TOO BRIGHT FOR THE NIGHT SKY

In just a few years, the number of satellites in orbit has risen sharply, driven almost entirely by giant constellations such as Starlink, OneWeb and Guowang. They bring internet to remote areas, but they also leave bright trails across astronomers’ images.

Within a year of the first Starlink launches in 2019, astronomers proposed a simple rule: a satellite seen in reflected sunlight should be no brighter than magnitude 7 — slightly fainter than the faintest stars a person can see under a dark sky — with a slightly fainter limit for higher orbits. The International Astronomical Union’s centre for protecting the dark sky adopted this recommendation. Until now, though, few measurements checked whether satellites actually comply: earlier campaigns followed small numbers of hand-picked targets.

A net for meteors, turned to satellites

Denis Vida, Peter Brown and colleagues at Western University in Canada, with Defence Research and Development Canada, used Luciole, a camera system derived from the Global Meteor Network. As the authors put it, satellites can be seen as very slow, long-lasting meteors. Instead of pointing at known satellites, Luciole casts an optical net over the sky and records everything that moves, 25 times per second.

From 7 October to 11 November 2024, six sensitive cameras at Elginfield Observatory, Ontario, logged 24,812 tracks of 5,118 different satellites in 230 clear hours. Each satellite’s brightness was calibrated against the stars in the same image.

Who breaks the limit

The observed population is systematically brighter than the recommended limit, above all in the crowded orbital layer between 500 and 600 kilometres. To compare operators fairly, the team counted how often each group exceeded the limit among all the passes where it could have been seen:

• Starlink: about 25% of possible passes — and, because of its sheer size, about two thirds of all the too-bright passes recorded.
• Spent rocket bodies: about 21%. Breeze and Long March rocket stages exceeded the limit on 97–99% of the passes actually observed.
• Guowang/Hulianwang, China’s emerging constellation: more than half of possible passes, and every single pass observed. With 13,000 satellites planned, the authors warn it could become the dominant source of optical sky pollution unless it is dimmed.
• OneWeb: only about 6%. Its higher orbit is not the whole explanation, since other constellations at similar or greater altitudes appear much brighter.

Progress is not guaranteed either. Starlink satellites had been getting dimmer from 2020 to early 2024, but the newer “Direct-to-Cell” models are almost as bright as the very first ones. How their brightness changes with the viewing angle points to strong mirror-like glints, which the authors plausibly attribute to Starlink’s large flat antennas — so darkening surfaces alone may not be enough.

The twilight bottleneck

Most low satellites are visible only when they are still in sunlight while the ground is already dark: in the two hours after sunset and before sunrise. A model of a 550-kilometre layer shows about 3.3 hours of intense visibility per night, all year round, between 35 degrees south and 35 degrees north. Higher satellites, around 1,200 kilometres, stay lit for most of the night.

Twilight is precious for astronomers: it is when they take calibration images and hunt asteroids that come close to Earth. The authors predict that a person under a dark sky sees around 44 bright satellites in their field of view at any moment of twilight. For the Vera C. Rubin Observatory, about one 30-second twilight image in three should carry a satellite trail.

Lower limits

The survey comes from a single site at mid-northern latitude over 36 nights, with fixed cameras. It also misses satellites still climbing to their final orbit, often the brightest ones — so the numbers are, if anything, underestimates. The authors propose their “satellite flux” as a standard that any observatory can use, and see OneWeb’s results as a benchmark the rest of the industry can already meet.