Abstract: Retrieving internal wave parameters from remote sensing data plays an important role in internal wave research. The generation and propagation mechanisms of such waves can be studied using the parameters extracted from remote sensing data. The authors use empirical mode decomposition, wavelet decomposition and high-order polynomial fitting in extracting internal waves’ (IWs) half-wave width from optical remote sensing images. With the method of Empirical mode decomposition and wavelet decomposition the remote sensing data is decomposed and the signal of IWs is extracted by the normalized variance of IWs. Polynomial fitting is based on the assumptions that bright and dark stripes completely change within the wave modulation and that the first derivative the half-wave width can be extracted. The three methods have been verified by the image of China-Brazil Earth Resources Satellite (CBERS), which was imaging the northern South China Sea near Dongsha Atoll on July 10^th, 2004. Results show that the three methods can effectively extract the needed parameters with all the results of half-wave width being in good agreement with each other. The above methods have obvious advantages in dealing with non-stationary and nonlinear remote sensing data. Using the extracted half-wave width data and other related data (water depth and mixed layer depth), the authors retrieve IWs’ amplitude base on the nonlinear internal wave theory.

Key words: empirical mode decomposition, wavelet decomposition, polynomial fitting, internal waves, half-wave width