地面附近光化學污染所產生的臭氧問題為台灣地區極待解決的問題，而三維化學 傳輸模式為空氣品質規劃與管理的重要工具，此類模式計算費時，所以過去無法用於長期模擬。大都只選擇短期高濃度的事件日來做模擬，常以改變參數或邊界條件的方式來修改模擬結果，以期待得到較佳的模擬結果，但是此種模擬方式卻不一定能成功模擬其他狀況，因此本研究採用光化模式進行一個月連續之長期模擬，以涵蓋不同的天氣型態，且不加任何修正，以評估模式可達到的準確性。 模擬結果顯示小時臭氧濃度非配對峰值常化偏差介於±0.71，配對值之常化偏差介於±0.67，配對值之絕對誤差小於0.73，結果表現比美國EPA建議結果略差；八小時平均臭氧濃度非配對峰值常化偏差介於±0.73，配對值之常化偏差介於±0.54，配對值之絕對誤差小於0.61。模擬期間污染物排放的不確定性，可能造成影響。有一兩天計算結果極不理想，如能就其原因加以分析，還可改善模式的表現。模擬結果雖稱合理，但由於模式非常複雜，還是有許多改善的空間。 The ground-level ozone produced by photochemical air pollution is a series environmental problem in Taiwan. Three dimensional chemical transport model (3-D CTM) is an important tool for air quality planning and management. The computation of CTM is a time-consuming task; their applications to long-term simulations were prohibited in the past. The short periods with high concentration, the so-called “episode”, were selected for simulations. Sometimes, the models were well tuned by adjusting parameters or boundary conditions so as to obtain better results. However, the performance may be poor while these models were applied to other situations. In this research, we use a 3D CTM to carry out one-month continuous simulation in order to determine the performance the model under various meteorological conditions. The simulation use same parameters for all period. The performances of the calculated 1-h concentrations are: the unpaired-in-time peak prediction accuracy ±0.71; the normalized bias ±0.67; the gross error, <0.73. This is somewhat worse than the performance criteria suggested by USEPA. The performances of the calculated 8-h concentrations are: the unpaired-in-time peak prediction accuracy ±0.73; the normalized bias ±0.54; the gross error, <0.61. The uncertainties in the emission database can generate some errors on the predictions. The performance of this model can be improved if the reasons that cause some extraordinarily flawed results can be identified. Since this system is so complex, there are numerous opportunities for improvements in the future.