Autoregressive modelling of measured sea waves off west coast of India

Abstract

The spectral analysis of measured sea waves can efficiently be carried out by the Fast Fourier Transform (FFT) technique. Even though many present techniques can be used for the generation of time series on sea waves from theoretical wave spectra (Pierson-Moskowitz, JONSWAP, etc.), they may not be computationally efficient in providing accurate results for a given application. A more efficient method is therefore described here to generate time series data on sea waves and to provide spectral relationships with minimum loss of information. The random process is expressed by the autoregressive (AR) algorithm which is a polynomial function. The optimal AR polynomial is obtained by using the Akaike information criterion method. The AR parameters of the estimated spectrum are very few in numbers as compared to the number of values of FFT spectrum or time series waves. Hence it provides a potential advantage of storing on-line long-term data on sea waves