Abstract: | 氣象狀況是空氣品質模式重要的輸入資料,台灣為一個山岳陡峭的海島,台灣的氣流受到地形和地面參數(如海水表面溫度、地面粗糙度等)影響極大,但許多氣象模式所需輸入的資料不容易取得準確的值,因此值得探討改變氣象模式輸入資料和演算法,對空氣品質模擬結果的影響。 本研究使用的氣象模式為RAMS而空氣品質模式則使用CAMx,氣象模擬改變四維資料逼近強度、地形處理、海水表面溫度等進行數次計算,求出這些改變對風場、溫度和濕度分佈的影響,然後這些不同的氣象模擬結果再輸入空氣品質模式進行計算,並將模擬的結果和實測資料比較,以評估其表現。 RAMS四維資料同化採用牛頓鬆弛法,進行分析場逼近時逼近項大小由時間尺度控制,過強的逼近必須避免,否則海陸和山谷風環流及地形效應會被削弱。至於地形處理,採用所謂的倒信封地形演算法所求出的結果比用傳統平均高程所求出的結果好。此一研究還用兩種海水面溫度資料,一個是氣候性的分佈,另一個為最佳差值的即時資料,然而兩者模擬結果相差不大。 It is well known that meteorological conditions are important input data for air quality models. Since Taiwan is an island with steep mountains, the air flows in Taiwan are significantly affected by local topography as well as surface parameters such as sea surface temperature and roughness. In general, it is a difficult task to collect accurate input data for a meteorological model. Hence, it is worth to investigate the effects of input data and computation algorithm of a meteorological model on the simulation results of a photochemical air quality model. The mesoscale meteorological model, RAMS, and photochemical air quality model, CAMx, was used in this study. Several runs were carried out by varying the parameters that control four-dimensional data assimilation (FDDA), topographic initialization methods, and sea surface temperature data. The effects of these modifications on the results of simulated meteorological fields, including the distributions of wind, temperature, and humidity, were studied. Then, the simulated meteorological fields were used as input of CAMx. Model. The performances of the CAMx model were evaluated by comparing the results with observations. The RAMS model employs Newtonian relaxation method for FDDA. While the analysis nudging is used, the nudging term will be controlled by a time scale parameter. Too strong nudging should be avoided, otherwise some circulations, such as land-sea breezes, mountain-valley winds, and topography effects will be diminished. For topographic initialization, the results obtained by the reflected envelop topography scheme are superior to that obtained by the conventional averaging approach. Two sea surface temperature data set were used, i.e., climatologically data and optimal interpolated real time data. The results of air quality simulation obtained by using two dataset are similar. |