Hacker News Reader

If you are looking for short period events, losing a single frame can mean losing critical data.

> Or they can just use that information for better scheduling: wait a minute or two to image a particular spot, so there won't be satellites in the field of view.

Which means more time and effort, and ultimately money needs to be spent on scheduling observations to avoid the thousands of these satellites.

And sometimes it isn't an option. What if a GRB happens in the same field of view as one of these satellites and you want to observe it as soon as possible?

Now there are some types of observations that won't be impacted that much. If you can tolerate scheduling around it, or throwing away some of your frames it is just a minor annoyance. But a lot of observations are also more sensitive to such disruptions.

> What if a GRB happens

Yeah, what if? I mean, you probably missed the GRB, sure, but apart from that, you'll be fine and you can wait for the next one, or find a data from somewhere else.

GRB - Gamma Ray Burst (I am guessing)

Yes

Forget satellites; you can't do raw long-exposure for "serious" terrestrial-based astronomy anymore because you'll get planes in the way.

> They're wide field of view, long-duration exposures. That used to be the way astronomers imaged space,

Sorry, but one of the two example images (https://noirlab.edu/public/images/iotw1946a/) is a single 333-second exposure with a modern survey camera, the Dark Energy Camera. This is not particularly long nor does it represent some outmoded observational strategy. Large, wide-field imaging sky surveys (such as the upcoming Rubin Observatory) are among the highest-profile ground-based astronomy projects today.

Masking and stacking can mitigate the problem but it does not of course compensate for the lost area and sensitivity. And the brightest satellites (like BlueWalker) saturate the readout electronics and spoil the whole exposure.

Narrow field instruments (such as spectrographs) have less geometric chance of seeing a satellite but tend to take longer exposures (tens of minutes), so there is a greater loss of telescope time when a streak does happen.

Even space telescopes are affected by streaks (https://www.space.com/hubble-images-spoiled-starlink-satelli...).

> It takes a bit more work and software

Equivalently, it takes more money and time. That just means less science, given flat to declining funding from Congress.

> Masking and stacking can mitigate the problem but it does not of course compensate for the lost area and sensitivity. And the brightest satellites (like BlueWalker) saturate the readout electronics and spoil the whole exposure.

Is this only a problem in systems that aren't aware of where the satellites will be? I naively assume that, if the system knew, it could start exposing in the next clear window. I naively assume that the window is almost always clear, especially for an individual sensor.

Even if they don’t know where the satellites will be, a simple motion detection algorithm in the stacker would delete the streak caused by the satellite. A given pixel will only be impacted for a few exposures, so just grab 5% more images.

I’d guess terrestrial light pollution is a much bigger problem.

  >so just grab 5% more images.

You say it like major observatory telescopes are sitting idle, but they're not. In practice what actually happens is you lose 5% of your scientific data return.

  >terrestrial light pollution

Problem is that it's global vs local, so our normal solution for terrestrial light pollution (build telescopes in a remotely populated area) doesn't work.

So just write all grant proposals with an extra 5% slop.

Why would they not already be trying to maximize grant size?

Budgets are fixed, so again this also boils down to "just do 5% less science."

Believe it or not, actually using 100% capacity on the big expensive telescope you paid for isn't some brilliant unheard-of suggestion like people seem to think. This Dunning-Kruger idea always seems to crop up whenever this particular topic is in the news.

There simply isn't any "slop in the system" that lets you get that 5% (and climbing!) back "for free." If there was, then that inefficiency should be fixed regardless of the situation with megaconstellations.

Doing 5% less science is an acceptable cost for full commercial space exploitation. I'd rather have internet service in 100% of the US than an extra few graduate theses.

And who knows, necessity is the mother of invention so one of those grad students could invent a way around it.

  >Doing 5% less science is an acceptable cost

That's certainly one choice we can make as a society.

However my point is we shouldn't delude ourselves that there's some "easy" fix, therefore the cost should actually by counted as zero. This defense mechanism is misguided and uninformed, yet it's shockingly common to have it (or some variation) crop up when discussing this particular topic.

Personally, I blame the current tribalize-all-the-things trend. You're either on Team Starlink or Team Astronomy. Only two options. Pick a side, we're at war!!!

  > And who knows,... one of those grad students could invent a way around it.

See? Shockingly common. :D

Even when we don't even have an idea, nevertheless we feel oddly compelled to suggest that maybe there's an easy ("grad student") solution that makes the scientific cost simply go away.

If we're willing to pay the cost in lost scientific data, then let's do it and say so. We shouldn't live in (oh so tempting!) denial about the downsides. That's all I'm saying.

Yes, but that's a system problem, not a user problem. The comment I replied to posited that a specific user would collect 5% less data. If you add ~5% to your grant proposal, you get the same amount of data.

As for the system problem, that is solved by building 5% more observing capacity. Or, more realistically, starting new telescope projects slightly earlier. This really isn't as complicated as you're trying make it out to be. Pot, meet kettle.

  >As for the system problem, that is solved by building 5% more observing capacity.

If this is to somehow come from existing funding sources, I'll need a moment to pick myself up off the floor laughing. ;D Obviously those sources are already at their budget appetite.

Or maybe this an Efficient Market / Pigouvian suggestion, where megaconstellaton operators pay (and pass on to their customers) into an Astronomical Reparation Fund worth ~5% of current global astronomy funding, to be distributed to astronomical grants and construction of new observatories? Because sure, that seems fair.

If you truly believe (as do I) that we value internet access more than the loss of scientific data, then you'll agree the internet users can easily reimburse scientists for the damage and still come out ahead. "You broke it, you buy it."

  > If you add ~5% to your grant proposal

..then I'll add 10% to mine!

It's a race to the bottom. Eventually folks get sick of it, so they enact rules and enforcement mechanisms to prevent padding. In the end all we accomplish is transforming an efficient & high-trust system into an overhead-laden, low-trust system.

if there were extra 5% budget they already would be using it

If it's just a matter of 5% inefficiency, I have to think that there must exist some possible mechanism for capturing the economic benefits of satellite internet (with a tax, perhaps) and redirecting a portion to astronomy to increase budgets there by 5% (or more) so that everybody wins.

Where do you see that the full well of those CCDs are fully saturated by a single satellite streak?

Even if that was the case, that seems a very poor design for a sensor and not really an issue of satellites. If a satellite streak can saturate your well, then so could a decently dense star field.

Do they blow out the individual pixels? Sure. But you just stack.

You don't lose sensitivity just because your stack rejected pixels, that's not how stacking works. You wind up with a tiny bit more noise, how much more depends on your capture. But not lower sensitivity, stacking isn't just average an area

It's not really that it saturates the entire image, it's that bright objects creates non-linear and ghosting artifacts which degrade the sensitivity of the entire exposure, even after post-processing in software.

https://www.space.com/bluewalker-3-prototype-satellite-brigh...

https://www.nature.com/articles/s41586-023-06672-7

https://iopscience.iop.org/article/10.3847/1538-3881/abba3e/...

Thanks for providing all those links. Indeed, the key there is peak magnitude. BlueWalker 3 has a giant array that reflects light like a mirror. So when you happen to get the angle of incidence exactly equal to the angle of reflection, it shines quite brightly. Of course since it's reflecting all sunlight in one direction, that means from all other spots on earth it will be extremely dim.

Also note that BlueWalker 3, like most low-earth orbit satellites, is only visible during "terminator conditions", when the satellite itself is illuminated by the sun but the telescope is still in darkness. Those times are typically an hour or so after sundown or an hour or so before sunrise. So one solution is to schedule astronomy for the middle of the night, when none of these low earth satellites will lit by the sun at all. Which certainly increases the costs of astronomy, since you can't use your telescope for as many hours per day.

Or, just use your space object catalog to look elsewhere in the sky when bright satellites happen to be in the sky during terminator illumination conditions. Which I see is one of the things your last link, "Mitigation of LEO Satellite Brightness and Trail Effects on the Rubin Observatory LSST", suggests!

The BlueWalker 3 antenna is white, not mirrored. It will reflect sunlight like a lambertian reflector, not like a specular reflector as you suggest.

https://ast-science.com/wp-content/uploads/2022/07/2022-03-1...

SpaceX has offered their space-rated mirror technology at cost to other satellite manufacturers, but it seems AST SpaceMobile hasn't taken them up on the offer. I can't see any mirrored surfaces on BlueWalker 3.

  >that means from all other spots on earth it will be extremely dim

The second paper I linked (see Figure 1) shows a minimum magnitude after array deployment of 5.5 magnitude, from multiple observatories at multiple locations on the Earth. That's still 10 times brighter than the IAU limit.

  >just look elsewhere in the sky

The hottest thing right now is whole-sky surveys like Vera Rubin (which are important because they can alert other telescopes to transient events), which can't really do that because they're already looking at... the whole sky.

> So one solution is to schedule astronomy for the middle of the night, when none of these low earth satellites will lit by the sun at all.

Let me translate that a bit: "So one solution is to just eliminate 20% of all available observing time."

Not really. Astronomical dusk and dawn are different than civil dusk or dawn, you'd normally skip an hour or so after sunset and before sunrise anyway.

> these days some of the best astronomy is performed by telescopes in space, so astronomy overall benefits by having easier access to space.

Public access satellites. The FCC requires broadcast licensees to demonstrate that they're acting "in the public interest". Should private rocket launches have to give a minimum amount of their payload to research and non-profit purposes?

Sure, why not? A Hubble-sized telescope is only about 12 tons, which is only about 70% the mass of what a Falcon 9 can deliver to low earth orbit, and SpaceX does that 100x per year (it's only May and they've launched 47 so far this year). For reference, Spitzer is 1 ton and Chandra is 6 tons.

The elephant in the room is that launching one Hubble per year to a Hubble-like orbit only uses about 1% of SpaceX's current annual demonstrated launch capacity, and the actual reason people are upset has little to do with the actual impact on astronomy and more do with wanting to punish a certain CEO. One new space telescope every year would completely revolutionize the field, yet he'd barely notice. Better think of something else.

The upset is also because urban people or people in many denser areas generally don’t really benefit from Starlink etc. You can bet if those types of satellites were the only readable way for them to get internet, astronomy would suddenly fall way down on their list of priorities.

Fun fact: in 2012 the Department of Defense donated 2 Hubble-class telescopes to NASA it had lying around and didn't need:

https://en.wikipedia.org/wiki/2012_National_Reconnaissance_O...

Space telescopes cost billions of dollars not because of launch costs but because of all the dedicated engineering required to fit the telescope into the size and weight restrictions required for launch. In the past we've always been willing to go to extreme lengths to meet launch requirements (hand-made parts with exotic materials, folding mirrors with intricate deployment procedures, etc.).

If we instead design something to be mass-produced at a cost more commensurate with that of a Falcon 9 launch, then it may well be much better than Hubble in terms of bang for the buck, but it won't be as capable as Hubble in absolute terms. Even so, although Falcon 9 is very cost effective per unit weight compared to past rockets, it does not have a large payload volume compared to launch vehicles used for previous large space telescopes, and volume is usually what you need for space telescopes (because big mirrors are constrained by volume, not weight).

Once Starship becomes fully operational it will completely transform the landscape because it is not only low cost but the interior payload bay volume is much larger than existing rockets in all dimensions.

The thing is, paying SpaceX's going rate for launch only adds a tiny sliver to the cost of a "Hubble-sized" telescope. Launch costs have become very cheap thanks to SpaceX - whether they donate it or not is somewhat like an American family getting a 100% discount on rice...it's not going to change their overall monthly food costs, because rice is already so cheap.

You ignore the entirety of the LSST which will take more image data in its first year than all of humanity.

It is wide field, and these constellations cannot just be stacked out.

You think you are more informed than you are.

OP is not wrong, there is no satellite constellation that can't be "stacked out" That's precisely what stacking does.

Trust me, I'm an astrophotographer, you can check my website in my profile, I know what I'm talking about here.

There's a bit of difference between the aims of astrophotography and professional astronomy.

Ones getting a subjectively good image where the other is getting an objectively good quantitative measurement.

Do the software "tricks" really translate that well. If they did, these terrestrial telescopes wouldn't engineered to the tune of billions.

Stacking isn't a "trick". It's mathematics.

I agree my aim is to make pretty pictures, and professional astronomy is to take accurate imagery, but up until I'm making subjective calls about colors, they're not too far apart before the image is stretched to a non linear histogram.

After all, I can't build pretty pictures on crappy data. Astrophotography is a heck of a lot of math

> the photo illustrations used in the article are selected to make things seem worse than they really are

Exactly. I do a LOT of astrophotography and satellite trails are actually very easy to get rid of. Even the stupidest, most naive outlier rejection techniques, such as taking a stack of tracked images, finding their mean and standard deviation, throwing out any data points outside of 2 standard deviations, and then re-averaging, will get rid of the satellites very cleanly. You can go to more advanced techniques such as doing linear fits and RANSAC and whatnot, but you get the idea.

NOT doing any outlier rejection, just taking the mean of all the images will show the satellite tracks, but very dimly. The people who are trying to make noise on social media deliberately and disingenuously take the max() instead of the mean() to make the problem seem worse than it is.

That said -- in all of my imaging sessions, aircraft are a much, much, much bigger problem than satellites, and still easy to deal with. Yet nobody makes any noise about aircraft.

Seriously, any time you have >3 images to stack satellites and planes are a complete non issue. The median pixel value is always going to exclude those. Five images is enough to eliminate any trails even if they overlap between images.

You're not imaging with cryogenic CCDs --- satellite trail can blow out whole rows.

Just do it in space and avoid the whole atmosphere and ground based light pollution. Space is cheaper than ever now, this is a problem for amateur astronomers, and hobbyists shouldn’t hold back progress.

Space imposes extraordinary costs and limitations which heavily precludes exploratory research. Instruments are built for space to achieve specific tasks which we know will be successful.

Instrument development has to first occur on earth-- including the research that lets us know which things will yield results deployed in space.

Astronomy aside, fleets of satellites once they turn to junk may eventually make further launching problematic.

SpaceX deliberately launches satellites into orbits which are too low to sustain on their own. Once the satellites run out of propellant, which happens after a few years of normal operation, they cannot remain in orbit indefinitely and eventually burn up in the atmosphere.

On very rare occasions, a launch malfunctions in such a way that the satellites end up in unintended orbits that do turn into long-term junk, but this is not the norm.

"just do it in space" has got to be one of the best hacker news solution to other profession's issue I've ever read.

This is equivalent to saying "Just do one percent the space science we do now"

Where is SpaceX offering to put up a billion dollars worth of space observatories to offset their pollution of a public good?

No, I'm not, I'm using cooled CMOS sensors. CCD is ancient tech these days.

> CCD is ancient tech these days.

Their performance at cryogenic temperatures is quite good which is one reason why currently in design/construction scientific instruments are still using them.

And in those instruments serious overload from shiny satellites is not just a little streak that can be easily excluded. https://www.aanda.org/articles/aa/full_html/2020/04/aa37501-...

Taking pretty pictures, and collecting scientific data are very different.

If a satellite is anywhere close to your target object in a frame you are using for photometry (measuring the amount of light), the frame is ruined, because it adds noise.

You ignore the entirety of the LSST which will take more image data in its first year than all of humanity.

It is wide field, and these constellations cannot just be stacked out.

You think you are more informed than you are.

Why is stacking not possible in this case?

The objects are so bright and so numerous that they corrupt an entire imagers data for the entire exposure. In a CCD rows and columns can become saturated and hot pixels can spill over into neighboring pixels (blooming). When the imagers for the LSST were being designed, it was not anticipated that they would have to deal with tens of thousands of hot fast moving objects.

https://www.lsst.org/content/lsst-statement-regarding-increa...

Not to mention this only happens at dusk and dawn when the satellites are in sunlight.