← the great storms · the live sky

THE HALLOWEEN STORMS · 2003

For two weeks in late October 2003, the Sun produced one of the most violent activity sequences of the space age — a parade of X-class flares and two back-to-back superstorms that blacked out a Swedish city and rerouted polar aviation. Until May 2024, this was the modern benchmark.

▶ Replay this storm in 3D. A simulation driven by the actual measured geomagnetic record of the storm — the AE and Dst indices from October 2003 drive the model's substorms and oval.

What happened

Two giant active regions crossed the disk in late October 2003, firing off X-class flares almost daily. On October 28 an X17 flare launched a CME that crossed to Earth in under a day; on November 4 came the biggest flare ever measured — it saturated the GOES X-ray sensors at X28, with ionospheric analysis placing it near X45 ± 51.

The storms of October 29–31 reached G54, with Dst bottoming near −353 and −383 nT on consecutive nights3. Geomagnetically induced currents tripped the grid in Malmö, Sweden, blacking out about 50,000 customers2; airlines rerouted polar flights; satellite operators reported anomalies fleet-wide. Aurora was photographed from Texas and the Mediterranean.

The simulation

The observatory renders Halloween as a measured-data simulation: the actual minute-scale geomagnetic record from October 2003 — every substorm onset in the AE index, the real Dst descent — drives the model's oval and substorm cycle5. It's labeled a simulation because no auroral-imaging satellite fleet of the 2024 kind existed to fit against; the driving, though, is the storm's own measured record.

advertisement

Two storms in three days

The Halloween sequence was not one storm but a barrage. Sunspot region 486 — one of the largest of the space age — fired an X17-class flare on October 28, 2003, with the CME leaving the Sun almost directly Earthward at extreme speed. It crossed the Sun–Earth distance in roughly nineteen hours, one of the fastest transits ever recorded, and drove a G5 extreme storm on October 29. Before that storm could recover, region 486 fired again: an X10 flare on October 29 whose CME arrived on October 30 and drove a second G5. Days later, on November 4, the same region delivered a flare so powerful it saturated the GOES X-ray sensors — off the top of the scale, estimated near X28 or beyond, still the strongest directly-observed flare of the satellite era.

What it did on the ground

Halloween 2003 remains the modern benchmark for space-weather consequences. A power blackout struck Malmö, Sweden when grid protection tripped; transformers in South Africa suffered damage that took units permanently offline. Airlines rerouted polar flights to keep radiation exposure and communication loss within limits. Dozens of satellites reported anomalies, instruments were safed across the fleet, and astronauts aboard the International Space Station sheltered in the most shielded parts of the ship. And through all of it, aurora was seen from Florida, Texas and Mediterranean Europe — red skies at latitudes where most witnesses had never considered the aurora possible.

The storm that blinded its own monitors

There is a reason this scene is labeled a simulation rather than a measured reconstruction, and it is one of the most honest facts in space weather: the storm was so intense that key solar-wind instruments saturated. The plasma sensor on the upstream ACE spacecraft could not return reliable speed and density through the worst hours — the very measurements that drive a reconstruction were blinded by the event they were built to measure. No amount of archival work recovers data that was never taken.

What does survive is the ground record: magnetometer networks worldwide measured the storm's full fury, minute by minute. The observatory's Halloween scene is driven by those measured geomagnetic indices — the storm-time ring-current index and the five-minute auroral-electrojet record — so every substorm intensification in the replay happens when the real one did, at the measured strength. The oval's placement and structure are then drawn by the same physics relations validated on the photo-anchored 2024 storms. Measured timing, modeled appearance, labeled as exactly that.

The sequence's legacy runs through this site in one more way: it is the canonical example of why the "replay any night" archive cannot reach every famous storm. For the greatest events, the monitors themselves are casualties — and the hand-built scenes in this storm library are the honest way back to them.

Frequently asked questions

How strong were the Halloween storms of 2003?
Two back-to-back G5 superstorms on October 29–31, with Dst near −353 and −383 nT3, driven by some of the fastest CMEs of the space age.
What was the biggest solar flare ever recorded?
The November 4, 2003 flare, part of the Halloween sequence — it saturated the GOES sensors at X28, and ionospheric measurements place its true size near X45 ± 51.
What damage did the Halloween storms cause?
Geomagnetically induced currents blacked out Malmö, Sweden2; polar flights were rerouted; dozens of satellites reported anomalies and one was lost. The full infrastructure story is on when the sky reaches the ground.

Sources

Numbered references for the claims above.

  1. Thomson, N. R., Rodger, C. J. & Dowden, R. L. (2004). Ionosphere gives size of greatest solar flare. Geophys. Res. Lett. 31, L06803. doi:10.1029/2003GL019345
  2. Pulkkinen, A., Lindahl, S., Viljanen, A. & Pirjola, R. (2005). Geomagnetic storm of 29–31 October 2003: geomagnetically induced currents and their relation to problems in the Swedish high-voltage power transmission system. Space Weather 3, S08C03. doi:10.1029/2004SW000123
  3. World Data Center for Geomagnetism, Kyoto — final/provisional Dst index. wdc.kugi.kyoto-u.ac.jp
  4. NOAA Space Weather Prediction Center — NOAA Space Weather Scales (G-scale). swpc.noaa.gov
  5. The Observatory’s own reconstruction pipeline and validation gates are documented on data & methods.
📖 Improve this page — spotted an error, or want a topic covered?
✍️ suggest a correction 💡 request a topic

By The Space Weather Observatory · checked against the data & methods record · Updated August 21, 2026

the sky · the sun · field guide · the great storms · data & methods · about · privacy · pro · ☁️ seasonmap · ☕ support