5 ms·

How long are the exposures? One would think that due to vibration etc. the exposures need to be extremely short, but long exposures are beneficial in astrophoto

by nawitus 12y ago

How long are the exposures? One would think that due to vibration etc. the exposures need to be extremely short, but long exposures are beneficial in astrophotography. EDIT: Wikipedia answers the vibration issues: http://en.wikipedia.org/wiki/Stratospheric_Observatory_for_Infrared_Astronomy#The_telescope http://en.wikipedia.org/wiki/Stratospheric_Observatory_for_I...

greglindahl 12y ago
It's normal to take many short exposures and stack them up to get a long one. The length is dictated by the amount of noise in your detector. From a look at the Wikipedia article, it looks like they need a lot more than just short exposures on Sofia.
- nawitus 12y ago
  Yes, you can stack short exposures, but long exposures are better in astrophotography (as a single 10 second exposure has a higher signal-to-noise ratio than 10x 1s exposures). If you take it to the extreme, it's very difficult to differentiate between the signal and the noise if your exposures are really short, which makes astrophotography in cities without a tracking quite mount difficult.
  - goodcanadian 12y ago
    SOFIA, like all major observatories, uses tracking. A guider camera tracks on the target itself if it is bright enough, or it does offset tracking on another star in the field. You don't need to have any visible signal from your target at all in order to stack the images. Note also, that not all instruments and observing modes are imaging. Often, spectroscopy is done instead.
    - goodcanadian 12y ago
      Maybe, I shouldn't reply to myself, but I feel like I have oversimplified. First, not all observatories use guiding cameras, but most do. An obvious exception are radio telescopes which do passive tracking, but will check their pointing on a bright source periodically. Second, the sky is very bright and variable in the infrared, so the dominant source of noise is from the sky. For this reason, you take an image on source followed shortly by an image off source to use to subtract the background sky from the on-source image. As the infrared sky can vary quite quickly, you want to sample the background quite often. If there is a "blank" patch of sky near enough to your target, this is accomplished by "chopping" the secondary mirror back and forth between the source and the off position. This occurs at frequencies on the order of 1Hz, so exposures are less than a second. If you have to move further to find "blank" sky, it is accomplished by "nodding" the telescope, i.e. moving the primary to the off position. This takes a bit longer and may only be done every ~30s. Because there is active guiding, there is no issue co-adding an arbitrarily large number of these images, and observations can actually be several hours long. And, because the background sky varies fairly rapidly, stacking short exposures (with the background subtracted) will actually give better noise performance than taking long exposures.
  - greglindahl 12y ago
    That's what I meant by "The length is dictated by the amount of noise in your detector." More noise, and you want longer exposures. Less noise, you'll want short exposures. It's easy to compute in advance, if you've measured the noise in your detector.
  - mikeash 12y ago
    Can you explain the cause of the difference between one long exposure and stacking short exposures? I'd naively expect one 10s exposure to produce equivalent data to 10x1s exposures.
    - goodcanadian 12y ago
      What I suspect nawitus is referring to involves the read noise of a CCD. There is always some low level noise involved in reading out a CCD. If this noise dominates, which may be the case for amateur astrophotography, it can explain the noise difference between a single exposure and stacking an equivalent number of shorter exposures. If for the sake of illustration, the read noise is 10 counts and the signal for a 1 second observation is 10 counts, you will have a signal to noise of 1. Stacking 10 of these will give you a signal to noise of ~3 (square root of 10). Conversely, for the 10 second observation, you will have 100 counts for a signal to noise of 10. However, there are other sources of noise, and instruments at major observatories are generally designed such that the read noise is not the dominant source of error either through more sensitive detectors or gathering more light (a bigger telescope or larger pixels) or some other means. If even a short exposure constitutes 1000000 counts, the difference between one 10s exposure and ten 1s exposures will be negligible.
      - mikeash 12y ago
        Exactly the sort of thing I was looking for. Thanks for the explanation, it makes perfect sense. I hadn't realized that there would be sources of noise that hit you for each read.
astrobokonon 12y ago
Vibration isn't the reason why exposures tend to be short, and at the longer wavelengths SOFIA operates at it's not the dominant source of the image quality. The real reason is that at the infrared wavelengths seen on SOFIA the sky that we're looking through is very bright, and can quickly saturate detectors. If you're interested in the nitty-gritty (with pictures!) there's a PDF that describes one of the methods: http://www.sofia.usra.edu/Science/instruments/FORCAST_observations_modes.pdf http://www.sofia.usra.edu/Science/instruments/FORCAST_observ... Edit: More nitty-gritty here: https://www.sofia.usra.edu/Science/ObserversHandbook/FORCAST.html#ImgPlanning https://www.sofia.usra.edu/Science/ObserversHandbook/FORCAST...
goodcanadian 12y ago
There have been a couple of good answers with regard to exposure time. With regard to vibration, the telescope is floated on an oil bearing and gyro stabilized. It is well isolated from aircraft vibrations.
- rbanffy 12y ago
  I imagine some interesting ideas of cabin noise management could be derived from research around this...

by nawitus 12y ago

greglindahl 12y ago

It's normal to take many short exposures and stack them up to get a long one. The length is dictated by the amount of noise in your detector. From a look at the Wikipedia article, it looks like they need a lot more than just short exposures on Sofia.

nawitus 12y ago
Yes, you can stack short exposures, but long exposures are better in astrophotography (as a single 10 second exposure has a higher signal-to-noise ratio than 10x 1s exposures). If you take it to the extreme, it's very difficult to differentiate between the signal and the noise if your exposures are really short, which makes astrophotography in cities without a tracking quite mount difficult.
- goodcanadian 12y ago
  SOFIA, like all major observatories, uses tracking. A guider camera tracks on the target itself if it is bright enough, or it does offset tracking on another star in the field. You don't need to have any visible signal from your target at all in order to stack the images. Note also, that not all instruments and observing modes are imaging. Often, spectroscopy is done instead.
  - goodcanadian 12y ago
    Maybe, I shouldn't reply to myself, but I feel like I have oversimplified. First, not all observatories use guiding cameras, but most do. An obvious exception are radio telescopes which do passive tracking, but will check their pointing on a bright source periodically. Second, the sky is very bright and variable in the infrared, so the dominant source of noise is from the sky. For this reason, you take an image on source followed shortly by an image off source to use to subtract the background sky from the on-source image. As the infrared sky can vary quite quickly, you want to sample the background quite often. If there is a "blank" patch of sky near enough to your target, this is accomplished by "chopping" the secondary mirror back and forth between the source and the off position. This occurs at frequencies on the order of 1Hz, so exposures are less than a second. If you have to move further to find "blank" sky, it is accomplished by "nodding" the telescope, i.e. moving the primary to the off position. This takes a bit longer and may only be done every ~30s. Because there is active guiding, there is no issue co-adding an arbitrarily large number of these images, and observations can actually be several hours long. And, because the background sky varies fairly rapidly, stacking short exposures (with the background subtracted) will actually give better noise performance than taking long exposures.
- greglindahl 12y ago
  That's what I meant by "The length is dictated by the amount of noise in your detector." More noise, and you want longer exposures. Less noise, you'll want short exposures. It's easy to compute in advance, if you've measured the noise in your detector.
- mikeash 12y ago
  Can you explain the cause of the difference between one long exposure and stacking short exposures? I'd naively expect one 10s exposure to produce equivalent data to 10x1s exposures.
  - goodcanadian 12y ago
    What I suspect nawitus is referring to involves the read noise of a CCD. There is always some low level noise involved in reading out a CCD. If this noise dominates, which may be the case for amateur astrophotography, it can explain the noise difference between a single exposure and stacking an equivalent number of shorter exposures. If for the sake of illustration, the read noise is 10 counts and the signal for a 1 second observation is 10 counts, you will have a signal to noise of 1. Stacking 10 of these will give you a signal to noise of ~3 (square root of 10). Conversely, for the 10 second observation, you will have 100 counts for a signal to noise of 10. However, there are other sources of noise, and instruments at major observatories are generally designed such that the read noise is not the dominant source of error either through more sensitive detectors or gathering more light (a bigger telescope or larger pixels) or some other means. If even a short exposure constitutes 1000000 counts, the difference between one 10s exposure and ten 1s exposures will be negligible.
    - mikeash 12y ago
      Exactly the sort of thing I was looking for. Thanks for the explanation, it makes perfect sense. I hadn't realized that there would be sources of noise that hit you for each read.

astrobokonon 12y ago

Vibration isn't the reason why exposures tend to be short, and at the longer wavelengths SOFIA operates at it's not the dominant source of the image quality. The real reason is that at the infrared wavelengths seen on SOFIA the sky that we're looking through is very bright, and can quickly saturate detectors. If you're interested in the nitty-gritty (with pictures!) there's a PDF that describes one of the methods: http://www.sofia.usra.edu/Science/instruments/FORCAST_observations_modes.pdf http://www.sofia.usra.edu/Science/instruments/FORCAST_observ... Edit: More nitty-gritty here: https://www.sofia.usra.edu/Science/ObserversHandbook/FORCAST.html#ImgPlanning https://www.sofia.usra.edu/Science/ObserversHandbook/FORCAST...

goodcanadian 12y ago

There have been a couple of good answers with regard to exposure time. With regard to vibration, the telescope is floated on an oil bearing and gyro stabilized. It is well isolated from aircraft vibrations.

rbanffy 12y ago
I imagine some interesting ideas of cabin noise management could be derived from research around this...