Hacker News Reader

Comment by a-wu | Hacker News Reader

I have no experience in astronomy or satellites, but here's my naive idea. The article suggested erasing trails from the images using software. If SpaceX made an open API that detailed the precise location of every satellite at every point in time, could the imaging software use this to know that at this location in the image there is definitively a satellite that can be erased? I'm not sure what kind of sensors these telescopes use, and it probably wouldn't solve the issue of the bright spot messing up the exposure, but at least you could get rid of the trails?

View on HN

SpaceX is publishing the precise location of each satellite continuously. It's really cool actually; most satellite operators don't do this. The raw data is available at Celestrak here: https://celestrak.com/NORAD/elements/supplemental/

I'm using this data to power my site that shows when you can see Starlink yourself: https://james.darpinian.com/satellites/?special=starlink-lat...

That's a really cool piece of software dude! Will definitely use this in the future to show some friends the starlink satellites.

This is really cool

That "where" feature based on street view is really nice

The overlay on a nearby Streetview showing where the satellite will likely appear, with a timeline-based trajectory, is especially impressive.

This site is awesome! Have you thought about extending it to check for the position of the ISS? I would assume that information is probably readily available, as well.

Thanks! Yes, it can track most of the other bright satellites in the sky as well, try the base page link: https://james.darpinian.com/satellites/

Wow, this is great. Thank you!

What's the difference between the SpaceX supplied TLEs and ones derived from other sources? More accurate? Updated more often?

> More accurate? Updated more often?

Yes. There's only one other (public) source, https://space-track.org/ which publishes tracking results from US military radars. But the tracking results aren't as accurate as the satellites' own telemetry data.

I've mostly dealt with amateur and semi-pro equipment and I'm not certain what the state-of-the-art is so someone else might be able to answer this better than me.

These telescopes essentially work by capturing photon counts on a sensor. The individual pixels on the sensor have a limit to the number of photons they can count. You could theoretically subtract the satellite pixel-count values from the photon counts to get rid of the trails. The two problems I see are: 1) You don't know the correct counts for the satellites and I'm not sure how you could get them. 2) The trails will probably saturate the pixels anyway (which can also cause bleeding into other pixels), in which case you just don't have the data of what's 'behind' the trails.

Tl;dr: Perfect subtraction is impossible due to physics.

Problem is: With photon count the uncertainty in the number of photons also goes up (the relative error goes down). So even if you know that you should have received 100 photons from the satellite (and have not reached the overflow of 256 in this example yet), Poisson statistics means you will actually get anything between 90 and 110 photons. So if you subtract 100 you have an uncertainty of plus or minus 10 photons left. That is deadly if you astronomical source only gave you 2 photons in that pixel in that time.

Are the sensors recording a timestamp with the photon counts? If so, it should be trivial to filter out bright fast moving objects with zero impact on image fidelity.

Photons are not individually time tagged (that is much harder and reduces quantum efficiency due to dead times), but rather are collected for some exposure time (think 30 seconds) and then their count is read.

one gets time signals from photomultipliers, but those are too bulky for traditional pixel sensors. And something as bright as a satellite flare might fry them at an unexpected moment. They are usually built and tuned to be able to observe few to single photons.

could they put an LCD in front of it, and block the relevant pixels as a satellite passes?

That would limit your resolution to the resolution of a LCD panel. Also just not how cameras work

it'd limit your dimming resolution, not your visual resolution.

I mean to use it like auto-dimming mirrors, or smart headlights. Block the bright stuff that you can easily predict, so you can protect the sensitive image sensors from over-saturating on garbage.

Nearly every object in orbit is already tracked and any specific object of interest can be tracked even more precisely, so no need for an additional API. That won't work as they use multiple hour exposures that accumulate the light, you can't "erase" something, as most of the info is already discarded, unless it's a deep stack. You can try recording the video of the same field of view with a smaller auxiliary telescope, accounting for the satellites that crossed the view of the big instrument in the resulting exposure. But that's still artifact-prone, and won't work in many cases (saturated pixels, very faint objects etc). You have to pause the exposure for every satellite that crosses the view, if that's possible at all.

It's possible with a shutter. Not easy, but feasible.

I mostly meant observational possibility, not a technical one. Astronomers rarely have photons to waste.

Telescope images are usually long exposures so that won't work. AFAIK there's no method of selectively dumping or damping the pixels only ass the Starlink satellites pass through the frame.

To do it you'd need a screen in front of the sensor that could occlude the pixels that the Starlink sats were passing over..

If you're doing image stacking, could you dump some frames that were ruined? It would take a bit longer to collect enough frames.

During twilight our section of EVERY frame would be ruined. Massively reducing the useful observation time during twilight. And keep in might that some sources are only visible to some telescopes during those hours. And we have neither the money to build arbitrarily many new telescope, nor the suitable locations with good conditions.

Astronomers do put a screen in front of the sensor to occlude pixels. This is to block the light of natural space objects that get troublesome.

Those are static though right? For starlink you'd need to have it moving.

They can be dynamic, with pixels.

One such technology is the optical vortex coronograph. Several types exist.

Going further, put a small space facing camera on each satellite. The images can be stiched together to improve resolution and maintain continuous observation of space at all angles. Image downloads can be limited to save bandwidth.

Are you joking?

Probably not. I see a lot of ridiculous "startup-hacks" for a hard science problem in this thread.

There is SW already to subtract satellite trails from pictures. And astronomers have stated that it's not a big deal yet. But 42000 extra satellites might be too much. Nobody knows, everybody is worried, lots of talk.

Reply on news.ycombinator.com