Internal wave signature detection and characterization from SAR images

This paper addresses the problem of the detection and characterization of oceanic internal wave (IW) packets. We describe an efficient method for automatic extraction of IW features present in ocean SAR images. The utility of wavelet analysis as a tool for IW detection is first examined using the 2-D wavelet transform (WT) based on the multiscale edge detection method. We show that the evolution of local maxima of the 2-D WT across scales characterize the local shape of these quasi-linear periodic structures. Then, the Continuous Wavelet Transform (CWT) is employed in a scale-space analysis of SAR image profiles to locate and characterize the local scaring properties of the IW packet crests. The parameters of the IW crests are caracterized and estimated using the minima lines given by the Wavelet Transform Modulus Minima (WTMM) method. We show that wavelet analysis provides a new strategy for examinig IW signatures and offers the flexibility which results from the introduction of well-suited basis functions. Results from this study illustrate that wavelet transform methods can be excellent tools to automatically detect and extract IW packet parameters from SAR images against background noise and IW lookalikes.