Starlink: low-earth orbit satellites could ruin radio astronomy

The grow­ing num­ber of satel­lites being launched into Earth’s low orbit is severely dis­rupt­ing astro­nom­ic­al obser­va­tions and, as a res­ult, space stud­ies as a whole. This is the latest warn­ing from research­ers at the Par­is Obser­vat­ory and the Nançay Radio Astro­nomy Obser­vat­ory. Without effect­ive mit­ig­a­tion strategies, par­tic­u­larly through col­lab­or­a­tion between astro­nomers and satel­lite oper­at­ors, the con­sequences for radio astro­nomy could be dis­astrous, they say.

It has long been known that sig­nals emit­ted by satel­lites inter­fere with astro­nom­ic­al obser­va­tions across dif­fer­ent wavelengths. In optic­al astro­nomy, satel­lites orbit­ing the Earth reflect sun­light, cre­at­ing bright streaks, or flares, in tele­scope images. In radio astro­nomy, glob­al nav­ig­a­tion satel­lite sys­tem (GNSS) satel­lites can con­trib­ute to radio inter­fer­ence, prob­ably through so-called ‘out-of-band’ emissions.

Xing Zhang. In our study, we noticed that SpaceX’s second-gen­er­a­tion Starlink satel­lites are emit­ting very intense out-of-band emis­sions (with flux dens­it­ies exceed­ing 500 Janskys, Jy) that light up the night sky in the radio fre­quency range. These emis­sions are start­ing to com­plic­ate our astro­nom­ic­al observations.

Indeed,Starlink aims to provide glob­al Inter­net cov­er­age with no delay between com­mu­nic­a­tions. This requires satel­lites orbit­ing at low alti­tude – typ­ic­ally about 500 kilo­met­ers – so that the sig­nal travel time between Earth and a satel­lite and back is very short, on the order of tens of mil­li­seconds. The prob­lem with low-alti­tude satel­lites, how­ever, is that they only see a small por­tion of the Earth at a time. This means that many satel­lites must be sim­ul­tan­eously put into orbit. Their num­ber is estim­ated at over 42,000 – equi­val­ent to the num­ber of square degrees of the whole sky. On aver­age, this trans­lates to one satel­lite per square degree; a colossal num­ber. Until now, we had some­thing like a few thou­sand satel­lites in oper­a­tion at any giv­en time. But this num­ber is grow­ing rap­idly, and Starlink is not the only play­er in this field: oth­er poten­tial large-scale con­stel­la­tions, such as OneWeb and Amazon’s Kuiper pro­ject, are also expec­ted to con­trib­ute to the grow­ing con­ges­tion in low Earth orbit.

This con­ges­tion could quickly lead to what is known as the Kessler syn­drome: there will be so much debris in orbit that sub­sequent launches will become less safe. The Earth will then become a kind of pris­on sur­roun­ded by debris pol­lut­ing not only the optic­al wavelength range, but in the radio band too.

This is dir­ectly impact­ing the work we do at Nançay, which involves search­ing for sig­nals from tar­geted exo­plan­ets. Instead of these, we recently observed some objects mov­ing across our radio images. We com­pared their tra­ject­or­ies with those of air­craft and satel­lites to find out what they were. It turned out that they were Starlink satellites.

Phil­ippe Zarka. Nor­mally, these satel­lites com­mu­nic­ate at very high fre­quen­cies, of above 12 giga­hertz, to send data. In prin­ciple, only radio astro­nomers work­ing in this fre­quency band should be dir­ectly affected. But the prob­lem is that the elec­tron­ic com­pon­ents on board these satel­lites seem to pro­duce their own radi­ation, which is out­side of this band.

XZ.  At Nançay, we wanted to test wheth­er we could detect the satel­lites in our images and if there was any­thing in the 10 to 85-mega­hertz range. We did this using obser­va­tions from Nen­u­FAR, a net­work of low-fre­quency radio tele­scopes optim­ized for these frequencies.

In 2023, we did­n’t detect any­thing. But a year later, we star­ted see­ing unusu­al sig­nals on our images, so we decided to track the Starlink satel­lites and com­pare the mov­ing objects on our images with these sig­nals. We then real­ised what was hap­pen­ing: while the first gen­er­a­tion of Starlink satel­lites also emit in radio bands, these emis­sions were not as bright at the fre­quen­cies we employ. The second-gen­er­a­tion emis­sions, on the oth­er hand, were much bright­er, which explained the unusu­al sig­nals and the source of the mov­ing objects in our images.

There were two main ver­sions of Starlink satel­lites: the early V0.9 to V1.5 mod­els and the more recent V2 Mini satel­lites. It has actu­ally been since the mass launch of the Mini V2 (around 40 satel­lites per week start­ing in 2023) that we have begun to observe a dra­mat­ic increase in light signals.

XZ. Indeed, their rep­res­ent­at­ives have begun par­ti­cip­at­ing in con­fer­ences on radio inter­fer­ence, which shows they are open to dia­logue with the sci­entif­ic com­munity. Some mit­ig­a­tion strategies have been tested, such as mak­ing satel­lites ‘invis­ible’ by tem­por­ar­ily dis­abling cer­tain onboard sys­tems when fly­ing over radio obser­vat­or­ies. How­ever, these approaches are cur­rently only real­ist­ic for large install­a­tions able to nego­ti­ate spe­cial agreements.

Some satel­lite oper­at­ors appear to be seek­ing to reduce out-of-band emis­sions in future ver­sions of their satel­lite sys­tems. While this would be a step in the right dir­ec­tion, these efforts have not yet been con­firmed or imple­men­ted in cur­rent fleets. In the mean­time, many hun­dreds of satel­lites will con­tin­ue to oper­ate in low orbit until they deor­bit, a pro­cess that typ­ic­ally takes between five and sev­en years. This poses a prob­lem in itself, as one can ima­gine the quant­ity of debris gen­er­ated when they do fall back to Earth.

The radio emis­sions from these satel­lites are 10 mil­lion times more intense than those emit­ted by the faintest celes­ti­al objects observ­able by LOFAR. They exceed the levels set by the Inter­na­tion­al Tele­com­mu­nic­a­tion Uni­on in the 150.5 to 153 MHz band alloc­ated to radio astro­nomy, but since they are form­ally clas­si­fied as ‘unin­ten­tion­al’ elec­tro­mag­net­ic pol­lu­tion, cur­rent legis­la­tion does not cov­er this infringement.

PZ. If these emis­sions were in a band that Starlink uses on pur­pose, only then would it be a leg­al issue. There are reg­u­lar inter­na­tion­al con­fer­ences that define by mutu­al agree­ment how the dif­fer­ent fre­quen­cies of the elec­tro­mag­net­ic spec­trum will be exploited. Some bands are reserved for defense, for example, while oth­ers are used for tele­com­mu­nic­a­tions, tele­vi­sion and radio. In the bands reserved for radio astro­nomy, no one is allowed to emit. How­ever, it turns out that when this emis­sion is unin­ten­tion­al, that is, acci­dent­al, you are pro­tec­ted by law. This is obvi­ously a loop­hole in the reg­u­la­tions. You could say that it is like inten­tion­al or unin­ten­tion­al hom­icide: these satel­lites are inad­vert­ently ‘killing’ cer­tain radio frequencies.

XZ. Per­haps these out-of-band trans­mis­sions were simply not anti­cip­ated dur­ing the devel­op­ment of these satel­lites. It is also pos­sible that no thor­ough emis­sion test­ing was car­ried out before launch. We just don’t know.

PZ. We agree that test­ing is no easy task. To obtain the entire spec­trum of a satel­lite’s elec­tro­mag­net­ic emis­sions, you have to put it in a Faraday cage and meas­ure all emit­ted fre­quen­cies. Since Starlink satel­lites are quite large (sev­er­al meters in dia­met­er), a very large Faraday cage is required, which is obvi­ously expens­ive to build. When we and oth­er radio astro­nomers pub­lish our find­ings, we are, in a sense, work­ing indir­ectly for Starlink and oth­er satel­lite com­pan­ies, as we are essen­tially doing their test­ing for them. How­ever, such pre­lim­in­ary test­ing should be man­dat­ory and included in reg­u­la­tions: no one should be allowed to launch objects into space without exhaust­ive pri­or analyses.

By pub­lish­ing our res­ults, we want to make astro­nomers aware that if they see any form of con­tam­in­a­tion in their radio images, they know where it comes from. Radio astro­nomy is pro­gress­ing at a fast pace across all fre­quen­cies, and there is a lot of research these days on ‘tran­si­ent’ events. These are phe­nom­ena that appear only for a short peri­od of time and include cata­clys­mic events and explo­sions, fast radio bursts and pulsars, to name but a few.

We can make an ana­logy with optic­al astro­nomy here: when you are look­ing for stars or galax­ies whose pos­i­tions are fixed in the sky and streaks of light from satel­lites appear in the image, this is a prob­lem, but you can ‘com­pensate’ for these trails. You can­not con­fuse them with the image of the object being observed. But, if you are observing trans­it­ory sig­nals, that is, sig­nals emit­ted by vari­able objects, you may of course con­fuse some satel­lite sig­nals with the object you are observing. The same applies to radio astro­nomy images.

PZ. Starlink satel­lites prob­ably emit their radio pol­lu­tion con­tinu­ously, but since they only cross your field of vis­ion for just a few seconds, you observe an event that actu­ally lasts for a very short time when you look at cer­tain objects. You there­fore need to be extremely cau­tious and com­pare all the emis­sions you detect with those of satel­lites to elim­in­ate them.

XZ. Ini­tially, we thought we would­n’t have any prob­lems with polar­ised emis­sions like those emit­ted by exo­plan­ets, because inter­fer­ence is mainly unpo­lar­ised. How­ever, Starlink inter­fer­ence is also polar­ised, which poses a real prob­lem and com­plic­ates the study of the tran­si­ent radio sky. With Nen­u­FAR, no mat­ter where you look in the sky, you typ­ic­ally have a Starlink cross­ing your radio beam every 15 minutes. It makes it com­plic­ated for us, and the situ­ation will only get worse in the future if noth­ing is done.

PZ. For me, this is one of the most ser­i­ous prob­lems with Starlink and oth­er such satel­lites. People had already lost their sense of won­der of the sky and the stars because of light pol­lu­tion in cit­ies, but they could always go to the coun­tryside or to a deser­ted place to redis­cov­er that magic. With Starlink, there are trails of light all over the night sky. You could say that Starlink is steal­ing the dark sky from human­ity. We no longer have a vir­gin sky. This, I would say, is a giant hold-up, and the whole of human­ity is the victim.

Interview by Isabelle Dumé