台灣地區的污染物問題相當嚴重，其中包含了本地排放的污染物，來自東亞的長程傳輸污染物，以及近年來受到很多討論的東南亞生質燃燒污染物。許多前人研究指出，污染物跟氣象的因子是密切相關的，因此本研究中使用群集分析的方法，針對亞洲地區的綜觀天氣進行分類，並進一步探討在何種天氣類型下，容易造成台灣高污染事件的發生。分類的期間選定2007到2010年的春季期間，使用Weather Research and Forecasting (WRF)模式的模擬結果，作為綜觀天氣型態的分類依據，其空間解析度為27公里。 除了針對天氣進行分類，我們同時也選取了台灣的空氣品質測站，並根據分類的結果探討不同天氣型態與污染物之間的關係。分類的結果顯示，萬里站的低濃度都是發生在與西伯利亞高壓有關的群集(C2和C5)，而綜觀天氣系統較弱的情形下(C3)，風速較小，同時濃度也比較高。斗六站的部份，CO高濃度則同時出現在弱綜觀與高壓較強的分類(C1 和C3)，顯示斗六的污染物不只是本地排放主導，同時也受到東亞地區長程傳輸污染物的影響。另外CO濃度的分佈情形，在鹿林山測站與恆春測站很類似，CO的濃度在高壓的天氣型態都是比較高的，濃度最高的群集同時也是高壓最強的群集(C5)。 為了了解東南亞生質燃燒的傳輸情形，我們根據鹿林山的CO濃度的高值，挑選2007年3月15到23日期間做為個案研究，嘗試了解東南亞生質燃燒的傳輸情形。在模擬期間，台灣受到高壓的影響，影響時間從17到22日，而鹿林山的CO高值則出現在3月16到18日。在此個案中，恆春站在3月16與19日也分別觀測到CO的高值。從軌跡分析可以發現，鹿林山CO的高值大部分都是來自於中南半島，顯示此個案的高值是受到東南亞生質燃燒之影響。分析結果顯示，當東南亞處於一個較遠離高壓影響的環境下，由於熱力的作用，生質燃燒的污染物較易被往上抬升，污染物被往上抬升後，便順著西風往東傳輸並影響台灣。我們同時也發現，在東北風的影響下，熱力作用所造成的抬升機制較不容易形成，生質燃燒污染物對其下風地區的影響也較小。此外恆春站的軌跡分析則指出，在此期間內，恆春站的CO高值是來自於東亞、菲律賓以及東南亞。Taiwan has been suffered from serious air pollution problem including the locally produced air pollutants, long-range transported pollutants from Eastern China, and recently the biomass burning from Indochina also draws attention. The air pollution in Taiwan is strongly associated with the meteorological conditions. In order to identify the synoptic weather patterns that are prone to cause high pollutions in Taiwan, a clustering approach by means of the two–stage method is applied using the 27-km Weather Research and Forecasting (WRF) meteorological model simulation result. The study period targets the springtime season from 2007 to 2010. The observed CO concentrations at several locations in Taiwan are used to identify the relationship between the meteorology and air pollution. The weather classification results showed that the low CO concentration appears mostly in the cluster that is associated with continental high-pressure systems at Wanli. And the high CO concentration is associated with the weaker wind speed in which the synoptic weather conditions are also weak. At Douliu station, the high CO concentration can appear in both weak synoptic weather pattern and strong high-pressure system. The distributions of CO concentration behave similarly at both LABS and Hengchun where the high CO concentrations are associated with the continental high-pressure systems. Furthermore, a 9-km resolution WRF modeling with the updated land use data is performed from Mar 15 to 23 in 2007 to understand the transport mechanisms of biomass burning species from Indochina. During the simulation period, the high CO concentration was observed from Mar 16 to 18 at Lulin Mountain site. The high CO concentration also appeared on Mar 16 and 19 at Hengchun station. The analyses of the backward trajectories at LABS were mostly from Southeast Asia, which indicated the high CO concentration observed at LABS was attributed to the biomass burning activities from Indochina. The biomass burning pollutants were uplifted to upper layer and transport to Taiwan along the westerly wind, and the weak synoptic weather pattern is favorable for the transportation of pollutants to Taiwan. Finally, trajectories analysis was also applied to Hengchun, where the high CO concentration is transported from the eastern Asia, Philippines and Indochina. For the non-event cases we found the strong prevailing northeasterly monsoonal flow would hinder the development of the thermal low in Indochina.
