On Oceanic Rogue Waves (1501.03370v5)

Abstract: We propose a new conceptual framework for the prediction of rogue waves and third-order space-time extremes of wind seas that relies on the Tayfun (1980) and Janssen (2009) models coupled with Adler-Taylor (2009) theory on the Euler characteristics of random fields. Extreme statistics of the Andrea rogue wave event are examined capitalizing on European Reanalysis (ERA)-interim data. A refinement of Janssen's (2003) theory suggests that in realistic oceanic seas characterized by short-crested multidirectional waves, homogeneous and Gaussian initial conditions become irrelevant as the wave field adjusts to a non-Gaussian state dominated by bound nonlinearities over time scales $t\gg t_{c}\approx0.13T_{0}/\nu\sigma_{\theta}$, where $T_{0}$, $\nu$ and $\sigma_{\theta}$ denote mean wave period, spectral bandwidth and angular spreading of dominant waves. For the Andrea storm, ERA-interim predictions yield $t_{c}/T_{0}\sim O(1)$ indicating that quasi-resonant interactions are negligible. Further, the mean maximum sea surface height expected over the Ekofisk platform's area is higher than that expected at a fixed point. However, both of these statistics underestimate the actual crest height $h_{obs}\sim1.63H_s$ observed at a point near the Ekofisk site, where $H_s$ is the significant wave height. To explain the nature of such extreme, we account for both skewness and kurtosis effects and consider the threshold $h_{q}$ exceeded with probability $q$ by the maximum surface height of a sea state over an area in time. We find that $h_{obs}$ nearly coincides with the threshold $h_{1/1000}\sim1.62H_s$ estimated at a point for a typical $3$-hour sea state, suggesting that the Andrea rogue wave is likely to be a rare occurrence in quasi-Gaussian seas.