Starlink Satellites Posing Issues For Astronomers

Spotting satellites from the ground is a popular pastime among amateur astronomers. Typically, the ISS and Iridium satellites have been common sightings, with their orbits and design causing them to appear sufficiently bright in the sky. More recently, SpaceX’s mass launches of Starlink satellites have been drawing attention for the wrong reasons.

A capture from the Cerro Telolo observatory, showing the many Starlink satellite tracks spoiling the exposure.

Starlink is a project run by SpaceX to provide internet via satellite, using a variety of techniques to keep latency down and bandwidth high. There’s talk of inter-satellite laser communications, autonomous obstacle avoidance, and special designs to limit the amount of space junk created. We’ve covered the technology in a comprehensive post earlier this year.

The Starlink craft have long worried astronomers, who rely on a dark and unobstructed view of the sky to carry out their work. There are now large numbers of the satellites in relatively low orbits, and the craft have a high albedo, meaning they reflect a significant amount of the sunlight that hits them. With the craft also launching in a closely-packed train formation, there have already been impacts on research operations.

There is some hope that as the craft move to higher orbits when they enter service, this problem will be reduced. SpaceX are also reportedly considering modifications to the design to reduce albedo, helping to keep the astronomy community onside. Regardless, with plans on the table to launch anywhere from 12,000 to 42,000 satellites, it’s likely this isn’t the last we’ll hear about the issue.

117 thoughts on “Starlink Satellites Posing Issues For Astronomers

  1. I’d rather they stayed in lower orbit to reduce the risks from space junk.

    Pointing camera’s at the skies and complaining about the crummy pictures is pretty silly. If it wasn’t for starlink it’d be light pollution, or the atmosphere.

    1. Considering we point cameras at the sky to detect Near Earth Objects, I wouldn’t call complaining about what amounts to photobombing silly. Unless SpaceX wants to fund and put into place orbital or lunar telescope platforms they have a responsibility to not become a burden to scientific discovery that dates back to Galileo.

      I’m not against using starlink for communications, but if the end results will be space junk and making the night sky less visible, we need to re-evaluate the cost/benefit equation.

      1. They don’t have that responsibility actually.

        The photobombing is entirely predictable which means it’s not photobombing.

        Star link will provide Internet I suspect astronomers who can be bothered to track predictable orbits of known satellites won’t be much use to anyone…

        1. Do you even have an idea how rare and expensive observation time on a big telescope is?
          Scientists all around the world have to wait for years to get their time slot. And if this slot is spoiled by Starlink, the preparation work of months goes down the drain.

          Does SpaceX pay for that?

          1. “Do you even have an idea how rare and expensive observation time on a big telescope is?”

            Of course he doesn’t. The authority with which he speaks on the subject is inversely proportional to the knowledge and experience he has with said subject.

            It’s an all too common trait today, and it’s doing devastating and long lasting damage to society.

          2. So why wouldn’t they check to make sure there’s no satellite in the way during their observation? The amount of sky a powerful telescope observes at once is tiny, it would take a fraction of a second for an object in a low orbit to pass through.

          3. @Shannon “The amount of sky a powerful telescope observes at once is tiny”

            the majority of earth telescopes are more wide than are long, its more about the exposure and less about the lengh/zoom

          4. Seems your objections moved from moral to financial.

            Which I personally don’t give two hoots about. I’d rather people have access to internet not to mention the cutting edge tech starlink is exploring.

            If astronomers paid for internet to rural areas starlink wouldn’t be needed that’s as silly as asking starlink to pay for astronomy time.

          5. >”So why wouldn’t they check to make sure there’s no satellite in the way during their observation?”

            Did you miss the part where the full Starlink constellation would be covering the entire sky with satellites in a grid approximately 400 miles apart? There would be no place on earth, no time of day or night, when you wouldn’t get satellites streaking past your telescope – except on the north and south poles where the orbits don’t reach.

        2. By your logic, if my dog dumps thousands of times a day on the part of your driveway that crosses the shared verge along the street according to a schedule it is not my responsibility, because it is a) predictable, b) you can step over it, and c) the run off is providing a benefit to nearby plants.

          1. Only if you consider space owned in the sense public land is.

            Thankfully this isn’t how we think of it (not thinking of it this way as prevented scary escalations during the Cold War as well as between DPRK AND Japan). As a consequence I consider it more important than the downsides of not externalising usage of space.

      2. I’m not sure why we’d want the desires of telescope operators to act as our gatekeepers to the development of space technology. Taking your argument to the logical end, we ban all space launches and all artificial lighting.

        There has to be a middle ground that allows for development. Suggesting that SpaceX eats the cost for the James Webb because somebody is annoyed that satellites exist is a bit too far.

        1. We’re not talking about a few satellites here. The amount for the whole constellation is staggering. Telescopes aside that orbit is going to get mighty crowded fast. That they take measures to make the birds less of an optical impairment is just one part they need to get right. The junk the constellation will cause is an even bigger issue.

      3. That’s true – if a Near Earth Object object where to track the exact orbit of one of the Starlink satellites in such a way that the Starlink satellite was constantly somehow between the object and all land based telescopes it could make it much harder to detect. Luckily such an orbit would also preclude it from being potentially harmful to us.

    2. “If it wasn’t for starlink it’d be light pollution, or the atmosphere.”

      as a Amateur astro photographer i already drive for several hours to avoid LP, guess i will have to kust avoid their orbit… and stay at home.

      1. It’s “great”, isn’t it? Light pollution in my area leaves the visible sky at a magnitude 3. I have to drive at least an hour and a half to get to a site to see. When we had the Northeast power outage, I spent most of the night explaining what “that thin, white, streaky cloud across the sky is”.

        I couldn’t imagine doing terrestrial astrophotography with SpaceX’s crap streaking across the sky, I imagine it’ll be even worse when they have 1,000 more satellites up there. So much for the professionals doing science…

      2. As an amateur astro photographer, you should know that any satellite is only visible when it’s dark on the ground but the satellite is in sunlight. This only happens just after sunset or just before sunrise. You will still get very dark very satellite free skies even if they don’t some the reflectivity problem.

    3. Using LEO satellites to provide internet is a stupid idea for an awful lot of reasons, and astronomy ain’t one of them (well maybe one of the minor ones).

      If you thought cellphone coverage sucked in the early 90s, just wait until you try to connect to a Starlink! The fundamental problem with this idea is the same one driving terrestrial network providers to build towers with smaller and smaller coverage areas. Smaller coverage allows more tower density, which equates to allow more users in the same area. Bell labs told us that in the 1960s when they first studied cellular services. Starlink will follow the same basic physical rules.

      I haven’t even gotten into the real problems Starlink is going to have. Let’s just say that Musks network speed and latency claims are going to be much much much harder to achieve than he makes it sound. Routing signals between satellites to save ground hops is an incredibly difficult engineering challenge, and it doesn’t scale linearly. It’s also incredibly unreliable under the best of circumstances, like when we tried it with the WGS birds.

      I wouldn’t put too much stock into Musks claims that it will be for poor folks either. Simple deduction tells us that these things aren’t free to construct, launch or maintain. We barely got the US to fund iridium and the US military ceases to function without access to that. Most of the time they’ll be covering huge expenses of nothing (Pacific ocean, Indian ocean, etc.) which means no income. This thing is either going to flop, and flop hard, or it’s going to cost an absolute for service. Nothings free.

  2. Wouldn’t it be relatively easy to turn off your light gathering while a satellite is passing overhead? I’m sure the orbits are published, and they pass overhead relatively quickly

      1. 42000!

        And Musk is not the only one planning to put a ridiculous amount of sats into orbit for intermet and such…
        Kessler Syndrome, here we come!

        I like that!
        When the Kessler Syndrome hits us (and it will!) we have no way any more to reach a plan(et)-b and need to clean up planet-a aka earth.

        That’s really good news!

        Go Elon!
        Throw us back into the early 20th century without satellites and space exploration!
        That’ll give us the time to clean the hut!

    1. Yeah exactly its hopeless trying to astronomy without keeping tract of the influence of known satellites

      I’d call into question the research chops of an astronomers who couldn’t deal with this.

      Unfair to deprive internet (targeted at already deprived people’s) because you’re too lazy to do your job properly.

      1. Heh. You call into question the physics of astronomy? If one is going to be mad about a group of people it should be all the oligopolies that created the “deprived” class in the first place. LEO is an end-run around their machinations (that included political) in ensuring people don’t have reasonably priced, capable broadband.

      2. Oh, you’d call into the question the research chops of astronomers who have concerns with this, would you?

        Mind sharing your credentials in the field of serious astronomy?

        Oh, that’s right, you have none. Zero. You’re just another example of the tragically common trend of feeling compelled to not just have an opinion on every subject that happens to grace one’s web browser, but to speak authoritatively on said subject.

        I think, therefore I am… an expert on everything.

        Richard Feynman once said “I have approximate answers and possible beliefs and different degrees of certainty about different things, but I’m not absolutely certain of anything and there’s many things I don’t know anything about… but I don’t HAVE to have an answer, I’m not threatened by not knowing something”.

        And in stark contrast to a man who knew so much… are people who know so little and, yet, purport to evaluate the knowledge of those who’ve dedicated their careers to a field they have zero credentials in.

        1. Ha! An appeal to authority that contains an appeal to authority.

          But still, we are talking about a LOT of sats. I don’t think there is anything to be done about it. The benefits are just too high when compared to some PhD candidate who wants to refine some data about central black holes in galaxies. Data that is millions of billions of years old when the researcher gets it. Higher priority jobs, like NEO detection, and eventually all serious astronomy, will just have to move to high orbit or solar orbits or the Moon. I think very good orbital telescopes are now affordable by groups of amateur astronomers. Check launch fees for Space-X.

          1. >”The benefits are just too high ”

            What benefits? Terrestrial cellular internet is already a thing, and the latency or bandwidth to the satellite constellation is going to be mediocre at best.

            The only thing the Starlink is good at is providing a US based company global access to evolving markets such as in developing countries, where they can become a monopoly by beating the local competition to the punch. With people using Elon’s space internet, the local competition can’t get off the ground because they face the cost of building the infrastructure first.

          2. 1) Calling into question the credibility of any astronomer who did complain about the matter.

            2) Demanding one to be a creditable astronomer in order to have a valid opinion or complaint on the matter.

            That’s actually a kind of no-true-scotsman. Basically what you’re saying is: if you’re not a renowned astronomer, you don’t get to complain. If you are, and you still complain, then your expertise would be questionable and therefore your opinions would be invalid. Devil if you do, damn if you don’t.

          1. ehmmm… I’m confused: is this “Starlink” (satellite internet) project for the benefit of human kind? Or is it just another business model (capable of producing a new generation of e-waste)?

            Somehow I’m not convinced this project is as noble as some people would like to believe.Sure the idea sound great, but the costs are huge. It needs to be maintained and devices on the ground need to connect to it and all those things aren’t free. Then if it works like intended, don’t expect to stream your latest youtube vid in 4K when you and the rest of the continent are connected to the limited group of available sat’s. But we’ll see. I’ve been wrong many times before.

          2. With one satellite serving an area of 200-400 miles across, the bandwidth allocation per customer is going to be tiny. This is not global wireless broadband – this is satellite phone in Africa level of communication.

            The major difference to other satellite internet providers is the low orbit, which reduces the signal latency to somewhere close to useful, but it’s still not going to be DSL fast. More like dialup fast.

    2. To have continuous satellite coverage you have to continuously have a satellite in view (view as in line of sight.) In the hours around sunrise and sunset then there will almost always be a satellite visible.

        1. If you look lower towards the horizon, you’re looking at the satellite orbits from the side, and the gaps between the orbits become very much narrower. Kinda like how you can look up and see a huge gap in the clouds, but look towards the horizon and the clouds fill the sky.

          And the orbits aren’t geosynchronous. The earth keeps spinning underneath the satellites, so you can’t aim “between the grid”. The grid sweeps in both directions, so if you want to take a long exposure photo of the sky, some of the satellites will pass through the frame.

    3. It is not that simple. Often astronomers only have a small window of opportunity to look at what they want to look at and with exposure times in the often up to 4 hour range, avoiding satellites is absolutely impossible. The best solution would be for SpaceX to help build and put into orbit a few Hubble like satellites. I do not think that will ever happen.

      1. A hubble like satellite is NOT a replacement for earth based scopes like the ELT, LSST, Large Binary Telescope, etc. While Hubble makes awesome photographs, it has very different applications and capabilities. One is not a replacement or alternative for the other.

        1. But a satellite telescope could be designed to do whatever you want within limits of size vs cost… why not look at this as an opportunity, design a satellite to fit in the spacex launcher so you can piggy back on one its (cheaper?) launches. If say a dozen of these were launched so they could go into a low energy earth moon transfer then you would end up with an array well outside earth influence that while it might not be perfect for some things could open up a whole new host of opportunities.

      2. I agree that ‘avoiding’ satellites is impossible, I’m suggesting coping with satellites. Given the absolutely amazing things modern astronomy is doing with adaptive optics, arrays of reflectors, etc. the relatively simple task of calculating orbits is easily within their grasp. Modern astronomical sensors aren’t photographic plates that couldn’t ignore some input, they’re computer inputs that stream their data in real time. The 4 hour exposure is the composite of all of the input during that time.

    4. Yes. It’s what they do now for frequently unpredictable airplanes, which are two orders of magnitude brighter than all the satellites in the sky combined. Grats, you’ve discovered why this story is nonsense. You can literally follow the money to the K Street PR firm pushing this fake story.

    1. Well judging by some of the comments it’s more “consumer interest vs scientific interests”. Not that the consumer is well equipped to make a suitable argument beyond “me want” which is how a lot of progress is justified.

  3. Some of these comments are just freaking tragic. If these satellites interrupted a damn football game the world would come to an end… But Astronomers – ahh screw ’em, who needs to look at the stars anyways..

    Amateur astronomers discover a great number of the “falling rocks” people are afraid of – not to mention, there are many more of us than people realize.

    As to the “well, just don’t collect data when they are overhead”… Let me ask you a question – when you practice your hobby – do you have to do hours of research prior to doing it? Then, do you have another hour or more of setup time and calibration time? Finally, once you are ready to actually perform the “hobby” – do you have to sit out in the wee hours of the morning for multiple nights during a specific time window to collect data on that one object you only get a chance to see for a month or two a year? yeah – now lets add another hour or two onto that so wee can plan for the damn swarm of satellites and likely lose data because the timing is bad.

    It amazes me – people will chain themselves to a tree to save a forest – but pollute the sky with thousands objects and no one gives a damn.

    To those people who have never even given a second though to astronomy – please do us all a favor and get educated before you lecture us on something you know nothing about.

          1. If you have one satellite for every 400 miles square, then you have potentially millions of users per each. No satellite has petabytes of bandwidth, and with an increasing number of users, the system efficiency goes down with more flow control messaging going on than actual payload data. It’s unlikely that any one connected customer would have more than dialup speed.

            If you have 1 gbps per satellite, and a thousand customers per satellite, you get 1 mbps per customer, which is at best entry level ADSL from the early 2000’s which is capable up to 24 mpbs. The practical maximum speed is around 100 kBps which is about as fast as 2G/EDGE cellular internet. You can’t even watch youtube on that, unless you like buffering.

          2. Also, wireless routing doesn’t work like wired routing where unused bandwidth can be taken up by someone else automatically. You have to reserve transmission slots in order to avoid collisions, so the bandwidth goes down when other people join the network even if they aren’t fully using their bandwidth.

          3. @Luke

            This is only the case in the uplink. The downlink in most VSAT systems is a single carrier with all the packets. No reservation is needed except for SLA purposes.

    1. I also thank you for your comment.

      “To those people who have never even given a second though to astronomy – please do us all a favor and get educated before you lecture us on something you know nothing about.”

      But that takes effort and interest… which the people you’re imploring have neither.

    2. I’m an astronomer myself, I see no issue with some more satellites passing overhead, I am among the ‘exclude the data from the time they’re in view’ bunch. Realistically for astronomical photography all of this will be handled by software, you’re not going to have to do any additional preparation. If you’re doing live observation you’ll get the briefest of flickers.
      I’m pretty sure that all hobbies that you get into in a serious way include hours of research and setup, what wouldn’t? Maybe pogs?

      1. the thing is that if you are making a 5 or 10min exposure most likely you will have several exposures that will be trow away by the constellation.

        I use APT and im sure it would be super easy to add that option to not shot if the exposure will have a satelite in view, but if its 10min expusure, its gonna be a hard time getting 5-10h integration.

  4. Alarmed by the realization that the sun was warming the earth and threatening rising oceans, Elon Musk devised a strategy to shroud the earth in a permanent blanket of sun blocking particles… launch a satellite network of thousands of satellites, and once they were evenly distributed above the earth, detonate them, releasing a cloud of fine, light blocking debris similar to the pollution created by daily air travel here on earth, but twice as dense. Predictably, this flung the earth into a new ice age, which could only be remedied by ceasing all air travel to reduce aerosol particulate pollution in the atmosphere to bring the net solar gain back up to a comfortable level. This also increased demand for Hyper Loop construction.

    I swear, it’s the kind of thing a joint toking billionaire would come up with.

    1. It would take a whole lot of fine particles to do that. Already the earth gets over 100 tons of space dust landing on it, Yet it has no effect and its not even really visible. When you got a nice bright blue sky day, position yourself so the sun is blocked by roof edge and look up. You can see a lot of stuff flying through the air, pollen, dust, trash and bird feathers being carried by high winds a few thousand feet up. It would take an amazing amount of material to accomplish that and cool the earth. Would basically have to have non stop flights dropping it!

  5. Just a thought. As I understand it these “long exposure times” are actually multiple images. They are stacked and an algorithm is used to eliminate the random noise. Couldn’t this method to be used to eliminate the satellite? May not work for all, but some.

    1. This is true – but for “deep sky” faint astro-photos, the images are still from several minutes to even hours in some cases… more than enough time for several of these streaks to appear and ruin the data.

      The problem is that the software is already working with very minute differences in gradient – and applying all kinds of smoothing and sharpening algorithms to try and eek out the details – which some times just makes a giant mess of things and you have to start over anyways.

      I am no expert in all the techniques – having only used 3 or 4 of the packages over the years.. but its a maddening process at times. You can literally spend hours trying to get more detail out of a series of images only to realize you’ll never get there and you need more data. So – throwing even a frame or two out is painful after all that work.

      I do think that the software will evolve to help with these type of issues – it has to, because I don’t think the issue is going away…

      1. The issue with stacking is that while it removes random uncorrelated noise, it amplifies the correlated noise or artifacts from the sensors.

        You always have “hot pixels” in your sensor. These can be compensated for by mapping the sensor, but this is never perfect. If you stack a thousand pictures, some residual error will remain and some pixels or groups of pixels will become lighter and darker. Each exposure adds the remaining offset error to the sum,

        This is why you can’t perfectly replace one long exposure with many short exposures.

    2. Yes absolutely. It is fairly easy to remove streaks for planes and satellites,. I have done it before many time. you just have to use a subtractive algorithm. There would be no need to throw the whole frame out.

    1. If you coat the whole sat with it, the albedo will go down a lot. The production of the solar panels (usually the most reflective part, then the antennas) on it probably also goes down a lot. It will also heat up more.

      1. Not to mention the fact that deep space objects that might be on a collision course with earth are usually spotted by the “shadow” as they cross in front of stars! So we could possibly lose important detection time of a impactor that could be heading our way. Not all astronomy is on the visible spectrum or imaged the way you think it is.

  6. Or in other words, just another company polluting the “environment” for their own benefit (profit) without thought or cause to anyone else.

    On earth, in space, it doesn’t matter.
    We just allow companies to do what they want with no concequences for them, but plenty for all of us.

  7. What we need are WiFi seeking rail guns. Where did I put the parabolic antenna and the copper rods..?

    Any how many do we have to take out before the debris causes a chain reaction, eliminating anything useful from the whole orbital plane.

  8. Yes, Starlink satellites will be a nuisance, to say the least. But let’s put some very rough numbers:
    • An object at 1200Km up can see an horizon 4000Km away, and thus can be seen from the Earth’s surface 4000Km around.
    • A (flat) circle 4000Km in radius have a surface area of 50 million Km²
    • The whole Earth has a surface area of 510 million Km.
    • So, assuming an uniform distribution of Starlink satelites at a given moment, 50/510 of the 12000 to 42000 total satellites would be visible from some point on the surface
    That’s 1176 to 4117 satellites would be visible. That’s A LOT.

    But let’s also have some perspective:
    • Satellites are very easy to differentiate from starts and planets
    • A considerable proportion of the sky is blocked by clouds anyway
    • What is considered clear, dark sky is in reality more or less murky with atmosphere
    • A similar debate has been raging for decades regarding public lightitng (astronomical observatories just went far away from cities and to space)

    So, the in-hindsight-unavoidable future has arrived: advanced astronomy must be moved to space just like it moved away from cities, and in fact a good part has done so. The benefits of global internet outweights the trouble to astronomy, just as the benefit of city lights did.

    1. I’m just realizing more numbers:
      • Less than 1/4 of the Earth’s surface is usable for astronomical observatories, on account of 3/4 being covered in oceans. That leaves us 294 to 1029 satellites visible from the ground at any given night. Still a lot, but not that impressive.
      • Advaced terrestrial astronomy, the one that can be affected by passing satellites, is performed long after disk and before dawn, let’s say from 8PM to 4AM. That’s 8 hours out of 24. That’s 1/3 of the time. That leaves us 98 to 343 satellites visible from the ground at the usable time of any given night. Starting to look like not that big of a problem.
      • During the post-dusk and pre-dawn time of every night, the satellites are in Earth’s shadow, and thus are NOT streaks of light! If anything, there would be a blink-and-you-miss-it occultation.

      No. The benefits of global internet far outweights the inconvenience to Earth-based astronomy. Despite being a star observer myself (not reaching extremely amateurish astronomer even), and often longing to darker skies, I think professional astronomers should seize this situation to push for more space telescopes.

  9. Just out of curiosity, but I seem to remember something called vanta black.
    It was supposed to be the blackest black there is. Why not paint the satellites black?
    The sun would heat them up more, but can’t these things be insulated against that?
    I’m no astronomer, but I do like to see the stars from time to time.
    With all the light pollution we have here in the USA, there are fewer and fewer places where one
    can see the Milky Way. Kind of sad really, people are so involved in themselves they don’t take the
    time to even look up and see the majesty of our universe.

  10. Snarky, oversimplified comment:
    Blame Comcast/Verizon/evil telecom co. for their crappy service and regulatory capture necessitating the use of space to provide decent broadband :P

  11. Wow, lots of drama here.

    The good news is that SpaceX has already announced that they’re working on the problem. Unlike most orbital assets, they are using what software people would call an “agile” approach, meaning that mistakes and problems are (relatively) short-lived.

    Hopefully, the next iteration will have better albedo characteristics.

  12. Let us not forget that the natural beauty of a night sky full of stars, without a constant
    flux of artificial satellites, has much esthetic and spiritual value. These values can not
    be measured in monetary ways. What is the value of seeing the Milky Way, or a nice
    meteor from a dark site away from the light pollution that plagues most natural urban
    nighttime environments already? It will further diminish our sense of being human and
    our oneness with nature.

    1. Yeah, astronomers may be able to work out a technical solution to avoid these satellites, but does SpaceX really have the right to just pollute the night sky for all the regular people out there?

      The night sky is a beautiful resource that everyone shares, and Elon is just pissing all over it for profits. We need actual regulation on light pollution and satellite visibility.

  13. It seems to me like there is a pretty straightforward technological fix for this problem, and it’s actually a really interesting problem to dig into and produce some imagery software that astronomers can use to get good imagery *despite* the increasing presence of LEO satellites. Even if Starlink weren’t happening right now, the problem is only going to get more relevant as putting objects into LEO is getting more and more accessible. The large number of starlink sats is just bringing the issue to everyone’s attention right now, because it is really hard to ignore several dozen streaks through a long exposure photograph.

    The funny thing is, astronomers are, er, astronomically good at dealing with artifacts caused by objects in between the telescope and the thing they want to look at. A lot of the field of adaptive optics was developed by astronomers trying to deal with things in the sky, usually water vapor, dust, and air, affecting the path of photons/radio waves coming from far-away targets. LEO satellites are just another thing.

    The key advantage we have with LEO satellites is that we know where they are to a high degree of precision. Here is tracking data for one of the recent starlink satellites that this article is discussing: https://www.n2yo.com/satellite/?s=44772 . It seems to me like this is an interesting (and solvable) math problem more than a cause for dismay.

    Let’s start with a simple truth: the long exposure trails from LEO satellites are inevitable. Sure — you can argue that Starlink isn’t really necessary, and plenty of people are wasting breath doing just that, but here is a log of all the launches that went up this year: http://www.spacelaunchreport.com/log2019.html#log . There are lots of interesting reasons to launch constellations into LEO, if it’s not Starlink this year, it’ll be another one next year, or just the aggregate number of LEO satellites in the sky as more and more low-cost launch providers come online (Rocketlab, Blue Origin, whoever else is currently realizing that rockets are imminently do-able as a private company). Sooner or later, this problem, which boils down to “from t = t0 to t1, something will cross my camera’s field of view” is going to become more and more relevant.

    What this really means is that “dumb” long exposure isn’t going to work for ground-based astronomy. As the astronomers in the article point out, it’s not really working now, and it’s just going to get worse as more stuff gets up there. So if dumb long-exposure (“sensor opens up and starts collecting photons, stops collecting photons after the exposure interval”) doesn’t work, how about smarter long-exposure?

    Smarter long exposure starts with the informational advantage — we know exactly when a LEO satellite is going to cross my telescope’s field of view, and we know exactly where it’s going to be in the sky. Why not just stop collecting data during the affected time?

    Consider this — ground-based telescopes are shooting long-exposure photographs, from dozens of seconds to hours, already, and they are already tracking their targets as they shoot to compensate for the earth’s rotation during the exposure. A LEO object crossing the tiny field of view of a telescope (it varies a lot, but let’s think about 0.1 – 3 degrees (6-180 arcmins) as a common range for nice, ground-based scopes) is going to cross the telescope’s field of view in a few seconds, so let’s use the ol’ s=r* theta formula to figure out transit time. We know that the ISS is moving at 10km/sec along its orbit, and it’s ~400km up, so what angle does it subtend in that second? Soo, theta (per second)=s/r = (206265 arcseconds/radian * ISS speed of 10 km/sec along orbit / 400km altitude) = 5156 arcseconds /sec or 86 arcminutes/sec or 1.4 degrees/sec. That means, if I’m taking a long-exposure photograph with my nice, 1 degree/60 arcmin field-of-view ground scope, and I know the ISS is going to pass directly through my shot, I know the ISS is going to be completely across the field of view in under a second. Even if I’m tracking an object in the solar system, the rotational speed of any object relative to something in LEO is going to be itty-bitty, and I’ll just have the LEO object in my FOV for a few seconds.
    I have a couple options for how to deal with the ISS: Let’s say that I’m taking a really long, tracked exposure because I’m looking through a particular wavelength filter at some dim stars that are quite far out, and I need to do a 10 minute exposure for my shot. If I know when the ISS is going to appear, the simplest thing to do is to pause integration on my sensor or save my entire exposure just before it appears, and resume the exposure/start exposing a new image a couple seconds later, once the ISS is out of my field of view. After my shot, I either treat the entire exposure like a continuous long exposure (even though I paused integration for a few seconds midway through) or I combine the two images to make a composite long exposure, which is no big deal for me because all astronomers are fluent in computational photography by necessity. This was an interruption, but it only affected me for 0.3% of my exposure time.

    So is that all that this fuss is about? Astronomers not wanting to turn off their long exposures for a couple seconds whenever a LEO object is crossing their path? Well, not exactly. Astronomers are right to realize that the times are a-changing. We only have 12,000 tracked objects in LEO right now, and Starlink is proposing adding 12,000-42,000 more satellites into LEO, increasing the headache for astronomers* by 2-4.5x. It’s not inconceivable that we’ll see a couple starlink-scale projects in the next decade or so (after all, what is cox without time warner? Verizon without Sprint? Batman without Joker?), so maybe this LEO object number balloons up to ~100,000 objects by 2040. ‘Democratized space’ sounds much cooler than ‘satellite traffic jams’

    *this will increase the headache, not only for astronomers, but for everyone. Consider also, the pile of space rubble and problems that emerge when two of those satellites smack into one another and turn into 50,000 more satellite bits, each of which can smack into the other 99,999 satellites zipping around. Neil Stephenson explores a similar math problem in his excellent book seveneves, and spoiler alert, the math has some real-world effects. Whoever invents the space-swiffer is going to be very wealthy in the future.

    So let’s say I’m an astronomer in the year 2040, and I’m all set up to take my nice, peaceful long exposure photograph for ten minutes. *Click* Oh no, there are 100,000 things zipping over my head. Now, most of those things aren’t in a position to reflect sunlight onto my telescope sensor, but they are just soaking up photons by being in between my subject and my lens for a moment. The really annoying bits are the satellites that are reflecting light from the environment onto some part of my sensor. That doesn’t happen every time an object passes through my shot, but only when the orbit, rotation of the earth, and orientation of the satellite are all in alignment. For the sake of a first crack at the problem, though, let’s assume that if the satellite is in my field of view, we need to pay attention to it.

    Well, LEO orbits are around 90 minutes, so I’m only ever going to see each satellite once, at most. The orbits aren’t evenly distributed (some orbits are more useful than others), but just to make the math easy, let’s pretend for a moment that they are. I’m also never going to see most of these objects — one of them, for example, is Ed White’s glove that he dropped during a spacewalk in the sixties. Great space memorabilia, not really going to mess up my shot. But what is? Let’s look at the 160 brightest objects in the sky right now:

    (sorry people who like using the internet, I can only paste links here, not images. https://share.getcloudapp.com/qGuzmeLl )

    That’s the path of the 160 brightest objects in the sky going over my head here in Rhode Island. There are about 2000 active satellites up there, not counting the secret military ones that are talking to us through the fillings in our teeth. My poor macbook air can’t handle all that data but here’s about half of the active satellite orbits.

    https://cl.ly/76f1f588034a

    This visualization is from Celestrak (http://celestrak.com/NORAD/elements/), and it’s pulling data from the space track dataset of all tracked orbital objects. These paths in the image are showing me a full orbit. That means that something is going to pass along there sometime in the next 90 minutes, and assuming even distribution of orbits, there’s about a 1/9 chance (10 minute exposure / 90 minute orbit) that something traveling along one of those paths will show up there during my exposure.

    Well, that’s still not too too crowded. I wish I could superimpose the field of view of my telescope onto the image, but the Celestrak visualizer doesn’t actually give me enough information to accurately calculate the projected area of my telescope, but I’m going to wave my hands and say, eyyyyy!, of the 2,336 active satellites, maybe one of those would pass across my 10-minute exposure. In reality, we have a lot of polar and equatorial orbits, which means that our friends in Southern Chile and Mauna Kea are going to have more of a problem than us plebes hanging out at middle latitudes. Sorry, astronomy friends. Yay plebes!

    Now, obviously, this problem is unavoidable if you’re hanging out all night taking pictures of the sky. You are 100% guaranteed to have some LEO objects cross your field of view at some point if you go out and take pictures tonight, and that’s only going to be more common when we’re at 10x the number of objects in the sky.

    If you think about it, though, this problem isn’t crazy — If I have to pause my long exposure photo ten times for 1-2 seconds during a 600-second exposure while I let satellites pass, it’s not so frequent that a modest computer can’t handle the problem. Pausing my exposure 100 times during the same exposure starts to get annoying. In ten years, astronomers at equatorial or polar locations are going to be looking at that kind of frequency. That might make a 600 second exposure into an 800 second exposure, and if you’re doing that kind of exposure all night, every thing, that really messes up your day. er, night. Can we do anything smarter than just pausing the entire image when a satellite goes overhead?

    This is where computational photography really, er, shines. We actually have enough information here to know when a LEO object could affect any pixel in our telescope at any time. We do a camera re-projection, knowing our telescope’s extrinsics (where it is on the planet, and where it is pointing), and the telescope’s intrinsics (FOV of the lens, lens distortion, etc), and we know, for a given satellite position at time t, which pixels in the telescope’s imager the satellite is going to be affected by the object. This opens up a whole world of adaptive exposure, where we can decide if a given pixel is compromised at a given moment, remove the compromised period from the exposure, and tack on a couple seconds to the end of the exposure for that pixel so every pixel in the image has an equal exposure time. It might add a couple seconds, total, onto a 600 second exposure, but it’s better than streaks in your image and better than shouting angrily into the void.

    I’ll end this comment-essay by saying that I am not an astronomer, nor do I work in space, but I enjoy thinking about both topics. If any astronomers or, uh, space-professionals happen to be reading this and notice any huge errors in accuracy or judgement in my math, let me know. I do a lot of work in computational photography, and am always surprised and delighted to see the same type of photography problem crop up across a lot of disciplines, and I like thinking about how the same type of computational photography thinking can apply to all those problems.

    See you out in the universe,
    Alex

    1. sooo what do you do for a living?

      excelente wall of text, absolutely cant spot anything out of the reality or possibility, other than pixel tracking, that will not gona happen, only bilion dollar mount could probably have such perdition of the path, sure you can do plate solving and not rely on the mount, but even so, avoiding the satélite is not gonna happen on these terms.

      1. I’m reading between the lines a bit here and I think you’re saying that it’s impossible to calculate which pixel on a sensor is going to be affected by a passing object. Why not? We don’t think twice about solving the problem in closer-distance computational photography — if you have a fixed calibration image (like the pattern of stars in the far distance in the telescope’s field of view) so that you can account for lens (and in this case, atmospheric) distortion, we don’t think twice about mapping the image of an object from 3D space onto the projected image on the sensor plane. Why do you think the problem is fundamentally different for a telescope?

        1. because there is still a big gap in accuracy.

          100ms in clock offset trows the position of the sattelite off
          3º temperature diferencial since calibration trow focus off
          platesolving still have some degree of error
          trackers get errors all the time, we (amateur astronomers at least), cant even keep a star(a fix point in the sky) for guiding in place.

          1. Sure, that makes sense — I think the per-pixel exposure control only really makes sense if you have a super-accurate setup, and is only really necessary if you’re shooting constant long-exposures, all the time, like a research telescope does. The time-based control could work for just about any setup, even simple amateur ones — you could work with off-the-shelf GPS and tracker data, and just increase your time window as your margin of error on tracking goes up. For most amateur uses, adding a few seconds of pausing per satellite is not a big deal.

  14. Pixel Tracking?

    You have the data, After the fact, detect and remove the affected pixels for the moment they are being affected. Even a home user has that much processing power. No need to track all the objects which you can clearly detect.

  15. Regarding the picture showing all the starlink satellites. Was this taken shortly after deployment while they are all clustered together? There just aren’t hat many satellites up there yet.

  16. Well of course chain of deadly rear poisonous orchids intended to vaporize and kill off large segments of humanity are visible. How else will Dr EvElon Muskovich extort a bazillion bitcoin from the world governments? Really. If it were as easy as saying theyre there everyone would have tried it by now.

      1. Yes confusion is good. Single raised eyebrow and Mad laughter echoes from cavernous spaceXtasy hanger as he removes his pinky gracefully from the side of smirking lips to the exquisitely soft fur of an extremely rare white persian long hair guinea pig. ’The fools.’ He thinks to himself.

  17. 20,000+ satellites at 60 per rocket. Can you imagine what all that pollution is doing to the ecosystem? Is it REALLY worth our health, and the health of the planet, so rural areas can surf FB all day and stare at their phones like zombies? Countries around the world are spending BILLIONS of dollars *per telescope* only to have the scientific data thrown out, because not a single image will be usable…what a waste. Stop Starlink NOW.

    1. That one’s easy.

      Falcon 9 first and second stage combined burn approximately 55,000 gallons of RP-1 (highly refined kerosene) per launch, and a Boeing 747-400 burns 22,000 gallons of jet fuel (lower grade kerosene) for a transatlantic flight.

      To take into account possible losses, let’s say it takes 340 launches to put 20,000 Starlink satellites into orbit. So that would mean launching the entire constellation on F9 would burn as much kerosene as nearly 700 transatlantic flights. This sounds like a lot, until you realize that there are between 2,000 and 3,000 transatlantic flights every day.

      In other words, launching 20,000 Starlink satellites will put less CO2 into the air than a single day’s worth of commercial air travel.

  18. For thousands of years everything was more or less OK, kingdoms rise and fell, but with a predictable consistency that always resulted in some manner of long term order and relatively decent quality of life for some percentage of the population.

    We have been polluting the earth with garbage for more than a hundred years in order to build things never built by anyone before, providing great comforts and entertainment unlike anything seen before.

    Now, over the past few decades, we have polluted airwaves in the sake of entertainment while preventing our precious bees from navigating, but at least the children can watch porn from their phones wherever they might be. Now the beautiful skies are being polluted in an effort to reach every single person with this “entertainment” in an effort to affect their minds, their ideals, their priorities with the propaganda and entertainment that detracts from the real beauty of life and provides nothing more than mindless entertainment for an individual and a revenue source for a corporation.

    This is not the beginning of the end; That beginning was a long time ago, but this is a sign of rapid acceleration towards that inevitable end.

    I think in 20 years the world resembles something like Half Life 2; Hong Kong certainly does right now.

    It appears that half the population of developed countries has made it clear that they value these meaningless entertainments and unhealthy comforts over the real beauties and meanings of life.

    The end surely must be near. At least the end for those who think independently and value the simple pleasures in life.

    Now you are faced with a choice. Do you care? If you do, you can fight it, which will not stop the inevitable, but could prolong things as they are. Or embrace it, and the inevitable end of the beauty of life; Bow to your corporate masters and be entertained.

    All of this just makes me want to move somewhere really beautiful and poor and live out my remaining days somewhere that people still value the real meaning of life. I want nothing of this nightmare, this revenue driven monster that will someday consume everything.

    A few weeks ago, I saw the string of satellites pass over me in the sky I have loved to watch for so long, and I realized the inevitability of even this simple beauty being polluted in the sake of revenue, under the guise of “improving” life. It made me hope that I die of natural causes soon, hopefully in a relatively painless way though a quick cancer wouldn’t be objectionable. I do not wish to partake in this disgusting entertainment-driven reality which results in nearly imperceptible loss of more and more personal liberties every single day. I am seeking employment in a remote area where I hope to live out my remaining days as far away from this nonsense as possible.

    That is all! Have a good night! I’ll go sit with my foil cap and my unreasonably cuddly cat that refuses to do anything useful other than be cute and cuddly. Definitely going to get the catnip out. This kitty won’t entertain his self.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.