傳統上，分析開徑式傅立葉轉換紅外光光譜儀（Open-Path Fourier Transform Infrared, OPFTIR）監測資料時，多針對所量測有害或異味物種逐一分析其可能污染來源方向。然而當污染源眾多且排放強度亦不同時，此法成效有限，特別是無法推估各污染源貢獻量。因此，本研究探討以絕對主成分分析法（Absolute Principal Component Analysis, APCA）搭配氣象資料，用於辨明污染物主要污染源與其貢獻量之可行性。本研究以架設於半導體廠附近之開徑式傅立葉轉換紅外光光譜儀多道監測測線資料為例，使用絕對主成分分析判斷主要異味物質氨其來源，並進一步區分半導體廠與鄰近光電廠及環境背景之貢獻量。本研究結果顯示在監測期間，氨主要排放源為半導體廠，以高於最小可解析之氨濃度值而言，在第一道（Path 1）與第二道監測線（Path 2）中半導體廠貢獻比例分別為71%與76%，且Path 2中來自於半導體廠的異味率（odor rate）高達23.78%，其他來源則無超過異味閾值（odor threshold）之樣本，顯示若可有效控制半導體廠所排放之異味物質，異味問題應可有效改善。本研究最後結論，若可有效運用本研究方法，在當異味物質有眾多污染源時，可辨別不同污染源的貢獻與其排放特性，有助於異味改善策略的擬定。

Most studies involving Open-Path Fourier Transform Infrared (OP-FTIR) measurements usually analyze the monitoring data on a compound-by-compound basis to investigate the source directions of toxic or odorous air pollutants. When there are multiple sources, this approach may not perform well and could not give estimates on source contributions. In this study, we evaluated the feasibility of employing Absolute Principal Component Analysis (APCA) along with meteorological data to identify major emission sources and quantify the associated contributions. Using OP-FTIR data collected at a semiconductor plant during an odor episode as an example, the receptor model of APCA is implemented to identify the sources of the main odorous pollutant, ammonia, and to further distinguish the contributions from the semiconductor plant, the nearby optoelectronic plant, and the background. The results indicated that ammonia was mainly emitted from the semiconductor plant during the monitoring period. For ammonia data above the minimum reported concentration values, the contribution of the semiconductor plant in the first (Path 1) and second monitoring locations (Path 2) was 71% and 76%, respectively, and the odor rate of Path 2 was 23.78%. On the other hand, none of the source-specific ammonia concentrations from the other sources was above the odor threshold in Path 2. This indicated that if the odorous pollutant associated with the semiconductor plant itself were controlled properly, the odor problem could be solved. It is concluded that the APCA model could be applied effectively to identify the sources of pollutants and quantify their contributions even when multiple sources exist. The modeling results also facilitate designing effective control strategies to address air quality and odor problems.