I love all of the comments about using software to filter out the images. This response is so obvious that it is from someone who has never worked with data from imaging the sky, and probably have never thought about it more than a drive-by comment. "Just filter it out" they say. And replace it with what? An average of near by pixels? An AI background replacement? That's just bollocks. Every single dot in the sky is another star, galaxy, planet, asteroid, etc. They are not blemishes in the image. To just write them off would be like drawing a map of the earth without New Zealand on it.

To reinforce this, there is a fair bit of astronomy that is looking for single photon differences. The optics and engineering here is very cool (pun intended) and can take months of time on an instrument to detect, if there is anything at all. Sources of noise are pathologically chased down. Adding in noise would be like trying to chop off your arm, easy to do once, but hard to undo.

As a reminder, astronomy is not as much about the 'image' as it is about the 'measurement'. Editing pictures like in Insta is not the point here, you can't fill things in.

Exactly. Modern day comet/asteroid hunters are taking images from night after night and treating them like time lapses to see what dots are moving. Removing these dots to fill in the void left by the satellite streaks sets this method of detection back to unusable/unreliable. Even filling in the removed satellite streaks in an image with data from previous/next images totally invalidates the tracking of objects in motion.

.. and I have given up chasing down the 'JuSt UsE SpaCE teLeScoPEs' comments. These people would write off the loss of New Zealand as the market simply finding the right benefit-cost ratio.

No, that's not how that works at all.

Remember, astronomy images generally have long exposure times, which already involve tracking the sky, and then averaging out values in the exposure across minutes/hours/days/whatever.

A satellite passing over a point in the sky means that area of sky is affected for only a tiny fraction of the total exposure. As long as you have a database of all satellite locations, you just don't include data from those time slices in the final average. Or if you don't know where satellites are in advance, you just do basic outlier detection -- the kind that already eliminates things like airplanes, birds, the ISS, etc.

Nobody's filling in anything with "black", and the area of satellite interference can be easily expanded to include the "glow" area.

In the end, the quality of the overall image is degraded by a tiny fraction of a percentage, which can be trivially overcome by extending the total exposure time by the same tiny fraction of a percentage.

I guess you could statistically impute that data. But it is the long tail of unlikely readings which are generally interesting, and imputation will not find those.

This is a logic flaw I see everywhere. Making a proposition for what might have been there is not a useful scientific instrument. And especially not when unseen anomalies are an integral part of pushing your field forward. If you're inferring what might have been there you're explicitly filling in based only on what you already know.

I agree, but isn't filing in what you already know a common technique when presenting things for human consumption? I'm thinking scanning microscopy, SEM, STM, etc.

The image implies that all the data is from the same instant, when it's not. Normally that's OK, but it's not always.

SEM fills in data?

Unfortunately, the 'glow' is from the optics, not the CCD (you'd see streaks or 'boxing' that way). You can't statistically impute the optics, as these issues are based on physics. A good starter article here; http://spiff.rit.edu/classes/phys312/workshops/w10c/telescop...

Thanks, that was a fun read. I think I have a better grasp of why you can't impute this now.

Imputing makes little sense when you are trying to _measure_ though.

It seems to be around 1-2%, which I would say is trivial. This can be further reduced if you use multiple exposures because then only 1 contains the streak. Getting satellite data from SpaceX about satellite positions and turning off the receiver for a few seconds would also mitigate the issue. I don't understand how this is treated as a problem that's worth halting satellite launches over.

In this sample image it's not just the white streak but the glow caused by it which extends about 5-8x it's width to either side. More like 20% of this image is unreliable data.

Taking multiple exposures isn't always possible, this may be a continuous multi-hour exposure. Additionally if you're looking for transients, NEOs, or star field mapping then there is definitely lost data that can't just be inferred.

The conversation isn't just about halting launches. It's about recognizing and working to mitigate the issue now before we get too far down this path.... and accepting that it's not a trivial "photoshop" problem.

> which I would say is trivial

I'm frankly amazed at the arrogance on display here. Do you honestly think you know better then the entire field of astronomy?

You point to saying that 2% is a small part of an image as a sign of arrogance? Isn't it a small part of an image?

You didn't say it's small, you said it's "trivial". This isn't about people taking pretty pictures, this is about people trying to do actual work. It doesn't matter if it's 1% or 100% if it still ruins what they're working on. So yes, you're being arrogant by saying that just because it's a "trivial" amount that's ruined, when you don't understand what they're doing.

2% is trivial. They lost 2% of information (as someone pointed out this might be closer to 20%). This means the cost to the astronomer was 2% (or 20%). Calling this "ruining" is an exaggeration. Of course it would ruin a pretty picture, but "This isn't about people taking pretty pictures, this is about people trying to do actual work." where you can choose the part of information you take into account.

Except if the thing you're trying to measure is in that 20%. Astronomy is time limited, you may never get the opportunity again.

It is a small part of an image, but that does not mean it's not a significant disturbance.

Think for instance about CO2 levels. Is it such a big deal if they reach 1% of the atmosphere? It's only a small part after all.

If the entire field of astronomy is so educated on the subject, why do most not know Starlink was never visible except at dawn or dusk, and it’s already implemented successful mitigations for those tones with visors?

...wow.

(1) astronomers are pretty darn aware of everything going on, and it's really shocking as an astronomer to see these sorts of claims. There have been close collaborations with Space X to try to mitigate some issues, but these are partial mitigations and the most important ones are only planned, not implemented. The "successful" mitigations I think you're referring to is their "DarkSat" coating, which is not actually a successful mitigation. It is still ~2x as bright as the ~maximum level of acceptable brightness for e.g. LSST.

(2) Space-X did not release any details about brightness characteristics, and they themselves claim that they were "surprised" by the brightness. To people that work in space rather routinely, that is staggeringly incompetent.

(3) > why do most not know Starlink was never visible except at dawn or dusk

Near earth object (NEO) detection only happens during these times. NEO includes "asteroids that threaten human life" so I would hope you agree that this is pretty important. Also, depending on latitude and time of year, "dawn or dusk" can constitute your entire observing window. So this doesn't really make things better.

(4) all of these mitigations should be done before companies can start doing ~10,000 satellite launches. The only reason this isn't done is outdated legislation that never conceived of this being possible.

SpaceX has just implemented sun Visors on the latest batch of satellites, so why aren’t you aware of that?

In space infrared detectors are the most effective way of searching for dangerous NEOs, and SpaceX has made them far cheaper to launch. How come you don’t know that?

https://www.nasa.gov/sites/default/files/atoms/files/nasem_r...

How about looking at what the username of the person you replied to spells out. Then click on it, look through their comment history and do some homework.

That doesn’t give them cart Blanche to post incorrect assertions. That would be called a “appeal to authority” argument.

I think you're mistaking "giving them carte blanche" for "highlighting one mistake they make in a long post containing a lot more information, context and nuance without acknowledging any of that in the slightest, and use that as if you caught them red-handed in a gotcha! that breaks down their entire argument".

I guess that's a specific version of the "cherry-picking data" fallacy.

how come you are so condescending?

> astronomers are pretty darn aware of everything going on

Then they claim SpaceX is only trying paint.

Unless Musk has secretly invented Romulan cloaking technology, being invisible doesn't help when you are trying to look at the stuff behind

What percentage of the night sky will be obscured if Starlink goes to 30,000 satellites?

Let me guess, if the visual radius of a one meter satellite 550,000 meters away is (1/X)^2. Thirty thousand would be 1/one trillion, or far less?

Its not just about Starlink though, its about setting a precedent before every company and country starts launching competing constellations. When should we start to care after the first few thousand or after hundred of thousands of eventual satellite's?

It would take trillions to appreciably dim the night sky. Wake me then.

CCDs are not really receivers that can be turned off for a couple seconds. A physical shutter moves in front of the detector and covers it (for most instruments as they exist right now).

Can you move the shutter in front of the receiver when you know a satellite will transit?

Sure. And for small field of view things that might be the future. But if you are looking at large chunks of the sky at some point with enough satellites the shutter will have to stay shut for the whole duration of the exposure.. so not really a good solution for that.

What if SpaceX funded research into shutters that cover part of the CCD? Is that a possible solution?

Shutters for astronomical instruments are specifically designed to open and close evenly, so that each pixel of the detector receives the same amount of exposure time (as good as is possible). Otherwise it becomes very difficult to calibrate the image, since some parts of the image received more light than others.

Even opening and closing the shutter multiple times during one exposure will cause issues - but probably still possible to get under control. But covering parts of the detector will create significant variants that will be extremely hard to control for. Keep in mind we don't really want pretty pictures, but we use these images to measure the brightness of objects in those images (photometry).

What if spacex just didn't pollute the sky for an entire planet?

We wouldn't get planet-scale internet

(can't reply below) If you had a localised shutter covering pattern X for 6 seconds, could you compensate for that by covering the inverse pattern of X for 6 seconds to get the same exposure for all pixels? So if you covered 3 satellite streaks of 6 seconds each using non-verlapping patterns (A,B,C) you could compensate for that by ending your exposure with covering the inverse of A+B+C for 6 seconds, adding only 6 seconds to your total exposure time and get uniform exposure.

Herein lies the problem. Your proposed solutions are ever more complicated as you begin to grasp the magnitude of the issue. Implementing this idea, and I'm not saying that it's good enough, would require modifications to pretty much every major telescope in the world. And there's no economy of scale because most of them are custom builds. Who's gonna pay for that?

That is only necessary in a small fraction of cases. You can just ignore the part with the streak or lower the shutter when you're using existing equipment.

If the conditions are quite stable that might work (or if one can live with limited accuracy), but conditions tend to change over time (moon position, thin clouds and of course the seeing - how much the atmosphere distorts the image) so nothing is as straightforward as one might hope for.

In space that approach would work quite well indeed, but then you wouldn't need it in the first place..

Edit: Even if one still has to mask the streak in the data in the end, it might still be useful to do, since it will protect the rest of the image from any excess blooming.

How much does the bloom extend in the image fom the streak? Can you measure it in arcminutes or is it optics dependent? Is the bloom present if you cover up the satellite exactly? (Is it caused by the sensor, or would diffraction around a shutter cause it as well?)

I think you could mitigate the problems with conditions changing if (instead of blocking out the streak in the image with some per-area shutter) you tracked the satellite in the image with a horizontal (or vertical) shutter stripe which has the right width to cover the entire satellite. This would make the data lost really minor, and changing conditions wouldn't be a problem, and the exposure would be even. Does that make sense?

If no light falls onto the detector then there is no blooming. And of course there will be diffraction from the shutter edges as well, so engineering it will be quite a challenge (one can take hints from coronographs). If the satellites does end up on the sensor then the dimension of the streak depends on the optics, how bright it is, the used detector, plate scale, etc. (if you are extremely unlucky you might have internal reflections too and they might end up who knows where on the detector - optics is hard).

what if we actually put the astronomical observation satellites above the spacex ones? that should solve the interference problems. Wasn't there a cool new telescope named after James Webb that launched recently?

JWST still hasn't be launched yet.

And space observatories are much much more expensive than ground-based ones. JWST for instance will cost around 10 billion dollars in the end - and they don't even have to pay for the launch cost (one of the contributions of ESA to the project is free launch on Ariane rocket). JWST will have a 8m mirror. Currently three 30m class ground-based observatories are being built (ELT is 39m, TMT 30m and GMT 24.5m) each with an estimated cost between 1 and 1.5 billion dollars. So you get three much much larger ground-based telescopes for the price of one large space observatory (and still have money left over).

1-2% of the size of the entire Universe is trivial to you?