Tornadoes in South Africa

Recently I had to estimate the probability of a tornado damaging a school in South Africa. The issue was made all the more topical by the appearance of two destructive tornadoes on 2 October. The one occurred in Duduza in the east rand and the other half way across the country in Ficksburg.  Was this a freak incident or are tornadoes a greater danger than we suppose? Drawing on a study by Milford and Goliger 1997 it would appear that approximately 15 destructive (greater than force F0) tornadoes occur each year on average.

The graphic below shows the Milford and Goliger's estimates of the likelihood a tornado occurs (in a given year) in any 1km square area of the country. The likelihood ranges from an effective 0 in the Western and Northern Cape, through one in 100 000 in the Eastern Cape and Free State (including Ficksburg). The likelihood increases to one in 10 000 in parts of Gauteng and the KwaZulu Natal midlands (including Duduza).

The vast majority of Tornadoes  go unreported as they occur in the sparsely populated areas - like large tracts of the Eastern Cape and Free State. Given that only about 15 destructive tornadoes occur each year  the occurrence of two tornadoes in populated areas on the same day would be highly unusual. However there is some doubt as to whether  the two could be classified as F1 or higher. F0 tornadoes are unlikely to damage permanent structures. Usually the  damage associated with F0 tornadoes are the result of peripheral damage - like tree branches falling onto houses. Unfortunately if a F0 tornado is able to topple an old tree or break  branches off they are more than likely to be able to level shacks in Duduza or force poorly built structures to collapse.

The two incidents may thus say more about the quality of housing than they do about the prevalence of tornadoes.