Wikipedia:Reference desk/Archives/Science/2022 July 12

= July 12 =

Diffraction spikes in the James Webb Space Telescope images
Either in the JWST’s first "promo" image (File:Webb's_First_Deep_Field.jpg) or in File:JWST Telescope alignment evaluation image labeled.jpg, one can see a strong hexagonal diffraction spike pattern around the brightest star.

I would assume that this is not very good for any serious image analysis (even if it looks pretty for the photo-op) and that this was not exactly a surprise to the telescope designers. Are there any countermeasures in place for when the scientific work begins, and if yes which? Tigraan Click here for my talk page ("private" contact) 09:14, 12 July 2022 (UTC)
 * This is clearly an important step in the image analysis, and there are methods of various levels of complexity of how to deal with the spikes and more generally the point spread function (PSF). The spikes from the brightest stars may well be beyond repair and the easiest way to deal with them is to mask them out, discarding all parts of the image covered by spikes and the stars from which they originate. There are, however, nice galaxies that sit right on top (or under?) a spike and it would be a shame to just throw them away. A heuristic model of the spike in that area may help in that case. More important is the question of how to deal with the PSF in general. The PSF is the image that a point source (i.e. a star) creates on the image. However, not only stars but all objects in the field are affected by the PSF. Some galaxies show distinct diffraction spikes, which are typically less sharp than those from stars. These originate from bright central regions of the galaxies that are extended yet compact enough for diffraction spikes to be discernible. Even if the spikes cannot be made out as such, the image of a galaxy is affected by the PSF. Mathematically, this is a convolution of the actual intrinsic image with the PSF. As long as the PSF is predictable (or can be measured by looking at the stars in the field), this is fairly simple to account for. In forward modelling, you build a model of the galaxy's intrinsic shape, then convolve that with the PSF and compare it to the image. Mathematicians usually think of deconvolution, but this is often problematic due to the noise in the image and artefacts that may be introduced through the algorithm. --Wrongfilter (talk) 10:17, 12 July 2022 (UTC)


 * Diffraction spikes happen in every telescope (except those with perfect cylindrical symmetry, like most refractors or Schmidt–Cassegrains). It's an unavoidable consequence of the wave-like nature of light. Fortunately, the amount of light in the diffraction spikes is very small compared to that in the core of the object around which you see the spikes. In the deep field image, there are about 50 objects with clearly visible diffraction spikes and I expect all of them have, at their centre, well saturated the CCD. Those are useless. One can attempt to subtract the diffraction spikes from the image, but the noise in the diffraction spikes will remain, which is bad news for the objects behind a diffraction spike of one of these bright objects (which must be foreground stars of bright quasars). Fortunately, there must still be (tens of) thousands of objects in this image not affected by these spikes. PiusImpavidus (talk) 10:44, 12 July 2022 (UTC)


 * If the field is important enough, the telescope could be rotated and another image taken. Saturated pixels will make it so that you cannot use an inverse PSF to remove the spike completely. If there was 32 bits of sampling, no noise or saturation, then perhaps they could be taken out. Graeme Bartlett (talk) 11:12, 12 July 2022 (UTC)


 * There's always noise, if only Poisson noise. Suppose that within the area of a distant galaxy you detect 100 photons from that galaxy. That's your signal. Then your signal-to-noise ratio is a useful 10. Now add a diffraction spike with another 300 photons in the same area. This contains $$\sqrt{300}$$ of noise. You can subtract the diffraction spike, if you can model it, but the noise in the diffraction spike will still be there. Your signal-to-noise ratio goes down to a marginal 5. PiusImpavidus (talk) 23:03, 12 July 2022 (UTC)
 * My point was that you will not be able to do it, due to the assumptions not being true. I write from personal experience in computationally removing image problems by using the inverse PSF. Without noise, great results are possible. But there is noise, so it is not so good. Graeme Bartlett (talk) 23:45, 14 July 2022 (UTC)

Where the building blocks/nutrients come from in plants (and all living things)
Hello, I was just curious, I had heard most of the mass of trees comes from the air. In that case, what is the branch of science I would look up to know exactly how trees, plants, fruit, etc. get their building blocks and gain their specific nutrients, etc. Say the vitamin c from an orange or the protein in beans, or lycopene in tomatoes. Thanks kindly. Stocktrain (talk) 21:04, 12 July 2022 (UTC)
 * You can read Plant physiology, Phytochemistry, Plant nutrition. Graeme Bartlett (talk) 22:00, 12 July 2022 (UTC)
 * Does anyone know the answer in a nutshell? Thanks! Stocktrain (talk) 18:00, 14 July 2022 (UTC)
 * Wikipedia might. Try reading the linked articles. --←Baseball Bugs What's up, Doc? carrots→ 21:22, 14 July 2022 (UTC)