Identifying Teleconnections between Southeastern US Tornado Outbreaks and Daily Climate Indices

Substantial recent literature has been dedicated to the improved understanding and prediction of southeastern US tornadoes due to their deviation from traditional storm characteristics in terms of timing (both diurnally and seasonally) and storm environment (specifically high-shear low-CAPE, or HSLC environments). Numerous studies have also attempted to relate global circulation patterns, such as El Niño Southern Oscillation (ENSO) and Madden-Julian Oscillation (MJO), to CONUS hail and tornado prevalence towards improving subseasonal and seasonal forecasting of these phenomena. Few studies, however, have considered the intersection of these two topics.

This study utilizes self-organizing maps (SOMs) to identify statistically significant patterns of several daily climate indices preceding days of severe convection in the Southeast, at varying lead times and across different seasons. The fraction of storm type (nontornadic, weakly/significantly tornadic, and outbreak) for the days associated with these patterns are used to identify robust patterns associated with southeastern US tornado outbreak days, and the spatiotemporal distributions of their associated storm reports are explored and compared with non-outbreak severe days. Lastly, North American Regional Reanalysis (NARR) data is used to develop composites of relevant environmental variables coinciding with these outbreak patterns in order to provide a physical relationship between the overarching teleconnection phase and the observed severe weather. Not only does this study reaffirm several teleconnection patterns identified in existing literature (i.e. negative PNA, and by extension La Niña conditions, in spring months) as being related to tornado likelihood via a new method, but it discovers several new patterns such as the counterintuitive result of strongly negative Gulf of Mexico SST anomalies in fall months (for which we provide physical reasoning). Some of these patterns are shown to favor more traditional storm environments with increased CAPE and shear, while others lead to distinct HSLC storm environments. These results suggest that climate-scale variability can modulate and potentially be used to predict Southeast storm environments and the likelihood of tornadic outbreaks.