Simulation of Wave Time Series with a Vector Autoregressive Method

Abstract

Joint time series of wave height, period and direction are essential input data to computational models which are used to simulate diachronic beach evolution in coastal engineering. However, it is often impractical to collect a large amount of the required input data due to the expense. Based on the nearshore wave records offshore of Littlehampton in Southeast England over the period from 1 September 2003 to 30 June 2016, this paper presents a statistical method to obtain simulated joint time series of wave height, period and direction covering an extended time span of a decade or more. The method is based on a vector auto-regressive moving average algorithm. The simulated times series shows a satisfactory degree of stochastic agreement between original and simulated time series, including average value, marginal distribution, autocorrelation and cross-correlation structure, which are important for Monte Carlo modelling of shoreline evolution, thereby allowing ensemble prediction of shoreline response to a variable wave climate.