近年來許多流行病學研究顯示,不論是短期或長期暴露在交通排放相關之空氣污染物-PM10、PM2.5 和 PMcoarse,皆會使人體健康產生不良的影響,尤其以呼吸與心肺功能傷害更為嚴重。為了評估大台北都會區懸浮微粒濃度空間分佈,本研究利用五種方法預測個案所暴露之污染物濃度,並比較實際值與預測值兩者之間的差異。 本研究主要從患有心血管疾病之世代研究中選擇24位研究對象。於研究對象之住家戶外陽臺放置Harvard Impactors。家戶採樣期間為2008年7月~12月、2009年4月~12月。所使用之土地利用模式(LUR)是以土地利用、道路長度和人口數為預測變項。而克利金法(kriging)則是利用監測站所處的位置汙染物濃度不同,利用空間統計計算出個案暴露濃度。 以建立模式結果而言,PM10土地利用模式之調整R2model值為0.58,PM2.5之調整R2model值為0.94,PMcoarse之調整R2model值為0.65。在五種預測方法之實際值與模擬值的R2Pred_H比較,PM10介於0.69-0.78,PM2.5介於0.58-0.77,而PMcoarse則是界於之0-0.19。在PM10, PM2.5方面,結果顯示克利金法的預測能力優於土地利用模式。而在PMcoarse方面無論以哪一種模式預測,其預測值與實際值的R2Pred_H值都顯著地低於PM10, PM2.5。 因為PM10 and PM2.5的R2Pred_H值皆高於0.5,以評估之地區具有密集空氣監測網絡的前提下,可以使用距離個案最近之監測站懸浮微粒濃度代替。然而,PMcoarse的R2Pred_H值皆低於0.5,因此未來研究須著重在如何提升PMcoarse的R2Pred_H值。
BACKGROUND: In recent years, many epidemiologic studies have shown that both short- and long-term exposures to traffic-related air pollutants such as PM10 (with aerodynamic diameters < 10 μm), PM2.5 (with aerodynamic diameters < 2.5 μm) and PMcoarse (with aerodynamic diameters between 2.5-10 μm) are associated with adverse effects on cardiovascular and respiratory systems. To estimate the spatial distributions of PM concentrations in Taipei metropolis, we applied five methods to predict subjects’ exposures and compared their R2 values between sampling and predicted data. METHODS: We chose twenty-four subjects from a cohort study and put Harvard Impactors at subjects’ home balconies to collect 24-hrs PM10 and PM2.5 home outdoor micro-environmental samples during July-December, 2008. A second campaign of PM collections was conducted at the same locations. However, we measure PM10 and PM2.5 from participants home outdoor micro-environmental after six months later. LUR models were constructed using land use, road-length, and population density as predictor variables. The kriging methods use coordinates of AQM stations combining with PM monitoring results to calculate PM exposure. RESULTS: The adjusted R2model of the final models was 0.58 for PM10, 0.94 for PM2.5, 0.65 for PMcoarse. For the validation dataset, the five methods gave R2Pred_H of 0.69-0.78 for PM10; 0.58-0.77 for PM2.5, and 0-0.19 for PMcoarse. The performance of the two kriging methods for PM10 and PM2.5 were better than those for PMcoarse. The R2Pred_H value of coarse particle was much lower than PM10 and PM2.5. CONCLUSIONS: The R2Pred_H value of PM10 and PM2.5 were higher than 0.5, so using the nearest AQM station might be satisfied for areas with dense air monitoring network. Nevertheless, the R2Pred_H values between PMcoarse were below 0.5. Future studies should focus on increasing the model performance for PMcoarse.
為了持續優化網站功能與使用者體驗,本網站將Cookies分析技術用於網站營運、分析和個人化服務之目的。
若您繼續瀏覽本網站,即表示您同意本網站使用Cookies。