Peak windspeeds of 125-144 m/s, based on DOW Doppler measurements, have been calculated for the 3 May 1999 Oklahoma City tornado, representing the highest windspeeds ever measured (with caveats concerning height and duration discussed below). Windspeeds of 125 m/s have been calculated for the Mulhall tornado during the same outbreak and 116 m/s for the Spencer, South Dakota tornado of 30 May 1998. Damage surveys in the wakes of these tornadoes have revealed F5, F4, and F4 damage respectively. While the DOWs have been recording extremely high windspeeds, the wind engineering community has been simultaneously reaching the conclusion that windspeeds of ~120 m/s are not necessary to produce the worst, F-5, damage that has been surveyed. However, comparison between DOW, and other radar, measurements with wind estimates derived from damage estimates, or even not-yet-existent surface in situ observations, is problematic. DOW radar measurements occur nearly instantaneously, representing integration periods of 0.01-0.02 s. Crude space-time conversions across the radar resolution volumes of the highest windspeed measurements result in time-scales of 0.2 s. 1/4 mile winds, corresponding to several second duration wind gusts are associated with F-scale estimates. Furthermore, DOW tornado measurements rarely are collected below 20 m agl. Standard surface measurements occur at 10 m agl, and Fujita-scale wind estimates apply at the height of damaged objects, usually about 3-5 m agl. Further complicating the issue is the fact that radars measure the motion of debris, which may be significantly slower than that of the surrounding air. However, it is likely that debris impacts are one major cause of damage. Finally, spectral width measurements of 15-20 m/s suggest that a significant fraction of the debris may be moving at speeds up to 30 m/s above the mean, perhaps over 150 m/s in the strongest measured tornadoes.

Careful interpretation of DOW windspeed measurements suggests that the radar observations and engineering estimates are not inconsistent.

In the Spencer, South Dakota tornado of 30 May 1998, DOW Doppler measurements at intervals of approximately 25-30 m, and a tornado windspeed model are compared to a house-by-house F-scale mapping. Observed damage is compared to reconstructed time histories of windspeed occurring at various points. Locations impacted by the core flow region experience two separate, but short duration, periods of extreme winds while locations just to the south of this region experience these winds for a longer time. Winds, and damage, are significantly lower on the north side of the track. Comparisons of observed winds and damage in the 3 May 1999 tornado may be presented if available.