Using Teleconnection Indices to Predict Seasonal Tornado Outbreak Frequency

While atmospheric teleconnections have been associated with extreme weather events globally, little research has considered their potential influence on US tornado outbreak frequency. The goal of this study is to improve seasonal tornado outbreak forecasting by creating a statistical model that forecasts tornado outbreak frequency in the US using teleconnection indices as predictors. For this study, a tornado outbreak is defined as 6 or more tornado reports associated with a single synoptic system and an event N15 rating index of 0.5 or higher. The predictive scheme will utilize a method known as support vector machines (SVM) and will be trained with teleconnection indices from 500 mb and 1000 mb geopotential height fields from 1960–2011, as well as monthly ENSO Niño 3.4 SST anomaly data. The tornado outbreak season is confined to all months after March for a given calendar year. Monthly/Seasonal teleconnection indices are derived from a rotated principal component analysis (RPCA) of the geopotential height fields. The results from this work reveal teleconnection indices that can be used to provide seasonal predictability for tornado outbreaks in the US. The outcome of this study could potentially allow forecasters the ability to predict tornado outbreak potential on a climatological scale with months of lead-time, allowing for better preparation strategies for tornado outbreak seasons.