Detection of solitary ocean internal waves from SAR images by using U-Net and KDV solver technique

Abstract

The internal waves are different phenomenon in oceanography and these are also waves but occur in the inside of the ocean. Many techniques exist for the detection of internal waves but every method has its own advantages and drawbacks. Earth observations systems keep an eye in monitoring ocean internal waves at regular intervals of time. But automatic detection of internal waves is still a challenging problem. This paper proposes a method which gives complete methodology for internal wave detection. The method is divided into three main stages that are input data pre-processing, parameter extraction and modeling. The proposed system will initially take Synthetic Aperture Radar (SAR) images, pre-process for noise. Augmentation techniques such as flip and rotation resolve occlusion caused by clouds. U-Net is used for segmentation and feature extraction of wave parameters such as frequency, amplitude, longitude, latitude. Finally, for ocean internal wave modeling, KdV (Korteweg–de Vries) solver is used. KdV solver takes the internal wave parameters as input and it gives the velocity, density plots of internal waves. The deep learning model is tested on SAR images and proved to give accurate results for the internal wave detection.