A method for artificially generating operational sea state histories has been developed, This is a distribution free method to simulate bivariate non stationary and non Gaussian random processes. This method is applied to the simulation of the bivariate process (Hs, Tp) of sea state parameters. The histories respect the physical constraints existing between the significant wave height and the peak period. Furthermore, we show that the persistence properties of the simulation data match to those of the observations.
In nature, the time sequence of sea conditions is random and not repeatable. Often, the required long-term informations can be estimated from observations. But, in this case we can only repeat already observed scenarios. An alternative to observations is the application of hindcasting techniques. However, it requires expensive and time consumming studies to apply theoritical models in order to generate wave data, and the generated sea state histories are often biased due to the inadequacies of the models. So we need alternative techniques to describe and predict the pattern of sea conditions for use in operational simulation studies. We propose in this paper a method of bivariate simulation for sea-state parameters such as significant wave height and peak period (Hs, Tp) with physical phenomena (sea state growth, wave breaking) generating dependance between them. Both parameters are known to be non stationary and non Gaussian processes. In the case of sea state parameters, non stationarity is mainly induced by season. The non Gaussian property is due to the definition of the parameters (Hs, Tp). If the artificially generated sea state histories exhibit statistical properties substancially the same as those of the observed data, they can be used as a substitute for real long term observations for estimating the feasability of offshore operations in areas where no experience exists.
