What Is a CME? Coronal Mass Ejections Explained

Almost every truly major aurora display — the kind visible from London, Berlin, or the northern United States — traces back to one specific event on the Sun: a coronal mass ejection, or CME. Understanding what a CME actually is, and why forecasting its arrival is so difficult, explains a lot about why aurora predictions carry so much uncertainty even a day or two out.

What a CME actually is

The Sun’s outer atmosphere, the corona, is threaded through with magnetic field lines that loop and twist as the Sun rotates and its internal dynamo churns. Occasionally, a region of that magnetic field becomes so twisted and unstable that it snaps and reconnects, hurling a huge bubble of plasma — billions of tonnes of solar material, embedded with its own tangled magnetic field — out into space at speeds ranging from a few hundred to several thousand kilometres per second.

That expanding bubble is the coronal mass ejection. It is a physically real cloud of material, distinct from the light and radiation released in a solar flare (a separate, faster phenomenon covered in the FAQ above), and it takes real time to cross the roughly 150 million kilometres between the Sun and Earth. A large CME can carry a mass on the order of a billion tonnes or more of solar plasma — for scale, that’s roughly comparable to the mass of a small mountain, flung outward at a meaningful fraction of the speed at which the International Space Station orbits Earth, though CME speed is still only a tiny fraction of the speed of light.

How CMEs are actually observed

CMEs are detected using coronagraphs — instruments that place an artificial disc over the Sun’s bright surface, blocking direct sunlight so the much fainter corona around it becomes visible. Historically, the LASCO instrument aboard the SOHO spacecraft (positioned near the L1 Lagrange point, the same general region used for solar wind monitoring) has provided the longest continuous coronagraph record; newer instruments such as CCOR aboard GOES satellites have added additional, more frequently updated coverage. When a CME appears in these images, it often shows a recognisable three-part structure: a bright, curved leading edge where the ejection is compressing and piling up the ambient solar wind ahead of it, a darker, lower-density cavity immediately behind that edge, and a brighter core within that cavity, corresponding to a dense filament or prominence of solar material being carried along by the eruption. Not every CME displays this structure cleanly, but it’s common enough to be the standard reference pattern forecasters look for.

A particularly important visual category is the halo CME: viewed face-on from a coronagraph pointed at the Sun, an eruption travelling either directly toward or directly away from Earth appears to expand outward as a full ring, or halo, around the blocked-out disc, rather than as an asymmetric plume heading off to one side. A halo CME is an immediate signal that an eruption may be Earth-directed — though whether it’s heading toward us or away from us (a rare “back-side” halo, associated with an eruption on the far side of the Sun as it currently faces Earth) has to be resolved using other data, including which side of the Sun the originating active region was on and, once available, images from spacecraft at different vantage points around the Sun.

Why some CMEs matter far more than others

Three factors decide whether a given CME turns into a major aurora event, and all three are genuinely difficult to know precisely until the CME is almost here:

Speed. Faster CMEs arrive sooner and hit with more energy, compressing Earth’s magnetosphere more violently and driving stronger geomagnetic disturbance. Recorded CME speeds span an enormous range — from a few hundred km/s for a slow, unremarkable ejection up to more than 2,500-3,000 km/s for the most extreme events ever measured, which is fast enough to cross the Sun-Earth distance in well under a day.

Direction. A CME has to be aimed roughly at Earth to matter. Many erupt from parts of the Sun facing away from us, or at an angle that carries them past Earth entirely — these can still be tracked and studied, but they produce no aurora here. Even a well-aimed CME can be a glancing blow rather than a direct hit, since the ejection itself has real width and structure, not just a single trajectory line.

Magnetic orientation — Bz. This is the single biggest source of uncertainty. Earth’s own magnetic field points northward near the equator. When a CME’s embedded field points southward (technically, a negative Bz in the GSM coordinate system) as it reaches Earth, it can efficiently connect with Earth’s field and pour energy into the magnetosphere — this is what actually drives a strong geomagnetic storm and bright aurora. If the CME’s field points northward instead, even a fast, well-aimed CME can pass by with comparatively little geomagnetic effect. See how to read a solar wind plot for exactly what Bz looks like in the data forecasters watch in real time.

Because Bz typically can’t be measured with real confidence until the CME’s leading edge passes a spacecraft stationed about 1.5 million kilometres upstream of Earth — roughly 30 to 90 minutes before the CME itself arrives — this is the fundamental reason no aurora forecast can promise a specific display days in advance. Forecasters know a CME is coming and roughly how fast; they generally do not know its magnetic orientation, the single most important factor, until it is almost here.

From eruption to aurora: a worked timeline

A typical sequence for a major storm looks something like this: a large sunspot region on the Sun becomes magnetically unstable and produces a solar flare, visible in X-ray data within minutes. Around the same time or shortly after, a CME is observed leaving the Sun, and its speed and apparent direction are estimated from coronagraph imagery. Suppose that CME is measured leaving the Sun at roughly 1,200 km/s and appears as a halo, suggesting an Earth-directed path — forecasters would estimate an arrival window of roughly 36 to 48 hours later, but expressed as a range rather than a single time, because CME speed can change as it travels and interacts with the ambient solar wind (a fast CME ploughing through slower wind ahead of it tends to decelerate somewhat before arrival, while a CME moving through unusually fast wind can hold its speed or even accelerate further). By contrast, a much slower CME measured at 400 km/s might carry an estimated arrival window of three to five days out, with correspondingly larger day-to-day uncertainty about exactly when it will hit.

When the CME’s leading edge finally reaches the L1 monitoring point upstream of Earth, real-time solar wind instruments measure its speed, density, and — critically — its magnetic field orientation. If Bz turns southward and stays there, geomagnetic activity ramps up, Kp climbs, and the auroral oval expands toward lower latitudes, sometimes within an hour or two of that measurement. This is also roughly the point where substorms — the sudden, dramatic brightening and motion aurora watchers actually see — become more frequent and intense.

Why “a big storm is coming” forecasts are still uncertain

News headlines sometimes announce a CME is “Earth-directed” and a storm is expected, based on the initial coronagraph observation, often a day or more before it arrives. That announcement is genuinely useful for deciding whether to keep a night free — worth checking Tromsø or Yellowknife’s live verdict, or for storm-only cities like London or New York, simply staying alert — but it is not a confirmed prediction of a strong aurora. The actual strength of the geomagnetic response depends heavily on that Bz orientation, which simply cannot be known with confidence until the CME’s leading edge is measured directly. This is exactly the gap between the multi-day “watch” alerts NOAA issues (see the NOAA G-scale explained) and the much shorter-range, higher-confidence warnings that follow once real-time solar wind data comes in.

Historical CMEs worth knowing

The 1859 Carrington Event remains the largest documented geomagnetic storm, driven by an exceptionally fast, well-aimed CME; it caused telegraph systems across Europe and North America to malfunction and, in some documented cases, to spark and shock operators, and aurora was reportedly visible at unusually low latitudes, with some accounts describing sky glow bright enough to read a newspaper by. A CME of comparable scale passed through Earth’s orbital position in July 2012, but Earth itself was not in its path at the time — a reminder that events of this magnitude are not purely historical curiosities. On a smaller but still significant scale, the March 1989 storm that caused the Hydro-Québec blackout (see do aurora affect power grids) and the May 2024 storm — widely photographed as far south as Melbourne in the Southern Hemisphere and across much of the contiguous United States and Europe in the north — were both CME-driven events, giving a useful sense of scale between “historic” and “the kind of storm that makes headlines every few years.”

Recurrent activity: CMEs aren’t the only driver

CMEs are episodic — they come from specific eruptions and don’t follow a fixed schedule. A second, steadier source of geomagnetic activity comes from coronal holes, which can drive milder but more predictable, recurring disturbances roughly every 27 days as the Sun rotates. That mechanism is different enough to deserve its own explanation — see coronal holes and recurrent storms for how it works and why it tends to produce a gentler, more forecastable kind of aurora activity than a major CME does, and how the two mechanisms combine differently depending on where the Sun sits in its roughly 11-year cycle (see solar cycle 25/26 and what they mean for aurora chasers).