Distortions Associated with Random Sea Simulators

Abstract

Some numerical techniques for simulating Gaussian ergodic stochastic sea models are described, analyzed, and contrasted. A general method for generating all numerical, linear, one‐dimensional simulators by wave superposition permits one to describe or create any of these numerical random sea simulators in five steps. The distortions associated with each numerical simulator by wave superposition are analyzed from a general point of view and the arbitrariness of some numerical simulation techniques commonly used is noted. The time‐consumed in these Monte Carlo experiments is an important factor. The numerical algorithms used can change indirectly the level of distortions associated with each numerical simulation technique. A special reference has been made to the use of the fast Fourier transform (FFT) for computing and to the second‐order autoregressive behavior of each wave component in order to reduce the time‐consumed. Three criteria are proposed for qualifying the numerical simulators in order to adapt the requirements of each numerical experiment considered. To explain the variability of random sea, the deterministic amplitude component simulators are rejected while a nondeterministic spectral amplitude simulator (NSA) using a FFT algorithm can be employed.