Strong thunderstorm windsThunderstorm winds causeTri-component thunderstorm wind damage to structures. HoweverDual-tree complex wavelet packet transform, the available number of tri-component thunderstormTri-component thunderstorm wind windThunderstorm winds records with a subsecond sampling interval for structural dynamic analysis is limited. In the present study, the use of the dual-tree complex wavelet packet transformDual-tree complex wavelet packet transform (DT- \( {\mathcal{C}} \) WPT) to generate tri-component nonstationary non-Gaussian thunderstorm windThunderstorm winds records is proposed. The use of DT- \( {\mathcal{C}} \) WPTDual-tree complex wavelet packet transform is to gain efficiency since it is a redundant transform with a low redundancy factor of 2 and provides phase information. The generation is seed-record- or data-driven and is based on the overall framework of the iterative power and amplitude correction algorithm. This ensures that the sampled records match the prescribed marginal (mixture) cumulative distribution of each record component as well as the crossed and non-crossed time-frequency-dependent power spectraSpectrum density functions. The use of the proposed approach is shown by numerical examples, illustrating its efficiency and showing that the statistics of the sampled nonstationary non-Gaussian processesGaussian process agree well with their targets.