以汙染源解析結果探討細懸浮微粒之數量濃度對於新莊地區學童肺功能影響

Abstract

近幾年來，許多流行病學研究以迴歸建模探討細懸浮微粒對於健康的危害效應，並顯示大氣中的細懸浮微粒(PM2.5)對於健康有不良影響。細懸浮微粒來源複雜，即便了解細懸浮微粒對於健康的危害，仍須進一步釐清造成危害的細懸浮微粒來源，並進行控管。因此本研究結合汙染源解析之結果，與國中小學童肺功能相關之測量數值，探討特定汙染源對於肺功能之健康危害。 本研究以臺灣新莊地區2008年2月至2008年6月超級測站細懸浮微粒各粒徑數量濃度資料，利用正矩陣因子分析法，進行汙染源的推估與解析。粒徑0.01 μm至2.5 μm小時數量濃度資料，由電移動度分析儀(SMPS)以及光學儀器(PMS)測量不同粒徑範圍，並將兩台儀器資料結合後得本研究之分析資料。本研究以正矩陣因子分析法解出不同汙染源的指紋，搭配各汙染源在不同時間下之貢獻，並參考各汙染源與細懸浮微粒上物種濃度之相關性，解析出不同汙染源可能的成因以及貢獻度。 此研究中共解出5個汙染源：地區型混和汙染源、新生(fresh)汽油引擎排放源、新生(fresh)柴油引擎排放源、老化(aged)交通汙染源、傳輸型混和汙染源。地區混和型汙染源及傳輸型混和汙染源，對超級測站地區總數量濃度貢獻分別為4.20%及13.89%，日夜變化無特別趨勢，且與二次反應物硝酸鹽及硫酸鹽有中度至高度相關，兩汙染源之差異在於風向條件概率函數(CPF)顯示傳輸型混和汙染源主要來自於西方。其餘三個汙染源日夜變化趨勢皆與台灣地區交通趨勢相似，新生汽油引擎排放主要反映台灣地區汽機車使用模式，而新生柴油引擎汙染源則反映新莊地區大型車輛(如公車)的尖峰與離峰時段差異。交通相關之汙染源對於新莊地區總貢獻約占80%以上，顯示交通汙染源於當地空氣品質之重要性。 汙染源解析各汙染源每小時貢獻值，計算其日平均後，以混合效果模式(Mixed-effects model)分析單一與多污染源之汙染源，對於當地五所國中小學2008年2月至6月肺功能數值FVC、FEV1、FEF25%、FEF50%、FEF75%、FEF25-75%之影響。地區型汙染源與新生汽油引擎排放源為地區型汙染源，因此不列入肺功能資料分析。 健康資料分析結果發現，傳輸型混和汙染源在1天延遲時會造成FVC數值下降0.00008 (升)，在2天延遲時會造成FEF25%及FEF50%各下降0.00034 (升/秒)以及0.00023 (升/秒)。除了與先前研究結果相呼應，更找出細懸浮微粒數量濃度對於其餘肺功能的不良影響。當考慮到多污染物共同效應時，傳輸型混和汙染源在1天及2天延遲時會造成FVC數值下降0.00013 (升)及 0.00016 (升)，並造成FEV1數值下降0.00010 (升)及0.00016 (升)。 本研究突破傳統對於細懸浮微粒健康效應研究之限制，除了區分特定粒徑之來源，更以汙染源解析結果取代細懸浮微粒總濃度分析健康效應。此研究方式可探討特定汙染源對於肺功能之危害程度，亦可利用汙染源之粒徑分布資訊，間接推估不同粒徑對於肺功能危害之可能效應。

Epidemiological studies have shown that particulate matter (PM) was associated with adverse health, especially for those with diameter ≤ 2.5μm. Previous studies have utilized the total concentration of PM2.5 to quantify the health effect of fine particles by using the regression model. The sources of PM2.5 are complicated. Although understanding the health effect of fine particle, it still needs to clarify the source of particles. Therefore, this study combined both results of source apportionment and lung function measurements of schoolchildren to investigate the relation between sources of PM2.5 and lung function indices. Number concentration of size-segregated PM in Xinzhuang area in Taiwan from 2008/06 to 2008/08 was utilized in this study. Hourly number concentration data measured by SMPS and PMS of each size bin from 0.01 μm to 2.5 μm were analyzed by positive matrix factorization (PMF). The size distribution of source profile and time series trend of each source from PMF output, as well as the correlation between source contribution and PM species were utilized to identify sources. In this study, five sources were identified and and they were local mix (Mix 1), fresh gasoline emission (Local gasoline), fresh diesel emission (Fresh diesel), aged vehicle emission (Aged vehicle) and transported mix (Mix 2) sources. Local mix and transported mix contributed 4.20% and 13.89% of total number concentration of PM2.5. Diurnal pattern of two mix sources did not show apparent time trend as traffic. Contribution of mix sources had moderate to high correlation with secondary species. The conditional probability function (CPF) showed that the direction of transported sources were mainly from west. Other three sources were related to traffic because their diurnal patterns were similar with the traffic trend in Taiwan. Gasoline emission reflected the usage of motor vehicles; diesel emission reflected the rush hour and off-peak times of large cars such as buses. Traffic related sources were estimated to contribute more than 80% to the PM2.5 number concentration in Xinzhuang District. Hourly source contributions were calculated into daily mean. Mixed-effects model was used to quantify the effects of single or multiple sources to lung function FVC, FEV1, FEF25%, FEF50%, FEF75% and FEF25-75% of schoolchildren in Xinzhuang District. Local mix and fresh gasoline emissions were the local sources which were not included in the mixed-effects model analysis. Results showed that increase of transported mix with 1-day-lag was significantly associated with 0.00008 (L) decreases of FVC; increase of transported mix with 2-day-lag was also associated with 0.00034 (L/s) decreases of FEF25% and 0.00023 (L/s) decreases of FEF50% in single-pollutant model. Considering the co-pollutants effect, increases of transported mix with 1- and 2-day-lag was significantly associated with the 0.00013 (L), 0.00016 (L) decreases of FVC; increases of transported mix with 1- and 2-day-lag was significantly associated with the 0.00010 (L) and 0.00016 (L) decreases of FEV1. This study overcame some limitation of traditional health effect study. Using source apportionment results in health data analysis can assist in quantifying the effect of different sources to lung function, and understanding the lung function effect of particulate matter with different sizes.