A Global Ensemble Forecast System (GEFS)-based synthetic event set of U.S. tornado outbreaks

Abstract

Severe convective storms (SCS) are important drivers of global insured losses, and tornado outbreaks – when many tornadoes occur within a short time span – cause extreme and localized loss of life and property. Tornado outbreak risk estimates from observations, either storm reports or reanalysis environments, are limited by meteorological conditions that have occurred in the historical period. A standard approach of addressing this inadequacy is to construct synthetic event sets that consist of unrealized but plausible events that better represent the full range of possible outcomes. In this study, we constructed and evaluated a synthetic event set of U.S. tornado outbreaks using Global Ensemble Forecast System (GEFS) environments and a tornado outbreak index. With over 800 000 daily maps of environments, over 200 000 synthetic events are generated. In a seamless framework, the synthetic event set includes “daughter events”, constructed from short-lead forecasts and resemble historical events, as well as independent physically plausible events, constructed from longer-lead forecasts. With the GEFS synthetic event set, we estimated that the 1-in-100-year and 1-in-1000-year U.S. tornado outbreak event has 150–250 and 275–400 (F/EF1+) tornadoes per day, respectively. The GEFS synthetic event set also shows robust shifts related to ENSO – higher outbreak activity during La Niña conditions – and trends – increased outbreak activity during 2010–2019 compared to 2000–2009 – consistent with reports. We also developed a subsampling procedure to estimate locally specific tornado outbreak risk, which we illustrate by generating return level curves for grid cells that cover Dallas, Nashville, and Chicago.