A Study on Characteristic Species of Collinear Sources by Using PMF as A Receptor Model

本研究主要探討共線性污染源中特有物種如何尋找以及對受體模式 PMF 之 重要性，特有物種為共線性污染源間造成差異的物種，該物種的主要貢獻來源為 一個或兩個污染源，所以在樣本中具有高分散性的特色。先利用四個河川揚塵的 污染源指紋資料 以自製樣本公式製作模擬樣本，並透過變異係數 (Coefficient of， Variation)來篩選特有物種，之後透過保留與刪除特有物種，來了解其重要性，最 後再利用四個農廢燃燒污染源、五個鋼鐵業污染源與混合污染源，驗證特有物種 的存在對共線性污染源模擬之重要性，以及特有物種篩選方法的適用性。 研究結果顯示，在四個河川揚塵的污染源中全部分析物種與只保留特有物種 之模擬結果一致，模擬因子皆可準確的比對到實際污染源，且總貢獻量的誤差約 在 4%之內。如果分析的物種缺乏某一污染源獨有的特有物種時，模擬結果會有 一個因子的總質量為零，必須降低分析因子數目，但此作法會使其他污染源之貢 獻量高估。而如果缺乏某一污染源之特有物種的情況下，則該污染源的總貢獻量 誤差約為 30%，最後，在解析共線性污染源的情況下，若某污染源之特有物種濃 度很低，則需進行敏感度分析，以了解物種之影響性。而再以四個農廢燃燒污染 源、五個電力業污染源與混合污染源加以驗證，發現全部分析物種與保留各污染 源部分的特有物種之模擬結果一致，模擬因子都能比對到實際污染源，且模擬貢 獻量與實際貢獻量並無差異，而總貢獻量的誤差不管高估或低估約在 4%之內； 而混合共線性與非共線性污染源的樣本，則無法以變異係數篩選出各污染源之特 有物種。This study mainly investigates the importance of the characteristic species of collinear sources with a receptor model of PMF and how to find the characteristic species. The characteristic species is caused by differences between the species of collinear sources, and the main contributor of the characteristic species is one or two sources, so it has a high dispersion in the sample. At first, we use the pollution fingerprint of sources of river dust to make homemade samples and that screened the characteristic species by coefficient of variation. Then the characteristic species have been retention and deletion to understand its importance. Finally, we use four sources of biomass burning, five sources of electric power industry and mixed pollution sources to verify the importance of the characteristic species of the simulation of collinear sources, and the applicability of the characteristic species screening method. The results showed the analyzed of simulation for all the species and only the characteristic species were not effect signification, simulation factor could be accurately compared to the actual sources of pollution, and the total contribution of the error would be about 4%. If the analysis of species lacked alone the characteristic species of collinear sources, the results would be a factor of the total mass of zero, the number of factor analysis must be reduced, but this approach would overestimate the contribution of other sources. If a case lacked the characteristic species of collinear sources, the total contribution of the error would be about 30%. Finally, if the concentration of certain sources of the characteristic species is very low, you needed to perform a sensitivity analysis and understand the impact of the species. Then we use four sources of biomass burning and five sources of electric power industry to verify the importance of the characteristic species of the simulation of collinear sources. The results would be the same with the sources of four rivers. And the mixed pollution sources couldn''t use the method of coefficient of variation to find the characteristic species.致謝..............I 摘要 ..............II Abstract..............III 目錄 ..............IV 表目錄 ..............VI 圖目錄 ..............VII 第一章 前言..............1 1.1 研究緣起..............1 1.2 研究目的..............2 第二章 文獻回顧..............3 2.1 大氣懸浮微粒描述..............3 2.2 受體模式相關理論..............5 2.2.1 化學質量平衡..............5 2.2.2 矩陣因子法..............7 2.3 受體模式之相關研究與應用..............13 第三章 研究方法..............13 3.1 研究架構..............13 3.2 汙染源指標物種之判定..............21 3.3 自製模擬樣本之建置..............22 3.4 共線性汙染源之特有物種篩選..............23 3.5 受體模式PMF之應用..............24 3.5.1 輸入檔之建置..............24 3.5.2 輸出檔之說明..............26 3.5.3 Fpeak..............28 第四章 結果與討論..............31 4.1 四組無共線性之汙染源..............31 4.2四個河川揚塵汙染源..............36 4.3 保留河川揚塵汙染源之特有物種..............41 4.4 刪除某個河川揚塵汙染源之特有物種..............47 4.4.1 刪除大安溪之特有物種..............49 4.4.2 刪除倒西之特有物種..............54 4.5低濃度之特有物種討論..............59 4.6其他汙染元組合之探討..............65 第五章 結論與建議..............94 5.1 結論..............94 5.2 建議..............95 參考文獻..............96 附錄A..............10