透過您的圖書館登入
IP:3.81.221.121
  • 學位論文

台北都會區大氣細懸浮微粒成份與來源之空間變異性模式分析

Modeling spatial variations of ambient PM2.5, compositions, and sources in Taipei metropolis

指導教授 : 吳章甫

摘要


背景: 細懸浮微粒(氣動粒徑小於等於2.5 微米的微粒,PM2.5)是重要的汙染物,對於人體健康有負面的影響。對於PM2.5和其成分而言,樓層高度的差異可能會對土地利用迴歸模式預測出來的汙染物濃度導致一些誤差。在評估汙染源貢獻與人體健康的相關性以及汙染源減量時,樓層高度也可能會導致不準確的評估。然而,在過去研究當中,樓層高度的影響仍未被廣泛地探討。此外受體模式已被應用來進行空氣污染之成因分析,但汙染源貢獻量的垂直分布仍未被廣泛地評估。另受體模式的解析結果有時包含混和型來源,此結果需要被其他模式進一步論證。 目的: 在第一篇研究議題中,建立細懸浮微粒與其成份之土地利用迴歸模式以評估居民的個人暴露,進一步地探討垂直分布特性之量測數據對於預測濃度之影響,以提升個人暴露預測之準確性。在第二篇研究議題中,應用受體模式來評估汙染源貢獻量的空間變異性,以及結合受體模式之結果與土地利用模式來改善受體模式之定性結果和釐清各污染源對於貢獻量推估值之影響。 方法:在30個家戶採樣點與1個參照測點利用哈佛衝擊器來量測細懸浮微粒濃度,並透過能量分散式-X 射線螢光分析儀分析其成份之濃度,此外結合具有水平與垂直分布特性的量測資料來建立土地利用迴歸模式,以預測不同樓層的細懸浮微粒濃度以及改善PM2.5之源解析結果。因素分析則被應用來定性與定量汙染物的來源及其貢獻量。 結果: 在第一篇研究議題中,細懸浮微粒與其成份之土地利用迴歸模式的R2介於0.46到0.80,家戶附近的交通資訊與在家戶半徑300到5000公尺之工業區面積是細懸浮微粒和其成份的主要預測因子。此外未考慮垂直分布上的差異時,細懸浮微粒、矽與鐵的預測濃度推估誤差介在5.6%到11.0%。在第二篇研究中,結果顯示主要的細懸浮微粒來源是因子3 (51.7 %),因子1 (31.8 %)與因子2 (12.9 %)。交通尾氣排放對於因子3是主要的來源,因子1為一混和型汙染源(非尾氣的交通排放、工業活動和二次氣膠)。對於因子2來說,在平常日時非尾氣的交通排放以及工業活動是主要來源,在宗教節日時生質燃燒(燃香和燃燒紙錢)是主要來源。另外土地利用迴歸模式可以改善受體模式對於汙染源之定性結果,並進一步地釐清各污染源對於混和型汙染源之影響。交通尾氣排放貢獻則具有垂直分布的些微差異。 結論: 在流行病學研究中評估個人暴露時,應該要考量垂直高度的影響(即離地表之高度)。為了有效率地進行細懸浮微粒減量(特別是針對那些居住於高樓層的居民),交通尾氣排放的垂直分布之差異是需要被關注的。

並列摘要


Background: Fine particulate matter (Particles with aerodynamic diameters of less than 2.5 μm, PM2.5), an important pollutant, has been investigated for their adverse influences on human health. For PM2.5 and compositions, the difference of building floor possibly causes a bias on predicted concentration of pollutant in land use regression (LUR) model. This influence may also lead to an inaccurate estimation on the association of PM2.5 source contributions and human health, as well as the effectiveness of PM2.5 reduction. This effect is not investigated widely in recent studies. In addition, although receptor models have been applied for source apportionment of air pollution, the vertical distributions of PM2.5 source contributions were still not evaluated extensively. Furthermore, the receptor models sometimes retrieved mixed sources. This needs to be further clarified by another statistical analysis. Objectives: In the first topic, LUR models of PM2.5 and compositions were developed for exposure assessment of residents and the influence of vertical distribution measurements on predicted concentrations was further evaluated in order to improve the estimation of individual exposure. In the second topic, receptor models were applied to estimate the spatial variations of PM2.5 source contributions. The receptor modeling results were combined with LUR models to improve the source identification from receptor model and to clarify the effect of each source to the estimated contributions. Methods: PM2.5 was measured at thirty residence sites and one reference site by Harvard impactors, and analyzed for its compositions by using energy-dispersive X-ray fluorescence spectrometry. LUR models using horizontal and vertical distribution measurements were developed to predict PM2.5 and composition concentrations at various building floors and improve the source resolution of PM2.5. Factor analysis was applied to identify the sources of ambient PM2.5 and apportion their contributions. Results: The results in the first topic showed that LUR models’ R2 ranged from 0.46 to 0.80. Traffic information in the vicinity of residential houses and industrial areas within large radii of 300m to 5000m were observed to be major predictors of PM2.5 and composition. The estimated bias of the predicted concentrations (i.e., PM2.5, Si, and Fe) ranged from 5.6% to 11.0%, if vertical contrasts were not considered. In the second topic, the major source to PM2.5 was Factor 3 (51.7 %), followed by Factor 1 (31.8 %) and Factor 2 (12.9 %). Tailpipe traffic emission is the main source in Factor 3. Factor 1 is a mixed source of non-tailpipe traffic emissions, industrial activities and secondary aerosols. For Factor 2, non-tailpipe traffic emissions combining industrial processing and biomass burning (incense or joss paper burning) are dominant on normal and scared days, respectively. LUR model can improve the identifications of the factors from receptor models, and further clarify the effect of individual source in mixed sources. Tailpipe traffic emissions had a slightly vertical contrast on the estimated contribution. Conclusions: The influence of vertical height from the ground should be considered in the assessment of individual exposure for future epidemiological studies. The vertical variation of tailpipe traffic emissions needed to be taken into consideration to reduce PM2.5 concentration in air, especially for the residents living at high building floor.

參考文獻


Lu, H.Y., Lin, S.L., Mwangi, J. K., Wang, L.C. and Lin, H.Y., 2016. Characteristics and Source Apportionment of Atmospheric PM2. 5 at a Coastal City in Southern Taiwan. Aerosol and Air Quality Research 16, 1022-1034.
Kao M. C. and Lung S. C., 2000. Personal Particulate Exposures in Buddhist Temples. Chinese Journal of Public Health, 19, 138-143.
Agrawal, H., Welch, W. A., Miller, J. W. and Cocker, D. R., 2008. Emission measurements from a crude oil tanker at sea. Environmental Science & Technology 42, 7098-7103.
Aguilera, I., Eeftens, M., Meier, R., Ducret-Stich, R. E., Schindler, C., Ineichen, A., Phuleria, H. C., Probst-Hensch, N., Tsai, M.Y. and Künzli, N., 2015. Land use regression models for crustal and traffic-related PM 2.5 constituents in four areas of the SAPALDIA study. Environmental Research 140, 377-384.
Andreae, M. O. and Merlet, P., 2001. Emission of trace gases and aerosols from biomass burning. Global biogeochemical cycles 15, 955-966.

延伸閱讀