Our reading of data from GFZ Helmholtz Centre for Geosciences (geomagnetic Kp index). Spot an error? Tell us at hello@adatastory.com.
In the 1950s the Earth's magnetic field was stirred to storm level on about 116 days a year. In the 2010s it was on about 38. Every day since 1932 of the index aurora watchers check before they look up: how often the sky storms, in which months, and how this year compares.
Kp measures how hard the solar wind shakes the Earth's magnetic field, on a scale from 0 to 9. At 5 NOAA calls it a minor storm, and the northern lights reach further from the poles. In the 1950s Kp got there on 115.5 days a year, 31.6% of days. In the 2010s it was 37.7 days, 10.3%, and in 2020–2025, 47.5.
The strong storms fell further. Days reaching Kp 7 went from 19.8 a year in the 1950s to 2.7 in the 2010s. The index has drifted too: its makers estimate Kp reads about a third of a unit lower than in 1932, because the Earth's own field has slowly shifted. Raise the 1950s bar by that third and the decade still had 91.3 storm days a year, 2.4 times the 2010s.
The Sun's activity rises and falls over about 11 years, and the storms follow. Cycle 25 began in December 2019. In its first 2,508 days it brought 336 storm days. By the same point cycle 24 had 212, cycle 23 had 485, and cycle 19, which began in 1954, had 756.
Cycle 24 was the quietest in the record: 381 storm days in 11.0 years, 34.6 a year; at this point it was running at 30.9. Cycle 25 is running at 48.9 a year. Its busiest full year so far was 2025, with 97 storm days. Of the nine cycles since 1933, only cycle 24 had fewer storm days at this point.
Over 1932–2026, March, April, September and October had a storm on 26.8% of days; June, July, December and January on 16.8%. The months around the equinoxes are 1.6 times as stormy as those around the solstices, and they hold 48.4% of all the days that reached Kp 7, though they make up 33.4% of the calendar. A typical October brings about 7.9 storm days in its 31; a December, 4.6.
The rhythm holds in quieter times. Since 2016, 17.6% of September days and 16.7% of March days had a storm, against 8.4% of December days. The data shows when storms come, not why.
On any day since 1932 the chance of a storm was 21.5%. Storms bunch together, so 13 or 14 days after a storm day the chance is already 27.0% or 26.3%. At 27 days it jumps to 36.7%. Storms come back after about one turn of the Sun as seen from Earth.
That is why Julius Bartels, who introduced Kp in 1949, drew it in rows of 27 days, one row per solar rotation. GFZ still calls these "musical diagrams". Below is 2003, when 43.3% of days had a storm, in his layout: read down a column and storms return to the same place, rotation after rotation. Within 2003, 54.1% of storm days had another 27 days later.
Hover or tap a day for its values.