本篇研究以52MHz特高頻測海雷達進行不同地點的海浪回波之觀測，先後分別將天線架設在枋寮及頭城地區，利用海浪回波頻譜的一階波和二階波回波功率代入不同的經驗模式來反演海面風速與示性波高(significant wave height)，並將其與中央氣象局小琉球和龜山島浮標資料做比對，進行經驗模式的校正。 回波訊號處理利用快速傅立葉轉換取得海浪回波頻譜，使用四次非同相積分及五根天線回波頻譜的平均來減少訊號擾動以及手動選取一階波回波的頻率範圍，利用高斯函數擬合並用動差法(Moment method)計算一階波回波功率，最後將其與頻譜圖的總回波功率相減，即可得到二階波回波功率。 由實驗結果可以看出一階波回波功率與風速呈正相關的趨勢。由三參數經驗模式反演風速的經驗模型得到的三個校正參數分別為α=26.62、β=0.84及γ=1.53，最終反演結果的相關係數約為0.62，方均根誤差約為1.4公尺/秒；由誤差函數(Error function)反演風速的經驗模型得到的校正參數分別為σ=3.37、μ=7.34和校正倍數=0.875，最終反演結果的相關係數約為0.68，方均根誤差約為1.6公尺/秒；而由二階波回波功率反演示性波高的經驗模型得到的校正參數α=2.08，此時反演出的結果相關係數為0.72，方均根誤差為30.71公分。 ;In thesis, a 52MHz VHF sea radar was used to observe sea echoes at different locations. Antennas were set up in Fangliao and Toucheng areas respectively, the first- and the second-order spectral power were substituted into varied empirical models to invert the sea surface wind speed and significant wave height, and were compared to datas from Central Weather Bureau fot XiaoLiuqiu and Guishandao Buoy data for the calibration of the experience models. The Fast Fourier Transform (FFT) algorithm was utilized to derive the Doppler spectrum of the sea echoes. To reduce the fluctuations of the random echo signals, a four times incoherent integration, the average of the sea spectrum retrieved from the five antennas and manually selecting the frequency range of the first-order spectral power were employed. A Gaussian function fitting and the moment method were used to calculate the first-order spectral power and by subtracting it from the total echo spectral power, the second-order spectral power can be determined. Results of this study revealed that the first-order spectral power is positively correlated to the sea surface wind speed. The calibration parameters obtained from the three parameters empirical model of the wind speed inversion are α=26.62,β=0.84 and γ=1.53, the correlation coefficient of the final inversion results is approximately 0.62, and the root mean square error is arround 1.4m/s. Based on the error function of the wind speed inversion, calibration parameters (i.e. σ=3.37,μ=7.34), calibration multiple (i.e. 0.875) and the correlation coefficient of the final inversion results (i.e. 0.68) were obtained with a root mean square error of 1.6 m/s. The calibration parameters obtained according to the second-order spectral power of the significant wave height inversion is α=0.28, the correlation coefficient of the final inversion results is approximately 0.72 with s root mean square error around 30.71 cm.
