Why you did not see the aurora, by symptom

Most failed aurora nights have one dominant cause, and it is usually identifiable from what you saw rather than from what the forecast said. This page is a diagnostic table: match the symptom, read the most likely cause, change one thing next time.

The order matters. The causes below are arranged roughly by how often they are the real explanation, and the first one accounts for more failed nights than every other combined. It is also the one that no space-weather product can help with.

The diagnostic table

Cloud: the cause that is not in any space-weather product

Cloud is the single most common reason an aurora night fails, and none of the indices, models or apps on this subject measure it. NOAA’s products describe the state of the magnetosphere. They contain no information about whether there is a stratus deck between you and it.

This produces a specific and very common failure: a genuinely excellent geomagnetic forecast, a correctly identified night, an accurate alert, and an observer standing under complete overcast. The forecast was right. The night was still lost.

The fix is procedural rather than technical. Treat the space-weather forecast and the cloud forecast as two separate gates that both have to open, and check the second one before committing to travel. Being willing to drive an hour to a hole in the cloud converts more failed nights into successful ones than any improvement in aurora forecasting.

Timing: you probably left during a growth phase

If the sky was clear, the activity was there, and you saw a faint arc for an hour and went home, you very likely watched a substorm growth phase and left before the expansion.

Aurora does not build gradually. Energy accumulates in the magnetotail for 30 to 60 minutes while the sky shows a quiet arc, then releases in an expansion lasting 10 to 30 minutes that produces everything worth photographing. What a substorm is sets out the full cycle.

Two signs are worth learning. An arc drifting slowly toward the equator is a tail that is loading. An arc that sharpens and brightens along one section is onset beginning, and you have a minute or two.

The second timing error is stopping after the first display fades. Recovery gives way to a new growth phase, and during active periods substorms recur every two to four hours. The second onset of the night is frequently the better one.

The map is not a probability

The OVATION map shows modelled energy deposited into the upper atmosphere at each point. It does not show the probability that a person standing at that point will see something.

Two consequences follow, and they point in opposite directions. Aurora inside the coloured region can be invisible from directly beneath it if the emission is faint — the model is reporting energy, not brightness against a moonlit sky. And aurora well outside the coloured region is frequently visible, because emission occurs around 100 kilometres up and can be seen from hundreds of kilometres away by looking toward the pole.

The second is the more expensive mistake, because it stops people going outside at all. How to read the OVATION map covers what the colours actually represent.

High Kp, empty sky

Kp is a planetary index averaged over three hours. Two things it cannot tell you explain most cases of a high number over a quiet sky.

It does not know your longitude. A storm can concentrate over one range of longitudes, and a planetary summary reports the disturbance regardless of where on Earth it occurred. Your sky can be quiet through a Kp 7 block that produced a superb display eight time zones away.

It does not resolve time within the block. A sharp 20-minute substorm and a steady three hours of moderate activity can produce similar values, and the reverse is also true — a high block can describe activity that had already faded by the time you were outside.

The signals that respond faster are Bz measured at L1, which leads by 20 to 60 minutes, and Hp30, which uses the Kp method over 30-minute windows. How the indices differ covers which one to reach for.

Latitude: the threshold you cannot wait out

If you are at a mid-latitude site and the geomagnetic activity was moderate, the honest answer is that the night was never going to work, and no amount of patience or dark sky changes that.

The threshold is set by corrected geomagnetic latitude, not by the latitude on a map, and the two can differ by several degrees. NOAA’s own storm scale states the lowest latitudes at which aurora has been reported at each level: about 55 degrees geomagnetic at G2, 50 at G3, 45 at G4 and 40 at G5. Those are the extremes reported, not the typical experience.

Work out your own coordinate before planning around a forecast — the geomagnetic latitude calculator does it, and what Kp you need where you live explains why a single universal threshold is misleading.

Your eyes are working correctly

A large share of disappointment is not a failure at all. It is the gap between what a camera records and what a dark-adapted eye sees.

In low light the retina switches to rod cells, which are far more sensitive than cones but do not distinguish colour. A camera, meanwhile, integrates light over several seconds. The result is that a display which photographs as vivid green genuinely appears as a pale grey-white arc to the eye, and this is normal rather than a sign of a weak display.

The gap is largest for diffuse aurora — the broad structureless glow that is often below the visual threshold entirely while appearing clearly in a five-second exposure. It closes during bright substorm expansions, where colour does become visible to the unaided eye.

Two habits help. Allow 20 to 30 minutes of dark adaptation before judging anything, and protect it: a phone screen at normal brightness resets much of it in seconds. Photographing the aurora with a phone covers the camera side of the same problem.

Sky brightness: moon, town and twilight

Three separate things brighten a sky, and they fail differently.

Moonlight raises the visual threshold. A bright display is fine under a full moon and the foreground lighting can improve a photograph; a faint one is washed out. If you have a choice of dates, prefer nights near new moon — how moon phase affects what you see covers the numbers.

Light pollution is worse than moonlight because it is concentrated toward the horizon, which is exactly where aurora appears from most latitudes. Driving 30 minutes away from a town does more for a marginal display than any equipment.

Twilight is the absolute one. Above the Arctic Circle the sun does not sink far enough below the horizon for months around midsummer, and no geomagnetic activity is visible through it. This is a calendar constraint, not a viewing skill — see why the aurora has a season and the dark-hours calculator for your latitude and date.

Bottom line

Work through the causes in this order, because it is roughly the order of frequency. Was there cloud? If yes, that is the answer and nothing else matters. Did you wait at least an hour after arriving, and did you stay after the first fade? If not, you probably met a substorm growth or recovery phase rather than a quiet night. Were you reading the coloured OVATION region as a viewing probability, and did you look toward the pole from outside it? Was your corrected geomagnetic latitude ever compatible with the activity level, given that NOAA reports aurora down to roughly 50 degrees geomagnetic at G3 and 45 at G4? Was the sky genuinely dark — no moon, no town glow, and past astronomical twilight for your latitude and date? And finally, did you allow 20 to 30 minutes of dark adaptation and keep a bright phone screen out of your eyes?

If the answer to all of those is yes and the sky still did nothing, then the activity was real but it happened at another longitude or inside a different part of the three-hour block. That is an ordinary outcome, not a mistake, and the correct response is more nights rather than better forecasting.