以衰減校正之正矩陣因子分析法解析台灣中部揮發性有機物來源

Abstract

揮發性有機化合物（Volatile Organic compound, VOC）具有高度反應性，且廣泛排放於各種人為與自然來源，在大氣化學過程中扮演重要角色。VOC亦為近地面臭氧（O₃）形成的重要前驅物，其可與氮氧化物（Nitrogen Oxides, NOₓ）進行光化學反應而促進臭氧生成。因此，釐清VOCs的污染特徵及其來源，對於空氣品質管理具有重要意義。 本研究探討光化學損失對臺灣中部工業區VOC來源解析結果之影響。研究於雲林縣三處特殊性工業區光化學監測站蒐集環境VOC資料，包括兩處鄰近大型石化工業區的上風處測站，以及一處距離工業區較遠的下風處測站。2024年全年每小時量測共54種VOC，並選取日間07:00 – 18:00之觀測資料進行分析，以涵蓋光化學反應較為活躍的時段。 本研究利用正矩陣因子分析法（Positive Matrix Factorization, PMF）解析上風處與下風處之VOC污染來源。傳統PMF分析出五類主要污染來源，包括石化排放、液化石油氣、老化氣團、生物源排放及溶劑使用。為考量VOC於傳輸過程中所發生之光化學降解，本研究進一步建立衰減校正之正矩陣因子分析法（decay-adjusted PMF），利用上風處測站資料建立經光化學老化之石化排放來源剖面，並將其作為具化學意義之約束條件，透過多線性引擎（Multilinear Engine, ME-2）應用於下風處測站資料。結果顯示，decay-adjusted PMF進一步解析出一個以高反應性烯類為特徵物種的當地污染源，顯示在傳統PMF分析中，當地排放與傳輸而來的石化排放可能難以有效區分。 在下風處測站，decay-adjusted PMF重新分配了各污染來源的貢獻，石化排放占總VOC濃度的比例由17% 降至11%，並另外解析出18% 的當地污染源。各污染來源臭氧生成潛勢進一步顯示VOC之化學反應性對臭氧生成貢獻的重要性：雖然老化氣團約占總VOC濃度的40%，其對總臭氧生成潛勢的貢獻僅約9%。另一方面，考量光化學損失後，傳輸性石化相關的排放對臭氧生成潛勢的貢獻由37% 降至21%。此外，透過robust regression進一步分解老化氣團來源後，將部分老化氣團的貢獻重新歸屬至其他污染來源。儘管老化氣團分解後各污染來源的VOC貢獻有所重新分配，各來源之整體臭氧生成潛勢分布大致仍維持不變，僅石化來源的臭氧生成潛勢略為增加。 綜合而言，本研究結果顯示，傳統PMF可能無法有效區分下風處工業環境中傳輸性與當地反應性VOC排放。將光化學損失納入受體模式分析後，可改善此類污染來源之分離，並進一步分解老化氣團因子，提供傳輸性VOC來源組成之更深入資訊。因此，decay-adjusted PMF架構可作為改善VOC來源解析之有效方法，並可為工業區VOC及臭氧污染之管制策略提供更適切的科學依據。

Volatile organic compounds (VOCs) play a critical role in atmospheric chemistry due to their high reactivity and widespread emissions from anthropogenic and natural sources. They are also important precursors of ground-level ozone (O₃) formation through photochemical reactions involving nitrogen oxides (NOₓ), making the characterization and source apportionment of VOCs essential for air quality management. This study investigates the influence of photochemical loss on VOC source apportionment in an industrial region of central Taiwan. Ambient VOC data were collected from three Photochemical Assessment Monitoring Stations (PAMS) in Yunlin County, including two upwind stations near a major petrochemical complex, and one downwind station. A total of 54 VOC species were measured hourly throughout 2024, with daytime observations (07:00–18:00) selected to capture periods of active photochemical processing. Positive Matrix Factorization (PMF) was applied to resolve VOC sources in the upwind and downwind datasets. The conventional PMF solution identified five major source categories, including petrochemical emissions, liquefied petroleum gas (LPG), aged air mass (AAM), biogenic emissions, and solvent usage. To account for the photochemical loss of reactive VOCs during transport, a decay-adjusted PMF approach was developed by incorporating petrochemical decay profiles derived from the upwind dataset into the Multilinear Engine (ME-2) as a chemically informed constraint for the downwind dataset. The decay-adjusted PMF resolved an additional local source characterized by reactive alkenes, indicating that local emissions were not distinguishable from transported petrochemical emissions in the conventional PMF solution. The decay-adjusted PMF redistributed the source contributions at the downwind site, with transported petrochemical emissions decreasing from 17% to 11% of total VOC concentration and a newly resolved local source accounting for 18%. The source-specific ozone formation potential (OFP) further demonstrated the importance of photochemical reactivity: although AAM accounted for approximately 40% of the VOC concentration, it contributed only 9% of total OFP, whereas the contribution of transported petrochemical-related emissions to OFP decreased from 37% to 21% after photochemical decay was considered. In addition, robust regression separation attributed portions of the AAM contribution to other source categories. Despite the redistribution of VOC contributions after AAM decomposition, the overall OFP distribution among sources remained largely unchanged, with only a slight increase in the petrochemical contribution. Overall, these results demonstrate that conventional PMF may not adequately distinguish between transported and local reactive VOC emissions in downwind industrial environments. Incorporating photochemical decay into receptor modeling improved the separation of these sources. Subsequent separation of the AAM factor provided further insight into the source origins of transported VOCs. The decay-adjusted PMF framework therefore provides a useful approach for improving source attribution and supporting more effective VOC and ozone control strategies in industrial regions.