INTRODUCTION

Freak waves are events that featuring a single, unusually large and steep crest and posting severe hazards to the mariners and offshore structures. Perhaps one of the weakest and most difficult aspects for the explorations of freak waves in the oceans is the ostensive scarcity of actual field measurements of rogue wave events. Because of the uncertain and unpredictable nature of the occurrence of rogue waves, the conventional, discrete kind of wave measurements at fixed time intervals have not been conducive in capturing actual freak wave episodes. A viable approach for making comprehensive freak waves measurement does not seem to be presently available.

The state of the art on freak waves was first summarized over a decade ago at the NATO Advanced Research Workshop (Torum and Gudmestad, 1990). A new workshop, the Rogue Wave 2000, took place in November of 2000 that further assessed the more recent results (Olagnon and Athanassoulis, 2000). There was an interesting explanation given earlier by Dean (1990) to indicate that both of nonlinearity and directionality are primarily possible causes of freak waves. The interactions between currents and waves have prominently studied to illustrate that it can also be responsible for generating rogue or extreme waves at a particular location (White and Fornberg, 1998; Lavrenov, 1998). Laboratory studies have demonstrated that freak waves can be generated through nonlinear wave-wave interactions in a two-dimensional wave flume (Stansberg, 1990). It has also been numerically modelled that a typical freak wave having a single and steep crest can be generated by the third order nonlinear interactions (Yasuda et al., 1992; Trulsen, 2000; Mori and Yasuda, 2001). One of the untenable issues for a tangible investigation of freak wave, however, is the fundamental difficulty in making field observations in order to verify the well-developed theories. Moreover, most of the conventional approach of short period statistical analysis are developed under the basic assumption of a stationary state, while the real sea state changes constantly in time and space, especially under stormy sea conditions. It is, therefore, rather thorny to compare and correlate observed data including freak waves with the conventional statistical analysis in a rigorous sense.

This paper presents an empirical analysis of available wave measurements collected during 1986 - 1990 in the Sea of Japan where freak waves are known to have observed. To explore the observed wave data with freak waves, the analysis consists of applications of the conventional stationa ry processes approach to examine their statistical properties such as frequency and directional spectra, probability density functions of surface elevations, distributions of wave height, crest height and run length, and also an application of the nonstationary time-frequency spectrum analysis based on the recently advanced wavelet transform method.