This paper addresses the problem of the characterization of oceanic internal wave packets from SAR images. Empirical Mode Decomposition (EMD) provides a new strategy for extracting internal wave signatures, and offers the precision, flexibility and accuracy which are proved by Huang et al. in 1998. The EMD is employed in a scale-space analysis of SAR image profiles to locate and characterize the wave parameters of internal waves. First, the EMD is applied for the nonlinear internal wave temporal series of synthetic aperture radar (SAR) decomposition. Second, a criterion based on the theory that the max normalized deflection stands for the largest energy is introduced to detect the internal wave component. The results show that EMD is an excellent tool to detect internal waves against background noise, and to estimate, with a good degree of precision, soliton wavelengths from SAR ocean image profiles.