在傳統海流估計方法中，以在海底設置ADCP(Acoustic Doppler Current Profiler)最為常見，但是為了獲得足夠的空間覆蓋率及時間解析度的海流資料，其耗費的人力、經費及時間相當龐大。 雷達衛星之所以可運用在海流偵測方面，是因為合成孔徑雷達與地面目標物之間的相對運動速度會造成回波訊號頻率中有都卜勒頻移的現象，並且記錄在方位方向訊號的相位歷史中，因此可藉由估計方位方向訊號的都卜勒中心頻率(Doppler centroid)，反演出在徑向方向的海洋表面流速。 本研究使用的衛星資料為ERS-2衛星，首先將衛星資料做前處理，再使用CDE(Correlation Doppler Estimator)做不同視窗大小的都卜勒頻率估計，之後對衛星造成的都卜勒效應做校正，以及檢查陸地的都卜勒偏移量是否為零，若否，則將其校正為零，並將校正過後的頻移量轉換成速度。 在研究中發現，CDE法對於均值(homogeneous)的影像，其估算出來的都卜勒中心頻率相當準確，例如水域的部分，相當適合用於海面的估測；而使用越大的估計視窗，流速估計值的標準差越低，但相對的是犧牲空間解析度。雷達衛星的確可以觀測得到海面上海流的變化，提供大尺度的觀測，改善傳統測量方式無法達到大面積偵測的缺點。 ADCP(Acoustic Doppler Current Profiler) is the most common method to obtain the ocean current information. However, in order to obtain sufficient spatial resolution of the current-data, it is rather expensive both in time and cost. Theoretically, it is possible to extract the surface target velocity from the Doppler shift embedded in SAR phase history in which the azimuth frequency shift is related to the motion of surface target in the radial direction. In this thesis, Correlation Doppler Estimation(CDE) was used to compute the baseband Doppler centroid by varying window size. The SAR data was acquired by ERS-2 satellite. It should be noted that the Doppler effects due to the satellite altitude variations and Earth rotation, among others, were correctly removed. By doing so, the Doppler shift is carefully checked against relative stationary ground surfaces which basically induce zero Doppler shift. The current velocity is then converted from Doppler shift. It is found that the CDE method is reasonably accurate over the homogeneous scene, like the ocean surface. The larger the window size we used, the lower the standard deviation is at cost of spatial resolution. Experimental results show that detection of the variations of current field using SAR is highly possible at relatively large scale observation.
