Real-world rogue wave probabilities

Despite several strong hypotheses on how rogue waves can be generated in idealized conditions, the real-world causes of these waves are still largely unknown. We credit this to insufficient amounts of observational data and a missing robust probabilistic framework to analyze the available data.

We address these issues by processing over 1 billion wave measurements from offshore buoys and organizing them into a comprehensive catalogue. Through a robust, machine-learning driven analysis, we then identify several characteristic sea conditions that lead to significantly higher risks to encounter a rogue wave. This yields quantitative evidence on the relative importance of the underlying physical mechanisms.

Specifically, we find that by far the most important factor is crest-trough correlation, a parameter related to spectral bandwidth, with other parameters acting as minor corrections. This has profound implications on the nature of "everyday" rogue waves, and suggests a path towards a more reliable extreme wave forecast.