Impacts of Large Satellite Constellations on Astronomy

[ncb1397]

What was totally surprising was how bright the starlinks were. You know, we thought if they were 9th or 10th....
hey, we could live with that. But third magnitude? 5th mangitude? That is far brighter than we had any reason to expect and it turned out SpaceX hadn't expected them to be that bright. So, yes, we should have been aware of that, but there are many many objects in earth orbit that are similar sizes (rocket bodies, defunct satellites) at anywhere from 400 to 1200 km that are much fainter. And what is really surprising about the SpaceX satellites is how bright they are.

Thank you for proving my point.  The quote you cited was explicitly talking about Starlink, then you applied it to OneWeb.  They aren't the same.

So, for SpaceX, the threshold is 9th and 10 magnitude is fine. For OneWeb, it is completely different? You do know that this was from January 2020, when Starlink was planned for 1000 km + altitudes. As such, his statement at that time, whether they were intended generally for all constellations or just for starlink, apply to ~1100-1200 km regardless. And the difference in angular velocity is only good for a difference of about .3 magnitude meaning you can interpret that as 9.3-10.3 if you wish. Regardless, the OneWeb satellites apparently are not far from the target while 5th magnitude starlinks are off by a factor of 40.

It's probably going to be?  So you want other people to believe your wild guess that the other factors don't matter, rather than what the studies of professional astronomers say?

Don't know what you mean. There are multiple references up thread of astronomers saying the 5th magnitude were a major problem. They also say the OneWeb altitude is a major problem (mainly in the context of untold thousands of such satellites). You are inserting your own interpretation of what is the bigger problem, and in a public forum, you are entitled to that (as is everyone else). Regarding VRO, my interpretation is because the 5th magnitude starlinks will cause the electronics on the observatory to go on the fritz while the OneWeb satellites won't cause the same ghosting effects, they are of lesser impact.

So now instead of careful studies by professional astronomers you want to base conclusions on popular media reports?

We know why there has been more reported in the popular press about the Starlinks -- because they form a highly-visible "string of pearls" that non-specialists can see with no equipment.  That attracts attention.  OneWeb doesn't have that visibility.  So it's not in the media as much.  That doesn't mean it isn't a problem.

No, I am not referring to popular media sources. Below are professional sources. Numerous anecdotes have been provided on astronomer twitter feeds or conference videos with examples of starlink impacts on observations. The narration on the "SpaceX photobombing" slide is the following:

Of course, you have all seen photo bombs by starlink. You just saw some more in past presentation. They continue to roll in from observatories around the world.

https://twitter.com/lcjohnso/status/1196370554414125056

edit: This could go back and forth forever, you get the last word (whatever that will be). For the time being anyways.

[daedalus1]

The last word is that astronomers will have to face a reality that has been building since 1957 and adapt. The economic well being of 7.5 billion people will win out easily. Yes try and mitigate the effects as much as possible but it won't be stopped whatever the argument.

[eeergo]

"Unescapable reality" (wonder if it's like a "manifest destiny" of some sort):

https://mobile.twitter.com/planet4589/status/1286114040490741762

https://mobile.twitter.com/hsocasnavarro/status/1285972780002021376

[meekGee]

These shots are annoying.

A satellite train (before dispersal) is only a single object.  You need to purposely wait for it in order for it to ruin your comet photo.  At any time, the bunch is only a problem for one observing location over the entire globe (in relation to a specific target).   Normally, astronomers use the publicly available data to AVOID problems, not to look for them.

After dispersal, each orbital plane presents only 2-3 satellites within line of sight of an observer (above approx 15 degrees) and only 2-3 planes are visible from a single location.  This makes for a total of 4-9 satellites in that entire dome.  The majority of them during the majority of the night, are dark.

Astronomers are worried, I get it, especially since other constellations intend to fly a lot higher (affecting lit time and elevation, but not # of sats in sight), but in the long run this sort of campaign is not in their best interest.  Working with SpaceX and other providers to better understand and better mitigate any problems is a lot more constructive. 

[Orbiter]

That's funny, because here's a shot I took of the comet and in this shot, I had a satellite in almost every single frame (most weren't even Starlink). I was able to stack them out using a stacking technique on a free, commonly known stacking software (DeepSkyStacker's kappa sigma stacking method for any astronomers out there). I'm willing to bet some of the people who are taking these photos know about this method too.

[envy887]

That's funny, because here's a shot I took of the comet and in this shot, I had a satellite in almost every single frame (most weren't even Starlink). I was able to stack them out using a stacking technique on a free, commonly known stacking software (DeepSkyStacker's kappa sigma stacking method for any astronomers out there). I'm willing to bet some of the people who are taking these photos know about this method too.

Yeah, that the other photo was intentionally stacked without subtraction, apparently for the sole purpose of complaining about Starlink. Which is dumb.

[su27k]

https://twitter.com/ralfvandebergh/status/1288232162089676801

Telescopic observations show dramatic reduction in brightness #Starlink satellites in parking orbit. Note the attitude change of the solar array and the difference in brightness character. False color processed to bring out better certain micro contrasts @spacex @SpaceXStarlink

[Eric Hedman]

If Astronomers are annoyed by Starlink, here's an update on Kuiper one of the other projects that will annoy them:

https://news.yahoo.com/amazon-invest-10-bn-space-based-internet-system-143020830.html

[Twark_Main]

At 590 km, Kuiper plans 28 orbital planes with 28 satellites per plane for a total of 784 satellites.  At 610 km, Kuiper plans 42 orbital planes with 36 satellites per plane for a total of 1296 satellites.  At 630 km, Kuiper plans 34 orbital planes with 34 satellites per plane for a total of 1156 satellites.

Kuiper states that its system... will be capable of providing continuous coverage to customers within approximately 56°N and 56°S latitude

source: https://docs.fcc.gov/public/attachments/FCC-20-102A1.pdf

[thirtyone]

At 590 km, Kuiper plans 28 orbital planes with 28 satellites per plane for a total of 784 satellites.  At 610 km, Kuiper plans 42 orbital planes with 36 satellites per plane for a total of 1296 satellites.  At 630 km, Kuiper plans 34 orbital planes with 34 satellites per plane for a total of 1156 satellites.

Kuiper states that its system... will be capable of providing continuous coverage to customers within approximately 56°N and 56°S latitude

source: https://docs.fcc.gov/public/attachments/FCC-20-102A1.pdf

Hmm. At least that's a better altitude. Depending on size they may need some more optical mitigation as well. Let's hope they actually spend some effort on that. Also a manageable constellation size on its own, at least strictly from a astronomy point of view. To be fair I have this feeling most constellations have astronomical numbers just to make sure they reserve FCC status or something.

[jg]

Rubin observatory (LSST) impact:

[RedLineTrain]

Visorsat at magnitude 7.0 observed.

https://twitter.com/richard_e_cole/status/1295604802966171649

[jg]

Rubin observatory (LSST) impact:

The Rubin observatory video I referenced is worth listening to. It isn't very long. It shows the issues that astronomers are facing.

1) if the satellite is too bright then you get into trouble with the detectors. Besides saturating the pixels, there are other nonlinear effects as you can see in the images in the video affecting the detector at very low levels, that cause streak artifacts to either side of the bright streak

2) The streaks are actually quite wide, for a reason that should have been obvious to me but I had missed. The telescope is focused to infinity and therefore the satellite is out of focus. It's too close to the Earth! It is probably just as well that this is true, because saturation would be much more of an issue if the streaks were narrower in some ways. Whenever a CCD saturates the charge no longer remains confined to the pixel in which the saturation occurs but spills over into other pixels in that column. And there is also some nonlinearity at saturation. So the streak actually affects a bunch of pixels much wider than I had naively thought.

3) number of satellites is also of concern at least in the long run. If these constellations are built out to the number of satellites that at least appear in the press, which is not a given, then it becomes really hard to avoid them, probably effectively impossible for the LSST with its wide field. You can see that in one of the images presented in Tony Tyson's presentation.

It sounds like there are productive conversations going on between the astronomical community and SpaceX and the situation is clearly better than when the first Starlink flight was launched. Whether they will be enough is an interesting question not yet completely obvious, though the general tenor of the presentation makes one pessimistic if you believe in no impact at all on astronomy. I think astronomers are very much aware that in the long run opening space for telescopes that could be built cheaply enough to be interesting and commonplace would be a major step forward, but that's probably still decades out. Their concern is clearly justified.

It is pretty clearly going to impact one of the current very high priority projects for astronomy. How much is what is not yet understood completely as mitigations are not complete, and they are also seeing what can be done in software to compensate and remove the artifacts to what is possible, but LSST operates in a regime a few other other telescopes. It is very wide-angle, surveys the entire sky every four days, and can detect items to very very faint levels. So it is likely to be affected worse than most telescopes.

The cost of the instrument is in the order of a billion dollars, this is not a trivial amount of taxpayer money being spent for LSST. It's actually a very interesting project and having done a small amount of consulting for them I can say that it presents some reasonable unique challenges on the computing side. They wish to be able to detect new objects in the sky and report them to astronomers in literally of order seconds or tens of seconds. This is to allow time to main astronomy so that new objects of unknown types can be studied almost immediately after discovery. That means that you have to do quite sophisticated image processing on a very near to real-time basis and the problem is identifying what is something really new and worth looking at. It's hard.

Jim

[meberbs]

It sounds like there are productive conversations going on between the astronomical community and SpaceX and the situation is clearly better than when the first Starlink flight was launched. Whether they will be enough is an interesting question not yet completely obvious, though the general tenor of the presentation makes one pessimistic if you believe in no impact at all on astronomy.

The visorsat results posted just above should change the outlook at least somewhat. He mentions in the presentation 10-20k electrons per pixel even with reductions from 0.9, but the 7.0 magnitude result based on his plot should be in the 1-5k range. This should also knock down the broad part of the trail shown on the later slide more into the noise. It wouldn't make the satellite invisible to the observatory by any means, but it would be a significant cut into the effects. There are mentioned at the end some techniques such as predictive brightness models for the satellites that give some hope of making the impacts on science data truly negligible. (More work would be done to get there, but the actual science results would be close to non-impacted.)

[jg]

Just to keep things in perspective, LSST detects things 24 magnitude and fainter.  These satellites are *bright*...

[ChrisWilson68]

Just to keep things in perspective, LSST detects things 24 magnitude and fainter.  These satellites are *bright*...

Yeah, they're bright by those standards at some times.  At times that they're not illuminated by the sun, they're not bright.

Even when they are illuminated, they only affect a very, very small percentage of total pixels.  The chart in the video clearly shows that at magnitude 7 they do not saturate any pixels and the crosstalk issues go away.

Toward the end of the video it mentions predictive brightness models.  This basically means that they can figure out exactly where the satellites will be and how bright they'll be, so they can eliminate the affected pixels from their data sets.

So, the astronomers will have to do some extra calculations, no doubt.  But there's every reason to believe that in the end, through a combination of reducing the brightness of the satellites, sharing data, and extra calculations, that the end effect on the astronomy being done will be pretty small.

[jg]

I think you're being a bit optimistic, to say the least. Its pretty clear in my mind from that talk there will be an impact particularly to LSST, exactly how much is yet to be known.

There is a good faith effort going on to mitigate the problem as best as can be arranged on both sides. I don't think we've heard of the end of this.

[jg]

You also missed Tony's statement that the signal from the satellites is present in the data to a hundred to one signal to noise ratio! That's a huge problem wherever light from the satellites are present. Say you can make things 10 times fainter, which is great. You're still going to have one hell of a single noise ratio contribution from the satellite. If a satellite crosses the area of interest then it effectively can destroy most of the information there. And as Tony noted noted, the effective width of the satellite for at least LSST is much wider than you might naively think, as I did, being as much as 30 or more arcseconds wide. Given the resolution of the images of LSST, which are small fractions of an arc second since they're observing from one of the best sites on the ground on planet Earth, that's a very wide track.

The problem is not only the brightness of the satellites, but that there are so many of them, meaning that a whole pile of images will be affected and it becomes essentially impossible to dodge them. For LSST the observing cadence is 4 days. Anything affected by a satellite will next be observed 4 days later. Each image is very large, as I remember it is something like the diameter of the moon, to put it into human context. They take pictures of all spots on the sky every 4 days, rinse and repeat, for the next 10 years.

For amateur astronomers, even with their smaller telescopes, these are very bright objects. Amateurs are not going to be even as capable as LSST for processing their data. Thankfully most telescopes have a much smaller field of view and the problem of number of satellites is significantly less severe for them. But it's going to be a problem no matter what.

Don't try to minimize that it's a problem. Let's try to work together to make it as small a problem as possible. The feeling I got from that talk is that a SpaceX is taking this as a serious problem and doing what they can to mitigate the problem. Likewise, the LSST project is taking the seriously, and doing what they can to reduce the problem.

[ChrisWilson68]

You also missed Tony's statement that the signal from the satellites is present in the data to a hundred to one signal to noise ratio. That's a huge problem wherever the light from the satellites are present. Say you can make things 10 times center which is great You're still going to have one hell of a single noise ratio contribution from the satellite.

Yes, a 100:1 signal to noise ratio.  But only for a miniscule percentage of pixels at any given time.

That's why the predictive brightness modeling holds so much promise.  SpaceX knows where the satellites are and where they will be to a very high degree of precision.  After some observation, they should be able to model their brightness fairly accurately.  So you know which pixels are compromised and throw those out.  You still have the data from the vast majority of your pixels.

This isn't unprecedented.  Lots of scientific observations in many fields have some of their data compromised and they have to throw out some data points.

[meberbs]

You also missed Tony's statement that the signal from the satellites is present in the data to a hundred to one signal to noise ratio. That's a huge problem wherever the light from the satellites are present. Say you can make things 10 times center which is great You're still going to have one hell of a single noise ratio contribution from the satellite.

It doesn't mean anything like what you just said. The only thing a Starlink having a significant SNR guarantees is that the satellite can be detected.

It says nothing directly about how much other signals could be detected. It has no effect on anything that is not the relatively small fraction of pixels directly affected. ChrisWilson68 actually left out part of the potential application of predicting the satellite trails. You don't have to just ignore the trails, you can subtract off the trails to the limits of your ability to predict the trail. We don't know what that would be yet, but the best case is to remove it down to the noise level, so there is no net effect on the science data.