US Tornado data is often dismissed as being unreliable due to sampling issues resulting from under-counting of smaller storms before the widespread use of doppler radar.
However, excluding EF1 category storms, the record of more powerful categories can be seen to be quite consistent.
There is little likelyhood of a EF3 of EF4 storm going unreported even before doppler
Here the various storm category groups have been “normalised” to have a mean of zero and std. deviation of 1 to allow a direct comparison.
It is clear that the ‘global cooling’ period from 1950 – 1975 had consistently higher tornado activity that the later ‘global warming’ period.
There would seem to be the beginnings of a turn around tornado count showing a small but steady increase since arount year 2000.