An aurora visible over France never arrives unannounced. It is the final link in a chain of events that starts on the Sun, 150 million kilometres away, one to three days before the sky lights up. Each link in that chain comes with its own observation, its own time horizon and its own level of certainty. Understand the mechanics and you know exactly what to watch at 48 hours, at 24 hours, at 1 hour, and in the final minute. Here is the full journey, from solar flare to your doorstep.
Step 1 - The solar flare: the starting gun
Everything begins in an active region of the Sun, around a sunspot group where the magnetic field is twisted like an over-stretched elastic band. When that band snaps, the stored energy is released all at once: a solar flare, a flash of radiation sweeping the whole spectrum, from radio waves to X-rays.
Flares are ranked by their peak X-ray output: classes C (modest), M (medium) and X (major), each letter representing a factor of 10 in intensity. An M5 is five times stronger than an M1; an X2 is twenty times stronger than an M1. This radiation travels at the speed of light and reaches Earth in about 8 minutes, well ahead of any particles.
The essential point: a flare alone is not enough to produce an aurora. The X-ray flash disturbs the dayside ionosphere (HF radio blackouts), but it does not shake Earth's magnetic field. What you should watch for is what sometimes accompanies the flare: an ejection of matter. An X1 with no ejection will deliver nothing the following night; an M2 with a well-aimed ejection heading our way can put on a memorable show two days later.
Step 2 - The CME: the cannonball
A coronal mass ejection (CME) is a cloud of magnetised plasma, billions of tonnes of matter torn out of the solar corona and hurled into space. It is the CME, not the flare itself, that triggers the geomagnetic storms behind low-latitude aurorae.
We detect it with coronagraphs, space telescopes that block out the solar disc with an occulting disc to reveal the corona, like a permanent artificial eclipse. The historic reference instrument is LASCO, aboard the SOHO satellite. On the images, the CME appears as a bright bubble escaping the Sun, one to a few hours after the flare.
The detail that matters for us is geometry. If the CME shoots off to the left or right of the frame, it will miss Earth. If it appears as a halo expanding all around the occulting disc, it is coming straight at us (or straight away from us, which analysts settle by checking whether an active region was facing Earth). A halo CME launched from a central region of the disc is the best 48-hour warning signal there is.
Analysts also measure the CME's initial speed from successive coronagraph frames. But the measurement is uncertain: the CME is seen side-on or in projection, its shape is irregular, and its speed will change during the trip. At this stage we know a projectile has been fired and roughly when it will arrive - not yet how hard it will hit.
Step 3 - Propagation: a 1 to 3 day journey
The CME then has to cover 150 million kilometres. Depending on its initial speed - typically 500 to 3000 km/s - the trip takes anywhere from under 24 hours for the fastest monsters to three days for the slowest. Most geoeffective CMEs arrive 36 to 72 hours after the flare.
Along the way, the CME interacts with the ambient solar wind: a fast CME is slowed by the medium it ploughs through, a slow one gets a gentle push. Forecasters estimate the arrival time with propagation models: full numerical models of the WSA-Enlil type, which simulate the inner heliosphere in 3D, and simpler drag-based models that capture this progressive braking.
Honesty about precision is required: even the best models quote an arrival window of ± 6 to 12 hours. A CME due Saturday at 18:00 may arrive Saturday morning or at dawn on Sunday. That is why a 48-hour forecast remains a probability forecast, never a promise. Its job is to make you keep the evening free and keep watching, not to guarantee the show.
Step 4 - Lagrange point L1: the sentinel at Earth's gate
Some 1.5 million kilometres from Earth, in the direction of the Sun, sits Lagrange point L1, a gravitational balance point where sentinel satellites keep permanent watch. Historically that role has been played by missions such as ACE and later DSCOVR. These probes dip their instruments directly into the solar wind and measure, in real time, exactly what is about to strike Earth.
This is THE decisive dataset in the whole chain. No more estimates from coronagraph images: at L1 we measure the real speed, the real density and, above all, the real orientation of the cloud's magnetic field. The lead time is short - the solar wind covers those 1.5 million kilometres in 15 to 60 minutes depending on its speed - but the information is near certain. When the L1 curves jump, impact on the magnetosphere is imminent.
Remember the hierarchy: the coronagraph says "something is coming in 1 to 3 days", the L1 point says "here is exactly what hits in under an hour".
Step 5 - Bz, speed, density: the parameters that decide everything
Once the CME reaches L1, four numbers decide what happens next. The first one dominates all the others.
- Bz, the north-south component of the magnetic field. Earth's magnetic field points north. If the field carried by the CME points south (negative Bz), the two fields are antiparallel and can splice together: this is magnetic reconnection. Picture two magnets offered pole-to-opposite-pole: they latch on. The magnetosphere then opens up like a zip and solar wind energy pours in. Bz holding south below -10 nT: the door is open; below -20 nT: a severe storm is likely. Bz north? The door stays shut, and even a fast CME may produce almost nothing.
- Speed. The faster the solar wind hits, the more energy is transferred. 400 km/s is routine; above 600 km/s, every nT of southward Bz counts double.
- Density. Dense plasma (tens of particles per cm³) compresses the magnetosphere and amplifies the initial shock.
- Bt, the total field strength. A high Bt (20, 30, 40 nT) signals a powerful magnetic cloud: if its orientation swings south, the energy reserve is enormous. Strong Bt with Bz still north is a situation to watch very closely.
This is why an aurora night is so often decided on the live L1 curves: a rotation of Bz from north to south can turn a quiet evening into a storm within half an hour.
Step 6 - From shock to aurora: substorms, Kp and hemispheric power
When the shock reaches the magnetosphere, the injected energy piles up in Earth's magnetotail, then discharges in episodes called substorms: these produce the sudden brightenings of the aurora, in waves lasting 30 minutes to 2 hours. A storm night is never uniformly active: it alternates spectacular peaks and lulls.
The best-known yardstick for all of this is the Kp index. An important reminder: official Kp is computed over 3-hour windows and is retrospective - it describes what has just happened, not what is happening this second. To steer a live observation, minute-by-minute nowcast estimates are preferred, derived continuously from L1 measurements and ground magnetometers. We wrote a full practical guide to the Kp index, with thresholds per French city.
Another precious indicator is hemispheric power, expressed in gigawatts, which estimates the total energy dumped by particle precipitation over one hemisphere. In quiet times it runs at 10 to 20 GW; above 50 GW the auroral oval is well fed; above 100 GW a major storm is under way and the oval can push down towards French latitudes.
Step 7 - The aurora chaser's checklist for France
Here is how all of this translates into practice, horizon by horizon.
At 48 h: the coronagraph signal
- An M- or X-class flare fired from a region facing Earth, followed by a halo CME on coronagraph imagery.
- Propagation models publish an arrival window: note it down, with its ± 6-12 h margin.
- In the Pulsar app: flare alerts warn you from class M upward, and the 3-day Kp forecast folds in the estimated arrival.
At 24 h: consolidate the plan
- Check the cloud forecast for your area and scout a clear northern horizon, away from lights.
- Check the lunar phase: a full Moon dims the visual show considerably.
- In the app: the 3-hour-bin forecast highlights the most likely windows for your city.
At 1 h: the L1 curves
- The shock shows up at L1: a simultaneous jump in speed, density and Bt.
- Watch Bz: if it dives south (-10 nT or lower) with high speed, get ready to head out.
- In the app: real-time solar wind charts and threshold push alerts do this monitoring for you.
In real time: the trigger
- Live Kp by the minute climbs, hemispheric power tops 50-100 GW: a substorm is under way.
- Go outside, give your eyes 15 minutes to adapt, shoot the northern horizon with long exposures: the sensor sees the aurora before the eye does.
- In the app: live Kp, the auroral oval map and the tonight page tailored to your city tell you whether this is the moment.
What to remember
Forecasting an aurora means following a relay race: the flare's flash in 8 minutes, the halo CME image within a few hours, the 1 to 3 day journey refined by models, then the raw truth measured at L1 less than an hour before impact, and finally the substorm discharges that live Kp and hemispheric power translate minute by minute. No single link is enough on its own; it is their sequence that turns a sunspot into green curtains above the French horizon. The chain is public and the data is open - all that is missing is your vigilance, and a clear sky.
Read more
- Understanding the solar wind - the medium the CME travels through.
- Understanding solar storms - flares, CMEs and their impacts.
- Understanding the Kp index - full illustrated explainer.
- Kp index: the practical guide - thresholds per French city.
- Tonight's forecast by city - the whole chain, condensed into one verdict.