# Early Detection of the Draupner Wave Using Deep Learning

## TL;DR

This paper introduces a deep learning approach using LSTM networks for early detection of Draupner rogue waves, enabling predictions minutes before peak occurrence and potentially improving safety in marine engineering and other fields.

## Contribution

The study demonstrates that LSTM-based deep learning models can predict rogue waves well in advance, surpassing existing warning time scales and applicable to various extreme time-series phenomena.

## Key findings

- LSTM models can predict Draupner waves minutes before peak.
- Early warning times are significantly improved over seconds.
- Results applicable to multiple fields beyond marine engineering.

## Abstract

In this paper, we propose and apply a deep learning strategy for the early detection of the Draupner rogue (freak) wave, which is also known as the New Year's wave. We use a long short term memory (LSTM) network and show that Draupner rogue wave could have been observed at least minutes before the catastrophically dangerous peak has appeared in the chaotic wave field using the available data. Compared to the existing early warning times scales on the order of seconds, this is a major step forward which will certainly enhance the safety and understanding of the marine engineering. As the rogue wave data sets get improved in the future, our results may be enhanced to increase the early warning time scales. Our results can be used to predict other rogue waves and extreme time-series phenomena in fields including but are not limited to hydrodynamics and marine engineering, optics, finance, and Bose-Einstein condensation, just to name a few.