As a new tool for ocean wave measurement interiorly, X-band radar can be used to provide sea state information and a wave field can be get form an image sequence. However, there are still some problems in the retrieval of wave frequency spectrum and significant wave height ( Hs). A nonlinear regression method was used to fit the in situ wave frequency spectrum and radar one-dimension image spectrum with standard PM, JONSWAP and TMA spectrum, and the basic form and the corresponding spectral parameters can be obtained accurately. Then, a generalized regression neural network model (GRNN) was introduced to retrieve the wave frequency spectral parameters from the one-dimensional radar image spectrum parameters. In the model, the signal — to — noise ratio (SNR) of the image sequence was added to set up a nonlinear relationship with Hs ,and the inversion results with the in situ data and the traditional algorithm result (the establishment of the linear regression equation between SNR and Hs) were compared. The results show that the mean error of spectral parameters and significant wave height are less than 20%, while the mean error of the traditional algorithm is more than 20%.%X-波段雷达作为国内海浪观测的一种新工具,在海浪频谱获取和有效波高反演方面仍存在较多问题.本文利用非线性回归方法,将现场实测浮标数据频谱和雷达一维图像谱分别与标准频谱模型进行拟合,发现浮标频谱和一维图像谱具有标准频谱的特征,能够较准确地获取相应的谱参数.提出了建立由雷达一维图像谱参数反演海浪频谱参数的神经网络模型,同时在模型中加入影像序列信噪比,进而反演有效波高,并将反演结果与现场实测数据和传统算法(建立影像序列信噪比与有效波高之间的线性回归方程)进行了对比,结果表明,获取谱参数的误差和反演有效波高的平均误差在20％以内,而传统算法计算有效波高平均误差在20％以上.
展开▼
