Spectral measurements and spectral data have long been used for describing the wave climate at specific sites, and as tools for dimensioning coastal structures plus, recently, wave power devices. Although the spectral approach will continue to be important, an ever growing number of time dependent problems have arisen in recent years, thus spurring the interest for the stochastic modelling of wave related time series.
A method to build wave climate simulation models has been presented and applied to 74 months of 3-hourly Waverider spectra, measured off Figueira da Foz, Portugal. The stochastic models used are based on previous works by Medina et al (Proceedings of the IAMR Conference, Madrid, 1991) who used 6-hour data for the coast of Oregon. Despite its simplicity, the method presented has been found to work well. The synthetic time series for significant wave height, wave energy and upward 0-crossing periods, and peak spectral frequency, seems to reproduce most of the features of the mean observed monthly average and variance trends, autocorrelations, and cross correlations with significant wave height.
The main weaknesses of the model seems to lie in the emulation of interannual variability. Preliminary analysis under way raise hopes that a change in the type of parameterisation (eg by average monthly values, instead of by month number) will produce better models, while preserving the simplicity and speed of the current methodology.