Z‐Transform Modeling of P‐M Wave Spectrum

The approximation of the Pierson‐Moskowitz (P‐M) wave spectrum as the output of digital filters to white noise excitation is considered. It is shown that the mathematical peculiarity of this spectrum is the source of the numerical difficulties encountered in approximating it by autoregressive (AR) models. Further, it is shown that a reasonably accurate initial AR approximation leads to quite efficient autoregressive moving average (ARMA) models. This is accomplished by employing two new techniques. Numerical data are given in a dimensionless form so that they are applicable to a variety of structural dynamics applications.