Vera C. Rubin Observatory/LSST
- Thread starter parejkoj
- Start date
This is a nice writeup about Rubin/LSST, with some quotes from a colleague about the satellite problem:
https://www.optica-opn.org/home/art...ry_2024/features/the_vera_c_rubin_observatory
https://www.optica-opn.org/home/art...ry_2024/features/the_vera_c_rubin_observatory
Although I think technically I'm not supposed to say too much about the details, a current annoyance (LSST will be very good at spotting satellites) on the project is featured in this article about "sensitive topics" research. This is the other half of the "there are too many satellites" problem: the ones that don't want to be seen in the first place.
https://www.science.org/content/art...earch-security-without-disrupting-peer-review
https://www.science.org/content/art...earch-security-without-disrupting-peer-review
Well, I've got a bit of news for the "don't want to be seen" satellite crowd... if you're standing in the room, unless you genuinely can become invisible, you're going to be seen. Demanding that everyone with eyes not look in your direction doesn't work too well, as everyone now knows that there's SOMETHING in that direction that doesn't want to be seen or known about.
Reality has an automatic Streisand Effect going on.
Reality has an automatic Streisand Effect going on.
Eh, I think there is a decent gap between the US government asking the observatories it directly funds not to share certain images; and suggesting that the USA is demanding that no-one look in certain directions. The former is at least within spitting distance of reasonable; the latter is entirely laughable. People have been publishing ephemera tables and photographs of NRO satellites for decades, so it seems implausible that anyone would expect anyone to take the latter suggestion seriously.
Everybody (that cares to and that the NRO would care about knowing) knows where NRO's satellites are, because they're too big to be concealed and when you rule out everybody that is more than happy to tell you all about their satellites you're only left with intelligence birds.
What really matters though isn't that they're up there. It's what they're looking at and what they can see.
What really matters though isn't that they're up there. It's what they're looking at and what they can see.
It's been tried a few times before, and the result of any kind of removal of satellites from public databases is that amateurs find them again (because satellites are really, really, really bright) and put them in another database.
I would imagine that if amateurs can find them its just confirming that Russia/China almost certainly have already
moosemaimer
Ars Scholae Palatinae
Trying to hide something's orbital parameters seems like a good way to ensure someone else puts something into an orbit that intersects without knowing.
Is the request to hide orbital parameters, or is it to censor the high quality images captured by one of the largest telescopes in the world?
It's not really possible to hide a satellite's orbital parameters unless it's literally completely quiet. Because it HAS to be moving at a very significant speed, even crappy SDRs are able to track the doppler shift of any signal that a satellite is putting out. And in case you're wondering 'but what if the signal is really really low?' - yeah, amateurs have been able to listen to the attowatt-level signals from Voyager. No matter how well you try to beamform and shield your signal, it requires nothing more than $200 worth of radio equipment and the free software Audacity to figure out the orbital parameters of an unknown satellite.
It might be plausible for this aspect of observability to change with radio communications replaced by optical communications, especially to relay satellites rather than to the ground. For all we know, this is the actual purpose of the Starshield project, and the broader secure communications stuff that's justifying the expenditure publicly is just a cover.
So, some clarifications: this is somewhat similar, but not quite as bad, as the restrictions that were put on Pan-STARRS early in their survey. Those restrictions were lifted after a few years.
We have a tech note describing the details. The main parts are an 80 hour release delay (previously full images from one night would become public the next morning) and not sending alerts on anything that could be a satellite. We didn't want to alert on satellites anyway (they're not a science goal, so they're just junk in the data stream), but there's the usual 80/20 problem of getting "everything" instead of just best effort. Plus, very near earth asteroids (the ones most likely to be dangerous and one of our top survey goals) can move fast enough to look the same as a satellite streak. "Oops, we missed that city killer because we can't alert on streaks" is not a good look, but we're stuck with it.
There is already a secret table of the orbital elements of recon satellites; I think it's held by Space Force now? That is consulted when checking for "dangerous" passes of commercial satellites, without ever releasing the list of secret ones. Similarly, when they say "don't release alerts on anything in these orbits", that list includes real things, non-secret satellites, and empty sky.
The thing that makes this so ridiculous to me is that amateur satellite hunters already do a pretty good job tracking supposedly hidden satellites with mostly binoculars and small telescopes. The Chinese and Iranian intelligence agencies would be absolute idiots (they're not!) to not know where all our stuff is with only slightly more sophisticated gear.
We have a tech note describing the details. The main parts are an 80 hour release delay (previously full images from one night would become public the next morning) and not sending alerts on anything that could be a satellite. We didn't want to alert on satellites anyway (they're not a science goal, so they're just junk in the data stream), but there's the usual 80/20 problem of getting "everything" instead of just best effort. Plus, very near earth asteroids (the ones most likely to be dangerous and one of our top survey goals) can move fast enough to look the same as a satellite streak. "Oops, we missed that city killer because we can't alert on streaks" is not a good look, but we're stuck with it.
There is already a secret table of the orbital elements of recon satellites; I think it's held by Space Force now? That is consulted when checking for "dangerous" passes of commercial satellites, without ever releasing the list of secret ones. Similarly, when they say "don't release alerts on anything in these orbits", that list includes real things, non-secret satellites, and empty sky.
The thing that makes this so ridiculous to me is that amateur satellite hunters already do a pretty good job tracking supposedly hidden satellites with mostly binoculars and small telescopes. The Chinese and Iranian intelligence agencies would be absolute idiots (they're not!) to not know where all our stuff is with only slightly more sophisticated gear.
Even all the 'optical' comms (from mmWave to THz to true optical) diffract and diverge enough to be easily detectable. Even the best modern optical beam devices still have like half a degree of divergence to 10% intensity.
It's an unwinnable physics puzzle - to get enough SNR for anywhere near useful amounts of communication you need to have fairly significant comms power, and once you have that you're spreading around so many photons that it becomes very easy to see stuff with even mundane electronics. Like, a DSLR sensor can detect single photons these days.
This is separate from the threat posed by possible military devices in space. You can know exactly how they move, but that won't tell you what they have inside or when they might deploy a deadly payload.
Close to where all the big observatories are is a small, very touristy town called San Pedro de Atacama. There lives a young French baker, who fell in love with a local and stayed, and promptly opened a genuine, no-holds-barred fantastic French bakery right there. I’ve been to it.
He called it "Franchutería" as "Franchute" is slang for "French guy." I guess a decent translation would be "The Frenchierie.”
It is therefore entirely possible not just to find a croissant, but a GOOD croissant, and a very genuine baguette, in the area.
Also, if gotten at the observatory itself, it’d probably be Nescafé, which (still, alas) is the default “coffee” in Chile. Even restaurants will serve you Nescafé unless you specifically ask for -and they have the equipment to make- espresso. Coffee shops are legit, but restaurants are a crapshoot.
—Chilean guy
The main optical observatories are mostly quite a bit further south, closer to La Serena than Antofagasta. But it's good to hear that there's a good bakery! I mostly wasn't impressed with the bread on my trips to Chile. The ice cream and empanadas though, those were top.
Yeah, all the coffee snobs I know who have gone down had a bit of a rude awakening because of this.
My memory is that the coffee at Paranal Observatory is pretty decent -- at least espresso and its variants, because Paranal is operated by the European Southern Observatory, and there are a lot of Italian astronomers, and they (at least) weren't going to put up with Nescafé.
Plus, the observatory Residencia was the "hotel" featured in the last act of the James Bond film Quantum of Solace. (Though the real rooms aren't as nice as the ones in the movie.)
I don't think LSST will cause any actual security issues. I find it hard to believe that china/russia are not successfully tracking all of our spy sats already.
Whether or not that's true (and for the record, I agree with you completely), we've been told we have to work extra hard to remove potential satellites.
As I said, it's the 80/20 problem: doing 80% of the work takes 20% of the time, it's the remaining 20% of the work that's really hard (whatever the exact fraction, I hope you get the idea). We've always planned to mask out things that are likely human-made (aka long streaks), but it's a lot harder to catch all of them. This also puts requirements on how we share data (see the details in the tech note I linked above) both internally and externally.
There are also subtle details like what do you do about something that might be a satellite streak that goes off the edge of a detector? It could be a satellite, it could be an asteroid, it could be a very nearby asteroid moving very fast. Those latter are things we really want to catch (its one of our design goals), but there's not much to distinguish between a barely-streaked satellite and a barely-streaked asteroid.
I realized I'd never linked the video of the mirror coating from May!
View: https://www.youtube.com/watch?v=yKqEDFvmYwY
Even having worked in person at a few other telescopes, seeing just how dang SHINY the mirror becomes when the silver layer is applied is still amazing.
I was reminded of this from an internal video of the M1M3 mirror being lifted up the elevator (it's a very big elevator!), in preparation for being mounted soon. If that is publicly posted some where, I'll share it here: it's weird seeing such a large reflective surface driving around on the floor, and being lifted past the camera.
View: https://www.youtube.com/watch?v=yKqEDFvmYwY
Even having worked in person at a few other telescopes, seeing just how dang SHINY the mirror becomes when the silver layer is applied is still amazing.
I was reminded of this from an internal video of the M1M3 mirror being lifted up the elevator (it's a very big elevator!), in preparation for being mounted soon. If that is publicly posted some where, I'll share it here: it's weird seeing such a large reflective surface driving around on the floor, and being lifted past the camera.
Last edited:
Ah, here we are. It doesn't have cool music to go with it, but it shows the process of moving the mirror from the coating chamber (as shown in the video above), up the elevator, into the dome (via outside!) and under the telescope. Some really neat reflections of the various dome support structures as the mirror moves around.
https://rubin.canto.com/v/Construct...ew&column=video&id=ij2su39odt4uf1radrmatkcs6m
M1M3 (the primary and tertiary, made from a single piece of glass: our telescope has a rather unique design!) was mounted in early October. ComCam (the "only 144 megapixel" commissioning camera) was mounted in August, and M2 sometime earlier this summer, so we now have a complete telescope.
And, uh, ... :itshappening.gif:
https://rubin.canto.com/v/Construct...ew&column=video&id=ij2su39odt4uf1radrmatkcs6m
M1M3 (the primary and tertiary, made from a single piece of glass: our telescope has a rather unique design!) was mounted in early October. ComCam (the "only 144 megapixel" commissioning camera) was mounted in August, and M2 sometime earlier this summer, so we now have a complete telescope.
And, uh, ... :itshappening.gif:
Ah, good, I don't have to hint about it any more! Here's a picture of the open dome from the first night of commissioning, which was last Thursday, October 24th! Not a bad desktop background (but boy does our current gallery software make the download button hard to find).
https://rubin.canto.com/v/Construct...ew&column=image&id=27r8l2ub491f168jpq41p8rb4s
And the commissioning "group photo", one of the first images taken through the full optics with ComCam mounted. The system wasn't designed to be focused at this very short distance, but we have literal pinhole camera "filter" that gets "good enough" focus.
https://rubin.canto.com/v/Construct...ew&column=image&id=vf95idcg5944p8aftabvc7r12g
And a timelapse of the first night's imaging and telescope motion. The telescope will spend the next few weeks moving much slower than its design speed, starting out at less than 1% speed/acceleration/jerk, but we know the telescope motors work as designed: see my video from a year or so ago earlier in the thread. The primary goal of these early observations is to calibrate the active optics system (AOS) so that they can correctly balance and support the mirror as the telescope structure moves. As that is refined, telescope speed will be increased, slowly.
https://rubin.canto.com/v/Construct...ew&column=video&id=sl3bvpcpkp0jvfj794hq50m562
The on-sky images are still under embargo, but I suspect there will be some public announcements about them before too long.
https://rubin.canto.com/v/Construct...ew&column=image&id=27r8l2ub491f168jpq41p8rb4s
And the commissioning "group photo", one of the first images taken through the full optics with ComCam mounted. The system wasn't designed to be focused at this very short distance, but we have literal pinhole camera "filter" that gets "good enough" focus.
https://rubin.canto.com/v/Construct...ew&column=image&id=vf95idcg5944p8aftabvc7r12g
And a timelapse of the first night's imaging and telescope motion. The telescope will spend the next few weeks moving much slower than its design speed, starting out at less than 1% speed/acceleration/jerk, but we know the telescope motors work as designed: see my video from a year or so ago earlier in the thread. The primary goal of these early observations is to calibrate the active optics system (AOS) so that they can correctly balance and support the mirror as the telescope structure moves. As that is refined, telescope speed will be increased, slowly.
https://rubin.canto.com/v/Construct...ew&column=video&id=sl3bvpcpkp0jvfj794hq50m562
The on-sky images are still under embargo, but I suspect there will be some public announcements about them before too long.
As part of Pan-STARRS, I'm envious that they've taken the biggest camera title from us, but that's progress for ya. Their camera is monstrous compared to ours, and it's not like they're the smallest things in the world to begin with. An utterly mental piece of work.
From a personal point of view, it'll be interesting to see how their shutter mechanism holds up over time.
From a personal point of view, it'll be interesting to see how their shutter mechanism holds up over time.
In addition to stealing the biggest camera from Pan-STARRS, we're also stealing the "three letter agency meddling" title from you as well. I noted this some above, but there's been some more public reporting about it recently, with my UW colleague Zeljko talking more openly about the "negotiation" process that occurred a few years ago.
@Smeghead : if you worked on Pan-STARRS before, you can appreciate the "reading out the camera is like having a hair dryer running inside the -100ºC dewer" problem we have, of keeping a consistent temperature in the dewer and backend electronics. Pan-STARRS readout power must be quite a bit lower than ours, with 7s readout (vs. 2s) and three times fewer detectors. ComCam is only 9 detectors, so doesn't really have that problem. ComCam also has a simpler shutter mechanism than the full camera, so nothing to report on that yet.
@Smeghead : if you worked on Pan-STARRS before, you can appreciate the "reading out the camera is like having a hair dryer running inside the -100ºC dewer" problem we have, of keeping a consistent temperature in the dewer and backend electronics. Pan-STARRS readout power must be quite a bit lower than ours, with 7s readout (vs. 2s) and three times fewer detectors. ComCam is only 9 detectors, so doesn't really have that problem. ComCam also has a simpler shutter mechanism than the full camera, so nothing to report on that yet.
@parejkoj : I'd be interested in swapping commissioning war stories in a couple of years once LSST is properly up and running. 
It's interesting to see how LSST is taking a similar approach to what we did mumblemumble years ago, with a scaled-down version of the camera for first light and initial commissioning. We started with a quarter-size version that then ran as a test system in the lab for us to continue development on when the first gigapixel camera went to the summit.
Actually, if I'm truthful, technically the first instrument on the telescope was a tiny homemade thing with a single example of the 60 CCDs in it, but that was only used for a very short time and wasn't all that useful for testing out all the backend stuff.
Looking at the animation for the shutter for the full-size camera, I'm wondering how the multi-segment blades will hold up over time. Our shutters have a pair of single blades that have proved to be a maintenance item over the years, and LSST triples the number of moving parts.
One of the things that simplified thermals with our camera design was that the vast majority of the readout electronics are outside the cryostat and are cooled by ambient air, with a system to take heat out of the exhausts. Being external, we don't have to worry about outgassing, either. I forget what the combined power consumption is for the CCDs, but there's a decent bump in the power supply outputs when biases are applied, and clocking charge to read them out does result in another sizeable increase; sustained back-to-back readouts eventually overwhelm the lift capacity of the cooling system, but that's not any sort of realistic observing mode.
The other thing that's always freaked me out about our cameras is L3. Ours is a little smaller than LSST's (around 600mm diameter), but it's highly convex, and it freaks me out to think of the stresses involved when the camera is pumped down; there's just short of 3 tonnes of air pressure on the outside of it once we're at vacuum (at sea level), and it's just the wrong damn shape to be able to do that.
It's interesting to see how LSST is taking a similar approach to what we did mumblemumble years ago, with a scaled-down version of the camera for first light and initial commissioning. We started with a quarter-size version that then ran as a test system in the lab for us to continue development on when the first gigapixel camera went to the summit.
Actually, if I'm truthful, technically the first instrument on the telescope was a tiny homemade thing with a single example of the 60 CCDs in it, but that was only used for a very short time and wasn't all that useful for testing out all the backend stuff.
Looking at the animation for the shutter for the full-size camera, I'm wondering how the multi-segment blades will hold up over time. Our shutters have a pair of single blades that have proved to be a maintenance item over the years, and LSST triples the number of moving parts.
One of the things that simplified thermals with our camera design was that the vast majority of the readout electronics are outside the cryostat and are cooled by ambient air, with a system to take heat out of the exhausts. Being external, we don't have to worry about outgassing, either. I forget what the combined power consumption is for the CCDs, but there's a decent bump in the power supply outputs when biases are applied, and clocking charge to read them out does result in another sizeable increase; sustained back-to-back readouts eventually overwhelm the lift capacity of the cooling system, but that's not any sort of realistic observing mode.
The other thing that's always freaked me out about our cameras is L3. Ours is a little smaller than LSST's (around 600mm diameter), but it's highly convex, and it freaks me out to think of the stresses involved when the camera is pumped down; there's just short of 3 tonnes of air pressure on the outside of it once we're at vacuum (at sea level), and it's just the wrong damn shape to be able to do that.
Certainly feel free to use this thread for that so the rest of us can follow along
I can do that up to a point. There's probably some stuff that would be inappropriate to share in a public forum, but I can probably come up with some stuff.
Looking at my previous post about our L3, I never do get concave vs convex right the first time around. L3 is very concave on the atmo side, with the centre being something like 10-15 cm recessed than the outer edge. I know the glass is stupid thick, but it still gives me the heebie jeebies thinking about the pressure differential.
Looking at my previous post about our L3, I never do get concave vs convex right the first time around.
L3 is very concave on the atmo side, with the centre being something like 10-15 cm recessed than the outer edge. I know the glass is stupid thick, but it still gives me the heebie jeebies thinking about the pressure differential.
I always just remember that conCAVE is CAVED in.Looking at my previous post about our L3, I never do get concave vs convex right the first time around.
Looking at my previous post about our L3, I never do get concave vs convex right the first time around.
Ah, sure. Our L3 is quite thick, to support the vacuum dewer. Our camera electronics are also outside the vacuum dewer, but need their own separate glycol cooling system, to extract all that heat. Full frame readout is a bit over 1kW inside the dewer, plus a bit more on the backend electronics. That's a lot of heat to disperse! The combination of those cooling systems is one of the most complicated pieces of engineering in this whole complicated system.
We have the other problem regarding lenses: L1 (the outer lens) is over 1.5m across, and quite convex. You can see that convexity pretty well in this image, from when it was being boxed up to ship. The largest, most expensive, purest lens ever made, and it sticks out so much that if you're standing in front of it, you can't tell where it is! I never went into the clean room where it was being stored at SLAC, but I looked in on it once. There was a big metal bar about a foot out from the front of it as a "NONE SHALL PASS!" zone. I've heard from the engineers that they were scared to be in the room with it when L1 wasn't covered, for fear of bumping anything into it, because they just couldn't see it!
I can't imagine trying to draw away a kilowatt of heat through the cryocooling system.
Our cameras run a helium-based system from CTI, with a pair of cold heads on two sides of the camera (the electronics feed through from the other two). I couldn't tell you which heads we use, but looking at various datasheets, the lift capacity per head is in the region of 50-60W or so, with a sharp temperature increase for every 10W above that.
I remember the L3 arriving in the lab being a big deal for both cameras. Being concave (see, @Bardon, I'm learning!) they were always easy to cover with a metal plate for protection in the lab after being mated up to the camera. Our biggest worry during testing was always making sure we purged the space above the lens with dry nitrogen and then kept it dry, as the humidity here meant we'd get condensation very quickly if the camera was cold.
At the summit, if the dry air supply stops, we risk not only condensation but icing. Most days it's stupidly dry, but at "only" 10k feet, we do get clouds at the summit for a number of days a year.
Then there was the storm in 2019, where everything was covered in ice; L3 was the least of our worries:
(Photo from here, but were taken by park rangers and our guys from IfA Maui, so using them is probably fine)
The big bugger centre frame is DKIST, the solar observatory. The little guys next door are ours, PS1 and PS2 going left to right. The Air Force's space surveillance facility is on the right.
We were down for weeks after that. Once the conditions improved, it took a lot of work to restore power and network to the summit, as the ice had brought down a lot of the lines and some of them are only accessible via helicopter.
Speaking of thermal power, DKIST's primary mirror is something like 4.2m in diameter, and they take that and point it at the sun...
Our cameras run a helium-based system from CTI, with a pair of cold heads on two sides of the camera (the electronics feed through from the other two). I couldn't tell you which heads we use, but looking at various datasheets, the lift capacity per head is in the region of 50-60W or so, with a sharp temperature increase for every 10W above that.
I remember the L3 arriving in the lab being a big deal for both cameras. Being concave (see, @Bardon, I'm learning!
) they were always easy to cover with a metal plate for protection in the lab after being mated up to the camera. Our biggest worry during testing was always making sure we purged the space above the lens with dry nitrogen and then kept it dry, as the humidity here meant we'd get condensation very quickly if the camera was cold.At the summit, if the dry air supply stops, we risk not only condensation but icing. Most days it's stupidly dry, but at "only" 10k feet, we do get clouds at the summit for a number of days a year.
Then there was the storm in 2019, where everything was covered in ice; L3 was the least of our worries:
(Photo from here, but were taken by park rangers and our guys from IfA Maui, so using them is probably fine)
The big bugger centre frame is DKIST, the solar observatory. The little guys next door are ours, PS1 and PS2 going left to right. The Air Force's space surveillance facility is on the right.
We were down for weeks after that. Once the conditions improved, it took a lot of work to restore power and network to the summit, as the ice had brought down a lot of the lines and some of them are only accessible via helicopter.
Speaking of thermal power, DKIST's primary mirror is something like 4.2m in diameter, and they take that and point it at the sun...
I've done the sunrise bike down Haleakala a couple times. It's definitely cold up there before sunrise but I never knew there was snow, let alone that much
Haleakalā sees snow on occasion in winter, but it's not a given. The temperature right now is just passing 9C (still several hours away from the diurnal peak at around 2-3pm depending on the time of the year) and last night's low was around 5C. It's comparatively rare that the temperature drops below freezing for any length of time. Even at this time of year the weather can be really nice - I was up there to do some maintenance the other weekend. It was a glorious day with what seemed like endless visibility and it was pushing 17C.
The summit of Mauna Kea gets more snow more often, given they're 4000 feet higher up and their temperatures are that much lower.
2019 wasn't snow - that was freezing rain, which is exceedingly unusual. A major storm came through, there were gusts to 100mph (very rare) and the conditions were just right/wrong for supercooled water to start coating everything. I've personally only ever seen that once before when I lived in Chicago.
The summit of Mauna Kea gets more snow more often, given they're 4000 feet higher up and their temperatures are that much lower.
2019 wasn't snow - that was freezing rain, which is exceedingly unusual. A major storm came through, there were gusts to 100mph (very rare) and the conditions were just right/wrong for supercooled water to start coating everything. I've personally only ever seen that once before when I lived in Chicago.
It does make the trees look very pretty when the freezing rain coats all the branches
I've been lax in updates...
In January, we released a single image from ComCam, from the very first night of observing (so, before we'd optimized focus or anything else), placed in context of the full camera. There was a press release about it, but I think the image is the thing you're all most interested in:
https://noirlab.edu/public/images/ann25002a/zoomable/
Also, just in the past few days, the full camera was mounted on the telescope, ready to start commissioning it over the next few months! There are a bunch of timelapse videos of the installation that people might find interesting:
I can't find the exact number, but I think there's about 1cm of clearance around the camera as it is moved into place inside the secondary. That's some precision crane operation!
In January, we released a single image from ComCam, from the very first night of observing (so, before we'd optimized focus or anything else), placed in context of the full camera. There was a press release about it, but I think the image is the thing you're all most interested in:
https://noirlab.edu/public/images/ann25002a/zoomable/
Also, just in the past few days, the full camera was mounted on the telescope, ready to start commissioning it over the next few months! There are a bunch of timelapse videos of the installation that people might find interesting:
- Summary timelapse of all steps: https://noirlab.edu/public/videos/noirlab2511a/
- Timelapse from outside the telescope, looking down: https://noirlab.edu/public/videos/noirlab2511b/
- Timelapse from inside the telescope itself (you can see the primary and secondary mirrors as the very shiny things on the right and left sides of the video): https://noirlab.edu/public/videos/noirlab2511d/
I can't find the exact number, but I think there's about 1cm of clearance around the camera as it is moved into place inside the secondary. That's some precision crane operation!