First light with a Star Analyzer 200 - Lots of filters

[Ags]

6 hours ago, robin_astro said:

You can approximately colourise the spectrum (the various spectrum processing programs eg Visual Spec, RSpec, BASS project etc can so this for you) but you first need to calibrate your spectrum in wavelength. Your colours are way off, here is your spectrum with some lines identified.

Thanks for the help,, I have not attempted to calibrate the spectrum yet. instead I have gone down a rabbit hole of working on a rabbit hole of working on a user interface.  I had my doubts that the spectrum showed anything real, so it is great to see two lines at least!

I have noticed that the blue end seems a bit faint, and I have been wondering why. I will check the spectra either side of the zero order next time I go out. To be honest, I was happy to find Vega at all, and get anything that looked like a spectrum. I had forgotten entirely that the SA is blazed.

Edited August 26 by Ags

[Ags]

@robin_astro here is my first attempt at autocalibrating the image based on detecting the deepest absorption line which in this case I now know to be Ha (obviously not a general solution and other spectra will be different). I then set the RGB region from 400 to 700 nm.

I'll do some research on a better algorithm to generate a flat (my current very simple smoothing approach generates false peaks either side of an absorption line), and change the autocalibration to a be a best fit based on a fingerprint of a number of common lines. And @vlaiv I also work on generating a more realistic rainbow.

Edited August 26 by Ags

[vlaiv]

54 minutes ago, Ags said:

And @vlaiv I also work on generating a more realistic rainbow.

Are you actually making a program to provide above output?

Generating rainbow is fairly easy in that case. You'll need XYZ matching functions and good deltaE function. First is well defined and for second you can use Lab color space deltaE (you can easily look up both online).

For each wavelength and intensity of your spectrum - you would:

1. create XYZ color space coordinate using XYZ matching functions

2. search sRGB color space in vicinity of original XYZ clipped to sRGB for color with smallest deltaE between it and original XYZ color and use that in rainbow.

[robin_astro]

3 hours ago, Ags said:

here is my first attempt at autocalibrating the image based on detecting the deepest absorption line which in this case I now know to be Ha (obviously not a general solution and other spectra will be different). I then set the RGB region from 400 to 700 nm.

 You don't need an algorithm to  find the lines for wavelength calibration in any star.  That would be impossible in general since you don't know in advance what you are going to find and at high redshifts the lines will not even be where you expect them !  eg

http://www.threehillsobservatory.co.uk/astro/spectra_3.htm

https://britastro.org/observations/observation.php?id=20210406_144443_9e1c6a4cf219d14d

You just need to calibrate the setup once (typically using a hot star like Vega with clear Balmer lines.) Once you have the dispersion from this in Angstrom/pixel  you can apply this to any star using the zero order as the 0 Angstrom reference point.

Before progressing your spectrum processing software further I would recommend first learning more about how spectra are processed from images to final spectrum  using the links I have posted and the already available tools. 

Cheers

Robin

[robin_astro]

Although you can use tools like Rspec, Vspec etc, calibrating Star Analyser spectra is actually very  simple. The wavelength is proportional to the distance along the spectrum from the zero order.

Guess the identity of one strong line somewhere in the middle of the spectrum of your A type star. This is usually H beta.  Calculate  the dispersion (4861/distance in pixels from the zero order) then apply this to other lines. These should then correspond to other Balmer line wavelengths eg H alpha 6563, H gamma 4340 etc. If they do not, try H gamma, or in your case H alpha as the line you guessed until the lines fall into place.

(You can also use the Telluric bands in the IR from our own atmosphere as approximate reference points. These will appear in all spectra but will not be obvious in cool stars which have their own very broad bands from molecules)

Cheers

Robin

 

[robin_astro]

4 hours ago, Ags said:

I'll do some research on a better algorithm to generate a flat (my current very simple smoothing approach generates false peaks either side of an absorption line)

Flat is not the correct term. In spectroscopy, flat correction is something completely different.  What you are doing is extracting the  continuum. (The underlying shape of the spectrum without absorption or emission lines) This relatively easy to do for hot stars like Vega which have few lines. 

Fitting a filtered curve to the whole spectrum including any lines will not work as you have seen. (It will give an even more inaccurate result with spectra with many lines). There are two common techniques used to extract the continuum. You can either  cut out the regions around the lines and fit a spline curve to what is left or you can pick a number of points in the continuum between the lines and again fit a spline curve through these.  You can see a demonstration of this in my "low resolution slitless spectroscopy" presentation and in my flux calibration document. (There it is used to calculate the response of instrument plus atmosphere but the same technique can be used for rectifying the spectrum to normalise to the continuum)

Cheers

Robin

[Ags]

Yes, I have been thinking about a way to prune out fine structure so I can extract the continuum more cleanly. It is an enjoyable programming challenge. 

Point taken about doing a bit more reading!

Anyway, I have a better colorization now, at least.

[MalcolmP]

A very interesting topic thanks  @Ags .

(Aside) I have seen refs to telluric bands before and lazily just assumed they were due to elements in the same group as Tellurium. 
Today I googled and found telluric means in, on or of the Earth, hence in our atmosphere.
More reading produced " ... named the new element in 1798 after the Latin tellus 'earth'  "
Ha! If only I had done Latin and the classics at school instead of Math, Phys & Chem I would have been on-message years ago  !!!  Every day there is something new to learn  
 

[Ags]

Its kind of twisted to call bands from atmospheric oxygen and water telluric bands isn't it!

Looking forward to getting out again and shooting a few more spectra! 

[robin_astro]

30 minutes ago, MalcolmP said:

More reading produced " ... named the new element in 1798 after the Latin tellus 'earth'  "

Similarly Helium after  the sun, named by Sir Norman Lockyer who discovered it in the spectrum of the solar chromosphere before it was found on earth.  The similarly named "Nebulium" found first  in the spectra of nebulae did not fare as well though. That turned out to be a "forbidden" line of ionised Oxygen [OIII] only found in very low density gasses so the atoms have long enough for the  low probability transition to occur before colliding. 

Cheers

Robin

[MalcolmP]

2 hours ago, robin_astro said:

Similarly Helium after  the sun, named by Sir Norman Lockyer who discovered it in the spectrum of the solar chromosphere before it was found on earth.  The similarly named "Nebulium" found first  in the spectra of nebulae

The story of helium did cross my path in my yoof, but Nebullium is new to me, thanks.

by Sir William Huggins, F.R.A.S in 1864
https://archive.org/details/philtrans01845403 8 pages + some illustrations (Plates) elsewhere in the volume
quote "[No. 4373. 37 H. IV. K.A. 17 h 58 m 20 s . N.P.D. 23° 22' 9"-5. A planetary nebula
;very bright; pretty small; suddenly brighter in the middle, very small nucleus.]
In Draco.
On August 29, 1864, I directed the telescope armed with the spectrum apparatus" etc,/quote

No. 4373 = NGC 6543 The Cat's Eye Nebula
It surprised him that it had no spectrum, just a single monochromatic line, no mean feat for 1864 !
Is this the next target for  @Ags  after Vega I wonder  ?
 

[Ags]

Well astronomers are going to be surprised when it turns out those OIII regions are solid Nebulium after all!

[robin_astro]

3 hours ago, MalcolmP said:

Is this the next target for  @Ags  after Vega I wonder  ?

Interestingly  Helium emission can be observed using a Star Analyser but you need a total eclipse to hide the bright photosphere (which does not show helium in its spectrum), by measuring the flash spectrum at 2nd and 3rd contact as here in 2006

details here

http://www.threehillsobservatory.co.uk/astro/spectra_27.htm

(If I remember correctly Lockyer observed it outside eclipse using some kind of occulting device, though Janssen who might be considered the co-discoverer (particularly if you are french !) did observe it earlier during an eclipse.)

Cheers

Robin 

Edited August 26 by robin_astro

[Ags]

The SA200 went out again tonight, and I was a bit more prepared this time. Firstly, I used the right spectrum this time. Secondly, I tried polar aligning and goto with the SA200 scope, and it worked, so no need after all to flip to another scope and camera to do polar aligning and plate solving, which is very convenient. I did Vega again, as well as Caph, Shedar, and V424 Lacertae. The last is a K or M star, so the spectrum should have more features and complexities to confound my processing pipeline. I was hoping to do Polaris as well to get a Sun-like spectrum, but it was obstructed. 

I really should switch to the RC6 for the depth of focus if nothing else, although gathering more light and higher resolution should be helpful too. 

Edited August 26 by Ags

[Ags]

I took a spectrum of Caph, and tried to identify some lines in it.

The red lines are O2 and H20 I believe. the green lines should be the Balmer series. Yellow marks the K and D1/2 lines.

As you might notice I have slightly improved my continuum extraction; I generate fewer false peaks although my approach breaks down in the violet area. I have a new glitch that I generate false broad absorptions around local peaks. In general I fail in any busy section of the spectrum. i will try another approach on the weekend.

[Ags]

I overexposed a new Vega spectrum, but still the Balmer lines show quite well.

[Ags]

I also snapped the obligatory shot of M57, showing its narrowband nature.

[robin_astro]

13 hours ago, Ags said:

In general I fail in any busy section of the spectrum

You approach will not work in spectra with many lines. You need a different approach and chose points between the lines that you identify as being in the continuum and join them up with a spline curve. This is particularly difficult and often impossible in low resolution spectra though. There is also the complication of ripples in the response of the camera sensor which can be mistaken for features in the spectrum. 

Only in hot O/A/B stars  do you actually see the continuum in low resolution spectra. In G-K stars there are so many lines that they blend together at low resolution to hide the continuum. In these stars you only see any continuum at much higher resolution and sometimes even  then it is  impossible. In cool M stars for example  the spectrum is just very broad molecular absorption bands with no continuum and in supernova the lines are so broad and blended due to the high velocities

See the section on this rectification process in my flux calibration document (see screen shot here)

 The best approach therefore is to correct for instrument response and atmospheric absorption using a hot star with few lines as a reference to produce a spectrum as it actually is without attempting to remove the continuum. This can then be used as the basis to produce a rectified spectrum if required

 

Cheers

Robin

[Ags]

@robin_astro Actually that makes my life easier. If there is no general objective solution to my goal of making a pretty spectrum, thenI can just provide aesthetic tools to get close enough. Then I'll switch to calibrating with a bright star for the more scientific view of things.

I have tried a few red stars and they are rather complicated 😀

[robin_astro]

38 minutes ago, robin_astro said:

Only in hot O/A/B stars  do you actually see the continuum in low resolution spectra. In G-K stars there are so many lines that they blend together at low resolution to hide the continuum. In these stars you only see any continuum at much higher resolution

for example if you look at the spectrum of Caph (beta Cas/HD432) which is an F2 giant so still quite hot

At  the resolution of the Star Analyser we might expect to get something like this. A relatively simple looking spectrum with mainly Balmer lines and we might think we can estimate where the continuum is reasonably well

If we look at high resolution though this is what we see

and if we zoom in on just 300 angstrom around the H gamma line, we see the true complexity of the spectrum and realise that we just dont see the continuum in the low resolution spectrum (The continuum actually runs along the top  through the points between the absorption lines )

This example spectrum of Caph comes from the MELCHIORS library of very high resolution spectra

https://www.royer.se/melchiors.html

specifically

https://set-p-stweb.bm.icts.kuleuven.be/quicklook/307828_F2III_VbetCas.png

 

Cheers

Robin

Edited August 29 by robin_astro
typo

[Xilman]

On 26/08/2024 at 13:33, Ags said:

Its kind of twisted to call bands from atmospheric oxygen and water telluric bands isn't it!

Looking forward to getting out again and shooting a few more spectra! 

Nowhere near as twisted as calling neon a metal, which is what astrophysicists do.

[Ags]

I think I have spline smoothing figured out. Caph spectrum again, with continuum extracted with cubic splines.

[Ags]

I found an old post somewhere by @robin_astro from the early days of Star Analyser about a star party in Patrick Moore's garden, where they shot a few interesting targets in Cassiopeia, including WR 1 and Navi, both objects with interesting emission features. WR1 is a bit faint for 30 mm aperture, but I am hoping guiding and exposure time can cure that. I know what I am doing on the next clear night!

I am thinking about upgrading my spectroguider scope. Maybe a small collimatable newt like the Ursa Major 80mm f6.3 might do well. With 500 mm focal length, plate solving with the ASI120MM would still be possible, guiding would be ok, and at f6.3 I could keep more of the spectrum in focus than at my current f4.5.

[robin_astro]

Yes WR stars make great targets for the Star Analyser with their very broad intense emission lines. Also because the light is concentrated in the emission lines they show up better than you might expect given the brightness. They could be an interesting challenge with just a 30mm aperture  though. Here are the observations I made from Sir Patrick's garden back in 2005

http://www.threehillsobservatory.co.uk/astro/spectra_12.htm

and another Star Analyser field with a WR star in it 

https://britastro.org/observations/observation.php?id=20201216_234948_8cabda965bfe692f

Cheers

Robin

[robin_astro]

WR140 is one of the brighter ones. It is a binary system with an O star in a very eccentric 8 year orbit and was the subject of pro-am spectroscopic studies in 2009 and 2017 observing the system at periastron  when the winds of the two stars collide. 

http://www.threehillsobservatory.co.uk/astro/spectra_41.htm

It is due to happen again in November this year.

There is a famous JWST image showing the rings of dust produced each time

https://webbtelescope.org/contents/media/images/01GEJB2906TM9VR2FSJ4TFMNQM