Accurate wave climate characterization, which is vital to understand wave-driven coastal processes and to design coastal and offshore structures, requires the availability of long term data series. Where existing data are sparse, synthetically generated time series offer a practical alternative. The main purpose of this paper is to propose a methodology to simulate multivariate hourly sea state time series that preserve the statistical characteristics of the existing empirical data. This methodology combines different techniques such as univariate ARMAs, autoregressive logistic regression and K-means clusterization algorithms, and is able to take into account different time and space scales. The proposed methodology can be broken down into three interrelated steps: i) simulation of sea level pressure fields, ii) simulation of daily mean sea conditions time series and iii) simulation of hourly sea state time series. Its effectiveness is demonstrated by synthetically generating multivariate hourly sea states from a specific location near the Spanish Coast. The direct comparison between simulated and empirical time series confirms the ability of the developed methodology to generate multivariate hourly time series of sea states. Finally, the potential of the proposed methodology to simulate multivariate time series at multiple locations and incorporate climate change issues is discussed.
Civil and Construction Engineering
autoregressive logistic regression; simulation; stochastic processes; time series; univariate arma