Numerical Analysis Of Wave Spectrum Density Function

In this paper, the instantaneous surface elevation data sample, which is from a recording project in executed in North Sea and has a six day's duration, is analyzed. The corresponding energy spectral density function (target spectrum) was estimated by using the transformation method based on FFT. Based on a set of recording data form observation platform called Alwyn, the empirical spectrum expression is modified in terms of its peak elevation parameter and peak frequency position parameter. The results shows that spectral moment m0 vary remarkablywith dependent variable y and peak frequency ωp vary remarkably with T1×ωp. When T1×ωp=5 and γ = 3.27 ,the predictive values of m0 and ωp tally with the analysis resultof actual measured data. It's should be pointed the mean up-cross period and zero spectral moment fluctuate a lot, and they don't belong to the same wave condition. It's because the data are observed in a 6-day storm. By compare the given 409 significant wave heights and statistical characteristics in a duration with that obtained by FFT, it's regard they are fitted very well.Traditional spectrum estimation method, Fast Fourier Transform, assumes that the covariance function out of the computation domain is regarded as zero, the period histogram is considered to be the whole period. The obtained spectrum function actually is the product of the real spectrum and a window function. In this paper, the maximum entropy method is compared with Fast Fourier Transform in their aspect of establishing a frequency spectrum for wave data. The algorithm of Maximum Entropy Method (MEM) is discussed and the computation results show that MEM method is better than FFT method in that it avoids some unrealistic assumption. Because it doesn't need window function, it doesn't have the problem of spectrum leakage. Even for some short wave record, MEM can return satisfactory result. For the storm wave data from North Sea, the combination of Akaike method and empirical method for determining the order is good. MEM method is another good option for spectrum establishment of the wave data.