| Abstract: | 近年來台灣社會蓬勃發展，人均所得不斷增加下，繁盛的工商活動卻造成了空氣汙染，嚴重地影響國人的健康。有鑑於此，我們以PM2.5濃度作為本研究的量化指標，並以十個環保署空氣品質監測站(大里、西屯、沙鹿、忠明、豐原、線西、埔里、三義、馬祖及永和站)於2015至2018年之間每小時的偵測數據來探討各種環境因素對PM_2.5濃度的影響。 本研究利用Catmull-Rom spline插值法和線性插值法來插補遺漏值，並進行單根檢定來確認數據的定態性質；其次將待分析的PM_2.5濃度區分成四個等級，並將各濃度等級以百分比表示；最後利用次序邏輯斯迴歸程式進行迴歸參數與勝算比的計算，並呈現出邊際效果。實證結果顯示： 化學因子：(1)在PM10濃度對PM2.5濃度的影響方面，所有測站都呈現出同向的關係。(2)在SO2濃度對PM2.5濃度的影響方面，除了埔里站外，其他測站都呈現出同向的關係。(3)在NO濃度對PM2.5濃度的影響方面，除了三義、馬祖、線西外，其他七測站都呈現出反向的關係。(4)在NO2濃度對PM2.5濃度的影響方面，除了大里、忠明及埔里站外，其它七測站都呈現出同向的關係。(5)在CO濃度對PM2.5濃度的影響方面，除了大里站外，其它九測站都呈現出同向的關係。(6)在O3對PM2.5濃度的影響方面，所有測站都呈現出同向的關係。 氣象因子：(1)在氣溫對PM2.5濃度的影響方面，除了沙鹿、豐原及三義站外，其它七測站之PM2.5濃度與溫度的關係都呈現U字型態。(2)在風速對PM2.5濃度的影響方面，除了大里、沙鹿、豐原、埔里及馬祖站外，其它五測站之PM2.5濃度與風速的關係都呈現U字型態。(3)在相對溼度對PM2.5濃度的影響方面，除了大里、西屯及三義站外，其它七測站都呈現出同向的關係。 季節性虛擬變數：大里、忠明、豐原、埔里、馬祖及永和站在東北季風期間之PM2.5濃度較高，而線西、西屯、沙鹿、三義站在東北季風期間之PM2.5濃度則較低。
In recent years, with vigorous economic development, the prosperous business activities have caused severe air pollution, which has seriously damages our health. In view of this, we use the hourly data from 2015 to 2018 of ten of the EPD air quality monitoring stations in Taiwan (Dali, Xitun, Shalu, Zhongming, Fengyuan, Xianxi, Puli, Sanyi, Matsu and Yonghe stations) to explore the impact of various environmental factors on PM2.5 concentration. We use the Catmull-Rom spline method and the linear interpolation method to interpolate missing values, and then perform unit root tests to check the stationarity of the data. Since the PM2.5 concentration was divided into four levels, we use the ordered logistic regression method to estimate the relationships between the PM2.5 concentration levels and the environmental factors. The empirical results are as follows：1. Chemical factors： (1) PM2.5 concentration level is positively related to PM10 concentration for all stations. (2) PM2.5 concentration level is positively related to SO2 for all stations except for Puli station. (3) Except Sanyi, Matsu, and Xianxi station, PM2.5 concentration level is negatively related to NO for the other seven stations. (4) Except Dali, Zhongming, and Puli station, PM2.5 concentration level is positively related to NO2 for the other seven stations. (5) PM2.5 concentration level is positively related to CO for all stations except for Dali station. (6) PM2.5 concentration level is positively related to O3 for all stations.2. Meteorological factors： (1) Except Shalu, Fengyuan, and Sanyi station, the relation between PM2.5 concentration and temperature for the other seven stations presents a U shape pattern. (2) Except Dali, Shalu, Fengyuan, Puli, and Matsu station, the relation between PM2.5 concentration and wind speed for the other five stations presents a U shape pattern. (3) Except Dali, Xitun, and Sanyi station, PM2.5 concentration level is positively related to relative humidity for the other seven stations.3. PM2.5 concentration level is higher during the northeast monsoon for Dali, Zhongming, Fengyuan, Puli, Matsu, and Yonghe station, while it is lower during the northeast monsoon for Xianxi, Xitun, Shalu, and Sanyi station. |
