Estimates of directional spreading function of fetch limited wind wave and swell are made from measurements observed with spatial gauge array at a wave tank by Bayesian directional method (BDM) and maximum likelihood method (MLM) respectively. Numerical simulation shows that BDM is superior to MLM in stability and accuracy, especially in the estimation of bimodal spreading. It is shown that bimodality of wind waves at high frequency range reported by Young et al. (1995) and Ewans (1998) Is an artifact of analysis technique. On the other hand, Both MLM and BDM demonstrate bimodal spreading of swell at frequencies lower than spectral peak frequency, and can be regarded as the existence in reality. The bimodality is speculated on the attribution to the resonance between air flow and wind wave, and the nonlinear wave-wave interaction is so weak that cannot transfer energy among different directions effectively.
Required by the study of satellite remote sensing, offshore structure design, wave prediction and air-sea interaction, directional spectrum has been the central problem about ocean wave research. The directional spectrum is, however, less well understood. The inclusion of directionality places significantly higher demands on observational systems, which in turn has limited the number of available data, and more stringent demands on estimating models. Like the omnidirectional case, our understanding of the nature of the directional wave spectrum is best studied when some of the many variables that contribute to the general sea state are constant. Fetch-limited sea states provide this situation, and has been the subject of a number of studies. The most significant are those of Mitsuyasu et al. (1975), Hasselmann et al. (1980) and Donelan et al. (1985), each providing parameterizations of a unimodal directional distribution.
