The data sources behind the dashboard
SolarStormAlert compiles its picture of near-Earth space weather from four public feeds. This page goes deeper than the dashboard's citation pills: what each source measures, what it's uniquely good at, where it falls short, and why we poll it on the cadence we do.
NOAA SWPC — Kp index and storm alerts
NOAA's Space Weather Prediction Center compiles the planetary Kp index — a 0–9 measure of global geomagnetic activity — from a worldwide network of ground-based magnetometers, and issues the official Watch/Warning/Alert bulletins the dashboard relays. It is the authoritative source for the G-scale rating that anchors the dashboard's severity language; see the G/R/S scales for what each level means.
We poll it every 15 minutes. Observed Kp is itself a 3-hour planetary index, so a faster poll would not produce a faster number — the fifteen-minute cadence is ample headroom against a value that only changes every three hours. A sub-three-hour Kp reading isn't something any of our current sources can provide.
NASA GOES — X-ray and proton flux
NASA's GOES satellites, in geostationary orbit, carry the instruments behind two of our feeds: X-ray flux, used to classify solar flares by intensity, and proton flux, the channel behind the dashboard's radiation-storm reading. GOES is a joint NOAA/NASA program; NOAA operates the satellites once they're in orbit. This is the only source in the dashboard for real-time flare classification, and the same X-ray data also feeds the dashboard's radio-blackout reading.
We poll it every 15 minutes, the same cadence as SWPC — flux data updates faster upstream than Kp, but it isn't the fastest-moving signal we track; that's reserved for the solar wind feed below.
ACE/DSCOVR (at L1) — solar wind
Spacecraft stationed at the L1 Lagrange point, about 1.5 million kilometers upstream of Earth toward the Sun, measure solar wind speed, density, and Bz — the north-south component of the interplanetary magnetic field. Because L1 sits upstream, it is the only source that can lead a geomagnetic event: everything else we poll reports on conditions already underway, or on forecasts built from earlier data.
We poll it every 2 minutes, by far our fastest cadence, tuned against the roughly 15–60 minute transit time from L1 to Earth.
Honestly: the L1 fleet has drifted since this feed was first built. DSCOVR is currently absent from NOAA's live stream; the spacecraft actually reporting today are ACE, IMAP, and SWFO-L1. IMAP's density readings run five to six times higher than ACE and SWFO-L1 during the same period, so we never average density across the three — we use NOAA's own selected-spacecraft flag instead. Speed and magnetic-field readings agree well across all three. We keep "ACE/DSCOVR (at L1)" as the label because it matches what NOAA's own feed is still called, even though DSCOVR itself isn't currently in the loop. Separately, the plasma sensors (speed and density) have intermittent dropouts distinct from the feed going stale — when a field is missing, the solar wind panel names exactly which one rather than showing a guessed number.
NASA DONKI — coronal mass ejections
NASA's DONKI service is a curated notification feed, not a single instrument: it aggregates NOAA-forecaster assessments of which coronal mass ejections are Earth-directed, with a modeled arrival window rather than an exact time.
We poll it hourly, which is well inside DONKI's own update rhythm — CMEs take one to several days to arrive, and DONKI's assessments only change as NASA forecasters issue or revise them. DONKI is enrichment, not part of the alert spine: it never drives an alert tier on its own, and a real multi-day DONKI outage proved the point — the CME panel correctly showed "unavailable" rather than a false healthy reading.
How these sources work together
Each source is polled on the cadence its role needs, not a uniform schedule. Along the early-warning path: an early heads-up comes from a spike in NOAA's Kp forecast; a firmer watch relays NOAA's own official product; a warning can come from the L1 solar wind feed's own shock detection — the fastest and most leading signal we have; and a confirmed alert follows the observed Kp value once NOAA publishes it. NASA DONKI sits outside this ladder entirely: it enriches the picture with CME context but never triggers a tier on its own.
Whichever upstream sensor sees a change first is the one that drives the read at that moment — that's why the polling cadences differ by source instead of matching a single interval.