(Updated 8 October 1999)
Issues associated with tornadoes including how the 1998 season stacked up, changes in death tolls, and some simple statistical modelling of tornadoes in the United States.
Estimates of the climatological threat of at least one event (hail, wind, and tornado separately) within 25 miles of a point in the United States, based on data from the Storm Prediction Center from 1980-1994. This includes animations of the annual cycle and the number of days per year that locations reported severe weather for the period, as well as changes from the 1980s to the 1990s in reporting frequency.
Using the severe thunderstorm climatology above, I've started putting out the annual cycles for threats at a series of points. The locations on the map that this links to are approximately right. I put them on the map from the grid by eyeball and gave a city name for reference to each point.
Important questions to consider when trying to develop a statistical model that describes the likelihood of tornadoes is how the reports have changed through time and how they differ in space. Using the 1920-1998 record of tornadoes by F-scale in the US and reports from around the world, some fairly simple models of the distribution can be developed. Hint: the reports from France look a lot like th 1920-1949 record in the US, in the period prior to the development of a system to collect reports systematically in the US.