How active is the Sun, and how strong have space weather storms been?

Steps

1. NOAA SWPC JSON: observed-solar-cycle-indices, predicted-solar-cycle, noaa-planetary-k-index and the daily solar data text; GFZ Potsdam Kp_ap_Ap_SN_F107_since_1932.txt (one file, 5.5 MB). Annual F10.7 is the mean of GFZ's adjusted daily flux (days with a value). A storm day is a UT day whose highest 3-hour Kp falls in a G level (G1 Kp 5 up to G5 Kp 9; 5- is 4.667, 8- is 7.667, 9- is 8.667 and counts as G4, 9o is G5).
2. Radio flux is used for the long record and for the comparison with NOAA's outlook because observation and outlook share one scale and both licences are clear. NOAA's own sunspot count is a different count from the international number the 2019 panel predicted, so the page compares the count only with itself. The record of storms is the GFZ Kp index, which starts in 1932.
3. Cycle comparison: monthly means of the GFZ adjusted flux, smoothed with the standard 13-month window, and the maximum in a window around each cycle. The NOAA outlook is the SWPC JSON as published; nothing is refit here.

80 yearly F10.7 rows, 357 monthly NOAA sunspot rows, 144 monthly rows for cycle 25 and the outlook, 95 storm years from 94.8 years of Kp, 12 strongest days. Unit: F10.7 in solar flux units (1 sfu = 10^-22 W m^-2 Hz^-1); sunspot count as NOAA's own number; Kp from 0 to 9 in thirds; Ap in the index's own units (multiply by 2 nT for an average disturbance); storms by NOAA G-scale from the highest 3-hour Kp of a UT day. Every derived figure on this page (changes, indexes, averages, counts) is computed from the published rows, and the table under each chart says how.

Checks

• The GFZ file is read whole: 34,615 consecutive days from 1932-01-01; its licence line (CC BY 4.0, sunspot numbers excepted) and citation are read and checked by the script.
• The GFZ monthly F10.7 and NOAA's monthly F10.7 (Penticton) agree: mean difference 0.53 sfu over 260 months, 96.9% within 3 sfu.
• Storm days add up; May 11, 2024 reaches G5 and the October 28 to November 1, 2003 storms reach G5 in the Kp data; the SWPC pages for May 2024, the NCEI page for 2003 and the NASA record for 1859 are fetched and their wording checked; the NOAA scale rows (G1 to G5, Kp 5 to 9, frequency per cycle) are parsed from SWPC's page.
• SILSO's licence page (CC BY-NC) and NOAA's field description (ssn is the S.I.D.C. number) are read and checked, which is why the long sunspot series is left out.
• The 2019 panel prediction (115 in July 2025, range 105 to 125) is read from SWPC's Solar Cycle 25 page and checked word for word.

Caveats

• The longest sunspot record (since 1749) is the international number, whose licence (CC BY-NC) does not clearly allow reuse here, so the sunspot chart starts in 1997 with NOAA's own count. Kp is a 3-hour index, so a storm day is a day with at least one such period at the level; the same storm can span two UT days. Values for the latest months are preliminary (NOAA: the newest 6 months; GFZ: the most recent days are nowcast). Kp before 1932 does not exist: storms such as the 1859 Carrington event are known from other records, with estimates that vary.
• Radio flux and sunspot count measure the activity of the Sun in different ways; neither says by itself how likely a storm at Earth is.
• A storm 'day' is a UT day, and the same storm can be counted on two days; Kp level G5 on a day does not give the storm's peak strength. Ap is a daily average of the planetary amplitude.

Releases and revisions

• NOAA updates the indices monthly and the outlook is refit monthly; GFZ updates Kp daily. Re-run space_solar_activity.py to refresh; --layer refreshes only the small file for the System and Now pages.
• NOAA treats the newest six months of the monthly indices as preliminary and GFZ replaces nowcast Kp with definitive values within weeks.