Planned satellite constellations may swamp future orbiting telescopes

Planned orbital observatories would see satellites cross nearly all of their images.

See full article...

 

[pagh]

The telescopes should be launched to higher orbits. They're discussing asteroid spotting, and needing to observe the horizon at dawn to capture these asteroids in similar orbits to Earth. That means it's a bad design. They should be in high orbit, or somewhere like L1 inside of Earth's orbit, rather than one so obviously ill suited to their needs.

"Oh, but it costs more to get there, and it costs more to communicate with there."

Cheaper launches are a thing, and they're getting cheaper continuously. Laser communications are a thing, and no longer require dedicated time on the DSN.

Orbital observatories have service lives measured in decades. Hubble was launched in 1990. The only people who even considered the possibility of satellite constellations back then were science fiction writers.

Sure, we can avoid the satellite constellation problem for future observatories (assuming we ever fund science again). But commercial providers do need to take some legal responsibility for maintaining the quality of science and other non-commercial activities in orbit.

 

[wagnerrp]

And now a second comment where I admit to not knowing how digital photography works. With these long exposures shots is it operating just like a film camera or is it taking hundreds of readings per second and then adding them together as a sort of timelapse composite? So each pixel is the sum of light received at that pixels location of the duration of the shot from thousands of samples or like one long single reading on that one pixel?

It works just like film. You open the shutter and expose the sensor. The sensor is hit by photons and accumulates charge. At some point, you read the charge from the sensor, which zeros out the charge. That's like developing and fixing the film.

Each time you read out the sensor, there is a fixed level of noise applied. The shorter your exposure time, the larger this noise floor is compared to the accumulated light level. There is some benefit to stacking multiple images, but this is not as good as just doing long exposures.

There is a saturation point at which the sensor is filled and cannot gain any further charge. Once you reach this point, you're destroying data. You want to read out before then, and that limits your effective SNR. A bright satellite crossing the field of view can cause this to happen very quickly, and depending on the electrical design of the chip, may cause bleed over to other cells or whole rows.

 

[Mad Klingon]

Future orbital observatories should be placed at or near geosync orbits. Fit them with laser xmit/receive to avoid taking valuable comms sat slots. Avoids almost all sat constellation issues. Make them serviceable. If Starship and/or Orion can go to the moon, either should work for in orbit geosync service missions. Especially since one of the early planned Orion missions was an asteroid visit.

But Earth based astronomers are likely out of luck.

 

[terrydactyl]

Could a workaround be made by changing 600-second exposures to 600 one second exposures, blacking out the satellites from each and then combining? Seems like a process could be made to automate that. Maybe the hardware is not equipped to save that many images.

Not sure if it was answered but, a single longer exposure has a better signal-to-noise ratio than an equal amount of short exposures. I.e. your 600 one second exposures will have more noise, meaning less data.

 

[archtop]

A one second exposure does not collect near the amount of light/photons as 600s...kind of the whole point of a long exposure.

But 600 one-second exposures do equal one 600 s exposure -- that's what I think the poster was saying. And if you knew the paths of all the satellites, you could time those exposures to avoid the satellite passage. In fact, keep the "shutter" open and only close it when a satellite passes by, until the total exposure equals 600 s.

btw I'm sure this was thought of long ago -- I wonder what its drawback is.

Edit: The drawback could be S/N, as the above poster points out. (Although I don't see how 600 1 s exposures combined would have a higher S/N than 1 600 s exposure.)

 

[Wickwick]

That is like asking the people in the post above that enjoy Yosemite to build their own park.

As David Byrne says:

"Most beautiful, most intelligent criminals you've ever seen. Now you're paying top dollar. For what you used to get for free."

Sure. It's exactly like that. If Yosemite covered the entire globe.

 

[George Moromisato]

And now a second comment where I admit to not knowing how digital photography works. With these long exposures shots is it operating just like a film camera or is it taking hundreds of readings per second and then adding them together as a sort of timelapse composite? So each pixel is the sum of light received at that pixels location of the duration of the shot from thousands of samples or like one long single reading on that one pixel?

Imagine you have a grid of receptors, let's say 1000x1000 for a 1 megapixel image. When a photon hits a receptor, it adds a charge (electrons) to the receptor. Photons keep hitting receptors and electrons keep building up.

At the end of the exposure, there is a circuit that "reads" the number of electrons in each receptor.

That's the theory. In practice, there is thermal noise which bumps up the electron count, not always uniformly. And there is error during read-out. I.e., the result may be +/- 10 electrons (or whatever).

If you only expose for a short time, the number of electrons that are from the signal is very small relative to the noise. Specifically, once the signal is below the read-out error, you're basically just getting noise, and no amount of stacking will remove it--you're just stacking noise at that point.

Caveat: This is my understanding from when I did a bunch of astrophotography. But my understanding could certainly be wrong/outdated.

Edit: After consulting an LLM, I realize that I made at least one mistake: the read-out error can be removed with stacking as long as it is random. But the problem is that you're paying the read-out noise for each frame, instead of only once at the end of a long exposure. Thus it takes much longer to get the same quality if you do many short exposures.

 

[Wickwick]

Orbital observatories have service lives measured in decades. Hubble was launched in 1990. The only people who even considered the possibility of satellite constellations back then were science fiction writers.

Sure, we can avoid the satellite constellation problem for future observatories (assuming we ever fund science again). But commercial providers do need to take some legal responsibility for maintaining the quality of science and other non-commercial activities in orbit.

The discussion here is for new observatories. Rather than set aside orbits for science, why not give them all of space above commercial orbits? If you send your telescopes to 1,000 km you're going to have fewer satellites above you than Hubble has had to deal with for almost all of its life.

 

[Wickwick]

It works just like film. You open the shutter and expose the sensor. The sensor is hit by photons and accumulates charge. At some point, you read the charge from the sensor, which zeros out the charge. That's like developing and fixing the film.

Each time you read out the sensor, there is a fixed level of noise applied. The shorter your exposure time, the larger this noise floor is compared to the accumulated light level. There is some benefit to stacking multiple images, but this is not as good as just doing long exposures.

There is a saturation point at which the sensor is filled and cannot gain any further charge. Once you reach this point, you're destroying data. You want to read out before then, and that limits your effective SNR. A bright satellite crossing the field of view can cause this to happen very quickly, and depending on the electrical design of the chip, may cause bleed over to other cells or whole rows.

Correct me if I'm wrong, but one can adjust the gain across the charge well during the exposure without read noise, right? So in theory, it should be possible to build a sensor that literally doesn't add any charge to a pixel when a satellite's (known) position comes by?

 

[wagnerrp]

If you only expose for a short time, the number of electrons that are from the signal is very small relative to the noise. Specifically, once the signal is below the read-out error, you're basically just getting noise, and no amount of stacking will remove it--you're just stacking noise at that point.

With enough stacking, good signal will still rise out of the noise floor. The problem is that a lot of the stuff they want to observe is not good signal. In particular, things like asteroids move, and once it moves to the next adjacent pixel, further stacking is detrimental.

 

[Steve austin]

The article indicates that a principal reason Hubble was placed in such a low orbit was launch by STS and the desire/need to allow crewed update, also via STS. As far as I know, no subsequent observatory has been designed with later access being a factor. Given current and imminent launch options, the demonstrated practicality of high bandwidth optical data links, and the apparent lack of interest in serviceable observatories, are there theoretical or practical reasons why future observatories couldn’t orbit much higher, either between the Van Allen belts or above them? That obviously doesn’t help anything already in orbit or is close to, but are there reasons why this couldn’t be done for things in the planning stages?

Edit: typos

 

[wagnerrp]

Correct me if I'm wrong, but one can adjust the gain across the charge well during the exposure without read noise, right?

That's beyond my youtube-explainer-level knowledge of the subject.

 

[George Moromisato]

Correct me if I'm wrong, but one can adjust the gain across the charge well during the exposure without read noise, right? So in theory, it should be possible to build a sensor that literally doesn't add any charge to a pixel when a satellites (known) position comes by?

I don't know. My guess is that it would be easier to physically cover the light-path so no photons reach the sensor. But maybe there is a way to do it electronically. I just don't know how low the gain can go. Can it go to zero? I don't know.

 

[Wickwick]

That's beyond my youtube-explainer-level knowledge of the subject.

I'm pretty sure that's how CID's worked. I realize they're not as fashionable as CCDs or CMOS detectors these days, but perhaps they have a Renaissance with the ability to selectively ignore bright sources.

 

[Wickwick]

I don't know. My guess is that it would be easier to physically cover the light-path so no photons reach the sensor. But maybe there is a way to do it electronically.

One can easily shutter the entire aperture of the telescope, but that wouldn't really help for whole-sky surveys. For those you'd have to have some sort of interposing layer where you could block tracks on a sub-sensor basis. Basically, you'd need an LCD screen overlaid on the actual sensor. That's totally do-able, but it leads to light losses - but minor ones like perhaps 10% rather than having the entire pixel blown out. However, if you do have pixel-level control of amplification, you can simply turn off gain where satellites are expected.

 

[Wickwick]

The article indicates that a principal reason Hubble was placed in such a low orbit was launch by STS and the desire/need to allow crewed update, also via STS. ASAs far as I know, no subsequent observatory has been designed with subsequent access being a factor. Given current and imminent launch options, the demonstrated practicality of high bandwidth optical data links, and the apparent lack of interest in serviceable observatories, are there theoretical or practical reasons why future observatories couldn’t orbit much higher, either between tha Van Allen belts or above them? That obviously doesn’t help anything already in orbit or is close to, but are there reasons why this couldn’t be done for things in the planning stages?

The Chinese observatory (Xuntian) mentioned in the article is intended to dock with their space station for upgrades, then go back on its own.

Of course, those upgrades are known well in advance. There's plenty of time to spiral to a higher operational orbit and to come back down. Adding more propellant could be included in each upgrade cycle.

 

[Dioptase]

The US and EU constellations provide internet access. The Chinese constellations provide Chinese network access. They're not duplication of capacity because they're not offering the same thing.

Hmmm, well every article I find says they are providing internet. One has signed an agreement with Brazil and is in talks with 30 countries as of February this year. Sounds like they are offering nearly the same thing to me. How successful they will be remains to be seen, but having two or more sets of satellites provide the same service in orbit because some need to be inactive over certain geographic areas for political reasons would seem to me to be causing duplication of capacity.

 

[Wickwick]

Hmmm, well every article I find says they are providing internet. One has signed an agreement with Brazil and is in talks with 30 countries as of February this year. Sounds like they are offering nearly the same thing to me. How successful they will be remains to be seen, but having two or more sets of satellites provide the same service in orbit because some need to be inactive over certain geographic areas for political reasons would seem to me to be causing duplication of capacity.

It's duplication of hardware, perhaps. It's not duplication of capacity. Starlink's capacity as it flies over Shanghai is zero.

 

[dtich]

A one second exposure does not collect near the amount of light/photons as 600s...kind of the whole point of a long exposure.

well. that's a false explication. the OP is suggesting 600 - 1s exposures as compared to one 600s exposure.. and the fact is the number of photons collected by each method is quite comparable, and given perfectly clean circumstances, fairly equivalent. however, the noise recorded by multiple exposures is also compounded and requires tailored algorithmic processing of the stack to filter out. even so, space-based astrophotography, eg. Hubble, JWST, etc, does use multiple short exposures as opposed to longer single exposures for this and several other reasons.


edit: haha, ok, didn't realize this had been 'answered' so many times already.

 

[normally butters]

Future orbital observatories should be placed at or near geosync orbits. Fit them with laser xmit/receive to avoid taking valuable comms sat slots. Avoids almost all sat constellation issues. Make them serviceable. If Starship and/or Orion can go to the moon, either should work for in orbit geosync service missions. Especially since one of the early planned Orion missions was an asteroid visit.

But Earth based astronomers are likely out of luck.

The only space-facing observatory in a geosynchronous orbit is NASA's Solar Dynamics Observatory. The reason why space observatories typically don't use geosynchronous orbits is that their precise pointing requirements conflict with the station-keeping requirements to play nice with other satellites in geosynchronous orbits maintaining their prescribed slots. Since observing the Sun doesn't require particularly precise pointing, this isn't a big deal for SDO, but it's a problem for looking at distant star systems.

Space observatories want to be in orbits where they'll rarely if ever need to maneuver to avoid a collision or maintain a prescribed orbital slot. Any other potentially conflicting spacecraft will be expected to maneuver to avoid the space observatory.

 

[wagnerrp]

Basically, you'd need an LCD screen overlaid on the actual sensor. That's totally do-able, but it leads to light losses - but minor ones like perhaps 10% rather than having the entire pixel blown out.

That's a 50% loss of light, and it's necessarily going to be linearly polarized, which could be a detriment (or potentially benefit) to applications that are specifically observing polarized sources. You might be better off with a DMD. Those are apparently seeing increased use as a programmable diffraction grating for spectroscopy.

 

[Bernedoodle]

I wish this article ended with a more upbeat conclusion, like the ARS article on space based telescopes from 2023:

https://meincmagazine.com/space/2023/...s-should-take-advantage-of-starship-paradigm/

Since most of these satellite constellations are dependent on launch costs continuing to fall dramatically, why not assume that in the near future it will become very affordable to launch telescopes far above the constellations? Already New Glenn can launch a telescope heavier than Hubble to a geostationary transfer orbit, and it’s a reasonable bet that Starship will figure out refueling and be able deliver something much bigger there and beyond.

Additionally there are dozens of companies planning to mass produce telescopes pointing downward; this will lower component costs for new science missions.

https://nova.space/press-release/commercial-earth-observation-market-surpasses-8-billion-by-2033/

https://www.esa.int/Applications/Ob...ends_to_watch_in_commercial_Earth_observation

I see no reason to be pessimistic about the future of astronomy.

 

[Wickwick]

That's a 50% loss of light, and it's necessarily going to be linearly polarized, which could be a detriment (or potentially benefit) to applications that are specifically observing polarized sources. You might be better off with a DMD. Those are apparently seeing increased use as a programmable diffraction grating for spectroscopy.

E-ink then

 

[wagnerrp]

Hmmm, well every article I find says they are providing internet. One has signed an agreement with Brazil and is in talks with 30 countries as of February this year. Sounds like they are offering nearly the same thing to me. How successful they will be remains to be seen, but having two or more sets of satellites provide the same service in orbit because some need to be inactive over certain geographic areas for political reasons would seem to me to be causing duplication of capacity.

Brazil has been turning increasingly authoritarian over the past decade, significantly increasing online censorship, and would likely welcome the sort of controls Chinese "internet access" might afford them.

 

[wagnerrp]

Additionally there are dozens of companies planning to mass produce telescopes pointing downward; this will lower component costs for new science missions.

Astronomers have stated time and again that they don't want a plethora of telescopes of existing capability. They want a handful of large, expensive flagship observatories that they then have to fight over for access.

 

[Wickwick]

The only space-facing observatory in a geosynchronous orbit is NASA's Solar Dynamics Observatory. The reason why space observatories typically don't use geosynchronous orbits is that their precise pointing requirements conflict with the station-keeping requirements to play nice with other satellites in geosynchronous orbits maintaining their prescribed slots. Since observing the Sun doesn't require particularly precise pointing, this isn't a big deal for SDO, but it's a problem for looking at distant star systems.

Space observatories want to be in orbits where they'll rarely if ever need to maneuver to avoid a collision or maintain a prescribed orbital slot. Any other potentially conflicting spacecraft will be expected to maneuver to avoid the space observatory.

The OP did sat at or near geosync. I don't see the need to go quite that high, but if one went to an orbit 1,000 km beyond geosync, I doubt you'd ever have to maneuver to avoid something. You'd be beyond the graveyard orbits of even the geosync birds. Of course, just going to 5,000 km orbital height (rather than 42k for geo) probably sets you up to never make an avoidance manuever.

 

[ranthog]

Of course the astronomers' solution is for everyone to adapt to their needs. That's the m.o. of astronomers across the world. Heaven forbid they acknowledge the solution is to just go above all the satellites.

One has to realize that the cat is out of the bag. Even if all the civilian megaconstellations were cancelled tomorrow, the military applications are enough to demand their growth. Astronomers may as well adapt to reality and start planning to fly above the noise sources or adapt to them on the ground

Hell, you have the ephemeris information for every noise source before its in your image. If you're dead set on imaging through them then find a way to deal with the noise - physical blockers or selective amplification drops as noise sources go past.

Saying go above the constellations is great, but who in the heck is going to pay for it? It is far more expensive, especially as it will require replacing existing equipment. The most obvious problem is that no one is forcing these constellation operations to pay for the additional cost to do astronomy. Instead, they're doing damage to current and future projects, without having to pay for that damage.

I would guess that astronomers already thought of trying to compensate for the additional noise, and if it were that easy they already would have.

 

[Wickwick]

Saying go above the constellations is great, but who in the heck is going to pay for it? It is far more expensive, especially as it will require replacing existing equipment. The most obvious problem is that no one is forcing these constellation operations to pay for the additional cost to do astronomy. Instead, they're doing damage to current and future projects, without having to pay for that damage.

I would guess that astronomers already thought of trying to compensate for the additional noise, and if it were that easy they already would have.

Thanks to the megaconstellations, it's going to be cheaper to launch above them than it would have been to launch prior to them. There's no "paying for" the added height except relative to the new, vastly cheaper market.

And I'm not so convinced that astronomers have actually worked to block the noise yet. The timescales for development of these observatories is such that the realities of tens of thousands of noise sources wasn't part of the consideration when the next generation of telescopes were designed. The next set of telescopes might very well take that into account, but the current crop of astronomers who are vested in the legacy hardware won't have the pristine conditions they designed for. Should infrastructure projects be put on hold until all the science projects that assumed their non-existence are done? Ignore the commercial aspects of these constellations. The military considerations are enough to establish they will exist.

 

[Chuckstar]

I am curious about one thing - they state that the constellations should be launched to lower orbits. Is there a reason why the observatories can't be raised to higher orbits? This is a serious question - do the observatories need a specific period that only occur with a limited array of orbits?

Especially considering that the article says this:

Unfortunately, even if we had an infinite budget, we couldn’t just solve this by increasing our reliance on space-based hardware.

With enough money, you just put the observatories in high orbits. They generally work better the further they are from Earth, anyway. And we wouldn’t even need infinite budgets to do that, considering how far launch costs have declined. And considering we’re not going to be servicing anything with the Shuttle, and don’t really know what a servicing spacecraft would look like, claiming you need low orbits for servicing is kind of a red herring.

 

[wagnerrp]

It is far more expensive, especially as it will require replacing existing equipment.

Yes. "future orbital telescopes", right there in the title. We're talking about new equipment. Getting new equipment into higher orbits will cost less than it took to put existing equipment in existing LEO, in absolute terms, nevermind accounting for inflation.

The most obvious problem is that no one is forcing these constellation operations to pay for the additional cost to do astronomy. Instead, they're doing damage to current and future projects, without having to pay for that damage.

We have protection for radio astronomy, but we have no protection for optical astronomy. You don't have to look any further than the repeated failures of dark sky initiatives to see that.

 

[Chuckstar]

I don't know. My guess is that it would be easier to physically cover the light-path so no photons reach the sensor. But maybe there is a way to do it electronically. I just don't know how low the gain can go. Can it go to zero? I don't know.

That introduces additional diffraction.

 

[Ecnhoffer]

I suspect the present U.S. administration has its standard solution to this potential problem. Fire the astronomers in NASA.

 

[Bernedoodle]

Astronomers have stated time and again that they don't want a plethora of telescopes of existing capability. They want a handful of large, expensive flagship observatories that they then have to fight over for access.

Perhaps someone would enjoy fighting over this:

“
The big advantage of using individual small telescopes is that they can be separated as far apart as we like. Since the finest detail in an image depends on the maximum separation between the mirrors collecting the light, this implies that we can make images of stars as though we had telescopes with mirrors hundreds of metres in size. It would be impossibly difficult to build a conventional telescope as large as that, so aperture synthesis is the only way to achieve the sharp images that astronomers really want.”

https://lambda.gsfc.nasa.gov/product/websites/AMI/mrao.cam.ac.uk/telescopes/coast/handout.html

 

[Wickwick]

Perhaps someone would enjoy fighting over this:

“
The big advantage of using individual small telescopes is that they can be separated as far apart as we like. Since the finest detail in an image depends on the maximum separation between the mirrors collecting the light, this implies that we can make images of stars as though we had telescopes with mirrors hundreds of metres in size. It would be impossibly difficult to build a conventional telescope as large as that, so aperture synthesis is the only way to achieve the sharp images that astronomers really want.”

https://lambda.gsfc.nasa.gov/product/websites/AMI/mrao.cam.ac.uk/telescopes/coast/handout.html

It's not just about having the small telescopes fart apart. It's also keeping their position to a small fraction of the wavelength of light that you're going to image. Orbital interferometers are being considered for an orbital version of LIGO (LISA), but even that's a rudimentary first step to what's required to achieve imaging that's diffraction limited by such a large separation.

 

[wagnerrp]

It's not just about having the small telescopes fart apart. It's also keeping their position to a small fraction of the wavelength of light that you're going to image. Orbital interferometers are being considered for an orbital version of LIGO (LISA), but even that's a rudimentary first step to what's required to achieve imaging that's diffraction limited by such a large separation.

And to some degree, you really do need the greater light collecting capability of a larger area.

 

[Chuckstar]

Perhaps someone would enjoy fighting over this:

“
The big advantage of using individual small telescopes is that they can be separated as far apart as we like. Since the finest detail in an image depends on the maximum separation between the mirrors collecting the light, this implies that we can make images of stars as though we had telescopes with mirrors hundreds of metres in size. It would be impossibly difficult to build a conventional telescope as large as that, so aperture synthesis is the only way to achieve the sharp images that astronomers really want.”

https://lambda.gsfc.nasa.gov/product/websites/AMI/mrao.cam.ac.uk/telescopes/coast/handout.html

That’s a 1999 paper describing a theoretical telescope with the equivalent resolution of a 100 m mirror. The VLT first operated in interferometry mode with the equivalent of a 200 meter mirror in 2002. This is quite old news.

 

[Wickwick]

And to some degree, you really do need the greater light collecting capability of a larger area.

I'm pretty sure it's more than that. I think you need multiple portions of the aperture between the extremes to reduce the shape of the diffraction pattern, but my understanding of optics doesn't carry to that level of detail.

 

[Wickwick]

That’s a 1999 paper describing a theoretical telescope with the equivalent resolution of a 100 m mirror. The VLT first operated in interferometry mode with the equivalent of a 200 meter mirror in 2002. This is quite old news.

On the ground - a far different proposition than doing it on orbit.

 

[Chuckstar]

On the ground - a far different proposition than doing it on orbit.

And who said otherwise? The comment I was replying to was about ground-based telescopes.