Digital generation of multivariate wind field processes

A very efficient procedure for the generation of multivariate wind velocity stochastic processes by wave superposition as well as autoregressive time series is proposed in this paper. The procedure starts by decomposing the wind velocity field into a summation of fully coherent independent vector processes using the frequency dependent eigenvectors of the Power Spectral Density matrix. It is shown that the application of the method allows to show some very interesting physical properties that allow to reduce drastically the computational effort. Moreover, using a standard finite element procedure for approximating the frequency dependent eigenvectors, the generation procedure requires the generation of a limited number of univariate fully coherent processes for describing the entire multivariate velocity processes independently of the number of components of the process