Abstract

abstract:

Five distinct records of US tornado outbreak history are created by applying separate criteria from published definitions of "outbreak" to the complete US tornado dataset covering 1975–2014. To identify statistical patterns in the resulting outbreak records, time trends were calculated using bivariate correlation against time (year), ANOVA was used to evaluate similarities and differences between the different records, and kernel density analysis was used to identify and map spatial patterns in outbreak variables in each record. The tornado record, and specifically the tornado outbreak record, shows a general increasing trend during the period of study for all five definitions, albeit only in the non-summer seasons. Results from ANOVA indicate statistically significant differences in outbreak records between definitions. According to our analyses, regardless of the definition, the total number of tornado outbreaks (TO), the total number of tornadoes that were part of an outbreak (Outbreak Tornadoes; OT), and outbreak tornadoes as a proportion of all tornadoes (OTP) exhibit their highest annual mean values in a tripole pattern including central Arkansas, central Mississippi, and northern Alabama. This is one of the few consistencies across outbreak records between the different definitions, and we propose that this region represents "Tornado Outbreak Alley," in contrast to the traditional "Tornado Alley" of the southern Great Plains.

pdf