Neon Line Calibration

[Keith Graham]

Hi,

I have a question about accurate neon line calibration. My equipment is the LhiresIII, 2400l/mm grating, ST8me acquisition camera, and a 12” SCT @ f/10.
I use ISIS for reducing my spectra. I generally take a neon frame immediately before and one after the target sequence. Once I have the final reduced spectrum, I do a telluric calibration in ISIS to tweak the calibration. I then use the BeSS feature in Vspec to check for accuracy before submitting.

The issue is that I occasionally get spectra where the telluric lines are simply incomprensible, so I cannot tweak the calibration using them. I therefore must rely on the neon lines for accurate calibration. But I am puzzled as to how I can be certain that my neon lines are so accurately located that they can be reliably used for accurate calibration.

With my setup, I get 6506.528, 6532.882, and 6598.953 lines on the neon spectra. I would think that, for accurate calibration, those lines should be at the exact pixel position on the neon spectrum as they would be on the target if that target spectrum had those lines. But as we all know, there are times when there is a shift in the neon lines between the before and after spectra.

So my questions are:
1. What must I do with the before and after neon spectra to me certain they are in their proper pixel positions?
2. How can I be certain that my neon calibrated spectrum is accurate when I cannot check it using telluric lines?

Thanks,
Keith Graham

[Andrew Smith]

Hi Keith, You might want to have a look at this http://www.astrosurf.com/buil/isis/He_c ... method.htm. You could use a set of Ne lamps rather than the Filly Lamp which I think are no longer available.

One test I have done with my echelle is to process the before and after calibration frame as target spectra. This give a view of the shift if not the absolute error say to a misalignment of Ne and Target on the slit. Buil method addresses that issue.

Others I am sure will be better placed to advise.

Regards Andrew

[Robin Leadbeater]

Hi Keith,

The calibration using the neon assumes that the position of the neon line exactly represents that wavelength in a star spectrum taken immediately before or after it. ie any shift you see in neon spectra from before to after a star measurement is caused by some drift in the spectrograph during the exposure eg due to flexure or temperature. If this is the case and assuming the drift is linear during the exposure, you can add the before and after neon exposures and use the result to calibrate the spectrum. (you might decide not to take neons between every exposure in a set of exposures but the more neons you take the more accurately you can compensate for any drift.) This seems to work pretty well for me but if you want to test this you can check your neon calibration compared with known lines in a star spectrum eg when the telluric lines are prominent and see if there is a constant offset. If your tweaks based on the tellurics are small and in both directions then your neon is good enough. if the variation between your "tweaks" is small but offset in the same direction then you could apply this offset to your neon calibration. If your tweaks are large though and vary significantly then you would need to look at the stability of your neon calibrator. I find with my LHIRES system with the 2400 grating the repeatability of calibration is ~2km/s using the internal neon or ~1km/s using the tellurics. (1km/s at H alpha ~0.02A)

Robin

[Keith Graham]

Thanks you Andrew and Robin for your suggestions.

Robin, when you say to add the before and after neons together, are you saying I should stack and sum them? Or would median or average combining work? I was under the impression that sum, average, and median combining dealt with intensity. I would think if the result we are looking for should be the median position of the neon lines between the before and after shift.

Cheers,

Keith

[Keith Graham]

Hi Again,

I went to this link,

http://www.astrosurf.com/buil/isis/guid ... to1_en.htm

and I found something that I remember reading a long time ago. Near the end of this tutorial, Christian refers to the wavelength registration feature of ISIS. If I am reading this correctly, if I use only the first neon frame in the processing and if I check the wavelength registration feature, would this not solve the problem of accurate calibration using only the neon lines? Or am I missing something? I always prefer using tellurics because I can actually see the telluric lines lining up, but of course this is not always possible.

Cheers,

Keith

[Keith Graham]

Hi Once Again,

The more I thought about this, I seemed to remember a discussion on ARAS about this very issue. With some digging, I found the link:

http://www.spectro-aras.com/forum/viewt ... ?f=8&t=458

Lo and behold I was the initiator of this very issue over 2 years ago, and both Robin and Christian gave some very good replies. So I think I now have something to go on that will hopefully get me some more answers.

But, Robin, would you please still clarify for me your comment about “adding” images?

Cheers,

Keith

[Robin Leadbeater]

Hi Keith,

But, Robin, would you please still clarify for me your comment about “adding” images?

I don't think it really matters how you stack the calibration lamp images. Ideally I suppose you might want to weight them according to the flux in the star spectrum they correspond to but that is probably going too far. I use the "addition of a sequence" feature in ISIS under "tools" "image processing 2"

Note that if there is significant movement during the observation this will reduce the resolution. Unless I have missed something in ISIS, the only way round that I can see is to either try using the "wavelength registration" feature, in which case you only need the first lamp spectrum in the sequence, or calibrate each star spectrum individually with its own lamp spectrum and then sum them after. It would be nice if ISIS could register spectra using a series of lamp spectra, rather like the "wavelength registration" feature but using the lamp spectra as the reference.

Robin

[Keith Graham]

Thanks for that information, Robin.

I started this thread because I had a calibration issue with a Be star in which the noise was so bad (another issue) that I could not get a telluric match. The BeSS check feature in Vspec showed a lot of **. I had forgotten about the wavelength registration feature in ISIS. When I used it, that seemed to have brought the calibration more in line as the Vspec BeSS check gave much better results. But when the telluric lines are so obscure, I wish there was some other way to check calibration results for accuracy even when the wavelength registration is used. Perhaps there is, but I am unaware of if it if there is one.

Cheers,

Keith

[Robin Leadbeater]

Hi Keith,

The internal LHIRES neon should be good enough for BeSS even if the Tellurics are not visible. I have recently been using the LHIRES to produce some very noisy Be star spectra of faint Be stars and could not use the tellurics to check - examples here
http://www.astrosurf.com/aras/be_candid ... idate.html

How big differences are you seeing between neon and telluric calibrtation when you can see the tellurics?

Robin

[Keith Graham]

Hi Robin,

When I tweak a Ne calibrated spec with the tellurics, I generally need to create a shift of between .1-.3. When I do that and then do a Vspec BeSS check, my errors, for the most part, are .0x with some noisy spots being higher.

When I submitted a recent spectrum, it was rejected because I was told it was a whopping 6A off. This took me by a big surprise because even the BeSS check did not show this much of an error. That is when I ran it again using Wavelength Registration. The BeSS check shows some ** but also some .0x errors. So I resubmitted this spectrum to see if it passes. A recent check in the Be DB shows the spec is still waiting to be validated. I have no idea what happened the first time because I submitted another noisy spec from another star taken that same evening and that one was validated.

Cheers,

Keith