SPECTRAL MODELS BASED ON BOUSSINESQ EQUATIONS

摘要

For a stationary wave field, spectral models represent the surface wave motion sufficiently accurately. There are however two drawbacks in using nonlinear spectral models: a) The computational time involved when simulating the random wave field; the number of operations needed is O(N2) (N is the number of frequency components). B) Spectral models are usually one-equation reduction of the two-equation time domain model, which does not then have the same characteristics of the original model. Here we explore two approaches to solve these problems. First, we use the hybrid FFT technique (Bredmose et al. 2004) both to speed up the calculations and to incorporate higher order nonlinear terms. Second, we look at frequency domain transformation without reducing the model to one equation. Significant improvement in computational speed was obtained with the hybrid FFT approach. The models also show good agreement to the data reported by Mase and Kirby (1992).