Tornadoes and hail by ENSO phase.
In 2015 Allen, Tippett and Sobel (Nature Geoscience) found that the southern U.S. sees fewer tornadoes and hailstorms in winter and spring under El Niño, and more under La Niña: El Niño's strong southern jet stream steers storms along the Gulf coast and cuts the warm, humid air the storms need. This route tests that against every January–May since 1950 in the Storm Prediction Center's record, and looks at the spring after a strong El Niño, when the Pacific often flips to La Niña.
Every southern spring, sorted by the winter before
Seventy-six springs in the South
State by state
The spring that ends a strong El Niño, and the spring after
Does the answer depend on the choices?
How this is counted
Why EF1 and stronger. The number of tornadoes reported each year nearly tripled between the 1950s and the 2000s, and most of the rise was in the weakest (F0/EF0) class: about 100 a year in the 1950s, about 800 in the 2000s, as Doppler radar, storm spotters, roads and phones spread. Tornadoes rated EF1 or stronger leave damage that was hard to miss in any decade, so their record is far steadier (between about 370 and 660 a year, decade by decade). Unrated tornadoes (EFU) are left out. The same check with EF2+ is in the table above.
The South is Texas, Oklahoma, Arkansas, Louisiana, Mississippi, Alabama, Tennessee and Georgia, chosen before looking at the data as the southern Plains and "Dixie Alley" where Allen and colleagues found the signal. Florida is shown on its own because El Niño's jet is known to bring it more winter tornadoes. A tornado counts in the state where it touched down.
ENSO phase of a spring is the December–February Oceanic Niño Index of the winter that ends it (El Niño ≥ +0.5 °C, La Niña ≤ −0.5 °C); a spring is labelled by its calendar year. Normal is a straight-line trend fitted through the springs after neutral winters only, and each spring is shown as a percentage of that line. On the map, every state shares the national trend shape, scaled to its own neutral-spring average; states averaging fewer than one EF1+ tornado a spring are greyed out. Each state's El Niño (or La Niña) departure is tested by shuffling the winter labels 2,000 times, and the Benjamini–Hochberg procedure controls false discoveries across the map (q = 0.10).
Hail is counted as days with at least one report of hail 2 inches or larger in the South, 1979–2025. Hail reports have grown several-fold since 1979, so hail is detrended in logarithms and should be read as a rough check, not a measurement.