Wave group statistics were analyzed using both measured and simulated data. Spectra of wave records measured at northern Taiwan were used as targets, and two simulation methods were applied to perform long series of runs to study the group statistics. Five statistical models were used to model the statistical distribution of total run and run length of the wave groups. Based" upon field and simulated data, it was found that the Weibull and the exponential distribution could be used to describe the statistics of run length and total run. For the statistics of the group run length and the group interval, where only two or more high waves must exist, it was found that, albeit not very satisfactorily, the exponential distribution could be used as an approximation.
Wave groups occur as a common phenomenon on the ocean surface. Both linear (Goda, 1983; Elgar et al., 1984) and nonlinear wave theories (Shemer et al., 1998) can be applied to explain their formation. Linearly, wave groups are formed through superposition of two or more free waves that accidentally run into each other. Nonlinearly, wave groups can be formed through a delicate balance of nonlinearity and dispersion. Nonlinear wave-wave interactions may lead to interchange of energies among wave components, and (linear) dispersion drives wave components with different frequencies away from each other. Only a balance between these "forces" can wave groups of permanent form such as solitons exists on the water surface (Yuen & Lake, 1982). Knowledge of wave groups is important for coastal engineering due to various reasons. They may have important effects on the stability of coastal structures (Medina et al., 1994), as well as on the tranquillity of a harbour (Ouellet & Thrriault, 1989). Wave groups can also excite long-period waves (Sch~iffer, 1993). Sand (1982) showed that
