The automated detection and recognition of internal waves

Abstract

A new framework is presented for an automated approach to identify possible internal wave packets in synthetic aperture radar (SAR) images as a means to infer primary information about the internal waves. An operational version of this framework is expected to be useful to both oceanographers and modellers for analysing the importance of internal waves for the mixing required to warm and advect deep-sea water to the surface. The framework is based on a combination of techniques using wavelets, edge discrimination, edge linking and edge parallelism analysis. Six satellite images of the Eastern Atlantic have been used to demonstrate and test the framework. The determination of the type of signature and the wavelength of the internal wave has been demonstrated and the accuracy of the approach is assessed.