Modeling Correlated Wind Speeds by Trigonometric Archimedean Copulas

Abstract

In this paper, the statistical features of correlated wind speeds at multiple sites are characterized by marginal distributions and Archimedean copulas. Firstly, the generalized lambda distribution (GLD), kappa distribution and Weibull distribution are employed to recover the quantile function of wind speed at each site. Then, three new Archimedean copula models are constructed to characterize the dependence structure of historical wind speed data. Based on Rosenblatt transformation, a generic algorithm is presented to produce sample realizations of correlated wind speeds, which would have the same statistical features as historical wind speed data. Finally, numerical examples are given to illustrate the proposed methods.