The subseasonal timescale, residing between the time scales of weather forecasts and seasonal outlooks, has long been considered a “predictability desert.” Prediction on this timescale is important for decision makers in a variety of sectors. While most research on subseasonal prediction has focused on temperature and precipitation, emergency management would greatly benefit from skillful prediction of tornado activity, which may lead to a high number of fatalities and property damage but remains a challenge for dynamic models.

In a paper recently published in the journal Geophysical Research Letters, Atmospheric Sciences graduate student Doug Miller and collaborators analyzed the relationship between tornado activity and weather regimes (i.e., the large‐scale, recurrent weather patterns) over the United States. They showed that weather regimes strongly modulate tornado activity and that persisting weather patterns (those lasting longer than 3-days) contribute to greater than 70% of tornado outbreak days. Based on the observational analysis, they developed a hybrid model to predict tornado activity on weekly timescales, and the model demonstrates skill out to week 3, which is an improvement of the current forecasting capability.

The paper can be found at https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2020GL087253