使用受體模式結合手動採樣有機化合物與連續監測資料探討臺北市細懸浮微粒污染源

Abstract

細懸浮微粒（PM2.5）對人體及生態的危害眾所皆知，因此掌握PM2.5的來源及貢獻量對空氣品質政策的制定十分重要。透過正矩陣因子分解受體模式（PMF）從受體點量測到的濃度資料追溯其來源和貢獻量，可以提供空氣品質政策科學上的建議。 PM2.5成分在線式監測儀器可以隨時監測元素、離子等化學物質以獲得成分小時值，而在過去PMF相關的研究中這些高時間解析度的資料能有效的解析污染源。也有研究表示加入有機化合物更可增強污染源的識別，特別是二次有機氣膠對於追蹤污染源有重要的價值。這些研究強調了有機化合物在PMF中的重要性，但大多受到採樣方法的限制使時間解析度較低。同時納入在線式監測高時間解析度以及有機化合物低時間解析度的研究較少，因此本研究嘗試結合兩者進行分析。 本研究於2022年11月至2023年4月，在台灣台北市大安空氣品質測站進行採樣。手動採樣使用石英濾紙，每次採樣12小時，並使用極致液相層析串聯質譜儀（UPLC-MS/MS）分析選定的有機化合物。結合測站連續監測的小時成分資料，使用多重線性引擎（ME-2）進行多重時間解析度的源解析。 使用PMF分析時分成模型一及模型二。模型一包含了24種成分測站的物種，模型二則額外增加7種作為特定污染源指標的有機化合物。模型一解析出六種污染源，分別是：交通（22.4%），揚塵（4.5%），燃油燃燒（17.4%），煤炭燃燒/工業（27.3%），工業（6.5%）和海鹽（13.1%）。模型二相較於模型一還額外辨識出名為「生物源」的污染源，特徵物種包含2-methylerythritol (2-MT)和arabitol。透過逆軌跡模式得到這些生物性氣膠可能來自台北市周邊山區或是宜蘭。另外，還透過levoglucosan和succinic acid辨識了可能潛在混合於交通源中的生質燃燒污染。 本研究凸顯了PMF分析在包含有機成分後能改善預測結果，多辨識出了一種污染源是與過去研究不同之處。此結果也讓未來在考量針對台北市的污染源進行管控時，提供一定的科學依據。

Air pollution, particularly fine particulate matter (PM2.5), has significant adverse health effects and contributes to atmospheric visibility reduction and global climate change. Understanding the distribution and sources of PM2.5 is crucial for effective air quality management. Receptor models, such as Positive Matrix Factorization (PMF), can help identify pollution sources by analyzing ambient concentration data at receptor sites. Online monitoring instruments for PM2.5 composition allow real-time measurement of elements, ions, organic carbon (OC), and elemental carbon (EC), enabling the detection in hourly patterns. PMF modeling, combined with these measurements, effectively explores PM2.5 contributions. Adding organic compounds enhances the identification of pollution sources, particularly secondary organic aerosols (SOA) in tracking pollution sources. Previous studies emphasized the importance of organic tracers in PMF modeling, but most have faced limitations in time resolution due to manual field sampling. Few investigations have incorporated both low time resolution data of organic compounds and high time resolution data from online monitoring. This study was conducted from November 2022 to April 2023 at the Daan Air Quality Monitoring Station in Taipei, Taiwan. Manual sampling was performed using quartz filters, with each sampling period lasting 12 hours. The selected organic compounds were analyzed using ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS). To integrate the hourly component data obtained from the continuous monitoring at the station, a multilinear engine (ME-2) was employed for source apportionment. The study aimed to identify and apportion pollution sources using the PMF model. Two models were implemented in this study: Model 1, which included 24 species, and Model 2, which included an additional 7 organic species. For Model 1, it was found that the optimal solution consisted of 6 factors: traffic (22.4%), dust (4.5%), oil combustion (17.4%), coal combustion/industry (27.3%), industry (6.5%), and sea salt (13.1%). In Model 2, an additional source called "Biogenic Source" was identified. This source was characterized by the presence of 2-methylerythritol (2-MT) and arabitol, which are indicators of biogenic aerosols. The backward trajectory analysis indicated that these aerosols originated from surrounding mountainous areas and peripheral regions. Furthermore, the potential biomass burning pollution in the identified traffic-related pollution source was identified through the presence of levoglucosan and succinic acid. The study highlighted the improved performance of the PMF model with the inclusion of organic components, as it allowed the identification of the biogenic source, which was not previously observed in field studies. The findings provide a scientific basis for future considerations and regulations regarding pollution in Taipei.