摘要 地表蒸發散量 (Evapotranspiration，簡稱ET)，又稱潛熱(Latent Heat)，單位為W/m2，於氣候動力學及生態系的生產率研究是不可或缺的因子之一，因為蒸發散資訊可表為記錄地表能量傳遞的過程之重要參數。雖然已有許多方式去估計蒸發散量，但需要氣象環境參數作輔助，故會受到地面觀測站分佈的影響，而且利用環境參數所反演的蒸發散量沒有辦法代表大區域範圍的平均。因此希望透過衛星遙測資料分析，將可有效取得環境大面積的蒸發散情形。 本篇研究以台灣為測區，使用Aqua/MODIS的資料與Neshida在2003年所提出的蒸發散比率（Evaporation fraction, EF）演算法反演全台的EF分布圖，EF為ET除上可用能量（Q），Q為可感熱與潛熱的總和。 驗證部分是看以MODIS衛星所反演的EF與嘉義跟宜蘭測站蒸發皿觀測值的相關性，並且將所反演出的EF乘上可用能量（Q）得到ET值與利用地面測站氣象環境參數反演出來的ET值做比較。2003~2005年嘉義測站使用環境參數與MODIS的影像資料所反演之蒸發散量的相關係數為0.083；2003~2005年嘉義測站利用MODIS觀測資料反演之EF與測站每日蒸發皿觀測量的相關係數為0.297。2003~2005年宜蘭測站使用環境參數與MODIS的影像資料所反演之蒸發散量的相關係數為0.502；2003~2005年宜蘭測站利用MODIS觀測資料反演之EF與測站每日蒸發皿觀測量的相關係數為0.740。結果顯示出兩種不同特性的資料在比對上有一定的困難，不管在時間以及空間的分布上，故需要進一步的研究佐證。 Abstract Evapotranspiration (ET, or latent heat flux) is one of the critical factors for understanding the climate dynamics and the terrestrial ecosystem productivity because of its close relation to energy transfer process. Although there are many approaches to estimate ET, most of the existing techniques of ET estimation require surface meteorological observations. Thus the area coverage of ET estimation is limited by the density of ground observation network, and it is difficult, if not impossible, to estimate ET at regional to global scale by means of traditional meteorological observations. Therefore, remote sensing is the one of the best solution for estimating ET at large scale. In this study, we used MODIS data on board NASA’s AQUA satellite over the Taiwan Island to map the distribution of evaporation fraction (EF). Validation was made by calculating the correlation coefficient between EF and the daily basin observation in Chiayi and Ilan stations. Quantitative comparison of ET between the satellite derived and the surface meteorological observation was also made. During the years of 2003 to 2005, the correlation coefficients were 0.083 and 0.502 in Chiayi and Inlan, respectively, while the correlation coefficients between EF and the daily basin observation in Chiayi is 0.297 and 0.740 in Inlan. Study indicated that direct comparison between these two types of observation may be difficult due to their distinctive observing characteristics of spatial-temporal patterns, among others. It should be further investigated.
