Saturday 26th July 2008

[NGC3314]

Still haven't found out, why SDSS uses green..

Well, that one I know. Many users of the SDSS data are mostly interested in galaxies. Most galaxy spectra and the sensitivity of the SDSS camera chips and filters mean that galaxies are best detected in the central three bands (gri) of the five SDSS filter bands. So this makes those three a natural choice for color renditions. And for most stars and galaxies, the spectral shapes are systematic enough that the colors aren't to different from what the eye would see (if our brightness sensitivity were a lot better). The oddities come in when we see things with strong emission lines. H-alpha falls in the r filter - it really does appear red. Likewise [O III] appears distinctly green, but is mapped to blue in the standard SDSS products.

(SergeNL - welkom aan onze hortus galacticus!)

[SergeNL]

Still haven't found out, why SDSS uses green..

Well, that one I know. Many users of the SDSS data are mostly interested in galaxies. Most galaxy spectra and the sensitivity of the SDSS camera chips and filters mean that galaxies are best detected in the central three bands (gri) of the five SDSS filter bands. So this makes those three a natural choice for color renditions. And for most stars and galaxies, the spectral shapes are systematic enough that the colors aren't to different from what the eye would see (if our brightness sensitivity were a lot better). The oddities come in when we see things with strong emission lines. H-alpha falls in the r filter - it really does appear red. Likewise [O III] appears distinctly green, but is mapped to blue in the standard SDSS products.

(SergeNL - welkom aan onze hortus galacticus!)

Hi NGC3314!
Thanks for the answer!

A small misunderstanding here about the use of the colour green. The point was not the use of green in average, but the use of green instead of red in the picture of M82. That point was not so clear in my text, but I responded to Nereid, who specifically asked: “Ey, that green stuff (the dust cloud in M82), isn’t that supposed to be red?”. He even added a picture, were the stuff indeed was red, while in the SDSS-image it was green.

So I replied, I had seen many of these SDSS-pictures, where the “stuff” was green, while on other pictures it was red. Then after a while I said, I still didn’t know why green was used (meaning instead of red.) So I didn’t mean the use of green in average.

I have two pictures of M82 in older books, but the cloud is nowhere to be seen. So perhaps it has invisible radiation (probably infrared), being now colour-mapped to a random visible colour, and added to the new pictures. To choose red would be better from the temperature point of view, but then you couldn’t distinguish between real red and made-up red. So to choose green would be better from a scientific point of view, then I hardly see any green in galaxies, so the made-up green doesn’t interfere with anything.
So perháps the dust colour is made up, and perhaps this is the reason they choose green.. I hope someone knows the answer..


Are the colours of galaxies on pictures about the same, what you would see with the naked eye, if nearby?
You say, you think so, and explain the use of filters. Of course you are partly right, but the difference can be big. Look at the two pictures above. The SDSS-image looks merely yellowish while the other one is really light blue. The older pictures I have: one is only yellow with dark clouds, and the other is light blue, with a yellow dot in the middle that I cant see on the light blue picture above.
So between the 4 different pictures there are already big differences. And in different lightening times all things may change.

So my question is, what do they do, to get a real nice picture? You can combine the red green and blue picture in different ways. And before that you can photoshop them, to accentuate details (why shouldn’t they?). So I am not so sure, you get the real (coulored) thing.

‘Hortus Galacticus’ was really nice found..   And I thought you were Dutch. In fact I thought aaahh.. half of the ‘bunch’ here can be Dutch. But I sort of took images of your remark in red green and blue, combined them again, finding out you used the words “Welkom Aan..”. That awoke the Sherlock Holmes in me.. Dutch would rather say “Welkom bij” (Welcome at..) So if you are not Dutch, it was very nice to welcome in Dutch! But in both cases: Thank you for the welcome..

[Nereid]

The relationship between 'what's in the SDSS/Subaru/HST/etc' image and 'what you'd see if you could view the object through an eyepiece at the 'end' of the SDSS/Subaru/HST/etc telescope' is a topic of much, much discussion, debate, confusion, and more ... 

But first, the objective facts (ha!).  If you obtain a spectrum of an object, paying very close attention to all sorts of subtle and not so subtle problems, you can plot (graph) the flux (energy per unit time per unit wavelength/frequency) against wavelength/frequency.

If that spectrum is 'dominated' by things called 'emission lines' (e.g. [OIII] (there are several), Hα), then what the object looks like in an image on a website and what you'd see with your own eye are very, very difficult to relate, in detail (IIRC, there is a lengthy Sky&Telescope article on this, based on some actual experiments, several years ago).

Part of the problem is that the emission lines occur at only one or two wavelengths (the widths are, usually, quite small), AND the response of humans' eyes to light of such narrow lines is by no means standard/fixed/unvarying!  Not only is there a well-known 'dark adaptation' physiological effect (crudely, as things get fainter, they lose colour), but humans have two, quite distinct, red cones (in the sense that the spectral response curves are not the same).  Men can only 'see' one kind of red (because the gene which codes for the cones is on the X chromosome), and so do many women (they have the same red cone gene on both X chromosomes).  But some women have both, and both are expressed ...

Put this together and what do you get?  Well, if you can get enough light from an object that dark adaptation is not an issue, and if the only emission line is Hα, and if the redshift is small, everyone will (colour blindness aside) agree that it is 'red' (though they may disagree on just what 'shade of red' it is).  Ditto if the only emission lines are [OIII] - 'green'. And so on.

Well, you can imagine the rest ... if the object is close to borderline 'colourless', if there are several emission lines, if the continuum is not weak, if ...

re M82: the Hα clouds are very real and very prominent ... provided whatever is doing the imaging is sufficiently sensitive to the Hα line!  In days past, the photographic plates that astronomers used to take images of galaxies were not (always) sensitive enough in the red part of the spectrum to show things like M82's Hα clouds.

[SergeNL]

..

All very informative!!

But I am still puzzled by the H-cloud, being green in one picture, and red in the other.

Being a blunt amateur I could think this: H has at least 2 emission lines, in red and in blue. The red seems dominant, then I have seen many pictures in red. But the 'doppler effect' in galaxy light may shift the Red emission line to infrared, and the blue one to green.. Eureka: H far away = green. And nearby = red. But talking about the same object, this is stupid.
But perhaps - if the red indeed shifts to infrared - they do map the colour to a random visible colour..
But it is you, that brought up the question.. grinn.. any idea of the answer?

Thanks for all the other info, Nereid!   Big fun, what "seen" exactly is? grinn.. Is there anything, that is NOT a problem?

But the point here was not, wat could be seen at the end of the telescope, and the difference with pictures, but the difference between these pictures, and what we would see, if we actually woud 'go' there, and watch the galaxy from nearby? My guess was, the difference is big anyway. To me that would be a reason not to worry too much about the pictures being 'manipulated' to get more details, or more beauty..

PS: Does H have 2 emission lines? I tried to do my homework with google, but I am not sure. This one here is nice about emission lines, but in Dutch (H = Waterstof):
http://virtueelpracticumlokaal.nl/spectrum_nl/spectrum_nl.htm

[Nereid]

Oh, the 'green in SDSS but red in Subaru' is easy to explain, as NGC3314 already did!

The SDSS images we are using map 3 (of 5) filters onto RGB (i -> R, r -> G, g -> B).  That means that Hα (restframe 656.3 nm; we see it as red) gets mapped to the G, and appears green in SDSS images.  You needn't worry about an object being redshifted so much that its emission lines move into a different filter, at least not for something as extended as M82 (see a very good set of posts by laihro, in the Give Peas a Chance thread for how objects dominated by one or two emission lines change (SDSS) colour with redshift).  Ditto the [OIII] lines (500.7 and 495.9 nm), 'real' green gets mapped to SDSS B (as they fall in the part of the spectrum covered by the g filter).

"Hα" is the "α" line in the Balmer series; there certainly are others, and they show up in the spectrum of Hanny's V (see here, courtesy of NGC3314; well some of them show up).  Unless something very peculiar is going on, the α line of each series in an astronomical object is the strongest (both emission and absorption) - Lymann, Balmer, Paschen, Brackett, ...