บทความวิจัยของมหาวิทยาลัยเชียงใหม่ล่าสุดตีพิมพ์ในวารสาร Chemosphere ตีพิมพ์ในวารสาร Chemosphere เป็นงานวิจัยที่ได้รับการสนับสนุนจาก สกว. (ปัจจุบัน สกสว.) ชื่อโครงการ “การติดตามตรวจสอบคุณภาพอากาศและสถานการณ์หมอกควันในภาคเหนือตอนบนของประเทศไทยเพื่อการประเมินผลกระทบต่อสุขภาพและสิ่งแวดล้อม”
Fresh and aged PM2.5 and their ion composition in rural and urban atmospheres of Northern Thailand in relation to source identification
โดยมี รศ.ดร.สมพร จันทระ หัวหน้าศูนย์วิจัยวิทยาศาสตร์สิ่งแวดล้อม คณะวิทยาศาสตร์ มหาวิทยาลัยเชียงใหม่ เป็นหัวหน้าโครงการ วัตถุประสงค์ เพื่อทำการเปรียบเทียบคุณภาพอากาศ โดยทำการเก็บฝุ่น PM2.5 เพื่อวิเคราะห์องค์ประกอบทางเคมี เปรียบเทียบระหว่างพื้นที่ในเมือง (เชียงใหม่ และลำปาง) กับพื้นที่นอกเมือง (เชียงดาว และแม่สะเรียง) และได้พบสารประกอบที่สามารถใช้เป็นตัวบ่งชี้อายุของฝุ่นในอากาศว่าเป็นฝุ่นใหม่จากกิจกรรมในพื้นที่เอง หรือเป็นฝุ่นเก่าที่เดินทางมาจากทางไกล หรือมีการสะสมในพื้นที่
3 สิงหาคม 2564
SaranaChansuebsriaPavidarinKraisitnitikulaWanWiriyaabSompornChantaraabaEnvironmental Science Research Center, Faculty of Science, Chiang Mai University, Chiang Mai, 50200, ThailandbEnvironmental Chemistry Research Laboratory, Department of Chemistry, Faculty of Science, Chiang Mai University, Chiang Mai, 50200, Thailand
Received 29 May 2021, Revised 30 July 2021, Accepted 3 August 2021, Available online 3 August 2021.
Handling Editor: R Ebinghaus
ในช่วงหน้าร้อน ซึ่งมีการเผาในที่โล่งจำนวนมากนั้นจะมีมลพิษ PM2.5 ในบริเวณนอกเมืองสูงกว่าในเมือง ส่วนช่วงนอกฤดูการเผา มลพิษในเมืองจะมากกว่าโดยเฉพาะบริเวณที่ใกล้แหล่งจราจร
นอกจากนี้ ยังพบสารประกอบที่สามารถใช้เป็นตัวบ่งชี้อายุของฝุ่น PM2.5 ว่าเป็นฝุ่นสดใหม่ (fresh) หรือเป็นฝุ่นเก่า (aged) ที่ค้างในอากาศนานแล้ว เพื่อประเมินว่าฝุ่นนั้นเกิดจากกิจกรรมในพื้นที่ (local) หรือมาจากที่อื่น (long range) หรืออาจจะถูกขังไว้ในพื้นที่ด้วยปัจจัยทางกายภาพหรือสภาวะทางอุตุที่เอื้อต่อการกักตัวของมลพิษ
• Fate of compounds of K bounded with PM2.5 as biomass burning tracer was reported.
• KCl is a main tracer found in fresh aerosols in the area with intensive open burning.
• KNO3 is a tracer for aged aerosols in ambient air caused by biomass open burning.
• High level of SIA was found in urban area due to mixed pollutant sources.
• Regional pollution causes the lack in spatial variation in pollutants composition.
This study aims to investigate ion composition of PM2.5 in various sites and seasons and to identify the main sources on spatial and temporal basis. PM2.5 compositions of two urban and two rural areas in Northern Thailand in 2019 were investigated to distinguish urban traffic and rural open burning sources. During the burning season, average PM2.5 concentrations in rural areas (104 ± 45 μg m−3) were slightly higher than those in urban areas (94 ± 39 μg m−3). Source identification of PM2.5 by cluster analysis during burning season in urban sites and one rural site revealed mixed sources of aged aerosols from biomass burning, traffic and transboundary pollution, characterized by (NH4)2SO4 and KNO3. Only PM2.5 in one rural area (Chiang Dao), where intense open burning activities observed, contained significant KCl level in addition to other compounds. KCl is being used as a tracer for fresh aerosols from biomass burning as opposes to KNO3 for aged aerosols. It was found that KNO3 proportion in total ions increased with PM2.5 concentrations both in urban and rural areas, indicating prominent open burning influences in regional scale. Source identification in other seasons was more distinguishable between urban and rural areas, and more varied depending on local emissions. Urban PM2.5 sources were secondary inorganic aerosols from traffic gas conversion in contrast with rural PM2.5 which were mainly from biomass burning.
PM2.5Water-soluble ionsSmoke hazeOpen burningKClAir pollution
Particulate matter with aerodynamic diameter less than 2.5 μm (PM2.5) is one of the prominent air pollutants in smoke haze, that persistently occur in South East Asia (SEA) for many years. It causes a broad range of harmful health effects (Wang et al., 2021) and therefore regarded as one of the indicators for measuring air quality. Open burning of agricultural waste and forest fire are the main sources of PM2.5 (Chantara et al., 2012). Simultaneously, meteorology in hot dry season (the lack of rain and air movement) and topography (as a basin area) of Northern Thailand further promoted the accumulation of pollutants (Kim Oanh and Leelasakultum, 2011). These factors cause an episodic of intense smoke-haze annually. The rising population and resource demand also increase the number of open burning and risk of forest fire. However, while most recognize open burning as the top polluters in dry season, traffic is also a well-known source of PM2.5 that emitted all year round. Many previous studies in Northern Thailand were focused on investigating PM2.5 from biomass burning alone but none have comparatively studied PM2.5 from urban traffic and biomass together. Although PM2.5 were contributed by biomass burning for 25–79 % (Pani et al., 2018), traffic was also blamed for its contribution to the occurrence of smoke haze pollution by many parties, especially the government sectors. Yet, ambient PM2.5 from traffic emission in Northern Thailand were still largely unexplored. Understanding the differences of its chemical characteristics in urban area and rural area as representative of traffic and open burning origin, respectively, is needed for future investigation of source contribution. In addition, PM2.5 chemical compositions among hot dry season, cool dry season and wet season are useful for the comparison with and without the influence of open burning.
Source identification is necessary for the development of reduction and mitigation strategies. Water-soluble ion (WSI) composition in PM2.5 can be used in Principal Component Analysis (PCA) to identify the source. For examples, WSI datasets were applied to PCA in Thepnuan et al., (2019), Khamkaew et al., (2016), Blaszczak (2018), Gao et al., (2011) and Bandpi et al., (2020). WSI of PM2.5 in Chiang Mai had been analyzed in earlier studies (Khamkaew et al., 2016; Thepnuan et al., 2019; Chantara et al., 2012). SO42−, NO3−, NH4+ or Secondary Inorganic Aerosol (SIA) were found to be the main component in PM2.5. K+, which are widely used as biomass burning tracer, were found in the fourth highest concentrations following the SIA in ambient PM2.5 in Chiang Mai-Lamphun Basin during smoke haze season. Ambient K+ was found to have high correlations with levoglucosan, a direct combustion product of cellulose, both in China and Northern Thailand during extensive open burning period (Tao et al., 2013; Khamkaew et al., 2016; Thepnuan et al., 2019). Thus, K+ can be an acceptable tracer for biomass burning in Northern Thailand. However, earlier studies in Northern Thailand only had one or two sampling sites and only characterized ambient PM2.5 collected during smoke haze season or dry season (Thepnuan et al., 2019; Khamkaew et al., 2016). This study was the first study in this region to investigate PM2.5-bound ions through different part of the year. We also considered spatial variability of PM2.5-bound ions by expanded the investigation through multiple urban and rural atmospheres of Northern Thailand.
In this study, we investigated WSI-bound PM2.5 compositions in urban and rural area both in wet and dry seasons to distinguish PM2.5 from urban traffic sources and rural open burning sources. We attempted to find the differences in ion compositions in various sites and seasons and to identify the main sources on spatial and temporal basis by principal component analysis. This research aims to know whether the changes in pollution sources in urban and rural areas on Northern Thailand have any effects on ion compositions in ambient PM2.5.
2.1. Sampling site description
Sampling sites were in Chiang Mai-Lamphun Basin, Lampang Basin and Mae Sariang Basin to represent Northern Thailand (Fig. S1). Urban sites represented traffic sources, while rural sites represented near-open burning sources. Urban sites are in cities of Chiang Mai (CM) and Lampang (LP), the largest and second largest metropolitan cities in Northern Thailand. Air sampler in CM was on the rooftop of the two-story Chiang Mai meteorological station building (~6 m high) while that of LP was ground level roadside (~1.5 m high) in front of the largest school in Lampang. Air samplers were set ~2 m distant from the road for LP and 15 m for CM sites. CM site was next to one of the busiest road in Chiang Mai (~100,000 vehicles per day in the nearby intersection). Rural sites are in small towns of Chiang Dao (CD) District in northern part of Chiang Mai Province and Mae Sariang (MS) District in southern part of Mae Hong Son Province, located near Myanmar border. Air samplers in CD and MS were 2.5 m aboveground on the rooftop of meteorological stations. The areas of districts were 152 km2 (CM), 1157 km2 (LP), 1882 km2 (CD) and 2587 km2 (MS) and the population density were 1539 km−2 (CM), 196 km−2 (LP), 51 km−2 (CD) and 21 km−2 (MS) (National Statistical Office, 2021). Seasons in Northern Thailand were categorized by the presence of rain as wet season (May–September) and dry season (October–April). Dry season was divided into cool dry (October–January) and hot dry (February–April) seasons. Each period has distinct meteorological patterns that influence the number of open burning activities. Hot dry season is characterized by lack of rain, low relative humidity (61 ± 6 %) and high temperature (27 ± 3 °C) while cool dry season is lack of rain, high relative humidity (77 ± 4 %) and low temperature (23 ± 2 °C) (Thailand Climatological Center, 2021). Intensive open burning activities are often detected in hot dry season because of the accumulated dry biomass. In cool dry season, the humidity in the air and in the biomass residue is still high after the passing of wet season. Thus, fire active counts are low, particularly in the forest area.
2.2. Sampling of PM2.5
Ambient PM2.5 samples in 2019 were collected for 24 h from 9 a.m. on quartz-fiber filter (QM/A, Whatman Inc., China) using mini-volume air samplers (MiniVol, Airmetric, USA) and low volume air samplers (PQ200, BGI, USA) with flow rates of 5.0 and 16.7 L min −1, respectively. During hot dry season (February to April), PM2.5 samples were collected every 3 days while during intense smoke haze period (13–19 March, 22–25 March, 6–9 April), samples were collected daily. During wet season and cool dry season, samples were collected once a week for 4 months (May to June and November to December). Field blanks were also collected on site using quartz filters with the same preparation as sampling filters for 5 min operation using the same air sampler. Filters are kept in a desiccator before pre- and post-weighted under controlled temperature and humidity to determine PM 2.5 mass concentration using MX5 microbalance (Mettler Toledo, Switzerland). Samples were kept in −20 °C until further analysis.
2.3. Analysis of PM2.5-bound water-soluble ions (WSI)
Sample filters and field blanks were extracted in 20 mL deionized water at 35 °C for 30 min using ultrasonicator (Elma, Germany). Extract solutions were filtered with cellulose acetate membrane (pore size 0.45 μm, Ø 13 mm) and kept in PE bottle at 4 °C. Analysis of cations (Na+, K+, NH4+, Ca2+, Mg2+) and anions (SO42−, NO3−, Cl−) was done by Ion Chromatograph (882 Compact IC plus, Metrohm, Switzerland) using the conditions from Sillapapiromsuk et al., (2013). Limit of Detection (LOD) of cations and anions were 0.005–0.027 and 0.001–0.004 μg mL−1, respectively. Accuracy was determined by analysis of mixed ion standard-spiked filters and ion recoveries ranged from 93 to 99 %. Standard reference material (low concentration of artificial rain sample; AR182) provided by the acid deposition monitoring network in East Asia (EANET), was also tested for an accuracy of ion analysis. Percent differences between measured values and prepared values less than 15 are considered well accurate, while those under 30 are moderately accurate. The result showed that different values of six ion species (SO42−, NO3−, NH4+, K+, Ca2+ and Mg2+) were less than 15 %, while those of Cl− (17 %) and Na+ (33 %) were slightly exceeded the limits but within acceptable ranges.
2.4. Data analysis
All statistical analysis was performed using SPSS. One-Way ANOVA was used to determine significant different among the sampling sites. Principal Component Analysis (PCA) was used to analyze multivariate WSI in PM2.5 to identify potential sources in urban and rural sampling sites. Datasets were analyzed with Varimax rotation. Components were extracted based on Eigenvalues higher than one.
2.5. Fire hotspot count
Fire hotspot count data was retrieved from NASA’s Fire Information for Resource Management System (FIRMS), using satellite data from Visible Infrared Imaging Radiometer Suite (VIIRS) instrument. The resolution of VIIRS sensor is 0.375 km × 0.375 km of the burned area to be counted as one (Choommanivong et al., 2019). Fire maps were plotted using ArcGIS. Daily active fire counts in Northern Thailand (9 provinces) were determined and compared with daily PM2.5 concentrations.
3. Results and discussion
3.1. PM2.5 mass concentrations and fire hotspot count
Ambient PM2.5 samples were collected in hot dry season 2019. Daily PM2.5 concentrations with fire hotspot counts (VIIRS) in 9 provinces of Northern Thailand are shown in Fig. 1. Overall, PM2.5 concentrations started to increase in mid-February and reached the highest concentration in March then gradually decreased in April. The highest daily concentration was 264 μg m−3 in rural CD site. Active fire count peak days often followed by PM2.5 peak days few days later. Only the active fire peak on 31 March (see arrow) followed by the increase in PM2.5 concentrations in MS. The heightened active fire counts on 31 March was mostly attributed from massive forest fire in MS district. The average PM2.5 concentrations in hot dry season in descending order were 115 ± 50 μg m−3 (CD), 110 ± 29 μg m−3 (LP), 80 ± 32 μg m−3 (MS) and 77 ± 41 μg m−3 (CM) (Table S1). The values of rural CD and urban LP were not significantly different as well as those of rural MS and urban CM. Rural CD had the highest average PM2.5 concentrations because of high open burning activities both in agricultural and forest areas. Urban LP had the second highest PM2.5 concentrations likely due to the combined effects from regional pollution from open burning and roadside traffic. Despite CM being the urban site, the average PM2.5 concentrations was the lowest, suggesting less open burning and traffic influences comparing to urban LP. Urban sites were more affected by microenvironment due to wide variation on factors such as the proximity to emission sources, microclimate and surrounding urban landscape (Tong et al., 2017; Zhao and Xu, 2019) Therefore, PM2.5 concentrations in urban sites tend to have high spatial variability. Lastly, the average PM2.5 concentrations in rural MS was higher than those of urban CM despite being a remote area with low population. PM2.5 in rural MS site was likely from local open burning sources and long-range transport from Myanmar, where fire activities were often greatest in SEA (Punsompong and Chantara, 2018; Sirimongkonlertkun, 2018).
Fig. 1. p.m.2.5 concentrations in urban sites (CM and LP) and rural sites (CD and MS) with active fire hotspot counts in Northern Thailand.
In wet and cool dry seasons, additional samples were also collected once a week. The concentration in wet season should be able to represent background concentration. PM2.5 concentrations were evidently lower than those of hot dry season, except at rural MS and urban LP sites where the concentrations were surprisingly high (37–60 μg m−3 and 39–71 μg m−3, respectively). The average PM2.5 concentrations in wet season in urban sites were 20 ± 8 and 53 ± 10 μg m−3 in CM and LP sites, respectively. While the concentrations in rural sites were 25 ± 7 and 47 ± 6 μg m−3 in CD and MS, respectively. High PM2.5 concentrations in the urban LP may resulted from being the roadside sampling and thus was heavily affected by traffic and road dust. Ca2+, an ion of crustal origin, was found in the highest average concentration in this site. Conversely, high PM2.5 concentrations in the rural MS may cause by high relative humidity (RH). Average RH value in wet season of MS (72 ± 12 %) was significantly higher than that of urban areas both CM (64 ± 8 %) and LP (66 ± 10 %). High RH can facilitate the formation of secondary inorganic salts (SO42−, NO3−) and increase particle size and density (Li et al., 2015; Zhang et al., 2015). Additionally, open burning season usually end at the end of April or the beginning of May (Yabueng et al., 2020). However, considerable number of open burning activities remained until mid-May 2019 and then subsided. The number gradually increased again throughout December. PM2.5 concentrations sharply increased in urban LP site in late December, due to relatively high local active fire hotspots. Cool dry season also had moderately high PM2.5 concentrations due to the lack of rain scavenging (Duan et al., 2016).
Fig. 2 shows fire hotspot distribution maps (VIIRS) by months in the district where the sampling sites located, including their surrounding area. The largest and second largest quantity of active fire hotspot counts were found in the districts of rural MS (4183 spots) and rural CD (2164 spots). However, most of the hotspot counts detected in MS District were massive forest fires during 24 March to 5 April (3072 spots) (Fig. 2 left). Thus, we can arguably state that only rural CD constantly had intensive open burning throughout the season while the rural MS only had a short period. Moreover, the burning activities in the area surrounding rural CD were relatively higher than those of the others. Additionally, active fire hotspot counts in urban LP area was quite high (895 spots), but most of the fire count was detected outside the main city. Therefore, high PM2.5 concentrations in urban LP might be mainly contributed from traffic, unlike urban CM. This might be factors of sampling point distance and elevation from the road.
Fig. 2. Fire hotspot distribution maps in the district of the sampling sites during hot dry season. Bottom left shows daily fire hotspot count in each district.
3.2. Water-soluble ion composition of PM2.5
WSI analyzed in this study were 3 anions (Cl−, SO42− and NO3−) and 5 cations (Na+, NH4+, K+, Ca2+ and Mg2+). The Mass percentage of WSI are shown in Fig. 3. Additionally, daily total ion and PM2.5 concentrations are shown in Fig. S2. Total ion concentrations in all sites and seasons were 11–15 μg m−3 (hot dry season) and 3–8 μg m−3 (wet and cool dry season) (Table S1). Total ion concentrations corresponded with PM2.5 concentrations, with strong correlations (r ~ 0.7–0.9) in hot dry season. The correlations in other seasons were lower (r ~ 0.4–0.8). Generally, ion composition in hot dry season were similar in all sites but those in wet and cool dry season have more variation particularly of minor ions such as Na+, Cl− and Ca2+. Cl− concentration in rural MS samples was unusually high (1.04 ± 0.56 μg m−3 average) in November and December. We speculated that Cl− may come from KCl and NH4Cl from biomass burning and fertilizers (Khamkaew et al., 2016). Another speculation was that they were sea salt, but high correlation between Na+ and Cl− (r = 0.851) was only observed in hot dry season. Ca2+ concentrations in urban LP in hot dry season (0.80 ± 0.40 μg m−3) and cool dry season (0.27 ± 0.17 μg m−3) were also relatively higher than other sites (1.6–9.0 fold, respectively). Ca2+ is associated with soil particles and wind speed and, without the influence of biomass burning, often found to be the most abundant ion seconded to SIA, particularly in urban area from road dust and/or area with low land cover (Ghosh et al., 2012; Kundu and Stone, 2014; Huang et al., 2018). Major ions in all sites were SO42−, NO3−, NH4+ and K+ and most contributed to total ion concentrations at approximately 75–95 %. SO42−, NO3− and NH4+ (SIA), followed by K+ in order of concentrations were common characteristics found in most studies regarding ambient PM2.5 ion components in Chiang Mai-Lamphun Basin during hot dry season (Khamkaew et al., 2016; Pani et al., 2018; Thepnuan et al., 2019). K+ was widely used as a tracer for biomass burning (Tao et al., 2013; Pachon et al., 2013). It has been found corresponding with levoglucosan, a direct combustion product of cellulose, in ambient PM2.5 during biomass burning season in China and Northern Thailand (Cheng et al., 2013; Thepnuan et al., 2019; Khamkaew et al., 2016). K+ is well known as a biomass burning tracer with high open burning activities mentioning in various publications (Zhang et al., 2010; Chantara et al., 2012; Thepnuan et al., 2019; Khamkaew et al., 2016). Also, an abundant concentration of K+ were found in biomass burning from the chamber, field and hood pile experiments (Chantara et al., 2019 and Kim Oanh et al., 2011). Apart from K+, Cl− was also emitted as much as those of K+ from these primary emission experiments. Some regions such as Fuling and Chengdu, China with biomass burning as main contributor, found K+ and Cl− in comparable concentrations in ambient PM2.5 (Tao et al., 2013; Qiao et al., 2019). In this study, only rural MS exhibited higher Cl− than K+ concentrations in December when open burning started. Thus, KCl were treated as a tracer for fresh or primary aerosols from biomass burning in this study, which was also agreeable with Wang et al., (2017). In addition, while fresh smoke from biomass burning contains more KCl particles, aged smoke had more KNO3 and K2SO4 particles (Li et al., 2003). During the aging process, Cl− was replaced by NO3− and SO42−, yielding KNO3 particles and HCl gases (Li et al., 2015). Regardless, only K+ and NO3− exhibited strong correlation (r ~ 0.8–0.9) while those of K+ and SO42− were only none to medium (r ~ 0.2–0.6). Thus, we acknowledged KNO3 as a tracer for aged aerosols or long-ranged particles from biomass burning in this study. Additionally, SO42− and NO3− were associated to both biomass burning and traffic emission. SO42− was reported as dominant ion in ambient PM2.5 in Chiang Mai-Lamphun Basin, while NO3− was in second. Unlike the trend in decreasing fraction of SO42− but increasing of NO3− in industrial-based countries in Asia (Singh et al., 2020; Xu et al., 2019), SO42− in SEA remained the greatest due to extensive biomass burning, uncontrolled SO2 emission from industrial sector, coal combustion and vehicle exhausts (Thepnuan et al., 2019; Lee et al., 2019). SO42− emission was also comparable to K+ emission in forest litter burning experiment (Chantara et al., 2019). Na+, Ca2+ and Mg2+ were found in low concentrations to non-detectable in some samples (0–10 % by mass). In wet season and cool dry season, SO42− and NO3− were still the main component. NH4+ percentage dramatically decreased, particularly in June. Na+ and Cl− in urban CM and rural CD had increased percentage. More obvious trend was observed in rural MS. Other ions such as Ca2+ and Mg2+ can only be detected in LP most likely a trace of road dust from traffic.
Fig. 3. Mass ion composition in urban sites (CM and LP) and rural sites (CD and MS).
Fig. 4 shows the spatial variation of PM2.5 and ions in various seasons. Urban LP mostly had the highest average PM2.5 and ion concentrations among the sampling sites, except in hot dry season when average PM2.5 concentrations of rural CD and urban LP were comparable. The pattern was even clearer in cool dry season because LP usually had the earliest occurrence of open burning activities in December. Higher ion concentrations in urban LP than urban CM also indicate that ground level and roadside sampling were significant factors. Sampler in urban LP was at ground level while sampler in urban CM was on the rooftop of two-story building and ~10 m far from the main road. Thus, at ground level and near roadside, traffic may contribute more to total PM2.5 in hot dry season than originally believed. Moreover, each average ion concentrations in rural CD were comparable to those of urban LP, except for K+ which rural CD was slightly higher. This agreed with rural CD being near open burning sources. In addition, NO3− was found to be relatively high in urban LP in wet and cool dry season, likely from NOx emission from traffic. Comparing between urban CM and rural CD in hot dry season, concentrations of K+ and NO3− in rural site were significantly higher, while that of Ca2+ in urban site was also significantly higher, suggesting the influences from open burning and road dust, respectively. In wet and cool dry season, all ions concentration in urban CM and rural CD were not significantly different.
High correlation between ions either means they come from the same sources or they were formed as a compound. NH4+ and SO42− were strongly correlated (r ~ 0.8–0.9) in all sites and seasons while NH4+ and NO3− was moderately correlated (r ~ 0.6–0.7). (NH4)2SO4 is the most common ammonium salt in the atmosphere because of its low vapor pressure compared to NH4NO3 (Kong et al., 2014 and Long et al., 2014). The forming mechanism involves oxidation of SO2 and NOx into H2SO4 and HNO3 through gas-phase oxidation and/or heterogeneous oxidation in aqueous phase by ambient oxidizing agents such as O3, H2O2, OH radical etc. H2SO4 and HNO3 then react with NH3 in acid-base neutralization into (NH4)2SO4 and NH4NO3 (Zhang et al., 2015). K+ and NO3− also had strong correlation (r ~ 0.8–0.9) in almost all sites and seasons, except for rural area in wet and cool dry season where correlation was only moderate to none (CD and MS, respectively). K+ and NO3− also had high affinity because KNO3 is non-volatile and the secondary product of other potassium salt (Galon-Negru et al., 2018). K+ and Cl− also had moderate correlation (r ~ 0.6–0.7) in rural sites in wet and cool dry seasons, indicating local biomass burning. Additionally, positive correlation between ratio value of ion/PM2.5 and PM2.5 concentrations can indicate the increasing ion contribution to the increasing PM2.5 (Fig. 5). It was found that correlation between K+ and NO3− was positive while NH4+ and SO42− was negative in hot dry (smoke haze) season. This suggested that KNO3 was progressively contributed to the increasing PM2.5. Conversely, contribution of (NH4)2SO4 became less. Even though, SO42− and SO2 can be emitted from biomass burning (Chantara et al., 2019), this finding suggests that contribution of biomass-related SO42− to ambient PM2.5 in Northern Thailand were relatively less than K+ and NO3−. Existing non-biomass-related SO42− may come from traffic and industrial-related emission which should be relatively stable in regional scale if not affecting by accumulation. However, pathways of SO42− are complex and many possible products can appear in the atmosphere (Wang et al., 2016; Guo et al., 2019). As KNO3 was a tracer for open burning, it could imply that increasing PM2.5 concentrations was contributed by aerosols from open burning. In wet and cool dry seasons, only NO3− exhibited moderate to strong correlations in urban LP and CM, indicating the contribution of traffic via NOx to NO3− conversion. However, SO42−/PM2.5 in urban area still had no relationship with PM2.5 despite traffic become more influential in these seasons.
3.3. Pollutant source identification by principal component analysis (PCA)
PCA was used in factor analysis to group all analyzed WSI species and to identify possible sources of pollutants. The results yielded three components during hot dry season in all sites. Datasets in wet season and cool dry season were analyzed together, yielding two and three components in urban and rural areas, respectively (Table 1).
Table 1. Factor loadings in principal component analysis and their associated sources.
Notes – Ions with factor loading in multiple PCs, only PC with maximum correlation is acknowledged.
Absent of Mg2+ and/or Ca2+ means they were not detected in any samples.a
Bold indicates negative correlation.b
Fresh or aged BB refer to fresh or aged aerosols from biomass burning.
Principal Component 1 (PC1) of the two urban sites (CM, LP) and one rural site (MS) in hot dry season had high loading of NH4+, SO42−, NO3− and K+, accounted for 35–42 %, identified as mixed of SIA and aged aerosols from biomass burning. The major ions were (NH4)2SO4 and KNO3. Urban area had high level of SIA because of various traffic-released gaseous precursors. These pollutants combined with the long-suspended smoke-haze from regional open burning, generating same composition profiles in all urban atmosphere. Although CD and MS were both rural sites, PC1 in MS had the same factor loading as in urban sites. We suspected that PM2.5 in MS was heavily influenced by trans-boundary pollution from Myanmar. Thus, its atmosphere was dominated by SIA and K+ much like in urban atmosphere. Backward trajectories (BWT) showed that air mass arriving at all sampling sites in hot dry season pass through many regions with exceeding number of open burning. CM, LP and MS mostly received the air mass from the Southwest at both 10 m and 1500 m altitudes (48–70 %) from western part of Northern Thailand, Myanmar and along the coast of Myanmar. Rural CD (the northernmost) also received addition air mass (29–37 %) from Bangladesh and India (Figure S3). However, only rural CD differed from the others with high loading of K+, NO3−, Cl− and Ca2+ in PC1, accounted for 34 %, identified as fresh and aged aerosols from biomass burning. KCl and KNO3 were key tracers for fresh and aged aerosols from biomass burning, respectively. This indicated that degree of local open burning activities affect WSI composition of ambient PM2.5. Originally, rural MS was believed to have high level of local open burning activities similar to rural CD. However, rural MS in 2019 only had a short period of high open burning activities due to forest fire as aforementioned (Fig. 2). Thus, rural CD was alone recognized as intensive open burning area throughout the season. Additionally, rural CD results were more like rural Doi Ang Khang (DAK), a mountain site surrounded by forest and agricultural fields near Thailand-Myanmar border (Khamkaew et al., 2016). PC1 of DAK and rural CD both had loading of K+ and NO3−, identified as biomass burning sources, while PC2 had NH4+ and SO42−, identified as SIA. Other minor sources identified in PC2 and PC3 included crustal materials and sea salts. Crustal materials may either come from soil combustion during open burning, road dust and soil suspension (Wang et al., 2016; Khamkaew et al., 2016). Sea salts (Na+ and Cl−) came from western and southwestern wind (Figure S3) and were loaded in PC2 (23.5 %) of rural MS which was closest to the sea. PC2 in urban CM and LP had high loading of Mg2+ and Ca2+, accounted for 22–29 %, indicating the influence of road dust.
In wet and cool dry season with the absent of regional pollution, urban and rural area gave more distinguishable results, indicating more roles from microenvironment. Urban area had NH4+ and SO42− in PC1, accounted for 44–47 %, whereas K+ and NO3− shifted to PC2, accounted for 35–39 %. Urban area with high factor loading of NH4+ and SO42− in PC1 was similar to the results reported by Ghosh et al., (2012). In rural area, K+ and Cl− were the common ions in PC1 of both rural areas, indicating local biomass burning sources. BWT showed that air masses were mostly local within Northern Thailand accounted for 80–90 % in cool dry season (Figure S3). Biomass burning in wet and cool dry season may include small-scale open burning of agricultural residues, meat cooking and firewood burning.
This study attempts to find the differences in ambient ion compositions in rural area and urban area both in wet and dry seasons, and to identify the main sources on spatial and temporal basis by PCA. The results showed that urban CM, urban LP and rural MS had similar ion compositions because of the existing regional pollution from open burning throughout Northern Thailand and neighboring countries to the west. Sources of pollution identified by PCA in these areas were a mixed of SIA and biomass burning from long range transport. The chemical composition of PM2.5 in the area were the combination of both locally and non-locally released pollutants suspended in the air for long time as secondary pollutants or aged aerosols. However, area with intensive open burning activities such as rural CD contained significant primary pollutants or fresh aerosols from biomass burning, with KCl as the tracer. Despite the nature of dominating SIA in ambient air in most studies, it suggests that extreme biomass burning in the area can affect ion composition of ambient PM2.5 by adding significant primary ionic pollutants to the atmosphere. Ambient PM2.5 chemicals characteristics were greatly influenced by proximity to open burning sources and its level of burning. The atmospheric fate of K+ compound between KCl and KNO3 were also revealed. KCl was relatively short-lived and turn into KNO3 during the aging process. This can help determine if pollution from biomass burning was local or from long distance. Additionally, traffic may also contribute a considerable proportion to total PM2.5 concentrations and affect its composition, especially at roadside as observed in urban LP.
Source identities in non-open burning season indicated more spatial varieties influenced by microenvironments due to the absent of regional pollution. Main sources of pollution in urban sites were SIA (NH4+ and SO42−) while those in rural sites were local biomass burning (K+ and Cl−) and agricultural-related emission (NH4+ and Cl−). In cool dry season when open burning started, freshly emitted pollutants contain K+ and Cl− concentrations comparable to those in open burning season.
Air mass movement showed that, while pollution typically come from the west and southwest during hot dry season as observed in previous studies in Northern Thailand, the air mass movements in cool dry season were mostly local. This reinforces the idea that KCl in cool dry season was from local biomass burning.
Credit author statement
Sarana Chansuebsri: Methodology, Validation, Formal analysis, Investigation, Writing – original draft, Visualization, Data curation. Pavidarin Kraisitnitikul: Data curation, Visualization. Wan Wiriya: Conceptualization, Project administration. Somporn Chantara: Conceptualization, Resources, Writing – review & editing, Visualization, Supervision, Project administration, Funding acquisition.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
This study was part of the project “Monitoring of air quality and smoke-haze in Upper Northern Thailand for environment and health impact assessment” funded by Thailand Science Research and Innovation (TSRI), grant number: RDG61A0031. This research work was partially supported by Chiang Mai University. Additionally, the authors gratefully acknowledge the NOAA Air Resources Laboratory (ARL) for the provision of the HYSPLIT transport and dispersion model used in this publication. We also grateful for the supports from Thailand Meteorological Department for manpower and sampling locations.
Appendix A. Supplementary data
The following is the Supplementary data to this article:Download : Download Word document (3
ขอบคุณคณะผู้วิจัย หน่วยงานและผู้เกี่ยวข้องทุกภาคส่วนที่เอื้อเฟื้อสถานที่ในการเก็บตัวอย่าง และแหล่งทุนที่เห็นความสำคัญของงานด้านนี้
งานชิ้นนี้เป็นส่วนหนึ่งของวิทยานิพนธ์ ของ นายสรณะ จรรย์สืบศรี นักศึกษาปริญญาโท สาขาวิทยาศาสตร์สิ่งแวดล้อม
สามารถ download ได้ฟรี 50 วัน จนถึงวันที่ 24 กันยายนนี้
ข้อมูลโดย : คณะทำงานด้านวิชาการเพื่อสนับสนุนการแก้ไขปัญหาหมอกควันภาคเหนือ มหาวิทยาลัยเชียงใหม่
Academic Center For Air Pollution in Northern Thailand, Chiang Mai University: AcAirCMU
CMU Model การจัดนิทรรศการ การดำเนินงานร่วมจังหวัดเชียงใหม่ การดำเนินงานร่วมภาคเอกชน การดำเนินงานร่วมสื่อสารองค์กร การประชุม การเป็นวิทยากร การให้ข้อมูลสื่อมวลชน การให้สัมภาษณ์ ข่าว ข่าวประชาสัมพันธ์ คณะทำงาน บทความ บทความมลพิษทางอากาศ มหาวิทยาลัยเชียงใหม่ วิจัย สภาลมหายใจเชียงใหม่ หลักสูตรการจัดการด้านมลพิษทางอากาศสำหรับภาคเหนือ โครงการอื่นๆ