The Night Sky Is Brightening 10% a Year, From Below and Above

Look up tonight from any mid-sized city and the sky has roughly a hundred fewer stars than it did when you were a teenager. Citizen-science data covering 2011 to 2022 show the night sky is brightening by an average of 9.6% a year, according to a Kyba-led study in Science, faster than every satellite-based estimate. The same decade has lifted more than 10,000 Starlink craft into low orbit. The familiar orange dome of a sodium-lit town is now competing with a second, fainter form of light pollution: sunlight bouncing off an armada moving overhead.

The two problems behave differently, demand different fixes, and arrive at the same place. The dark sky is being eaten from below and from above at once.

Two Sources, One Darker Sky

Terrestrial light pollution acts like fog. It lifts the ambient brightness of the whole sky dome, washing out the faintest stars first, then the brighter ones, until only the Moon, a handful of planets, and a few first-magnitude markers remain. A reader in central Glasgow sees maybe 30 stars on a clear night where a reader in the Cairngorms sees several thousand.

Space-based light pollution is the opposite. It does not raise the background. It punches discrete, moving points across the field. A single Starlink train rolling overhead at dusk does not stop you from finding Orion, but it changes the texture of the sky from quiet to busy, with brief flashes that the brain reads as meteors before correcting itself.

The two combine badly for astronomy. Telescopes already fight a brighter background; now they must also reject streaks that resemble the transient objects they were built to find.

The LED Swap That Hid Itself From Satellites

For two decades the standard satellite measurement of light pollution came from NASA’s Black Marble product, which uses the Visible Infrared Imaging Radiometer Suite on the Suomi NPP satellite. The product is excellent, with one blind spot. Its detectors are weak at blue wavelengths, which is exactly where modern LED street lighting has migrated.

Western Europe spent the last ten years replacing high-pressure sodium with cool-white LEDs. The orange haze faded on satellite imagery, and headlines duly reported Europe getting darker. The citizen-science work led by Christopher Kyba then asked observers to compare their own night skies to a reference set on the Bortle Scale. The eyes told a different story.

What the Eyes Caught That Sensors Missed

Eyes are more sensitive to blue light in dark-adapted conditions than the VIIRS sensor is. So the new LEDs read as roughly the same brightness on instruments, but noticeably brighter to a person standing on the ground. The Kyba paper put the global brightening rate at about 9.6% a year, with North America rising at 10.4%. Satellite measurements over the same period saw a roughly 2% rise. The difference is not a sensor error. It is the colour of the light shifting under the sensor’s floor.

Why the Shift Matters Beyond Astronomy

Bluer skyglow is not just an aesthetic loss. Short-wavelength light suppresses melatonin in humans and disorients night-active wildlife more strongly than the sodium glow it replaced. Nocturnal insects, sea-turtle hatchlings, and migratory birds key off cues that LED conversions disrupt at a different and more disruptive part of the spectrum.

Why Satellites Glow After Sunset

A satellite has no light of its own bright enough to matter from the ground. What you see is reflected sunlight. So a satellite in Earth’s shadow goes dark, which suggests night-sky observers should be safe. The geometry says otherwise.

The further you sit from a sphere, the more of its surface you can see at once. From the International Space Station you can see day and night on Earth in the same frame, with the terminator line cutting through the middle. The same effect lets a satellite a few hundred kilometres up still catch the Sun while sitting above a point in twilight or early night below.

That is why the worst satellite streaks for amateurs and observatories cluster in the hour after sunset and the hour before sunrise. Two windows, every clear night, anywhere on Earth. The lower the orbit, the shorter the illuminated window. The higher the orbit, the deeper into the night the satellite remains visible.

• Below 600 km, satellites typically fade within two hours of local sunset, the threshold astronomers have repeatedly asked operators to respect.
• Between 600 km and 1,200 km, satellites can stay lit through much of a winter night.
• Above 1,200 km, like OneWeb’s first generation, satellites remain sunlit nearly all night long.

Modelling Reflections With the BRDF

You cannot fly a satellite up, photograph it, then redesign it. The only way to know whether a bus-sized antenna array will streak across someone’s exposure is to predict it before launch. That prediction lives in a function called the Bidirectional Reflectance Distribution Function, or BRDF, which expresses the ratio of incoming light from one direction to outgoing light in any other.

From Mirrors to Paper

A mirror has a BRDF of zero everywhere except at the classic angle of reflection, where it spikes to one. A sheet of A4 paper has a BRDF that is roughly flat in every direction, the textbook diffuse reflector. Real satellites sit somewhere between, with a sharp specular peak rising out of a diffuse baseline, separated by a transitional blur whose shape depends on surface roughness and material.

The Spherical-Cow Tradeoff

Pick a slow, accurate BRDF model and you can simulate a single satellite over an entire orbit in minutes. Pick a fast, simplified model and you can sweep a constellation of 10,000 craft in roughly the same time. The astronomy community needs the constellation-scale answer, so most operational tools use the spherical-cow approach: solar panels modelled as flat plates, bus bodies as boxes, a single composite material per panel. The International Astronomical Union’s Centre for the Protection of the Dark and Quiet Skies collates ground observations of magnitudes that test these models in retrospect.

Why Modelling Beats Measuring

Measured brightness from a finished satellite is useful and unavoidable. It is also too late. Once the craft is on orbit, no operator is bringing it back to recoat or retrofit. The BRDF lets engineers ask design questions while the satellite is still on a workbench: shift the panel angle by 12 degrees, replace the antenna laminate, push the orbit 100 km lower, and watch the predicted magnitude curve respond. That feedback loop is what regulation can hook onto.

Painting It Black, and Other Tradeoffs

The most obvious mitigation is to paint the satellite a darker colour. SpaceX tried this with the experimental DarkSat in early Starlink launches. The optical result was real: peer-reviewed measurements of the Darksat magnitude found the coating roughly halved its reflected brightness. The thermal result was a problem.

Energy that no longer leaves the satellite as reflected light still has to go somewhere. It heats the structure, raising temperatures by tens of degrees, stressing electronics, and lighting up the infrared band where instruments like the James Webb operate. SpaceX retired the approach and moved to a sunshade design called VisorSat, then to a dielectric mirror film that scatters light away from the ground without absorbing the full thermal load.

Other levers exist and each one carries a cost:

• Specular over diffuse: polishing surfaces narrows the bright reflection to a thin angular slice. Most observers see a fainter satellite, but the unlucky ones inside that slice see a blinding flash.
• Diffuse over specular: a uniformly matte surface is visible to almost everyone, but no one sees a glare. Telescopes prefer this because consistency lets software subtract the streak.
• Operational dimming: tilting solar panels, dropping orbit altitude, and yawing the bus through twilight passes. The cheapest fix and the easiest to undo if a customer complains.

The Numbers Above Our Heads

The pace at which the sky is filling up is the part of the story that has changed fastest. Starlink alone has roughly 10,400 satellites in orbit as of mid-May, according to Jonathan McDowell’s running Starlink statistics, with the operational subset above 8,600. ESA’s Space Environment Statistics page puts the full operational population in low Earth orbit several thousand higher when other operators are included.

The Filing Backlog

Not every satellite that is announced launches. But filings with the International Telecommunication Union do tell you what spectrum is being claimed. Two Chinese filings lodged in late December, CTC-1 and CTC-2, each request 96,714 satellites across 3,660 orbital planes, on top of the existing Guowang and Qianfan national systems. Amazon’s Kuiper, OneWeb’s second generation, AST SpaceMobile, and Planet’s Pelican are all in queue.

What the Space Force Sees Coming

The US Space Force Future Operating Environment 2040 report projects the active-satellite population rising from about 12,000 today to roughly 60,000 by 2040. That is the conservative line. The same document tags US-government craft at 30,246 and Chinese craft at 20,913 by that date.

An exponential rise in the number of objects in orbit will contribute to emissions, sunlight reflections, and radio transmissions that negatively impact Earth observation and increase the risk of collisions with other satellites.

That language comes from the Future Operating Environment 2040 document itself, published in April by the United States Space Force, which is not a body usually given to talking down its own customer base. The connection between commercial growth in LEO satellite services such as the ScotRail Class 222 WiFi fit and the brightness of the sky above is now part of the official planning baseline.

Tucson, Magnitude Seven, and the Dimming Toolkit

The good news is that none of this is happening to a powerless audience. The city of Tucson, Arizona, replaced its sodium streetlamps with shielded 3000 K LEDs in a programme audited by Dark Sky International. The audit found a 34% reduction in blue-light emissions, a 7% measured reduction in sky glow over the city, and a 70% drop in lighting-related electricity use that the city reports as around $180,000 a month saved. Streetlights were also dimmed by 30% after midnight, which is the lever that does most of the visible work.

On orbit, the levers are starting to be codified. France’s national space agency, CNES, published technical regulation RT 48-10 which sets a brightest-magnitude limit of seven for any French-licensed mega-constellation craft. The IAU’s Centre for the Protection of the Dark and Quiet Skies argues for the same threshold globally, since seven is the magnitude above which the Vera C. Rubin Observatory’s main camera saturates. ESA’s Zero Debris office is collating dimming technologies into a public booklet, and a growing number of national space agencies treat brightness mitigation as a licensing prerequisite, not a courtesy.

The Rubin Observatory itself, which began science operations in mid-2025, is the immediate test case. Its Legacy Survey of Space and Time depends on ten years of clean wide-field imaging. Simulations published in The Astrophysical Journal Letters by Hu and colleagues estimated 10% of LSST exposures would carry a satellite streak under the constellation sizes available when the paper was written. More recent modelling against the post-launch Starlink population pushes that figure as high as 30%. Sacrificing 10% of observing time to schedule around predicted satellite passes roughly halves the fraction. None of this is invisible to a stargazer either: the same wide-field night-sky photography of events like the Scottish wolf supermoon increasingly carries faint diagonals that were not there a decade ago.

If the orbital filing rate slows, the Space Force’s 60,000 number for 2040 is roughly the ceiling. If it does not slow, that is the floor.

Frequently Asked Questions

How much darker has the night sky actually become?

Citizen-science data published in Science in 2023 by Kyba and colleagues found the night sky is brightening at an average rate of 9.6% per year globally, and 10.4% per year in North America, based on more than 50,000 naked-eye observations between 2011 and 2022. A child born in a place where 250 stars were visible would see roughly 100 by their 18th birthday at that rate.

Why do satellite measurements give a lower figure than citizen science?

The VIIRS sensor used for NASA’s Black Marble product is insensitive to short-wavelength blue light, which is exactly where modern LED street lighting has shifted. The result is a roughly 2% per-year rise from satellite remote sensing against the 9.6% rise that human eyes report.

Do satellites add measurably to sky brightness?

For now, only marginally for unaided eyes, but significantly for large research telescopes. A 2021 paper estimated that the diffuse reflected light from all tracked objects already raises the natural-sky background near zenith by about 10%. Streaks across individual exposures, especially at twilight, are the bigger immediate problem for ground-based astronomy.

Can a satellite be made invisible from the ground?

Not entirely while it remains sunlit. Black paint reduces reflected light but heats the satellite, mirror films scatter light into less harmful directions, and operational tricks like lowering orbit and angling solar panels cut the worst flashes. The realistic goal in current regulations is magnitude seven or fainter, which is below the saturation threshold of large research cameras.

What can individuals do about light pollution?

Three things matter most: shield outdoor fixtures so the light points down, use warmer colour temperatures of 2700 K or less for residential lighting, and switch lights off when not in use. Municipal advocacy for post-midnight dimming, as in Tucson, has produced double-digit-percent reductions in skyglow at low cost.

Where can I see a genuinely dark sky in the UK?

The Galloway Forest, Northumberland International Dark Sky Park, Snowdonia, the Brecon Beacons, and parts of Cairngorms National Park retain Bortle 2 conditions on clear nights. All are accredited under the Dark Sky Places programme run by DarkSky International.