High intensity winds in many parts of the world are caused by severely convective thunderstorm downbursts, which are responsible for the failures of many slender lattice structures, such as transmission line towers. Therefore, it is essential to build up an applicable and appropriate model to quantitatively assess the safety of these structures subjected to this sort of extreme wind loading through analyzing the corresponding dynamic responses. With the help of time-frequency analysis tools, the non-stationary component embedded in downburst winds can be detected so that downburst winds are characterized as a summation of time-varying mean and coherent fluctuating components. In the present paper, based on Holmes’ empirical model and Vicroy’s vertical profile used to model the deterministic time-varying mean wind speed, the vertical distribution wind velocity field of a definite position in the course of a thunderstorm downburst may be simulated by resorting to the uniformly modulating non-stationary stochastic process approach by Deodatis. In the simulation, the cubic spine interpolation technique is introduced here to reduce the times of Cholesky decomposition so as to further improve the efficiency of the simulation.6748