In considering the impact of satellites on astronomical observations, we have to bear in mind that individual sources of light pollution may be billions or even trillions of times brighter than the objects that astronomers study, and that many of the most scientifically important observations concern unrepeatable time -sensitive or transient events—such as the detection of near-Earth objects, supernovae or fast radio bursts.
ASOs can be seen from Earth because they reflect sunlight. Their brightness depends on numerous factors, such as the size of the satellite, its reflective properties, its height above the Earth and its orientation. As satellites move across the field of view of an astronomical exposure they leave streaks across the image (Fig. 4). For damage already caused by satellites in 2021, see refs.6,7,8,21 and references therein. To grasp the likely impact in the near future, consider our simplified 2030-era LEO satellite population of 100,000 satellites at a height of 600 km.
Only some satellites are visible above the horizon at a given time. For our 2030-era population, roughly 4,300 are above the horizon at any one time, and they cross the sky in about 13 min. For a small field of view, there may be only a few per hundred chance of being affected by a streak, but the observation could be completely wasted and need to be repeated22. More serious impacts would occur on wide field survey instruments. The Zwicky Transient Facility has already seen an increase in affected images from 0.5% in late 2019 to 18% in August 202121. The 3.5-degree-wide field imager of the Vera C. Rubin Observatory nearing completion in Chile will contain at least one streak in the majority of exposures23. Laboratory experiments using the Rubin Observatory camera detectors show that electronic crosstalk causes streaks to cascade and create additional fainter streaks; this effect can render some scientific analyzes impossible because the statistics of the background sky brightness are irrevocably altered. To avoid the crosstalk problem, the satellites would need to be no brighter than seventh magnitude, fainter than the faintest stars visible to the unaided eye at the darkest sites23.
Furthermore, as an object in space rotates, a brief bright flash or ‘glint’ can occur as a facet or particularly reflective component of the satellite briefly reflects more sunlight to an observer on the ground24.25. For example, Starlink satellites have been seen to change rapidly from fader than sixth magnitude to almost third magnitude26. These extremely bright and short duration (transient) events can mimic some of the most exciting phenomena in modern astronomy. A study in 2020 identified such a flash as the sign of a gamma-ray burst at the edge of the Universe—potentially an extremely exciting discovery. However, a year later it was found that this flash was actually caused by sunlight reflecting off an old Russian Proton rocket part27. We do not yet know how frequent this kind of problem will become as the LEO population grows.
When the Earth eclipses a satellite, the satellite is no longer illuminated from the perspective of an observer on Earth. (However, ASOs do emit thermal photons and so affect IR sensing even when in eclipse.) As a result, the impact of satellite constellations on astronomical observations is worst near the beginning and end of the night. However, some types of observation must be done at those times; and the fraction of the night affected depends strongly upon the height of the constellation, the geographic latitude of the observatory and the time of year3,21,28. In addition, observations near twilight will see the most streaks, and that is the same time period when it is preferable to search for near-Earth objects. As a result, our 2030-era satellite population would yield fewer discoveries of near-Earth asteroids, including ones that may cross Earth’s orbit. These are all factors that must be considered carefully in an environmental assessment.
Radio astronomy is affected by satellites using radio signals to relay data back and forth with ground stations and end-user antennas. Detecting faint celestial objects against this anthropogenic background can be potentially very problematic, as the emissions from satellites can easily be a trillion times louder than the astronomical targets3.4. In some observations, finely detailed maps are made by combining signals from many interlinked antennas, but the noise problem affects each antenna individually, which physics dictates will always be sensitive to a broad range of directions and frequencies. Unlike optical images, the effect is not a localized streak, but a complex effect across the whole map, which can be hard to recognize and remove—it is like trying to listen to very quiet music in a noisy room. A radio astronomy antenna is sensitive to a range of directions typically less than a degree across (the ‘main beam’) but also has reduced sensitivity in very different directions (the ‘sidelobes’). Likewise satellite antennas emit most of their power in the main beam, but also some in sidelobes. The worst effects, which can potentially damage sensitive electronic receiver systems, are seen for an alignment between the astronomical and satellite main beams—this rules out radio observations close to GSO targets, and should be avoided even for fast-moving LEO satellites. Sidelobe–sidelobe alignments are much harder to avoid, however, as there may be tens or hundreds of LEO satellites in the sky at any one time, and they are all moving quickly across the sky. The net effect is extremely hard to calculate, but a simulation by the Square Kilometer Array project29 suggests that once the mature Starlink population is in orbit, every observation in the relevant bands will take on average 70% longer.
International regulation of the use of the radio spectrum designates some protected frequency bands for radio astronomy. This approach was originally a great success. However the protected bands were chosen many decades ago when receiver systems were intrinsically narrow-band. Most modern radio astronomy is carried out with state-of-the-art broadband systems, which allow the detection of much weaker natural signals. As a consequence, protection of radio astronomy now relies on geographical radio quiet zones, which some nations provide and some do not. Where available, this zoning can protect against terrestrial interference, but not against satellite interference. When such interference was dominated by a small number of slowly moving GSO satellites, this was acceptable, but the new LEO constellations could lead to very serious issues. The new systems inevitably overlap with satellite communication bands. Furthermore, the volume manufacture and deployment of large numbers of relatively low-cost satellites is likely to increase the chance of sideband leakage into protected bands.
As for the spatial interference issue, the assignment of protected bands sets a previous, as frequency interference is implicitly recognized as an environmental effect. Recognizing that the issues should be subject to environmental laws such as the US National Environmental Policy Act (NEPA) is the logical next step as the problems get much worse.
Some spacecraft used for astronomy are placed at very large distances from the Earth, and are not affected by LEO satellites. Many, however, like the Hubble Space Telescope, are in LEO, and can certainly suffer from streaking. Occasionally a satellite may pass relatively close by (<100 km), in which case the streak caused is an extremely bright out-of-focus stripe, obliterating a significant fraction of the image. An example is shown in Fig. 5. A recent study by S. Kruk et al. (manuscript in preparation) showed that, depending on the instrument and observational parameters being used, between 2% and 8% of Hubble Space Telescope images were affected by satellite streaks, but also that the frequency was changing with time, reflecting the growth of the LEO satellite population. Our 2030-era population indicates that by the end of the decade a third of Hubble Space Telescope images will be affected, as will future LEO-based science missions, such as the Xuntian wide field observatory being built for the Chinese Space Station.
Mitigation, damage and their costs
The international astronomical community has had multiple meetings to discuss how to address the new landscape of increasing numbers of bright LEO satellites, leading to key reports6,7,8,9,10. A report by the US government independent advisory body JASON was also commissioned by the National Science Foundation4. Astronomers have engaged with satellite companies to discuss ways to mitigate the problems. For optical astronomy, this has included ideas such as painting satellites black, changing their orbits and orientations, adding sun visors and providing detailed positions and trajectories so that observatories can avoid pointing at them. For radio astronomy, key mitigations include redirecting beams away from major observatory facilities and employing sophisticated signal filtering. However, none of these mitigations can fully avoid LEO satellite constellations harming astronomical science7,8,10; launching significantly fewer satellites is the only mitigation that could do this.
The consequences of the current and proposed growth of satellite constellations have a direct cost from repeating or extending observations, wasting scientist’s time and even negatively affecting their careers. Implementing mitigations will also impose significant costs, on the astronomical community (and so the taxpayer), the satellite operator companies or both. We do not attempt to assess those costs here. Instead, we point out that this is a classic example of environmental damage, externalizing true costs. To give an example, one significant conclusion from the Observations Working Group of the Satellite Constellations 2 Workshop (SATCON2) held in 20218 was the need to establish a coordinated satellite observation hub under the umbrella of a larger International Astronomical Union Center for the Protection of the Dark and Quiet Sky from Satellite Constellation Interference30. Such a long-term mitigation activity will require significant sustained resources.
We note that the US Federal Communications Commission order under current legal discussion has, quite correctly, encouraged SpaceX to continue engagement with the astronomical community. However, these productive collaborations ought to proceed within the context and guidance of environmental assessment.