The vertical coherence function for near-ground windfields provides a critical measure of the wind environment for tall structures. Vertical coherence is generally defined only for stationary signals, but real windfields are rarely stationary. In fact, periods of highest wind speeds and turbulence are often very localized in time, making traditional models and estimates of vertical turbulence problematic. In this paper, we define a localized vertical coherence function. The standard (stationary) narrowband Fourier signal representation is replaced by a Morlet wavelet transform which captures time-localized narrowband energy. The localized coherence function is then defined from this time-frequency signal representation, and estimated using standard statistical approaches. Those time intervals with highest wind levels and turbulence are then used to characterize the vertical coherence during extreme events. Analysis is performed on data from severe storms collected at Texas Tech’s Wind Engineering Research Field Laboratory