Study On Statistical Characteristics And Prediction Methods Of Long-crested Extreme Waves

Extreme waves,which are much larger than those of the background waves.Randomness,strong nonlinearity and transient performance are main characteristics of extreme waves.The waves have caused serious threat to the safety of navigation and marine structures in both open seas and offshore regions.Investigating characteristics of extreme waves and then predicting where and when they will occur are keys to avoiding potential disasters.Long-crested waves,which are not considered wave direction,are the foundation of analyzing extreme waves.Many studies have demonstrated that extreme waves usually emerge from intensive wave groups.It is an efficient approach to study extreme waves based on wave groups.However,traditional methods to discriminate wave groups are affected by artificial definitions and mainly used to detect continuous large wave groups,whereas,wave groups that could pose a threat in the actual ocean may not be detected by the method.Therefore,it is important to propose a new method to identify whole wave groups accurately with less affected by artificial definition.The purpose of the dissertation is to propose a new method to identify whole wave groups in long-crested random time series,then analyze statistical characteristics of extreme wave groups in both stable(wave evolution is not considered in stable random time series)and unstable random time series and predict extreme waves in different water depths based on wave groups over shortterms.First,a new fully non-hydrostatic wave model is developed based on 3D incompressible Euler equations,which is able to provide numerious reliable evolutions of wave time series to analyze characteristics and short-term prediction of extreme waves.Second,the scale non-uniformity wavelet power is defined based on the wavelet transform,after sensitivity testing,two ends of wave groups can be identified based on the location of adjacent local minimums of the scale non-uniformity wavelet power.The method could quantitatively discriminate whole preliminary wave groups in random waves with different spectra and spectra bandwidths.Furthermore,for strong dispersive wave cases,if the dominant frequency of a wave group is higher than that of the adjacent trailing wave group,these can be considered together and called "coalesced wave group".Third,statistical characteristics of extreme wave groups in both stable and unstable random time series are analyzed.It is found that all wave shapes of extreme wave groups are contained within the ±2σL confidence intervals of the NewWave theory,where σL is the Lindgren standard deviation.Besides,good agreements are obtained between the wave shapes of the averaged extreme wave groups and based on the NewWave theory.Generalized Extreme Value Distribution is obeyed for the distributions of time lengths of extreme wave groups.Fourth,by investigating evolution of extreme waves over sloping bathymetry,it is found that the majority of extreme waves occurred at the seaward part of the bar.Besides,interaction of adjacent wave groups is neglected due to the weak dispersion.The maximum of the scaled non-uniformity wavelet power of wave groups can be used as a precursor to predict the occurrence of extreme waves over sloping bottoms.It is found that the precursor predict most extreme waves for different wave steepness and spectra widths successfully and effectively,and could be applied to various distances between the locations of discriminating wave groups and occurring extreme waves.Last,by investigating evolution of extreme waves in deep water,it is found that locations of generating extreme waves are uncertain.Besides,interactions of adjacent wave groups should not be neglected due to the strong dispersion.The maximum and the summation of the maximum of the scaled non-uniformity wavelet power of preliminary wave groups are used as precursors to predict the occurrence of extreme waves due to the nonlinear self-focusing of independent wave groups and interactions of adjacent wave groups,respectively,then the locations of forming extreme waves are calculated based on the theory models.The methods could predict most extreme waves for different wave steepness and spectra widths successfully and effectively within 40 peak wave lengths.