利用改良版受體模式分析都市中細懸浮微粒之污染來源與垂直貢獻比例

Vertical Distribution and Source Apportionment of Urban Particulate Matter Using a Modified Receptor Model

李建霖(Chien-Lin Lee)

指導教授 ： 吳章甫

國立臺灣大學/公共衛生學院/環境與職業健康科學研究所/碩士(2020年)

https://doi.org/10.6342/NTU202000580

摘要

越來越多研究證實細懸浮微粒(fine particle matter, PM2.5)與各種慢性及急性疾病有相關，空氣污染成為民眾極度關心的議題，PM2.5的成分以及污染來源也開始備受討論。PM2.5及其成分的濃度具有水平及垂直空間上的變異，過往較多針對水平空間變異的研究；較少研究探討污染物及污染源的垂直變異。本研究透過長時間的樓層採樣，量測PM2.5及其成分在不同季節及垂直高度下的濃度變化，並藉由受體模式分析台北都會區的主要污染源以及各污染源的貢獻比例、成份組成以及垂直變異。
採樣使用哈佛採樣器搭配鐵氟龍與石英濾紙採集24小時的樣本，以一週兩次的頻率在台北市辛亥路上某建築物的三樓、七樓、十一樓之陽台進行採樣，採樣時間為10個月(2018年6月-2019年3月)，觀察PM2.5與其成份在不同高度的濃度以探討台北都會區的空氣污染的垂直分布狀況。分析物種包含PM2.5濃度、16種無機元素、有機碳與元素碳濃度以及吸收係數。
本研究使用正矩陣因子法(Positive Matrix Factorization model, PMF)，並透過多重線性引擎(Multilinear Engine)建立改良版PMF源解析模式，使樓層資料解析時，能得出一致的污染源圖譜(source profile)，並假設不同的污染源在不同樓層貢獻之時間序列有相同趨勢，但濃度具一定比例關係。藉由模式模擬的方法，驗證改良版ME-2源解析模型能比美國環保署PMF軟體解出更精確的污染源解。另進一步探討將此模型應用在同時具水平變異與垂直變異的採樣點時，將不同棟的資料合併或獨立解析的結果差異，得出只要水平變異中污染源圖譜變異在一定比例以下，將資料合併解析會更適合。
在實地研究部分，於低樓層可以觀察到最高的PM2.5濃度(12.66 μg/m3)，其次依序為中樓層(12.61 μg/m3)、高樓層(12.38 μg/m3)。接著應用改良版源解析模型在實際樓層採樣的資料，結果發現貢獻比例最高的為二次氣膠污染物以及交通排放，其次分別為重油燃燒、工廠排放、土壤粉塵以及一個由生質燃燒與海洋飛沫組成的混合解。其中，交通排放的指標元素如Cu與有機碳、元素碳，以及模式所解之交通污染源皆觀察到在低樓層有較高濃度的顯著垂直變異關係，而其餘污染源相較沒有垂直上的變異，顯示交通污染可能是細懸浮微粒具有垂直變異的關鍵因素之一。

關鍵字

細懸浮微粒 ； 成份元素 ； 正矩陣因子模型 ； 垂直變異

並列摘要

Several studies have demonstrated that exposure to fine particles (PM2.5) could cause both acute and chronic adverse health effects. It is necessary to identify and quantify the contributions of PM2.5 sources. The concentration of air pollution shows considerable spatial variations. However, most studies only evaluated horizontal variability of PM2.5 and its elemental composition. The vertical variation of PM concentrations has not been widely studied.
In this study, vertical variations of PM2.5 were measured by sampling at three different floors (3F, 7F, 11F) at a building on a major road in Taipei metropolis twice a week for ten months. PM2.5 samples were collected to analyze mass concentrations, absorption coefficient and concentrations of 16 inorganic elements, organic carbon and elemental carbon.
Positive matrix factorization (PMF) model was used to quantify the contribution of the sources at the receptor site. The Multilinear Engine (ME) program, a flexible tool for solving PMF problems, was applied for receptor modeling. It was assumed that time series of resolved factor contributions at different floors should have similar trends while source profiles should be the same. The evaluation of the modified source apportionment model was first conducted by using simulation data. It was showed that the improved model performed better than the EPA PMF software. Furthermore, the average absolute error (AAE) was calculated to examine how profile variation affected the model performance.
In the field study, the highest average PM2.5 mass concentration was observed at the low-level floor (12.66 μg/m3), followed by the mid-level floor (12.61 μg/m3¬¬) and the high-level floor (12.38 μg/m3). With the modified source apportionment model, it was found that secondary aerosol and traffic related pollution were the primary sources in Taipei city, followed by oil combustion, industry emission, soil dust, and a mixed source composed of biomass burning and marine aerosol. Traffic related source and its tracer such as Cu, organic and elemental carbon showed significant difference between high-level sites and other two altitudes. However, apparent vertical variation was not observed for most components and sources. Therefore, traffic related source might be a key factor for vertical variation of PM2.5.

並列關鍵字

fine particulate matter ； elemental components ； positive matrix factorization ； vertical variation

參考文獻

Alastuey, A., et al. (2006). "Identification and chemical characterization of industrial particulate matter sources in southwest Spain." Journal of the Air & Waste Management Association 56(7): 993-1006.

Amato, F. and P. K. Hopke (2012). "Source apportionment of the ambient PM2.5 across St. Louis using constrained positive matrix factorization." Atmospheric Environment 46: 329-337.

Beelen, R., et al. (2014). "Effects of long-term exposure to air pollution on natural-cause mortality: an analysis of 22 European cohorts within the multicentre ESCAPE project." Lancet 383(9919): 785-795.

Begum, B. A., et al. (2005). "Investigation of sources of atmospheric aerosol at a hot spot area in Dhaka, Bangladesh." Journal of the Air & Waste Management Association 55(2): 227-240.

Belis, C. A., et al. (2013). "Critical review and meta-analysis of ambient particulate matter source apportionment using receptor models in Europe." Atmospheric Environment 69: 94-108.