51. ENSO‐based probabilistic forecasts of March–May U.S. tornado and hail activity
- Author
-
John T. Allen, Michael K. Tippett, and Chiara Lepore
- Subjects
010504 meteorology & atmospheric sciences ,Severe weather ,Meteorology ,Southern oscillation ,Probabilistic logic ,Storm ,010502 geochemistry & geophysics ,01 natural sciences ,Geophysics ,El Niño Southern Oscillation ,Climatology ,Convective storm detection ,Thunderstorm ,General Earth and Planetary Sciences ,Environmental science ,Tornado ,0105 earth and related environmental sciences - Abstract
Extended Logistic Regression is used to predict March-May severe convective storm (SCS) activity based on the preceding December-February (DJF) 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 Nina-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.
- Published
- 2017
- Full Text
- View/download PDF