Vast satellite constellations are alarming astronomers


ON MAY 24TH 2019 a Falcon 9 rocket built by SpaceX launched 60 communication satellites into a low orbit around the Earth. That evening they appeared as a string of sunlit dots moving across the sky, many of them as bright as the brightest stars, a source of passing wonder and mystery to casual observers—and a portent of doom to astronomers.

“There were all these panic messages,” remembers Olivier Hainaut of the European Southern Observatory (ESO): “‘Oh my God, it's the end, it's the end of astronomy as we know it!’” Jonathan McDowell, an astronomer at the Harvard-Smithsonian Centre for Astrophysics, says he was “gobsmacked” by how bright the satellites were. “I did some quick mental arithmetic and realised that thousands of satellites that bright would be a substantial fraction of the visible things in the sky…I felt an increasingly large pit in my stomach.”

The panicked messages and the abdominal unease stemmed from the knowledge that those 60 lights were just the beginning. Up until that point communication satellites dealing with large amounts of data had been, for the most part, few and distant, sitting high and invisible over the equator. The largest “constellation” in low Earth orbit was that of Iridium, a satellite phone company, which had around 70 of the things. With 60 satellites SpaceX had almost equalled that with one launch. And there were a lot more launches to come.

As originally proposed, the company’s plan for a system that could provide fast internet access to almost all parts of the globe called for 12,000 satellites. In subsequent plans a second phase has been added which brings the total to 40,000 satellites arranged in orbital “shells” at altitudes from 335km to 614km.

The company’s reusable rockets have allowed it to act on those plans at unprecedented speed. The first shell, consisting of 1,584 satellites, was completed in just two years; its services have beta testers over much of the world. SpaceX plans to put up more than a thousand a year from now on, and to pick up the pace when it replaces its Falcons with its next-generation Starships.

The satellites seen on that first evening were, astronomers now know, anomalously bright. And the company has gone some way to making its more recently launched satellites less visible. But Starlink is not the only mega-constellation around town. According to filings made with the International Telecommunication Union (ITU), which regulates the use of different radio-frequency bands, and with national regulatory bodies, 100,000 non-SpaceX communication satellites could be launched into low orbits by 2030.

OneWeb, a firm the British government recently took over, has launched more than 200 already and has filed a request to America’s Federal Communications Commission (FCC) for a total of almost 6,400. The FCC has also approved 3,000 satellites as part of Amazon’s Project Kuiper. China, which has added “satellite internet” to its list of needed infrastructures, has filed with the ITU for 13,000 satellites at altitudes from 500km to 1,145km. The largest ITU filing was made in early November by Greg Wyler, founder and former executive chairman of OneWeb. In partnership with the Rwanda Space Agency he has applied to send up a constellation of 327,000 communication satellites.

The reason that providing high-bandwidth services from space requires satellites in such large numbers is basically geometrical. Moving large amounts of data quickly is much easier if the receiver and transmitter are quite close. But satellites close to the surface move across the sky very quickly. So to be sure there are always a few in the sky over every user means you have to have a very large number of them.

Big constellations offering broadband-internet services were first proposed in the 1990s. But there was neither the technological base to mass-produce satellites of sufficient sophistication nor the launch capacity to spread them across the sky in sufficient profusion. Now there is. And the smartphone revolution which is partly responsible for producing the robust and highly capable electronics such satellites need has also increased the world’s appetite for their services. According to the UN Development Programme only one in five people in developing countries is online. Morgan Stanley, a bank which predicts that the global space economy, worth $350bn in 2016, will be worth $1trn by 2040, thinks that 50-70% of that growth will be in satellite-internet services.

That does not mean all the schemes can make a profit, or even get off the ground. But even if only some go ahead, the boom in the satellite population could transform the appearance of the evening and morning sky. How visible they would be to the naked eye and thus the general population is hard to say. If all the operators follow SpaceX’s example, maybe not too much; if not, the hours after sunset and before dawn could see dozens of faint lights racing through the sky.

But in either case they will transform the practice of amateur and professional astronomy. A study by the ESO shows that at Paranal, a site in Chile which is home to the organisation’s magnificent, if prosaically named, Very Large Telescope (VLT) there might typically be over 500 satellites visible in the sky at the beginning of the night (see chart 1). In long astronomical exposures, as pictured above, the paths of such satellites streak the sky like the bars of a jail cell.

The flash that split the night

“It’s very much a paradigm shift,” says Robert Massey of the Royal Astronomical Society, a British institution. “[The] large-scale utilisation of low Earth orbit is very different from anything we’ve had in the previous six decades in the space age.” That new paradigm raises the urgency of a range of problems: concerns about collisions and debris, the removal of defunct satellites, the effects of that much launch activity on the stratosphere, the allocation of radio frequencies, the need for arms control and so on. As Dr McDowell puts it, “Every time humanity moves into a new domain—the oceans or the air or space—we go, ‘Wow, this is enormous and really empty, we can throw as much garbage here as we want, and it’ll never fill up, right?’ Then pretty soon we go, ‘Oops, that wasn’t quite true.’ And we’re reaching that point in space.”

Because the worries concerned are not entirely novel, for most of the issues raised by the satellite boom there is at least some sort of pre-existing setting for negotiation or arbitration, and some sense of who is responsible to whom. The Outer Space Treaty requires countries to take responsibility for objects launched from their territory while they are in orbit. Satellite radio emissions are regulated by the ITU. The Vienna convention is where you go to talk about any possibility of damage to the stratosphere. There are evolving procedures for avoiding collisions, and norms for how to dispose of satellites after their lives are over. But there is no forum for discussing what satellites look like from the surface. It is an externality which has never previously been a worry.

This may seem a small thing. Astronomy is not a matter of life or death. But governments and philanthropists spend billions of dollars on it, thereby satisfying some of the world’s welcome appetite for wonder. Amateurs in their unknown tens or hundreds of thousands derive great joy from it. Astronomers may not have an inalienable right to an unobstructed sky. But a meeting on the subject, Dark and Quiet Skies, which was organised by the International Astronomical Union (IAU) and the UN last month, made clear their dismay at companies being able to take it away from them without so much as a by-your-leave.

Conflict between star-gazing and communications is not entirely new. Radio astronomers have had to deal with interference from broadcasters for decades. But radio waves beamed from an antenna, whether on Earth or in orbit, can be regulated. Radio telescopes often sit in “radio-quiet zones” where the use of mobile phones and other devices is either limited or banned to ensure that incoming celestial signals are not drowned out; the ITU sets aside some wavelengths specifically for astronomical use.

This does not mean that radio astronomers do not moan, nor that they do not have cause to. Emissions can often leak out of the intended frequency range. Scientists working on radio astronomy’s next big thing, the Square Kilometre Array, are very worried about the new constellations. The SKA is a €1.8bn ($2bn) project of 14 countries which will combine thousands of antennae of various types spread across South Africa and Australia into a single vast instrument.

A study released by SKA scientists in 2020 showed that a constellation of 6,400 satellites using frequencies between 10.7 gigahertz and 12.7GHz—the frequencies Starlink is licensed to use—would reduce that instrument’s sensitivity in a neighbouring frequency band allocated to radio astronomy by 70%. By the time the number of satellites reached 100,000, that band would be unusable. That would make some distinctive signals the array will look for impossible to spot. The scientists want to get the satellites’ beams directed away from their antennae.

Radio astronomers have to worry about the satellites all day and all night. Their optical peers focus their concerns on the twilight, or rather the twilights: night-sky professionals recognise a civil twilight, a darker nautical twilight, and a more velvet still astronomical twilight. At these times satellites remain illuminated by a sun which, from the observatory’s point of view, is already well below the horizon. The higher the satellite, the longer it remains illuminated, which makes OneWeb and the Chinese plans, both of which use orbits above 1,000km, more worrying than Starlink (see chart 2).

Dr Hainaut reckons that the VLT could lose about 2% of its overall data to satellite streaks during the evening and morning: galling, but not incapacitating. But a revolutionary instrument elsewhere in the Andes will fare considerably worse. America’s Vera C. Rubin Observatory is designed not just to survey the whole of the sky in more detail than ever before, but to do so on a weekly basis, thereby capturing telltale changes and one-off events. Looking at lots of sky with a sensitive eye maximises the problems posed by satellites. “All of the features that make Rubin Observatory amazing for discovering unknown things are the exact same features that make it highly vulnerable to lots of bright satellite constellations,” says Meredith Rawls, an astronomer at the University of Washington who works at the observatory.

Dr Hainaut reckons up to half of the Rubin Observatory’s images at twilight and dawn could feature streaks. Software should be able to remove them without losing more than 30% of the data, and possibly as little as 10%. But the streaks themselves are not the only problem. The chips in the camera’s light sensitive focal array—the largest in the world—can get overloaded if too much light hits them, creating electronic overtones and echoes that spoil the whole frame, not just the bit with the streak in it. For a programme devoted to looking for brief glimpses of rare events, that is a serious problem.

Dialogues between astronomers and SpaceX engineers have mitigated the worst of the problems by tweaking the satellites’ design. In operation each of the Starlinks looks like a cross between a windsurfer and a junk. They consist of a three-metre-long rectangular board, known as the bus, on the bottom side of which are the antennae used to pick up and transmit signals, and a nine-metre rectangular solar panel that stands above the bus like a sail.

Words like silent raindrops

The reason the first Starlinks looked so bright was that when deployed from their Falcon 9 at an altitude of about 300km—they climb to their final orbits under their own steam—their solar panels spread out horizontally, thus maximising the reflecting surface as seen from Earth. Now the satellites keep their solar panels out of sight as much as they can. To minimise reflections from the bottom of the bus once the satellites are in their final orbit SpaceX engineers came up with the idea of a visor mounted to one side that casts a shadow over most of the Earth-facing side. This makes them considerably less visible.

Dr McDowell welcomes these workarounds, but notes that they are not perfect, and that other operators may not be as amenable. The possibility of bright satellites in truly sky-spoiling numbers persists, he says, “even if none of the currently proposed constellations are going to do it. We have to plan for it and we have to regulate against it.”

Regulation, though, will take time. The IAU plans to make recommendations—possibly stipulating that the brightness of space objects in future should be dimmer than magnitude 7 (ie, invisible to the naked eye)—to the UN’s Committee on Peaceful Uses of Outer Space. That body may, one day, turn the package into a recommendation for a vote to better protect astronomy at the UN’s General Assembly. Such a vote might then lead to a new protocol under the Outer Space Treaty. But even if that all happens, it will not do so any time soon—not least because other issues raised by the mega constellations, such as risks from debris, will doubtless seem more pressing.

For professional astronomers an alternative solution to the problem might be to get over it not metaphorically, but physically: make ever more of their observations from space. This is not a perfect defence. If the telescope is in a low-ish orbit itself, satellites can still get in the way. A team led by Mark McCaughrean of the European Space Agency has found that in the 2010s the number of 11-minute exposures on the Hubble Space Telescope’s widest-field camera scarred by a satellite streak was 3.6%. In the first half of 2021 it was 6.6%. “It's gone up 80% in the space of a year,” says Dr McCaughrean. “And that's just at the beginning.” And because the Hubble is closer to the satellites in question, the streaks are out of focus; some show up as bands of light which fill up half the image.

Higher orbits are available. The James Webb Space Telescope (JWST), due to launch next month, will sit 1.5m kilometres from Earth free of all such concerns. Radio astronomers have long fantasised about setting up shop on the far side of the Moon, the only place in the solar system which none of Earth’s radio chatter can reach. But these are terrifically pricey projects. The JWST cost around $10bn—putting it in the same ball park as the entire Starlink constellation.

The same factors which allow commercial operators to launch many more and cheaper satellites could in time help science, too. But astronomy tends to depend on one-off flagship missions that simply do not respond to the same economies of scale. Changing its culture, and the expectations of its professional practitioners, would be a hard task.

And no satellite can replace the delight of star-gazing from your own back garden, or a local hilltop, with just a telescope or a pair of binoculars and the whole universe to look at. The intrusion of space-age infrastructure will not worry all such observers all the time—to see a satellite can be a thrilling thing. But the more routine the sight gets, the less likely it will be to elicit a frisson of the technological sublime. Eventually, it will become just another of the mundane ways in which the world obscures its wonder.