Impacts of Large Satellite Constellations on Astronomy

[Danderman]

It will be interesting to see the impact of AST’s 900 square meter antenna on astronomy, since these will be flying horizontally. I assume the nadir side will be dark, as mitigation for the impact on astronomy.

[Star One]

Selected tweets from this Twitter thread.

The coming weeks are likely to be important for the future of ground-based astronomy in the face of the threat of satellite constellations. How is it a threat, what are astronomers doing, why are these weeks important? Thread

https://twitter.com/EricLagadec/status/1381966978009591813

What are  astronomers doing? They met to discuss the impact of these satellites, and of light pollution in general, on astronomy and biodiversity. The conference was attended by ~900 participants, with a summary (of 279 pages) here: http://iau.org/static/publica...

https://twitter.com/EricLagadec/status/1381966999404695552

Why is this important today? The results of this work will be presented to the COPUOS (Committee on the Peaceful Uses of Outer Space) of the UN from April 19.

https://twitter.com/EricLagadec/status/1381967002483400713

What do the astronomers want? That these constellations limit as little as possible our observations of the Universe and our quest for answers to fundamental questions for humanity: how was the Universe formed, does life exist elsewhere? etc...

https://twitter.com/EricLagadec/status/1381967005163515909

International working groups have been formed to quantitatively measure the impact of these satellites on astronomy and to be able to discuss constructively with industry to mitigate the impact on astronomy.

https://twitter.com/EricLagadec/status/1381967028030894083

[Robotbeat]

Being able to discuss constructively is really important. And, companies are eager to do that. SpaceX in particular.

[Robotbeat]

No, it doesn’t really help high frequency traders. They already use microwave networks or other radio for that. Even lower latency.

Gamers and voice and audio, though, all benefit from low latency. Remote work is best done with low latency for responsiveness. High latency is a massive pain in the neck.

[su27k]

I believe this is the report they'll present to UN: Dark and Quiet Skies for Science and Society

I browsed the Mitigations and Recommendations section of chapter 6 "Satellite Constellation Report", seems reasonable.

[Dizzy_RHESSI]

Some initial estimates of the brightness of the satellites with visors were posted to the arXiv a few months ago, while the thread was locked. Past measurements by the same author were reasonably consistent with what other groups found.

https://arxiv.org/abs/2101.00374

The Brightness of VisorSat-Design Starlink Satellites
Anthony Mallama

The mean of 430 visual magnitudes of VisorSats adjusted to a distance of 550-km (the operational altitude) is 5.92 +/-0.04. This is the characteristic brightness of these satellites when observed at zenith. VisorSats average 1.29 magnitudes fainter than the original-design Starlink satellites and, thus, they are 31% as bright.

If confirmed it would mean that the visors are about as effective as DarkSat's coating, and further adjustments may be necessary to reach the target of 7th mag or fainter. A caveat is that this is a wide visible band and is not the same filter system as LSST/Rubin, but I wouldn't expect a huge difference.


Secondly, ROE astronomer Andy Lawrence has written a book on the topic.

https://andyxlastro.me/losing-the-sky/

[Lars-J]

Secondly, ROE astronomer Andy Lawrence has written a book on the topic.

https://andyxlastro.me/losing-the-sky/

From that link I see the author is one of the ones that promotes the demonstrably false theory that the LEO constellations are primarily about low latency stock trading. Sigh. There are good arguments to make, so make them instead of starting out by misrepresenting the other side.

[Robotbeat]

Yeah, it'd dumb. LEO satellite constellations are much higher latency than shortwave radio or the microwave radio networks HFTers *already* use. There's no money in it for SpaceX.

Any time someone makes that argument, to me it immediately marks them as either low-information types (i.e. that haven't really looked into any of this much) or folks who are actively willing to use dishonest arguments.

[rsdavis9]

Yeah, it'd dumb. LEO satellite constellations are much higher latency than shortwave radio or the microwave radio networks HFTers *already* use. There's no money in it for SpaceX.

Any time someone makes that argument, to me it immediately marks them as either low-information types (i.e. that haven't really looked into any of this much) or folks who are actively willing to use dishonest arguments.

Ok I may be missing something.
microwave dish chains is not used across the pond? Yes?
Is there an example of shortwave radio being used across the pond?

Not saying its less latency to use LEO satellites. But it is a close competition across the pond for fiber versus LEO laser linked.

EDIT: found this
https://www.bloomberg.com/news/articles/2020-06-17/companies-pitch-shortwave-radio-to-shave-milliseconds-off-trades

doesn't sound like anybody is currently doing it.

[Robotbeat]

Those using it aren’t gonna tell you about it. But there’s also hollow fiber.

[Asteroza]

Yeah, it'd dumb. LEO satellite constellations are much higher latency than shortwave radio or the microwave radio networks HFTers *already* use. There's no money in it for SpaceX.

Any time someone makes that argument, to me it immediately marks them as either low-information types (i.e. that haven't really looked into any of this much) or folks who are actively willing to use dishonest arguments.

Ok I may be missing something.
microwave dish chains is not used across the pond? Yes?
Is there an example of shortwave radio being used across the pond?

Not saying its less latency to use LEO satellites. But it is a close competition across the pond for fiber versus LEO laser linked.

EDIT: found this
https://www.bloomberg.com/news/articles/2020-06-17/companies-pitch-shortwave-radio-to-shave-milliseconds-off-trades

doesn't sound like anybody is currently doing it.

I think the info from 2019 regarding HFT shortwave should be still current and implying many camouflaged HFT shortwave stations, but the costs are significant. HFT will pay for the edge if it's a pretty clear win, but they naturally don't want to have those communication assets on their books when possible, and are open to outsourcing.

https://sniperinmahwah.wordpress.com/

That author still says the quality/quantity of data is still critical, thus shortwave is a stopgap for high priority small data. If you can beat oceanic fiber plus associated terrestrial fiber with a three or four bounce ISL relay on Starlink with RF uplink/downlink, the latency may be in your favor at a much better data quantity than shortwave. Potentially still vulnerable to solar flare activity though, same as shortwave.

[thirtyone]

doesn't sound like anybody is currently doing it.

They are already doing it. They're extremely secretive about it.

[FutureSpaceTourist]

https://twitter.com/deicherstar/status/1385243264765341697

Many of you are acutely aware of the growing problem of Elon Musk’s Starlink satellites. They’re creating havoc in the sky for professional researchers and astronomy enthusiasts, and the problem is growing worse. Jamie Cooper recorded this group passing in front of the Moon.

[Danderman]

It will be interesting to see the impact of AST’s 900 square meter antenna on astronomy, since these will be flying horizontally. I assume the nadir side will be dark, as mitigation for the impact on astronomy.

Let's see how AST mitigates the impact of 80 of these in LEO on astronomy. Maybe the nadir side could be very dark, but astronomers are then going to see wide black streaks. I am sure that AST has thought of this.


.

[vanbrua]

https://twitter.com/deicherstar/status/1385243264765341697

Many of you are acutely aware of the growing problem of Elon Musk’s Starlink satellites. They’re creating havoc in the sky for professional researchers and astronomy enthusiasts, and the problem is growing worse. Jamie Cooper recorded this group passing in front of the Moon.

I thought this was fake at first because of how bright the moon is compared to everything but the sun.

On closer inspection, that is a new moon. The moon is being lit by the Earth and it was taken at sunset or sunrise (the only time such a new moon is visible).

I'm still suspicious because the moon looks WAY too grainy to be from a single exposure long enough to capture that many Starling transits. I get a less grainy image with a 100 ms exposure than that (I did optimize my setup to image dim objects as quickly as possible though).

The most honest way that I can see this being made is multiple exposures stacked on top of each other . Then instead of choosing the good exposures, the astrophotographer purposely choose frames with a transit. And/or the astrophotographer used a neutral density filter to to artificially increase the exposure time??

Edit to add: to get 14 Starling satellites offset by half moon diameter while tracking the moon requires about a minute of exposures (if anyone know how far apart the planes and RAN of satellites were at the new moon, I could.gove an exact number). In that much time, my phone would give less grain than that photo.

[vanbrua]

Two more suspicious features of that photo:

1. Zooming in shows that the streak is about three pixels wide including 1 fuzzy pixel on each side. The pixels of the background adjacent to the streak are frequently darker than the rest of the background. By comparison, the edge of the moon is fuzzy for about 4-5 pixels and has no dark ring around it.

A long exposure does tend to smear the image because of atmospheric distortion. The streaks would be moving too fast for that to blur them. But, the streaks should have wiggles in it too.

It is possible that the dark edges are a JPEG artifact caused by DCT terms being quantized. But, there should not be a bias toward dark. There should also be some extra light areas.

2. One of the streaks has a vertical end near the left side of the image. This looks like it was composited into the final image. If it was the start or end of an exposure, the edge of the streak would be round instead of a vertical edge.

There is nothing wrong with stacking a bunch of frames of the same thing on top of each other. It is a great way to get good images of dark objects, small bright objects (lucky imaging), increase SNR, and increase the dynamic rage. But, this looks like a deliberate attempt to enhance the streaks.

[su27k]

Repost this here since this video is a pretty good summary of the current state of affairs (with regard to astronomy that is, the part about Kessler Syndrome is overplayed).

Actually, the impact of satellite constellations (not just Starlink) to optical astronomy and radio astronomy will be substantial when all active or proposed projects are complete.  There was a recent excellent talk at PSW Science Society by Dr.  Tony Tyson who is the Rubin Observatory Chief Scientist and Prof at UC-Davis.  In this talk he goes into substantial detail with real empirical data and theory about how and why certain constellations cause certain kinds of problems.  Some can be dealt with at least in part by scheduling astronomical observations and/or real-time shutter openings & closings but others cannot. 

He is in general laudable about the cooperative attempts of SpaceX to work with astronomers and to experiment with modification to Starlink satellite design and operation to reduce the impact.  Also that the lower Starlink orbit (around 550 km) is preferred to the 1200 km orbit of some other constellations (primarily the shorter active lifetime of lower satellites).  I don't know if anyone has started to model the impact of multiple Starships in orbit.  The numbers will be far smaller and the heatshield may or may not offer relief at least at certain wavelengths.  Reasonable for someone to take a look but the huge number of satellites is currently the biggest concern. 

I highly recommend this talk to people interested in these issues as it really lays out the concerns, what is known currently,  what options do or do not readily exist to mitigate, and projected impacts to astronomy. 

 PSW Science Society, lecture 2440:   Satellite Constellations and Astronomy 

[eeergo]

Hitting Nature now. Some of the arguments/estimates were advanced here a few months back and fiercely attacked. Note current statistics (incomplete and necessarily biased due to the small amount of time elapsed since launch) show a failure rate of v1.0 satellites at around 3%, stable around last year's estimates that were argued to be far too pessimistic. [sarc] I'm sure the impeccable peer review here will demolish this FUD. [/sarc]

https://www.nature.com/articles/s41598-021-89909-7

Disclaimer: I'm not one of the authors, nor know them or have contributed to the study in any way.

[envy887]

Hitting Nature now. Some of the arguments/estimates were advanced here a few months back and fiercely attacked. Note current statistics (incomplete and necessarily biased due to the small amount of time elapsed since launch) show a failure rate of v1.0 satellites at around 3%, stable around last year's estimates that were argued to be far too pessimistic. [sarc] I'm sure the impeccable peer review here will demolish this FUD. [/sarc]

https://www.nature.com/articles/s41598-021-89909-7

Disclaimer: I'm not one of the authors, nor know them or have contributed to the study in any way.

3%? According to whom?

It's not peer reviewed, but I'm pretty sure Jonathan's data is good. He's showing that 22 out of 1615 V1.0 satellites have lost maneuvering capability and either have decayed or are stranded in orbit. That's only a 1.4% failure rate. (A few more have been intentionally deorbited, but unless you're trying to make some kind of economics argument, those are irrelevant).

And most of those failures are in early launches L1 to L7. Only 5 of the 1195 satellites launched since L7 have lost maneuverability, a 0.4% failure rate. This suggests that SpaceX is still making significant improvements to the design for reliability, although it would take some more math to show the improvement in MTBF.

https://planet4589.org/space/stats/star/starstats.html

Also, from the paper:

SpaceX will actively de-orbit its satellites at the end of their 5–6-year operational lives. However, this process takes 6 months, so roughly 10% will be de-orbiting at any time.

This is simply wrong. SpaceX has demonstrated deorbits from operation altitude in about 1 month of active maneuvering. For example:

https://planet4589.org/space/stats/megacon/bfig/s1158.jpg

Thus, if at any time there are 230 pieces of untracked debris decaying through the 550 km orbital shell, there is a 50% chance that there will be one or more collisions between satellites in the shell and the debris. As discussed further in “Methods”, such a situation is plausible. Depending on the balance between the de-orbit and the collision rates, if subsequent fragmentation events lead to similar amounts of debris within that orbital shell, a runaway cascade of collisions could occur.

Leaping from a 50% chance of 1 collision to a runaway cascade by handwaving a "balance" is wholly unsupported. Even 50 collisions involving around 1000x more mass than that, in orbits that last hundreds of times longer, would not be likely to result is a cascade:

https://twitter.com/ProfHughLewis/status/1387532062446456835

[eeergo]

Hitting Nature now. Some of the arguments/estimates were advanced here a few months back and fiercely attacked. Note current statistics (incomplete and necessarily biased due to the small amount of time elapsed since launch) show a failure rate of v1.0 satellites at around 3%, stable around last year's estimates that were argued to be far too pessimistic. [sarc] I'm sure the impeccable peer review here will demolish this FUD. [/sarc]

https://www.nature.com/articles/s41598-021-89909-7

Disclaimer: I'm not one of the authors, nor know them or have contributed to the study in any way.

3%? According to whom?

It's not peer reviewed, but I'm pretty sure Jonathan's data is good. He's showing that 22 out of 1615 V1.0 satellites have lost maneuvering capability and either have decayed or are stranded in orbit. That's only a 1.4% failure rate. (A few more have been intentionally deorbited, but unless you're trying to make some kind of economics argument, those are irrelevant).

And most of those failures are in early launches L1 to L7. Only 5 of the 1195 satellites launched since L7 have lost maneuverability, a 0.4% failure rate. This suggests that SpaceX is still making significant improvements to the design for reliability, although it would take some more math to show the improvement in MTBF.

https://planet4589.org/space/stats/star/starstats.html

I was indeed using his data, and given most Starlink have been launched a few months back at maximum, there's an extreme bias towards underrepresenting the true failure rate (for example, cherripicking L7 onwards to maximize numbers but water down recent early failures) even if improvements have probably been implemented from the more 'vintage' early units (although visors/darkeners might have also negatively impacted reliability, as the recently implemented new thermal system might show).

Anyway, working with the numbers we have and trying not to bias the statistics even further:

1615 v1.0 as you say have been launched so far.

22 are stillborn (screened, 4), or failed at operational altitude (more dangerous, 18). Both categories aren't yet deorbited. That's the *ongoing* 1.4% you quote.

PLUS 14 which were deorbited early because of malfunctions, they were abandoned right away or after some lowering. Only 5 v1.0 sats are classified as "disposal complete", or actively deorbited in semicontrolled ways, which might constitute the "graceful" failures you say are only relevant to economics (but also could as well be considered to have been "lucky" ones).
Finally, 7 more are classified as "anomalous" under debugging but with potential serious problems.
That's 22+14=2.3% at the very least. If you count the 7 anomalous ones too, it's 2.7%.
Finally, if you care about all visible failures, including the ones benefiting fron active disposal, it's almost exacly 3%.