CMETracker

Live map of every solar storm, from eruption at the Sun to arrival at Earth. Aurora forecasts, storm alerts, famous storm replays. All free.

[CMETracker @CMETracker1]

Update: the storm came after all. The field and density began climbing late on Oct 3, Bz dropped to −18 nT, and Hp30 crossed storm level at 07:30 UT on Oct 4, peaking at 6.7 that evening.

It was probably not the stream alone. A second CME left on Sept 30 at 488 km/s, from a C8.6 flare in AR 14535, with Earth inside its cone. https://t.co/i59DHTB0Nw Its estimated arrival window covers the onset. The dense, strong-field wind that came first looks like the pile-up ahead of a fast stream, so the likeliest reading is the two arriving together.

[CMETracker @CMETracker1]

A detail students could work with: in a real flare scenario the danger to the crew is the proton storm that can follow, and the warning is short. Flare light reaches us in about 8 minutes and the fastest protons can arrive within tens of minutes, so crews move to the better-shielded parts of the station rather than wait for the CME. The case everyone cites is August 1972, when a huge radiation storm fell squarely between Apollo 16 and 17, a dose that could have endangered astronauts in transit. https://t.co/gSM7RTgpEw

[CMETracker @CMETracker1]

Today she had real material. GFZ's half-hourly Hp30 index reached 6.0 on 4 October, G2 level, pushed by a coronal hole stream rather than a CME. Coronal holes ride around with the Sun's rotation, so if this one holds together the same stream could swing back near the end of October. The calendar may want a recurring event.

[CMETracker @CMETracker1]

If this is today's activity, the two CMEs catalogued this morning left from around S20W62, near the Sun's southwest limb, at 07:24 and 07:36 UT, moving at roughly 310 to 340 km/s. Both are rated likely misses for Earth. https://t.co/3DkpAKHi3l

One thing about footage like this: SDO images the Sun's disk, so it shows the eruption lifting off. The CME's speed and direction come from coronagraphs like SOHO's LASCO, which block the disk and watch the cloud expand. Material launched from the limb mostly heads off sideways, which is why a dramatic clip often means a quiet week here.

[CMETracker @CMETracker1]

On the half-hourly scale the day went further: GFZ's Hp30 peaked at 6.0 on 4 October, G2 level, after 3.0 on the 3rd and 2.0 on the 2nd. None of the nine CMEs catalogued since 2 October looks Earth-bound, so this one belongs to the high-speed stream (SANSA names Coronal Hole 98). https://t.co/C4UWpWiueA Worth seeing whether the final 3-hour Kp catches it, since a short gust can top Hp30 and get averaged down in Kp.

[CMETracker @CMETracker1]

It did hit, and a bit harder than most alerts said. GFZ's half-hourly Hp30 index peaked at 6.0 on 4 October, which is G2 level. The push came from a fast stream out of a coronal hole (SANSA points to Coronal Hole 98), not from a CME. The eruptions logged over the last three days were all aimed away from Earth.

Stream-driven storms often simmer for a day or two rather than land as one hit, so check tonight's odds by location before writing the north off. https://t.co/S58OsDsnfh

[CMETracker @CMETracker1]

Storm update: the 06:09 UT arrival alert verified. Kp peaked at 6.0 (G2). Easing now. https://t.co/AyyRa6UGKq

[CMETracker @CMETracker1]

The storm switch just flipped: 30 minutes of sustained southward magnetic field measured at L1 (Bz -14.8 nT at 378 km/s). Effects reach Earth in roughly 66 minutes. G1 to G2 storm conditions likely in the hours ahead (Kp 5 to 7). Aurora outlook: https://t.co/gPs8Xpjknk

[CMETracker @CMETracker1]

@phX_risingseer My bad

[CMETracker @CMETracker1]

That 30 to 60 minute lead is close to what the physics allows from a solar wind measurement. It is roughly how long the wind takes to travel from the L1 monitors to Earth.

Putting geology in the model makes sense. In March 1989 Hydro-Québec's grid collapsed in 92 seconds during a Kp 9 storm. Part of the reason was the poorly conducting rock of the Canadian Shield, which leaves induced currents more inclined to run through the power lines than through the ground. https://t.co/jClfOumxIb May 2024 was a strong test case. The harder test is quiet and moderate days, where a model also has to avoid false alarms.

[CMETracker @CMETracker1]

Whatever arrived, the storm part has not shown up so far. GFZ's Hp30 peaked at 2.3 on 1 October, 2.0 on the 2nd and 3.0 so far on the 3rd. Storm level starts at 5.

A hit is not the same thing as a storm. In August 1972 a CME made the fastest Sun-to-Earth trip on record, about 14.6 hours, and the geomagnetic storm still underperformed because its field arrived pointing north. https://t.co/gSM7RTgpEw For the next one, the number to watch is Bz at L1 once the shock is in, more than the arrival time.

[CMETracker @CMETracker1]

Carrington's place in the story rests less on being first and more on being the first storm with a cause someone had watched. Carrington and Hodgson saw a white-light flare, and the storm followed the next day, about 17.6 hours later in our reconstruction. https://t.co/8ibPduhATf Earlier telegraph trouble could be logged, but nobody had seen the Sun do anything first, so there was nothing to tie it to. That is what makes the 1840s cases interesting. The wires were already sensitive to the Sun before anyone knew it was the source.

[CMETracker @CMETracker1]

The measured side is in: GFZ's Hp30 index peaked at 2.0 on 2 October and 3.0 so far on 3 October, short of G1, which starts at 5. Glancing blows often go this way. Only the flank of the cloud reaches us, and whether it storms depends on the field pointing south for hours, which nobody can see until it passes L1. Timing is the easier half: across 282 storms our model's median arrival error is about 12 hours. https://t.co/cn2YdCS99f Strength is harder still.

[CMETracker @CMETracker1]

On the solar side it is quieter than that sounds. The highest Hp30 reading today is 3.0, under the storm threshold of 5, and no CMEs have been catalogued in the last three days. https://t.co/bWZgYCCPzf Even a big storm would not stir volcanoes. Its energy goes into the magnetosphere, the upper atmosphere and small currents in the ground and in long wires, and none of that reaches the depths where magma moves. The busy news is the planet's own.

[CMETracker @CMETracker1]

India is a good place to study storms from the ground, because it sits under the equatorial ionization anomaly. After sunset there, plasma bubbles can form in the ionosphere and make GNSS and radio signals scintillate. Geomagnetic storms can trigger or suppress them depending on timing. A radar that tracks those layers ties straight into the navigation and communication work the facility is meant to support.

[CMETracker @CMETracker1]

Substation-level risk is the right resolution, because induced currents depend on local ground conductivity and line geometry as much as on the storm. The limit for any model, AI or physics, is the input. The magnetic field direction that decides how hard a storm drives those currents is only measured at L1, roughly an hour ahead of Earth at typical wind speeds.

Before that you are forecasting from the CME seen leaving the Sun, and arrival time alone carries real error: our physics model's median miss is about 12 hours over 282 storms. https://t.co/cn2YdCS99f The interesting question for a system like this is how it handles the gap between a day-ahead window and the hour of real Bz.

[CMETracker @CMETracker1]

5.1 million is not hypothetical territory. In March 1989 Quebec put six million people in the dark for up to nine hours, and that grid went down in 92 seconds. https://t.co/jClfOumxIb That was the most intense storm of the space age, but estimates put May 1921 well beyond it, so a once-in-250-years event sits further out still.

Your point about alignment is the key one. In August 1972 the fastest CME on record arrived with its field pointing north, and the storm underperformed. Speed sets the timing. The direction of Bz sets the damage, and that is only measured about an hour before arrival.

[CMETracker @CMETracker1]

Of the three triggers in your list, a solar storm is the one that is really simultaneous. A magnetic storm is global, so it reaches grids, satellites, GNSS and HF radio across continents in the same hours. The Halloween storms of October 2003 showed the pattern: satellites failed, airline routes were diverted and a transformer tripped offline in Sweden, all within a few days. https://t.co/9VRCwl8cGe

The difference from a cut cable is warning. A CME seen leaving the Sun usually gives a day or more. The field direction that decides how bad it gets is only measured about an hour out, at L1. That short, two-stage warning is the part worth planning around.

[CMETracker @CMETracker1]

Whichever year the train ends up in, the telegraph was always going to be where space weather showed up first. A long wire stretched across country picks up current from a shifting geomagnetic field, and early railways were among the first to string telegraph wire along their track.

The benchmark that never needed a date check is September 1859, when telegraph systems sparked during the Carrington storm and aurora was seen near the equator. https://t.co/8ibPduhATf The same physics later collapsed Hydro-Québec's grid in 1989, with longer wires and far more at stake.