How CuriousPilot Predicts Satellite Brightness
Where the brightness numbers come from, how they are calculated, and what they do — and do not — account for.
1. What “magnitude” means
Brightness is given in magnitude, astronomy’s brightness scale. Two things to remember:
- Smaller numbers are brighter. Negative numbers are very bright.
- Each step of 1 magnitude is about 2.5 times the brightness.
| Object | Magnitude |
|---|---|
| Full Moon | −12.7 |
| Venus at its best | −4.6 |
| Space Station on a good pass | −5 to −3 |
| Brightest star (Sirius) | −1.5 |
| A typical Starlink on a good pass | +3 to +4 |
| Faintest star the eye can see, dark sky | +6.0 to +6.5 |
| Faintest star the eye can see, suburbs | +4 to +5 |
The magnitude slider in the app hides anything fainter than the number you set.
2. The starting point: a measured “standard brightness”
Every satellite starts with one number, its standard brightness — how bright it would look at a fixed reference: 1,000 km away, with the Sun off to one side (half-lit).
That number lets satellites be compared fairly, no matter who observed them or when. It is the same convention used by Stellarium, Heavens-Above, and the amateur satellite-observing community.
Where CuriousPilot gets it: from the Stellarium satellite database. Stellarium’s values descend largely from the catalog built over decades by Mike McCants and other visual observers, who measured satellite brightness by eye and by camera against known reference stars.
So for most satellites, the base number is an observation, not a calculation.
3. When nobody has measured it
Stellarium does not have a value for every object. Newly launched satellites, obscure debris, and old rocket bodies are often missing. In that case CuriousPilot falls back to an estimate, in this order:
- A hand-set value for a few famous objects (Space Station, Hubble, the Chinese station modules).
- A guess based on the satellite’s name, if it belongs to a known constellation — Starlink, OneWeb, Iridium each get a typical value.
- A generic default for everything else, with rocket bodies and debris treated as somewhat brighter than average.
These fallbacks are the weakest part of the chain. They are rough by nature: a real object could be much fainter than the default, especially small debris. If a satellite in the app looks implausibly bright and it is an obscure object, this is the likely reason.
4. What the app calculates for each moment
The standard brightness is only the starting point. For every satellite at every instant, CuriousPilot adjusts it for the real geometry:
Distance. A satellite twice as far away is four times fainter. Overhead passes are much brighter than passes low on the horizon, purely because they are closer.
Sun angle (phase). Like the Moon’s phases, a satellite can be fully lit, half lit, or a thin sliver, depending on where the Sun is relative to your line of sight. Fully lit is brightest.
Air. Light is dimmed on the way down through the atmosphere. Straight overhead costs about 0.2 magnitudes. Low on the horizon it costs far more — roughly 1 magnitude at 10° elevation and 2 magnitudes at 5°. This alone can push a satellite that would be visible overhead below your detection threshold near the horizon. If you are at altitude (on a mountain or in an aircraft), you are above part of the atmosphere and the penalty shrinks.
Sunlight. A satellite in the Earth’s shadow is not lit at all, and is not shown as visible regardless of brightness.
5. Starlink gets extra treatment
Starlink satellites have a large flat, mirror-like surface facing the Earth. When the geometry lines up, that surface reflects the Sun straight at you and the satellite briefly becomes far brighter than the ordinary calculation would suggest — a flare. CuriousPilot models this specifically:
- Flare brightness curves come from Mallama & Cole (2024), “Extreme Flaring of Starlink Satellites,” with a separate curve for each Starlink generation (v1.0, VisorSat, v1.5, Mini). One caveat we are open about: the paper plots these curves but does not publish the underlying numbers, so ours were read off the published figures. The v1.5 curve is good to about ±0.1 magnitude; the others to about ±0.5.
- Direct-to-Cell Starlinks (the ones that talk to phones) are intrinsically brighter than the standard Mini. Following Mallama et al. (2024), the app makes them about 1.5 magnitudes brighter, both in their steady brightness and in their flares.
Flares are modeled for Starlink only. No comparable published brightness curves exist for other constellations, so the app does not attempt to predict their flares.
6. What the model does not account for
Being clear about the limits matters more than the numbers themselves.
- Mirror-like glints from non-Starlink satellites. Every other object is treated as a dull, evenly-scattering surface. Real satellites have solar panels and flat metal faces that can glint. The app will miss those brightenings.
- Tumbling. Dead satellites and spent rocket bodies often tumble, flashing as different faces catch the Sun. The app shows one steady average brightness instead.
- Which way the satellite is pointing. Orientation strongly affects brightness and is generally not published for anyone outside the operator. The app assumes an average.
- Weather. Cloud, haze, smoke, and humidity are not modeled. Every satellite tool makes this same assumption.
- Your sky. Light pollution changes what you can actually see by two magnitudes or more. The app does not know how dark your site is.
- Old or estimated catalog values. A standard brightness measured years ago may no longer describe a satellite that has since degraded, been reoriented, or been replaced by a different design under the same name.
7. How to use the predictions
- Treat the number as an estimate with real uncertainty, not a measurement. For well-observed satellites it is good. For obscure debris it may be off by several magnitudes.
- Real satellites can be brighter than predicted, briefly, from glints the model does not cover.
- Real satellites can also be fainter than predicted, especially small or poorly-catalogued objects that fell back to a generic estimate.
- When you are trying to identify something you saw, open the magnitude slider wide so candidates are not hidden — then judge the individual numbers.
- When you are comparing against a photograph, remember that a moving satellite smears its light across a trail. A trailed satellite is harder for a camera to record than a stationary star of the same magnitude.
8. Sources
- Stellarium — satellite database, and the brightness formula it uses.
- Mike McCants — the foundational visual brightness catalog most values descend from (the qsmag magnitude file).
- Kasten & Young (1989) — the atmospheric dimming formula. Applied Optics 28(22), 4735–4738.
- Mallama & Cole (2024) — Starlink flare brightness curves. arXiv:2405.13091.
- Mallama et al. (2024) — Direct-to-Cell Starlink brightness. arXiv:2407.03092.