22
$\begingroup$

On March 16, 2022 we were treated to the first image from the JWST of a star, where all 18 segments were aligned and in common focus. What a wonderful photo:

Telescope alignment evaluation image

(Source: Webb Twitter. Full 18 MP resolution at NASA web page)

I am curious why this central star has six very obvious “rays” in the image. An offhand comment I read (or heard) somewhere said this is because of the hexagonal shape of the JWST’s primary mirror.

This must refer to the outer perimeter of the primary mirror, I reckon.

So,

  1. If the perimeter of the primary mirror was circular, would these artifact rays not be present?

  2. How are these artifacts removed?

It occurs to me that every light source in the image must have these artifacts in some form (not just at the center of the image), so these quirky hexagonal light rays (including those from IR sources, of course) must be (potentially) anywhere (everywhere?) in the image.

Is there some magnificent optics software involved in the processing of the raw image data, removing these artifacts, even including where they overlap other (non-artifact) light sources (and leaving the non-artifactual “good” data)?

More generally, is there a good (general) write-up on the image processing pipeline for data from the JWST?

Improve this question
$\endgroup$
6
  • 1
    $\begingroup$ It is indeed a beautiful image. Just FYI, the image here in the focal plane is the 2D Fourier transform of the primary mirror. As asdfex points out the central star is heavily overexposed, so that the diffraction pattern stands out. The pattern is primarily hexagonal corresponding to the hexagonal symmetry of the overall shape, the narrow gaps between segments, and the 'missing' segment in the middle. But there is also one major feature (horizontal bar) not on the hexagonal pattern and that presumably is due to the top support arm holding the secondary mirror. $\endgroup$
    – Roger Wood
    Mar 17 at 18:12
  • 2
    $\begingroup$ @RogerWood Yes, that could be the top arm. The other two seem to align with the hex shape. But could also be some other optical element closer to the instruments. $\endgroup$
    – asdfex
    Mar 17 at 19:20
  • 3
    $\begingroup$ See also astronomy.stackexchange.com/q/48836/15228, which goes into a deeper write-up. $\endgroup$
    – Bear
    Mar 18 at 15:49
  • 1
    $\begingroup$ Keep in mind that the target star used for this alignment is too bright to be a useful science target. Those big bright spikes are far, far fainter than the star itself, which is overwhelmingly oversaturated and obscured by bloom. Look at the other objects in the image: there's only a couple instances where the spikes are even visible. JWST's not meant to look at stars like this, it's just a convenient test case for mirror alignment. $\endgroup$
    – Christopher James Huff
    Mar 18 at 19:15
  • $\begingroup$ Maybe part of this question should be: if the diffraction spikes can be removed, why weren't they removed in this image? $\endgroup$
    – Wyck
    Mar 24 at 13:55

2 Answers 2

Reset to default
23
$\begingroup$

These star-like structures are caused by diffraction of light at any structure inside the optical path and are called diffraction spikes. In case of JWST it is caused by, as you say, the borders between mirror segments and the outer shape of the mirror. You can experience the same effect with any camera that has mechanical aperture blades (search for "aperture stars")

If the perimeter of the primary mirror was circular, would these artifact rays not be present?

Yes and no. The diffraction would be equal in every direction and therefore a lot less visible. But, just having circular mirrors doesn't help - there would still be some gaps between mirrors that have a pointy, star-like shape causing similar effects.

It occurs to me that every light source in the image must have these artifacts in some form, so these quirky hexagonal light rays must be (everywhere?) in the image.

Yes, they are there. You can see a couple more objects that show the same artifacts.

How are these artifacts removed?

As we know the exact optical structure of the telescope, we can derive a deconvolution function that describes how light from a point-like source gets diffracted that can be used to revert parts of the effect.

But, this is not necessary as much as it seems. The main reason for this effect in todays' image is that the central star is grossly overexposed. Remember, it's a star and should be just 1 pixel in size if exposed (and focused) correctly. That's also the reason why only a few objects in the image show the artifacts: It's a very faint pattern only visible for very bright objects. All the objects that got correctly exposed (i.e. those which don't have a white core) don't show it.

Improve this answer
$\endgroup$
2
  • $\begingroup$ Re "These stars...": Don't you mean "These spikes..."? $\endgroup$
    – Peter Mortensen
    Mar 20 at 23:04
  • $\begingroup$ @PeterMortensen We need more different words. That was meant to be stars like the geometrical structure. $\endgroup$
    – asdfex
    Mar 21 at 9:23
15
$\begingroup$

This is just an addendum to @asdfex' excellent answer - and for my own curiosity. This shows the FFT of the aperture 'selfie' from NASA's website. The FFT does show the horizontal strip arising from the shadow of the vertical strut. But there are other differences that I can't explain. Perhaps some arise from the relatively poor resolution of the 'selfie' image that I worked with and perhaps some arise from the aliasing (downsampling) in creating this image.

JWST aperture and image

Improve this answer
$\endgroup$
6
  • $\begingroup$ The selfie is the obvious choice for that.. I was looking for something yesterday but couldn't find anything nice. Which other differences do you mean? The dotted structure? You can see this in the stars around the less overexposed objects below and to the left of the center. $\endgroup$
    – asdfex
    Mar 18 at 19:19
  • $\begingroup$ @asdfex the hexagonal pattern is much more dominant in the original NASA image. I realise my image is abs(fft) and it should be abs(fft)^2 power. Maybe that will help. I can have another look this afternoon. $\endgroup$
    – Roger Wood
    Mar 18 at 19:45
  • 1
    $\begingroup$ @asdfex I see there's a very nice answer by uhoh on astronomy stack-exchange to a similar question. astronomy.stackexchange.com/questions/48836/… It has a similar 'speckle' pattern. $\endgroup$
    – Roger Wood
    Mar 18 at 22:44
  • 2
    $\begingroup$ The speckle pattern is because the Fourier Transform is taken at one particular time, getting one particular phase. In the actual image, the phase of the source is oscillating and this would cause the speckles to vary over the period of the light wave (the reciprocal of its frequency). The various phases of the speckling are averaged in an image, whose exposure is typically much longer than this period, giving the non-speckled spikes. $\endgroup$
    – robjohn
    Mar 19 at 20:38
  • 2
    $\begingroup$ I should add that there will also be standing wave artifacts, where interference cancels the light, as in the rings of an Airy disk. $\endgroup$
    – robjohn
    Mar 19 at 21:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.