Suppose that you have a camera mounted on a telescope at a remote site; you can command it to perform tasks -- slew to the Moon, take a 5-second image, and so forth, and you can examine the resulting images ... but you can't sit in the dome itself.

After using the telescope for a few months, you become suspicious of the times written into the FITS headers of the images. You suspect that the time written into the header is, say, delayed by 10 seconds. In other words, the time written into the header is always 10 seconds after the shutter actually opened.

How would you test this notion? In other words, what sort of observation would you make to verify the time recorded in the FITS header? Even a rough estimate, good to 1 or 2 seconds, would be fine. Remember, you can't go into the dome itself -- all you can do is command the telescope and camera.

One idea might be to take pictures of near-earth objects, since their motion is very rapid. But how many have orbits known well enough to provide times good to 1 second? And how frequently do such good candidates pass by the Earth?

Mutual events of Jupiter's satellites might work, but they won't occur this year.

An occultation of a star by an asteroid might work, but how often will a good one occur for a fixed location on Earth?

Any more ideas?

And of COURSE this is a purely hypothetical question :-)

Though it won't likely give you 10s precision, Astrometry.net might help.

Otherwise, the answer depends on the resolution of the camera, depth of the images, and how large the field of view is. Certainly an image of the edge of the moon with a starfield to V~17 in the background aught to provide enough astrometric information to pin down the time the image was taken. Alternately, an image of Jupiter's moons should do it as well: their ephemeridies are very well determined (check out Xephem, where you can get the orbital positions to very high accuracy).

What happened to make this purely hypothetical situation come to be?

__________________
"What do you care what other people think?" -- Richard Feynman
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Feynman, at the conclusion of his Challenger report

The camera has a field of view of about 40 arcminutes, with a plate scale of about 1 arcsec per pixel. A one-second exposure through a wide-band filter probably gives high S/N down to 10th mag or so.

The moon is really bright, so that taking pictures of it and stars of even 6th mag is very hard.

I've looked into that. It's not as easy as one might think. First, the moons don't move that fast -- Io, for example, will at is fastest take several minutes to move one arcsec. That means that it sits at effectively the same place for many, many seconds ... which makes it hard to pin down the exact time of a picture to the second. In addition, the _absolute_ positions of the satellites aren't known all that well (or perhaps they are just hard to find), at the precision of one arcsecond.

Well, if you operate a remote telescope with a camera which someone else put together a while ago, and you've never worked with it in person, how would you KNOW exactly how the FITS header time is determined? And if you started a project which did require time at the level of one second, how would you check it?

Thanks for the ideas. Any others?

If you are concerned that the camera's clock is not reliable, why not just have a master clock displaying official time signals at the observing site, and compare them side by side over your data link. Even if the latter has an unknown time delay, it would affect your view of both devices equally.

By "near Earth" objects, are you including man made satellites? Though I'm not sure how hard it would be to get second or two accuracy with that camera.

__________________
I say there is an invisible elf in my backyard. How do you prove that I am wrong?

Disclaimer: Avatar is not an official NASA image and does not imply any specific interplanetary or interstellar capability.

The Leif Ericson Cruiser

Alas, in this remote operation, the camera is attached to a telescope. Unless someone places a master clock on a nearby mountaintop, I can't take an image showing the time displayed on a clock with the camera. The telescope won't focus on objects within the dome.

Aha. Yes, I had not been considering man-made satellites, but there are plenty of those, and they move fast, and some are very bright .... Hmm. I don't know off the top of my head just how well the orbits are known, but that is something I can check out.

Thanks for the good idea.

You're welcome. Two thoughts . . . You might want to check out Iridium flares, and you might want to ask about this in the Astronomical Observing or Astrophotography sections, where I think there are some people there that do a fair bit of satellite watching and photography, and might have some good pointers.

__________________
I say there is an invisible elf in my backyard. How do you prove that I am wrong?

Disclaimer: Avatar is not an official NASA image and does not imply any specific interplanetary or interstellar capability.

The Leif Ericson Cruiser

Thinking about this - could someone get somewhere to shine a light that the telescope could see? And can you do a series of one second exposures? For that, you wouldn't need a focus: Just turn on a light at a known time for no more than a second, based on a master clock, and take a series of one second images within a 20 or 30 second range, and look for the lit up one.

__________________
I say there is an invisible elf in my backyard. How do you prove that I am wrong?

Disclaimer: Avatar is not an official NASA image and does not imply any specific interplanetary or interstellar capability.

The Leif Ericson Cruiser

These are good ideas. One factor that makes them, well, a bit less easy, is that the readout time for the chip is around 30 seconds. I don't think we can change the readout properties without also changing the possible delay between exposure and FITS header time, which is the whole point of the exercise.

Stellar near-occultations at the Earthlit limb? The moon moves close to 1 arcsecond/second of time. Using a 0.4m on campus, I can expose 10 seconds or more in broadband without saturating either Earthshine or scattered light from the bright limb, and get stars to 10th magnitude or fainter. If you had an upcoming Pleiades occultation, that would do it (as long as this time offset is constant or had a relation to exposure tie that you can work out independently). (Every time I have to take an excursion into rapidly moving or changing targets I remember why I went into work on galaxies).

What is your conceptual problem in this hypothetical mission?

If you have the two-way telecommunication link needed to command the telescope and receive the images, it should be child's play by comparison to monitor an official clock alongside the telescope. Your scenario looks like seeking a solution which is in search of a problem.

You really have to make an observation with the telescope? You can't modify the camera? Why not use a trusted clock (GPS maybe) to time the pulses in the circuit that trigger the shutter? You might even be able to measure how long the shutter is really open that way, if there is a pulse on both sides. I don't think that's a big deal to do for a competent technician, and it sounds like the kind of modification that is useful enough to justify.

__________________
Don't try this at home - We're what you call "professionals" - MythBusters.

I replied to an earlier question of yours by saying that the camera is attached to a telescope, which does not focus on objects within the dome. Did you miss that?

For the hypothetical situation I posed, neither I nor anyone else can go into the dome and make any modification. The telescope is assumed to be so remote that all one can do is order it to take pictures at some given time, and analyze the result.

For the slightly less hypothetical real-life situation, yes, your suggestion is a good one. A technician who can take the camera off the telescope and just look at the darn thing as the shutter opens and closes, and write down the times of opening and closing on a piece of paper, would solve the whole problem.

Unfortunately, I'm not going to the telescope, and it's not clear that a technician authorized to remove the camera from the telescope will have the time to make the simple test ....

Sounds like it's time for a road trip!

But, if you don't have it already, try xephem: it can download ephemerides for satellites and asteroids, so you can see what objects might be up at a given time. Though I do wonder whether they are precise enough for what you need. Alternately, send me the lat/long of the observatory and the time of an observation that you're going to do, and I'll see what I can find.

Your plate scale is 1"/px, but do you know how good the astrometry is?

Did you try astrometry.net, just to see what it produces?

__________________
"What do you care what other people think?" -- Richard Feynman
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." -- Feynman, at the conclusion of his Challenger report

Now I understand the hypothetical predicament better. You need immediate error correction with no immediate means of upgrading your system. Yes, you need an event whose time is well determined ahead of time.

One of my colleagues is going out to the telescope someone in the next few months. I'm trying to find all the possible methods he might use when he gets there, or that we might use before and after.

Yes, depending on the location of the object, it's good to around 0.2 arcseconds.

Take a picture using a tiny field of view at some RA and Dec, compare the position of stars on it at the time in the fits file to what it should be using a computer program. If the wrong time is written, the stars in the picture will have shifted a little due to the earth's rotation.