Skillful forecasts of springtime CONUS tornado activity up to a year in advance / Prof. Kai-Chih Tseng (Department of Atmospheric Science, National Taiwan University) [曾開治 教授 (台大大氣)]

Abstract: Despite recent progress in the seasonal forecasting of continental United States (CONUS) severe thunderstorm activity, including severe hail storms and tornadoes, the prospects of developing operational seasonal severe weather outlooks remain uncertain due to intrinsic predictability limits and the correspondingly limited skill demonstrated to date. While many studies have demonstrated a connection between tropical sea surface temperature, particularly from the El Niño-Southern Oscillation (ENSO), and boreal spring CONUS tornado activity, it is still unclear if there are additional sources of reliable skill for seasonal CONUS tornado predictions. If such sources exist, we also must determine whether state-of-the-art dynamical seasonal forecast models can capitalize on such sources to produce skillful seasonal tornado forecasts at long lead times. Here, we develop a dynamical-statistical model for March-April CONUS tornado activity with the use of a global climate model recently developed at the Geophysical Fluid Dynamics Laboratory. We demonstrate the potential of leveraging the leading modes of global sea surface temperature (SST) for skillful seasonal tornado forecasts with a lead time of 9-11 months, with skill that surpasses models that only use ENSO as a predictor. We further perform an optimal pattern analysis to support that surface high pressure anomalies over the Caribbean Sea associated with ENSO and the long-term SST trend are the dominant predictability sources. These findings advance the potential for developing operational outlooks for CONUS springtime tornado activity.