Premium
ENSO‐based probabilistic forecasts of March–May U.S. tornado and hail activity
Author(s) -
Lepore Chiara,
Tippett Michael K.,
Allen John T.
Publication year - 2017
Publication title -
geophysical research letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.007
H-Index - 273
eISSN - 1944-8007
pISSN - 0094-8276
DOI - 10.1002/2017gl074781
Subject(s) - tornado , climatology , storm , el niño southern oscillation , environmental science , severe weather , forecast skill , probabilistic logic , supercell , meteorology , convective storm detection , convection , atmospheric sciences , geography , mathematics , statistics , geology
Extended logistic regression is used to predict March–May severe convective storm (SCS) activity based on the preceding December–February (DJF) El Niño–Southern Oscillation (ENSO) state. The spatially resolved probabilistic forecasts are verified against U.S. tornado counts, hail events, and two environmental indices for severe convection. The cross‐validated skill is positive for roughly a quarter of the U.S. Overall, indices are predicted with more skill than are storm reports, and hail events are predicted with more skill than tornado counts. Skill is higher in the cool phase of ENSO (La Niña like) when overall SCS activity is higher. SCS forecasts based on the predicted DJF ENSO state from coupled dynamical models initialized in October of the previous year extend the lead time with only a modest reduction in skill compared to forecasts based on the observed DJF ENSO state.