Monthly US temperatures and tornado occurrence

The question of the impact of global climate change on tornadoes, particularly in the United States, has received a great deal of attention in the popular press, but little in terms of definitive results in the scientific literature. The challenges associated with modelling future climate, particularly with respect to the details associated with the tornado environment, are many. Here, we consider a “natural experiment” that could potentially provide insight. We look at the 10 warmest and 10 coldest months for each month of the year from 1954-2013 and the (E)F1+ tornadoes that occur during those warm and cold months. In winter, warmer than normal months are associated with more tornadoes than normal and, in summer, warmer than normal months are associated with fewer than normal tornadoes. Statistical significance is particularly strong in the summer. Given the larger overall mean number of tornadoes in the summer than in the winter, the results show fewer tornadoes in the composite of warm months (10 warmest Jan, Feb, March, etc.) than in the composite of cold months.

A comparison of the centroid of tornado locations for warm and cold months shows differences of ~400 km in some months of the year. Overall, however, the difference in the location of the centroid for the composite for the 10 warmest months and the composite for the 10 coldest months is only ~80 km, suggesting little overall change in the average location of tornadoes.

The mean temperature for our warm months is >3 C warmer than the cold months, so that it is tempting to use it as a possible proxy for future average months. How well using historical warm months as a proxy for future average months is an open question. It is possible that atmospheric circulation may be sufficiently different that pattern changes could make the future look different than the warm past