Dark Window

Your deep sky companion

Light pollution and satellites

What brightens the sky, how the Bortle scale rates it, where to find darker skies, and how satellite trails end up in your subs and out of your stacks.

Where light pollution comes from

Light pollution is artificial light that goes where it is not needed. For astrophotography the worst kind is skyglow: light that escapes upward, scatters off air molecules, dust and moisture, and brightens the whole sky. It lifts the background in every frame, which buries faint detail and adds noise.

1234UnshieldedFully shieldedlight goes everywherelight goes where needed
  1. 1 Uplight: shines straight into the sky, where nobody needs it.
  2. 2 Glare and trespass: spills sideways into eyes, windows and yards.
  3. 3 Reflection: pavement and walls bounce part of the light back up.
  4. 4 Scattering: air molecules and haze spread it all into a glow that dims the stars over your whole horizon.

The usual suspects

Why LEDs changed things

Air scatters blue light far more than red: scattering strength rises steeply as the wavelength gets shorter (roughly with the inverse fourth power). Cool-white LEDs (4000 K and up) contain a lot of blue, so for the same brightness they brighten the sky more than warm amber lamps. LEDs are also cheap to run, which has encouraged more lighting overall. Warm (3000 K or lower), fully shielded, dimmable fixtures are the dark-sky-friendly choice.

How big is the problem?

What it means for you: a brighter background means a worse signal-to-noise ratio for the same exposure. The fixes, in order of power, are a darker site, narrowband or dual-band filters for emission targets (see the filters guide), and more total integration time (try the exposure calculator).

The Bortle scale

Amateur astronomer John Bortle published a nine-step scale in 2001 to describe how dark a site is, from Class 1 (the darkest skies on Earth) to Class 9 (an inner-city sky). It started as a description of what you can see with your eyes and has since been matched to instrument readings. The sky quality column is the brightness of the sky overhead in magnitudes per square arcsecond: a higher number is a darker sky.

  1. Class 1: Excellent dark site
    Sky 21.9 or darker mag/arcsec² · naked-eye limit 7.6 to 8.0

    Zodiacal light, gegenschein and airglow are obvious. The Milky Way casts faint shadows.

    1.0× natural
  2. Class 2: Typical truly dark site
    Sky 21.7 to 21.9 mag/arcsec² · naked-eye limit 7.1 to 7.5

    Airglow faintly visible. The Milky Way is richly detailed; clouds look like black holes.

    1.2× natural
  3. Class 3: Rural sky
    Sky 21.3 to 21.7 mag/arcsec² · naked-eye limit 6.6 to 7.0

    Some light domes on the horizon. Milky Way still shows lots of structure.

    1.6× natural
  4. Class 4: Rural / suburban transition
    Sky 20.8 to 21.3 mag/arcsec² · naked-eye limit 6.1 to 6.5

    Light domes over nearby towns. The Milky Way is clear overhead but washed out low down.

    2.5× natural
  5. Class 5: Suburban sky
    Sky 20.0 to 20.8 mag/arcsec² · naked-eye limit 5.6 to 6.0

    The Milky Way is weak or invisible near the horizon. Clouds are lit grey.

    4.4× natural
  6. Class 6: Bright suburban sky
    Sky 19.1 to 20.0 mag/arcsec² · naked-eye limit 5.1 to 5.5

    The Milky Way is only seen near the zenith. The sky glows grey-white at the horizon.

    10× natural
  7. Class 7: Suburban / urban transition
    Sky 18.0 to 19.1 mag/arcsec² · naked-eye limit 4.6 to 5.0

    The whole sky has a grey or orange wash. The Milky Way is invisible.

    25× natural
  8. Class 8: City sky
    Sky 17.0 to 18.0 mag/arcsec² · naked-eye limit 4.1 to 4.5

    The sky glows orange-white. Only bright constellations and some Messier objects show.

    63× natural
  9. Class 9: Inner-city sky
    Sky Brighter than 17.0 mag/arcsec² · naked-eye limit 4.0 or worse

    Only the Moon, planets and a few brightest stars and clusters are visible.

    158× natural

Bars use a log scale: each step to the right is a multiple, not an addition. Sky brightness bands vary a little between published tables, so treat the boundaries as approximate. The "× natural" figure compares the zenith sky with the darkest natural sky (22.0 mag/arcsec²).

What it means for imaging

When the sky glow is the main noise source, the exposure you need for the same image quality grows roughly in step with sky brightness. A Bortle 8 sky is around 60 times brighter than a natural dark sky, which is why the same target can take many nights from the city and a single night in the country. Class 4 and darker skies make galaxies and faint nebulae comfortable; Class 7 and brighter push you toward narrowband imaging of bright emission nebulae, star clusters, the Moon and planets.

Dark Window does not yet rate your own site’s Bortle class; use one of the maps below to find it.

Finding a dark site

Picking a site

Satellite trails in your subs

Sooner or later every long exposure catches something crossing the field. A satellite in orbit shines by reflected sunlight, so it appears as a straight, steady line as it moves across the sensor during the exposure. The most common culprits:

What a trail does to a sub

A trail adds a line of extra brightness across the frame, covering the pixels it passes over. It does not damage the sensor and affects only the sub that caught it. How bad it looks depends on how bright the satellite is compared with the sky and your target, and on how long the exposure is: the longer the sub, the higher the odds one will cross it. Wide fields and sub-exposures near dusk and dawn catch the most, because satellites are sunlit then.

How stacking removes them

Stacking software lines up all your subs, then looks at each pixel position across the whole set. A star or nebula pixel has about the same value in every sub. A trail pixel is far brighter in one sub than in the others, so it stands out as an outlier. Rejection algorithms (sigma clipping, winsorized sigma clipping, linear fit clipping and similar) throw that one value away and average the rest.

5 subsStackno trailOne pixel (the ring), sub by subsub 1sub 2sub 3rejectedsub 5Keep thefour thatagree

Starlink and other constellations

Counts below are a snapshot from October 2026 and keep changing as launches continue and old satellites come down.

Starlink satellites in orbit
11,137
As of Oct 10, 2026. 11,123 operational.
Starlinks launched since 2019
12,988
Older ones have already re-entered.
All satellites tracked in orbit
16,700+
Starlink is about two thirds of them.
// Largest constellations, July 2026
  • Starlink SpaceX10,783
  • OneWeb Eutelsat OneWeb654
  • Amazon Leo Amazon (formerly Kuiper)394
  • Qianfan Spacesail, China238
  • Guowang China SatNet186
Satellites in orbit. Bars share one scale, so the size gap is real. Source: KeepTrack satellite catalog.

Questions

How do I find the Bortle class of my backyard?
Three ways, from simplest to best. Look up your address on a light pollution map and read the color. Count the faintest stars you can see in a known constellation (the naked-eye limit column in the table). Or use a Sky Quality Meter, a small device that reads sky brightness in magnitudes per square arcsecond, and compare it to the table.
Do filters fix light pollution?
They help with the right target. Narrowband and dual-band filters block most artificial light because streetlights glow in a broad range of colors, while emission nebulae glow in only a few narrow ones. Broadband targets such as galaxies and reflection nebulae gain much less. Read the filters guide for what each type does.
Can I fix a bright, gradient-filled sky in processing?
Gradient removal tools can flatten the sky background and make a bright-sky image look clean, but they cannot bring back the signal that the sky glow drowned in noise. Light pollution adds noise as well as brightness, so a Bortle 8 image needs far more total exposure than a Bortle 3 one to look equally smooth.
Should I use shorter exposures to avoid satellite trails?
Shorter subs mean each trail ruins less of any single frame, and you can discard an affected frame without losing much. But a trail is still one frame's worth of rejected pixels, and the odds that a given sub is crossed grow with its length. In practice, stack enough subs with rejection turned on and the trails vanish whatever the exposure.
Is Starlink really a problem for deep-sky imaging?
For most backyard work it is an annoyance, not a ruin: a trail through a few subs that stacking removes. It matters far more for professional survey telescopes with wide fields and long exposures, and for imaging near twilight, when satellites are sunlit and brightest. The count keeps growing, so the number of affected frames will too.

References

Open the Dark Window planner to check tonight’s sky.

Support Dark WindowFree, no ads. A tip covers hosting and keeps updates coming.