本研究使用Lagrangian質點擴散模式Windmodel進行2003到2006年八月每天的模擬，藉由質點分布探討台灣夏季近地面的環流分布與對流之關係。 為瞭解質點擴散過程中的質點累積情形，選擇2003到2006年八月閃電頻率最高的地區作為質點累計區域（120.35°E-120.90°E，23.6°N-24.0°N），同時對質點累計範圍做分析，對等寬的狹長方形方格，計算質點在不同區域內的變化，以瞭解質點在累計區域一天內的變化情形。 本研究首先從2006年八月找出有午後對流系統在台灣生成，閃電多的個案、以及2006年八月整日閃電次數最少的個案作為參考，計算閃電多（少）的24小時逐時質點累計分布曲線，再將逐時質點累計分布曲線與其他天的質點累計分布曲線計算相關係數，算得之值為Ra（Rb）。結果顯示Ra值高的個案中有在台灣陸地出現對流雲生成、發生數千次的閃電，但這些個案的Rb值亦高。 接續的討論使用不同質點累計區域的質點隨時間變化特徵探討，定出R1（R2）為閃電多（少）個案的質點累計區域A01（120.35°E-120.40°E，23.6°N-24.0°N）質點累計曲線與其他天質點累計曲線的相關。將R1（R2）與2003至2006年逐年八月的每日累積閃電次數作相關，顯示在2006年8月的R1與120°E-121°E，23°N-24°N範圍內所發生之全閃電的相關係數為最好，其值達0.5。表示使用R1作為判斷2006年8月當日閃電次數多與否，準確性約為五成。 個案分析中針對一個質點累計區域總質點數下午出現峰值時，質點累計區域附近發生閃電的個案作討論，在質點二維分布圖看到海陸風環流對質點運動的影響；使用小區域的質點累計區域比較，顯示位於較內陸的區域出現峰值的時間與閃電在累計區域發生峰值的時間較為接近。 This study uses the Lagrangian particle dispersion model – Windmodel to simulate each day during August 2003 to 2006. Through the distribution of the particles we discuss the relationship between the summer surface circulation and convection in Taiwan. In order to understand the accumulation of the particles in dispersion process, we choose the particle calculating domain (120.35°E-120.90°E，23.6°N-24.0°N) , because there is the area of most highly frequency of lightning occurrences during August 2003 to 2006. We divide the particle calculating domain into the rectangle areas which have the same width of longitude. We can know the particles variation in the calculating domain in a day through the calculation of the particles variation in different areas. First, this study chooses the cases which have more lightning occurrences because of the convection systems development in the afternoon and have the least lightning occurrences of August 2006. Then we calculate the particles variation of the case that has more (lower) lightning occurrences in twenty-four hours. Next, we calculate the correlation coefficient Ra(Rb) of the particles variation of the case that has more (lower) lightning occurrences and the particles variation of the other days in whole calculating domain. The result reveals that some cases with high Ra value have the development of the convection system and several thousand lightning occurrences in the afternoon in Taiwan. However, these cases with high Ra value also have high Rb value. In the following discussion, we use the several particles variation of the different particle calculating domains. To define the correlation coefficient R1(R2) is the correlation of the particles variation of the case that has more (lower) lightning occurrences and the particles variation of the other days in A01 domain(120.35°E-120.40°E，23.6°N-24.0°N). Then we calculate the correlation of R1(R2) and daily lightning occurrences during August 2003 to 2006. The best correlation is 0.5 between R1 and the daily lightning occurrences in the area of 120°E-121°E, 23°N-24°N. It reveals that the accuracy is about 50% if we use R1 to judge the frequency of daily lightning occurrences in a day. In case study, we discuss a case that the particles increase in the afternoon, and it has more occurrences of lightning near the particle calculating domain at the same time. We see the impact of the sea-breeze circulation on particles through the two-dimensional particles distribution figures. It shows the most occurrences of lightning near the particle calculating domain in a day, almost the same time the particles have a peak in the divided particle calculating domain which the location is in the land and far away the ocean.
