Where aurora forecast data comes from: the instruments

Every aurora app, website and alert service in existence reads from the same short list of public instruments. There is no proprietary aurora data. Knowing which instrument produces which number tells you what each figure can and cannot support, and it explains most of the disagreements between apps.

The chain runs from the Sun to your screen through four layers: spacecraft watching the Sun itself, spacecraft sampling the solar wind upstream of Earth, satellites and ground stations measuring Earth’s response, and the agencies that turn those measurements into published products.

The four layers, and what each one can tell you

Read the warning-time column first. It contains the whole argument about forecast accuracy: the layer that sees days ahead cannot measure the variable that matters, and the layer that measures it sees less than an hour ahead. How accurate the aurora forecast is by lead time works through the consequences.

Watching the Sun: SOHO and GOES

SOHO, the Solar and Heliospheric Observatory, is a joint NASA and ESA mission operating from the L1 region. Its LASCO coronagraphs block the solar disc to image the faint corona around it, which is how coronal mass ejections are detected and their speed estimated.

A coronagraph sees a CME as it leaves the Sun. From a sequence of images, forecasters estimate its speed and whether it is directed toward Earth, and models project an arrival time. Transit typically takes 15 to 90 hours.

What a coronagraph cannot see is the magnetic field inside the CME. That orientation determines whether the arrival produces a major storm or very little, and it is not measurable remotely by any current instrument. This single limitation is why a three-day aurora forecast is a statement about probability rather than about outcome.

The GOES satellites carry X-ray sensors that detect flares directly. Flares matter for radio propagation, and a large flare is often the first sign that an active region is capable of launching a CME — but a flare on its own produces no aurora.

Measuring the wind: DSCOVR and ACE at L1

DSCOVR, the Deep Space Climate Observatory, is the primary real-time solar wind monitor. It operates near L1, the Sun-Earth Lagrange point about 1.5 million kilometres sunward of Earth, where a spacecraft can hold station upstream in the flow. It measures solar wind speed, density and the vector interplanetary magnetic field, including the Bz component.

NASA’s ACE, the Advanced Composition Explorer, launched in 1997 and occupies a similar region. It has served as the real-time solar wind source and now provides backup and complementary measurements. Between them they have kept a continuous upstream watch for decades.

The travel time from L1 to Earth is simple arithmetic and it sets the forecast horizon. At a solar wind speed of 800 kilometres per second, 1.5 million kilometres takes about 31 minutes. At 400 kilometres per second it takes about 62 minutes. Everything anyone knows about the incoming magnetic field orientation comes from that window.

This is the measurement that matters most for a decision tonight. A sustained southward Bz means energy is being transferred efficiently into the magnetosphere; the same wind speed with a northward field means much less. How to read solar wind speed, density and Bz covers the interpretation, and the live aurora forecast shows the current values.

There is a practical caveat worth stating. L1 is a single point in a very large flow. Structures in the solar wind are not uniform across the distance between L1 and Earth, so the wind that passes the spacecraft is a good but imperfect sample of the wind that arrives.

Measuring Earth’s response

Once the wind arrives, the measurement problem changes from prediction to observation, and two networks handle it.

In space, the GOES satellites sit in geostationary orbit about 35,800 kilometres up — inside the magnetosphere rather than outside it. Their magnetometers record the local field, which compresses when solar wind pressure rises and stretches during substorm growth. This is a direct look at the magnetosphere responding, and it is concurrent rather than predictive.

On the ground, magnetometer observatories record the surface field. The 13 observatories behind Kp are sited between 44 and 60 degrees geomagnetic latitude, deliberately below the auroral zone so the index reflects planetary conditions rather than local extremes. Dst comes from four near-equatorial observatories, and AE from a chain inside the auroral zone.

SuperMAG, run from the Johns Hopkins Applied Physics Laboratory, collates data from a far larger worldwide collection of magnetometers into a common format. It is the resource to reach for when the standard indices are too coarse, either in space or in time.

Who publishes what

Instrument ownership and data publication are separate things, which is why several agencies appear in any honest source list.

• NOAA’s Space Weather Prediction Center is the operational forecaster for the United States and the origin of most feeds an app reads: real-time solar wind, the estimated planetary K-index, the 3-day geomagnetic forecast, the 27-day outlook, and the OVATION 30-minute aurora model. All of it is public domain.
• GFZ in Potsdam is the official service for the definitive Kp and ap indices under IAGA, and it also publishes the high-cadence Hp30 and Hp60 series.
• The World Data Center for Geomagnetism in Kyoto publishes Dst and AE.
• The International Service of Geomagnetic Indices coordinates the index services under IAGA.
• NOAA’s NCEI archives the long-term space-weather record and publishes the International Geomagnetic Reference Field, which is what geomagnetic coordinate calculations are computed against.
• Natural Resources Canada and ESA operate their own forecast services, and AuroraWatch UK at Lancaster University runs a UK-specific magnetometer alert.

The practical consequence is that when two aurora apps disagree about tonight, they are almost never reading different instruments. They are applying different thresholds, different location models and different presentation choices to identical public numbers. That is worth knowing before treating one app’s confidence as independent confirmation of another’s.

What none of these instruments measure

Four things decide whether a person sees aurora, and the instrument chain above measures exactly one of them.

Cloud. No space-weather instrument sees it. A perfect forecast above complete overcast produces nothing, and this remains the most common single reason an aurora night fails.

Darkness at your location. Astronomical twilight is a matter of solar geometry, not space weather. Above the Arctic Circle the sky does not get dark enough for months around midsummer regardless of activity — see why the aurora has a season and the dark-hours calculator.

Your position relative to the oval. The instruments measure planetary and upstream conditions. Turning that into an answer for one address requires corrected geomagnetic latitude, which is computed rather than measured — the geomagnetic latitude calculator does it.

Whether the display is bright enough for human vision. Emission energy is not visual brightness. A camera routinely records displays that a dark-adapted eye cannot distinguish from grey sky, which is the source of a great deal of disappointment covered in the diagnostic table.

Any product presenting a single confident number has combined a measured quantity with several modelled and unmeasured ones. That is legitimate — it is what our own tools do — but the confidence of the presentation should not exceed the confidence of the weakest input.

Bottom line

Aurora forecasting runs on public instruments and there is no private data. Solar imaging by SOHO coronagraphs and GOES X-ray sensors detects coronal mass ejections 15 to 90 hours before arrival but cannot measure the magnetic orientation that decides their effect. DSCOVR and ACE at the L1 point, about 1.5 million kilometres sunward, measure solar wind speed, density and the interplanetary magnetic field including Bz, giving 20 to 60 minutes of warning depending on wind speed — that interval is the hard physical ceiling on short-range forecasting. GOES magnetometers in geostationary orbit and ground observatories record Earth’s response as it happens: 13 stations between 44 and 60 degrees geomagnetic latitude produce Kp, four near-equatorial stations produce Dst, an auroral-zone chain produces AE, and SuperMAG collates a far wider network.

NOAA SWPC, GFZ Potsdam and the Kyoto World Data Center publish the results, all free. None of these instruments measures cloud, darkness, your geomagnetic position or whether a display is bright enough to see — and those four decide the night.