Comment by avmich | Hacker News Reader

Comment by avmich | Hacker News Reader

Dense satellite groups are a real game-changer in how we observe the sky from the surface of Earth.

We're at the choosing point right now, what's better for us - a good uniform global communications or the pristine sky. The sky was always the norm; but it ceases to be so.

In order to have some compensation to astronomical society - that is, the whole population of Earth, as even looking to bight sky is getting be different - it would be good for SpaceX to create a program of high-orbit astronomical observatories. Working together with Earth astronomers, the company can participate in changes to observation possibilities in a way which has some upside as well, in addition to anticipated degradation. Hubble space telescope was launched some 30 years ago; an observatory of similar class or better can be significantly more affordable today, there can be several - or many - of them and that could be a worthy trade-off to the degradation of Earth-visible sky. I'm sure astronomical society would be very interested in participating and helping with building such a group of space observatories.

I think the key part here is "how we observe the sky from the surface of Earth."

The same economics that is putting cheap satellites in orbit should be usable to put cheap telescopes in orbit. What is needed to do that? Because I'm imagining an array telescope with the diameter of an Earth orbit.

I don't think SpaceX can lift LSST-class instruments to orbit (and transport the students/maintenance staff to visit/upgrade monthly), move radio telescopes like VLA/SKA/GreenBank to orbit, transport kids to dark-sky locations for the cost of a school-bus ride, make it possible to watch unfettered nebulae through a pair of 10x50s, or make it possible to photograph the Milky Way rising over a national park without streaks in the sky.

I don't have a solution to the constellation problem -- I just want to point out that, while launching JWST-class instruments would help to mitigate the impact for some professional astronomers, such a strategy cannot the broad impact of artificial stars in the sky.

Constellations can do undeniable good for the ground, but I think it is really a choice between dark skies and global satellite internet.

> Constellations can do undeniable good for the ground, but I think it is really a choice between dark skies and global satellite internet.

It's a good summary of the situation, yes.

> I don't think SpaceX can lift LSST-class instruments to orbit...

Literally - not soon, but the launch industry is at a point of fast evolution, so it might make sense to use it and device different kinds of astronomy tools - not multibillion gems, one-of-a-kinds, practically irreplaceable in case of launch failure, but numerous, with fast turnaround, sensors enjoying short design loops from conception to using and benefiting from decreasing prices of kilogram on orbit.

Frankly, some LSST-class instruments can only be based in space, and SpaceX can well be a good helper in getting them there.

> or make it possible to photograph the Milky Way rising over a national park without streaks in the sky.

With today's photographic equipment, yes. With naked eyes, unfortunately, yes. But there could be a non-insignificant replacement.

SpaceX's next-gen rocket, Starship, will have a 9 meter fairing and aims to be able to launch 100 tons into orbit for ~$10-100/kg. They also plan to support 3 launches per day. LSST has a primary mirror that's 8.4 meters wide, so something like it could potentially fit. Enormous radio telescopes can be constructed from a large array of smaller satellites via radio interferometry.

Two more ideas here. First, it would help if each satellite can have an astronomical payload; this is likely hard to achieve due to quite conflicting requirements between these two different goals.

Here is the second one. Earth observatories demonstrate streaks of satellites across astronomical images taken over comparatively long time. We know that at each moment satellite is in just one place in the sky; more, those constellations tend to have rather predictable orbits. Image sensing devices can use knowledge of satellite orbits to suppress reading a particular pixel at particular time if a satellite is expected to shine towards that pixel.

It's quite delicate task, I understand, but I also know that practical astronomy is known for inventing solutions to very diverse problems. This approach won't completely restore the abilities to observe the sky, but can significantly diminish the impact, discussed in the article. Especially if Earth-based devices would be constructed with such "shielding" (remember star shielding invention when working on exoplanetary discoveries?) in mind.

They're already doing that.

The streaky images in the press releases aren't meant to communicate "we don't know how to clean data," they're meant to communicate the severity of the interfering signal which must now be removed. Imperfectly, with imperfections in proportion to the severity of the interfering signal. Which is large.

> Image sensing devices can use knowledge of satellite orbits to suppress reading a particular pixel at particular time if a satellite is expected to shine towards that pixel

This is functionally the same, but a less flexible technique, as reading the pixel values anyway, and then discarding or covering up the values during data analysis/preparation. IRAF already contains these tools.

Getting shorter exposure times is actually helpful to mitigating satellite contamination. In the last 5 years, we're down to 1-2 electron noise per read, which is a full order of magnitude improvement.

Satellites are not new, so many techniques have been used and tested. https://aas.org/sites/default/files/2020-07/MitigationWG_Pre... is a relatively nontechnical survey of some of the techniques that Rubin plans to employ.

Suppressing pixel readings is not plausible with the current CCD designs, as far as I'm aware. These devices are so sensitive that any wiring needs to be done _extremely_ carefully to minimize crosstalk and keep temperatures extremely low (think under -100C).

Masking in data processing is the more likely technique.

> Suppressing pixel readings is not plausible with the current CCD designs

Better than suppressing pixel reading would be suppressing pixel changes. The first process lets pixel to change as a result of incoming photons, the second one maintains the "unread" state.

Yes, current CCD devices can't do that. Advantage of this change is also in reducing crosstalk - if we need, say, to remove accumulated charge and for that we need to draw the current, the current may influence our reading of neighboring pixels. I'm tempted to look towards "smart mirrors" - MEMS systems, even for mirrors of 1 sq. micron - but purely electrical solutions might be possible.

Monopolistic communication network. Sorry, it's hard to see the benefits.

As compared to every other ISP? This won't (can't) replace terrestrial ISPs.

Of course astronomical societies would be interested in someone gifting them some wildly expensive equipment.

To get better - much better sometimes - view of the sky, yes. And not only gifting - working together with world-class practical space engineers on building space observatories with potential benefits not expected before this issue becoming significant.

I'm serious here - seeing Elon Musk traditions, it would be not ridiculous to test such an approach.

Reply on news.ycombinator.com