Annual and seasonal tornado activity in the United States and the global wind oscillation

Abstract

Previous studies have searched for relationships between tornado activity and atmospheric teleconnections to provide insight on the relationship between tornadoes, their environments, and larger scale patterns in the climate system. Knowledge of these relationships is practical because it can improve seasonal and sub-seasonal predictions of tornado probability and, therefore, help mitigate tornado-related losses. This study explores the relationships between the annual and seasonal tornado activity in the United States and the Global Wind Oscillation. Time series herein show that phases of the Global Wind Oscillation, and atmospheric angular momentum anomalies, vary over a period of roughly 20–25 years. Rank correlations indicate that tornado activity is weakly correlated with phases 2, 3, and 4 (positive) and 6, 7, and 8 (negative) of the Global Wind Oscillation in winter, spring, and fall. The correlation is not as clear in summer or at the annual scale. Non-parametric Mann–Whitney U tests indicate that winters and springs with more phase 2, 3, and 4 and fewer phase 6, 7, and 8 days tend to have more tornadoes. Lastly, logistic regression models indicate that winters and springs with more phase 2, 3, and 4 days have greater likelihoods of having more than normal tornado activity. Combined, these analyses suggest that seasons with more low atmospheric angular momentum days, or phase 2, 3, and 4 days, tend to have greater tornado activity than those with fewer days, and that this relationship is most evident in winter and spring.