Examining the Capability of Machine Learning Methods for Unraveling Ocean Dynamics from Long-Range M-Sequence Data

Accurately representing ocean dynamics is crucial for naval monitoring and has garnered significant interest. The complexity of the ocean, akin to other intricate physical systems, poses challenges due to the absence of a closed-form equation of the dynamics. Additionally, environmental nonlinearities further complicate modeling efforts. However, recent advancements in data-driven machine learning methods offer promising solutions for unraveling some of the complex patterns in the ocean. In this work, we use long range m-sequence waveform data to investigate the ocean's hidden dynamics. The methods employed are inherently data-driven, enabling the discovery of low dimensional representations of the data without supervision. Preliminary results from applying these methods to m-sequence waveform data from the Wake Island Station are presented.