Empirical Relationship between U.S. Tornado Outbreak Probabilities and the Large-scale Environment

Tornado outbreaks – when six or more tornadoes occur within a short period of time – are rare yet impactful events. It would be advantageous to understand how tornado outbreak activity varies on the subseasonal-to-seasonal (S2S) timescale. We examined the question of whether conditions favorable for outbreaks could be distinguished from those favorable for non-outbreak tornadoes. We have developed a 6-hourly tornado outbreak index, which uses logistic regression to relate the probability of a tornado outbreak at every grid point over the U.S. to collocated values of convective precipitation, storm relative helicity (SRH), and convective available potential energy (CAPE). Storm report data from the Storm Prediction Center is used to train the model and evaluate its performance. We found that this tornado outbreak index represents the climatology, seasonal cycle, and interannual variability of tornado outbreak activity well. Overall, greatest skill of the index is exhibited over regions and seasons when tornado outbreaks occur most often, especially in the Tennessee River Valley and Southeast U.S. for late winter and spring. In terms of reliability, the index performs well for probability forecasts up to 10% but is overconfident for forecasts greater than 10%. We found that the El Niño-Southern Oscillation (ENSO) modulates the tornado outbreak index values via its influence on precipitation, SRH, and CAPE patterns. Therefore, ENSO may be a valuable source of S2S predictability for tornado outbreak activity.