A data-driven approach for simulation of full-scale downburst wind speeds

Recognition of the significance of wind loading on low-rise buildings and transmission line structures during the passage of thunderstorm downbursts has prompted research aimed at modeling downburst wind fields and simulating wind flows. The non-uniformity and non-stationarity of thunderstorms in both time and space pose considerable challenges for effective simulation. In this paper, the evolutionary behavior of downburst winds is examined in the time–frequency domain using both stationary wavelet transform and Hilbert transform. Stationary wavelet transform first decomposes a sample of a multi-component non-stationary random process into a set of mono-component signals. These signals are subsequently transformed into analytic signals with the Hilbert transform, which yields the instantaneous amplitudes and frequencies. An efficient simulation approach is then proposed for thunderstorm downburst winds using the instantaneous properties as the basis. Simulation is performed based on a sample realization of the process without the customary assumptions of piecewise stationarity or parametric models. The method is extended to the simulation of multivariate random processes utilizing proper orthogonal decomposition. Analytical expressions for the statistical properties of the underlying random processes are formulated. Example simulations of measured full-scale downburst wind data are presented to demonstrate the efficacy of the proposed method.