Problems with the historical undercounting of tornadoes is limited to EF1 and lower.
Deployment of more recent detection systems such as Doppler, growing population coverage and more sophisticated personal communication devices have made marked reduction in the number of smaller events that may go unnoticed.
However, even thiry years ago it is unlikey an EF3 event would go unnoticed.
Here the U.S. tornado data is grouped to show EF2 or greater, EF3 or greater and EF4 or greater. All three groupings show very similar accounts of the historic variation, indicating that there was consistent counting of all categories over this period.
There is a very clear change after 1974 in all categories. The post war cooling period was far more active than the later warming period.
The data shows the fallacy of the frequent claims that a warming of the climate will lead to more frequent and more intense storms. At least the evidence is clear that this is not true for tornadoes in the continental U.S.A.