbalance (MX5 and AX205; precision of 1 µg, METTLER TOLEDO, USA) at a temperature of 23°C and relative humidity of 30 ± 5%.
For the PCDD/F analysis, the PUF and post-weighed TSP samples were Soxhlet extracted with toluene for 24 hours, treated with concentrated sulfuric acid, and then passed through a series of cleanup columns containing sulfuric acid on silica gel, acidic aluminum oxide, celite, and carbon.The cleaned solution was spiked with known amounts of the "Method 23" recovery standard solution.High-resolution gas chromatography (TRACE GC) and high-resolution mass spectrometry (DFS), both from Thermo Scientific, USA, were used to identify the mass of the seventeen dioxin homologues.The fused -silica capillary column (DB-5 MS, 60 m × 0.25 mm × 0.25 μm) came from Agilent J&W, USA.The mass spectrometer was operated with a resolution greater than 10,000 under positive electron ionization conditions, and the data for seventeen 2, 3, 7, 8-substituted PCDD/F homologues were analyzed in the selected ion monitoring mode.To determine the PCDD/F homologue, the Gas Chromatography (GC) Retention Time Window Defining Solution and Isomer Specificity Test Standard was used to define the beginning and end of the retention period for PCDD/F isomers, and to demonstrate the isomer specificity of the GC columns.
Measurements of trace gases (CO and Ozone) and particles (PM 10 ) have been continuously collected at Lulin station since 16 April 2006 as a part of the 7-SEAS project.These data are also considered.In addition, CO (APMA-360 CO analyzer, Horiba, Japan) and ozone (EC9810B Ozone Analyser, Ecotech, Australia) instruments were operated on the second floor of Lulin station.Eight Dekoron tubes (1/4" O.D., Synflex, USA), bundled together as an air intake line, were extruded to the roof with an inlet point approximately 10 m above the ground.An oil-free sirocco fan (JSD-30S, Jouning Blower Co., Taiwan) provided a gas flow rate of approximately 2.0 m 3 min -1 to the gas analyzers.The room temperature inside Lulin station is kept at approximately 25°C.

Trajectory and Tracer and Dynamic Models
The Hybrid Single-Particle Lagrangian-integrated trajectory (HYSPLIT) model, developed by Draxler and Rolph (2003), computes the trajectories, dispersion and deposition simulations for hypothetical air parcels (http://www.arl.noaa.gov/HYSPLIT_info.php).In this study, HYSPLIT was used to estimate the fate of emissions from Southeast Asia biomass burning and the transport pathway affecting PCDD/Fs concentrations in Central Taiwan.5day backward trajectories were solved above the sampling point at Lulin Mountain at an altitude of 3,000 meters.We used Moderate Resolution Imaging Spectroradiometer (MODIS) satellite data to identify fire hotspots.At each fire location, tracer simulations were performed and analyzed from the WRF-Chem (ver 3.0) (Grell et al., 2005) modeling system, using a tracer module (Lin et al., 2009) to identify long-range transport associated with biomass burning in Indochina.Our model followed the planetary boundary layer scheme developed by Yonsei University (Hong and Dudhia, 2003).The tracer simulations were performed with a horizontal resolution of 27 km and 200 × 200 points grid.The model was divided into 35 vertical levels, in which the lowest level was approximately 20 m above the surface.

RESULTS AND DISCUSSION
During biomass burning season in 2011, we carried out atmospheric monitoring simultaneously at Lulin observatory and the northeastern regional hydro meteorological observatory in Son La province.Since there are neither dioxin emission sources nor combustion sources within a distance of 50 km around Lulin observatory station, concentrations of background PCDD/Fs detected were relatively low.Previous monitoring campaigns at Lulin station in 2008 and 2010 (Chi et al., 2010(Chi et al., , 2013a) ) found PCDD/F concentrations of 0.710-3.41fg I-TEQ m -3 and 0.232-4.12fg I-TEQ m -3 , respectively.TSP concentrations detected ranged from 5.32 µg m -3 to 55.6 µg m -3 in 2008 and from 8.58 µg m -3 to 45.2 µg m -3 in 2010.All PCDD/F and TSP samples were taken at Lulin station during the regular sampling periods in March, June, September and November in the years 2008, 2009 and 2010 for the analysis of PCDD/Fs.Each month, 5 to 10 samples (24 hours sampling) were taken.
During 2011, concentrations of CO, O 3 and PM 10 were lowest in the summer.The background concentration was thus estimated by taking the average concentration of each pollutant during this period.When sampling at Lulin station in 2011, the recorded concentrations of CO (146 ± 54 ppb), O 3 (37 ± 7.3 ppb), PM 10 (8.9 ± 4.5 µg m -3 ) (Fig. 2), were all higher than their background concentration of 84 ppb, 26 ppb and 5.9 µg m -3 , respectively and thus by about 60%.In particular, concentrations measured in March and April (231 ± 91 ppb for CO, 37 ± 12 ppb for O 3 , and PM 10 : 13 ± 9.2 µg m -3 ) were as high as that of background values.
Annual meteorological data revealed that dominant winds during this period of time at Lulin station were westerly and southerly.From this, we hypothesized long-range transport of Southeast Asian biomass burning as contributing to the higher concentrations of anthropogenic pollutants observed in Lulin station during spring.Wai et al. (2008) found an association between biomass burning activity from Southern and Southeast Asia and the elevation of some chemical concentrations at Lulin station during peak Indochinese biomass burning season.Obrist et al. (2008) also found both gaseous and solid elemental mercury correlated to biomass burning.Combustion of rich-mercury compounds would release mercury into the atmosphere through a process called "wildlife-related mercury emission".Based on the meteorological data at Mt. Lulin, westerly and southerly winds prevailed beginning on 23 March.Therefore, it is likely that the impact of Southeast Asia biomass burning was influencing PCDD/F concentration over Taiwan at that time.
The HYSPLIT model was applied to calculate a 5-day back trajectory for the event on 26 March 2011 (Fig. 4).The calculation was begun at 3.0 km above Lulin station.The HYSPLIT result found that the trajectory of air masses passed 22-23 March over Indochina.These results suggested additional context linking pollutants in Central Taiwan and biomass burning in Southeast Asia.
Fig. 4 shows the diurnal variation of atmospheric PCDD/Fs, CO, ozone, and PM 10 measured at Lulin station from 19 to 31 March, 2011.The atmospheric PCDD/F concentrations during this period ranged from 2.55 to 19.1 fg I-TEQ m -3 .The average value (6.78 ± 4.5 fg I-TEQ m -3 ) of atmospheric PCDD/Fs was higher than our previous measurements (0.232-4.12 fg I-TEQ m -3 ), during periodic sampling campaign at Lulin station in 2008 and 2010 (Chi et al., 2010;Chi et al., 2013a).The results of the intensive observation program show the change in atmospheric PCDD/Fs concentrations at Lulin station.They underwent a significant increase from 3.69 to 11.1 fg I-TEQ m -3 and 3.32 to 19.1 fg I-TEQ m -3 , reaching their peaks on 23 and 26 March, before dropping markedly on the following days.Concentrations of CO and O 3 measured in February were 202 ppb and 38.9 ppb, respectively.Because of its long lifetime (15 to 60 days in atmosphere, Novelli et al., 1998), it is crucial to use CO as a pollution tracer.Large amounts of CO are generated by biomass burning, making CO a distinct sign of this process (Warneke et al., 2006).Concentrations of CO, O 3 and PM 10 measured at Lulin station from 23 to 26 March ranged from 194-371 ppb, 38-43 ppb, and 13-20 µg m -3 for each compound.The highest CO and PM 10 concentrations, 371 ppb and 20 µg m -3 , were observed on 25-26 March.Increasing trends of the two pollution tracers were similar to that of the atmospheric PCDD/Fs measured at Lulin station.
PCDD/F concentrations measured at the Lulin and Son La stations during the Southeast Asia biomass burning events (18 to 31 March, 2011) are shown in Fig. 5.Many active fire points were detected by the MODIS satellite (Fig. 6) during 21-23 March and 25-27 March in Indochina.Measurements in Central Taiwan found that the CO and PM 10 concentrations observed at Lulin station during the same periods ranged from 117-371 ppb and 3.8-20 µg m -3 .The combined results of air mass path and satellite data support the hypothesis that the peak of PCDD/F concentrations in Central Taiwan originated from biomass-burning regions.Comparison of PCDD/F concentrations between Taiwan and Vietnam from 19-30 March (Fig. 5) shows that the patterns of concentrations varying in the two areas share a similar tendency.
When concentrations increased at Lulin station from 21 to 23 March (from 3.69 to 11.1 fg I-TEQ m -3 ) and 23 to 26 March (from 3.32 to 19.1 fg I-TEQ m -3 ), corresponding increases at Son La were identified on 18 to 21 March (from 15.1 to 30.1 fg I-TEQ m -3 ).Our previous study (Chi et al., 2013a) also reported a similar finding when examining variation in PCDD/F concentration between Chiang Mai, Thailand and Lulin station.Even the patterns of variability at both locations were similar; there was a time lag of 2 to 3 days between them.
Peak of PCDD/F concentrations during 23 to 26 March at Lulin that do not appear to correspond with those at Son La were the result of long-range transport from other areas.Pollutant concentration measured at a receptor region is dependent on emission levels in the source region, transport pathway, and also the production and loss processes between the source and the receptor (Kuribayashi et al., 2012).
Applying WRF-Chem for tracer simulation, we sought to determine whether or not biomass burning activity in Southeast Asian could influence air quality in East Asia.Tracers were studied from 19 to 28 March according to the fire spots reported in MODIS satellite data.Above each fire spot, 10,000 units day -1 in tracers were simulated.Every six hours, meteorological information was updated from National Centers for Environmental Protection Final Operational Global Analysis datasets.The distribution of tracers and the wind field at 700 hPa (approximately 3 km) are described in Fig. 6.Winds at 10 m s -1 were observed between 20 and 30°N, before merging toward higher latitudes.According to the model, high concentrations of tracers were found in Taiwan beginning at 00:00 UTC on 24 until 27 March.This modeling result is similar to the data collected at Lulin sampling station (Fig. 3).Moreover, comparing the concentration of PCDD/Fs measured at Son La and Lulin station, the highest concentration of 59.5 fg I-TEQ m -3 made Indochina one of the major sources for pollutants in Central Taiwan at this time.
Distributions of atmospheric PCDD/F homologues also reveal some connection between ambient air at Lulin and Son La stations.This is different from urban and industrial  areas of Taiwan where over 60% of dioxins are PCDFs.During normal background conditions, PCDFs only contribute 50 ± 4.4% of total PCDD/Fs at Lulin station.However, during biomass burning events in Southeast Asia, 64% of the total dioxins were PCDFs.The same scenario also happened in results from Son La sampling station, where the percentages of PCDFs during significant burning events were much higher than that of non-significant burning events.Gullett et al. (2006) found that even PCDDs (particularly octa-CDD) dominated in raw biomass, whereas the emissions are dominated by PCDFs (mostly penta-PCDFs).The difference between composition of raw biomass and air emissions from combustion suggested that PCDFs in emissions are formed through in situ processes.Therefore, the higher percentages of PCDFs found in this study are an attribute of Southeast Asia biomass burning influenced by long-range transport.In the previous study, we found that long range transport can bring pollutants from the Southeast Asia to Mt. Lulin in Taiwan during the biomass burning events in spring due to southwest monsoon activity (Chi et al., 2010).The following study in Chiang Mai, Thailand confirmed this linkage (Chi et al., 2013a).In this study, the data collected from Son La, Vietnam also described the same connection.The finding from the two studies validated the significant relationship between atmospheric conditions in Taiwan and anthropogenic activities in Indochina.

CONCLUSIONS
This study presents the first detailed measurements of atmospheric PCDD/F concentration in northern Vietnam.In this paper, we conceptualized how with faster wind, PCDD/Fs originating from biomass burning in Southeast Asia can affect the air quality in downwind areas such as Taiwan.Both air sampling results at the field, trajectory modeling, and model simulation results using WRF-Chem, depicts higher PCDD/F concentrations in Taiwan through long-range transport during late March 2011.Analyzing the characteristics of air components in the receptor region, we conclude that a significant increase in atmospheric PCDD/F concentrations at a high-altitude background station in Central Taiwan occurred as a result of the Southeast Asia biomass burning events.Moreover, one unmatched point between peak of PCDD/F concentrations in Lulin with those at Son La could come from other areas such as Thailand.Therefore, a simultaneous study in Thailand and Vietnam could be proposed for future study to elucidate the sources of PCDD/Fs in Taiwan during spring time.A future study including more Southeast Asian countries could clearly illustrate this relationship.Our study not only plays a role in understanding the regional long-range transport between Taiwan and Southeast Asia but also contributes to the knowledge of global transportation of atmospheric pollutants.

Fig. 1 .
Fig. 1.Relative locations of two high-altitude sampling sites in Southeast Asia (satellite image provided by http://m aps.google.com).

Fig. 2 .Fig. 3 .
Fig. 2. Average concentrations of CO, ozone and PM 10 measured at Lulin station during different month in 2011.0