自1960年代起，衛星成為監測熱帶氣旋的利器，而熱帶氣旋的生成與否，更是衛星應用上一個令人關注的研究議題。本篇研究即為應用SSM/I衛星觀測資料來推估各年份的海氣能量，並嘗試歸納出可用以預測熱帶氣旋是否生成之條件。 劉（2006）曾以2002及2003年的颱風個案為樣本母體，分別計算出這兩年颱風生成的閾值，其結果顯示此兩年的閾值有很大的差異。本研究進一步利用1998到2006年的衛星觀測值及Niño 3.4聖嬰年的分類方法，在熱帶氣旋生成的相近時間，分別計算I.晴空、II.由晴空狀況發展成雲簇而後再消散及III.由晴空發展成熱帶氣旋等三種情況下的能量值，並將觀測結果分成聖嬰年、正常年及反聖嬰年三類別，分別求得每類熱帶氣旋的生成閾值，最後以2007年（正常年）熱帶氣旋個案做驗證。 結果顯示，這三類環境能量場所計算出形成熱帶氣旋的閾值有明顯的差異：聖嬰年最高，正常年次之而反聖嬰年則最低，以這樣的分類結果做的個案驗證可預報度達95％並平均可在JTWC發佈前31.3小時之前預測出熱帶氣旋的形成，若是沒有分類則可預報度為92％，平均可提早28.6小時，證明了以分類後的結果做為預測熱帶氣旋的閾值是較好的標準。此外驗證結果亦顯示，2007年個案幾乎都能由本研究所得之閾值成功提前預測出來。換言之，除了等待JTWC所發佈的颱風警報外，此方法進一步地提前幫助JTWC預測熱帶氣旋生成與否，以利颱風的監控及減災救災。 Since the 1960s, satellites have become a powerful tool in monitoring tropical cyclones. The formation of such storms has become a key research area in satellite remote sensing. The aim of this research is to employ SSM/I data in retrieving the air-sea parameters for different typhoons in finding the possible signals or thresholds as whether tropical cyclones will occur or not. Liu (2006) analyzed the typhoons that occurred in 2002 and 2003, and calculated their formation thresholds. Result revealed that threshold values for the two years were quite different. This research extended the time period to eight years (1998 to 2006). The air-sea energy parameters for three cyclongensis conditions --- (I) stable clear sky, (II) clear skies developing into cloud clusters but subsequently dispersing, (III) clear skies developing into typhoons. Afterwards, the typhoon cases were categorized by the Niño 3.4 indices into three groups---Niño years, normal years, and Niña years. Results show that this research was able to accurately obtain the thresholds in predicting whether or not tropical cyclones will occur. This research discovered that the threshold values were significantly different for three different categories. The threshold for the Niño years was the highest, while the Niña years were the lowest. Meanwhile, the case study reveals that when the classification process is performed, the prediction accuracy reaches 95% and allows the warning time to be 31.3 hours earlier before the official JTWC warning is announced. On the contrary, if the process is not used, the two values reach only 92% and 28.6 hours, respectively. This is proof that the classification process does indeed improve the accuracy rate. Finally the typhoon cases in 2007 were used to verify the threshold values’ effectiveness. Most typhoons in 2007 could be predicted successfully before their actual formation, revealing a potential opportunity for researchers to develop to predict when a typhoon could form. In other words, we could have extra preparation time instead of only waiting until the JTWC issues a typhoon warning
