Questions to the campaign

[Robin Leadbeater]

The original poster has made an important point though which affects all measurements of EW. To measure the EW of a line you need to establish where the continuum lies so you can measure EW relative to it. This needs to be done in a systematic, repeatable and reproduceable way. The problem here is that at this resolution we do not see the continuum. The overall shape you are removing here are actually bands of very fine unresolved molecular lines. If these change in intensity your pseudo continuum, and consequently your EW will change even though the strength of the line you are measuring may be unaltered. The solution is generally not to try to remove the continuum across the whole spectrum but to try to define a local continuum for the particular line of interest. Ernst has been plotting the EW of the H alpha line, perhaps he can describe the method he has been using in this particular case so others can replicate it.

Robin

[Peter Somogyi]

Just let me add (without knowing how Ernst addressing the problem), in case you don't model and remove the M component anyhow (e.g. shoot mu Cep or Alp Ori and do the RV align), then the EW measurement will highly dependent on the coice of continuum interval selection.
My method was subtracting a well multiplied alpha Ori, usable only for a local place (short wavelength range) around a given feature set (because did not deredden for the whole range).
Determining the M ratio can be tricky, for the CaII region I was using an algorithm to detect that ratio automatically. At H-alpha, such a problem (B comp. almost negligible) less prominent than in the blue where you get the EW dependent on the B+M light ratio.
- Peter

[Christof Wiedemair]

Thank you for all your answers,
I see that there are several different approaches.

For Ernst, attached to this post you can find three of our 1d spectra.
I am looking forward to hear back from you.
Kind regards,
Patrick Franzinelli

180118.zip

(9.01 KiB) Downloaded 89 times

171116.zip

(8.9 KiB) Downloaded 81 times

170927.zip

(9 KiB) Downloaded 84 times

[Ernst Pollmann]

Patrick,
if I understand right, your main aim should be the evaluation of the Halpha EW.
But first, I wanted to get information of the pseudo-continuum. That´s why I compared your (170927) spectrum with a flux-calibrated ones of the MILES data base (No. s0199, HD039801, spectral type M2Iab).

This comparison shows a well agreement in relative flux within the Halpha range. It seems there is no appreciable influence of instrumental response in that area. But of course, for EW evaluation it requires the process of normalization.

EW_evaluation-1.png (17.07 KiB) Viewed 1424 times

EW determinations have been carried out since 1996 always within the spectral section 6551 to 6573 A. To execute that, a normalization has been carried out always within the section 6540 to 6590, how it is to see in the Figure above. You will recognize that a certain (minimum) of spectral resolution is necessary to determine plausible values of EW.

EW_evaluation-2.png (11.03 KiB) Viewed 1424 times

I have zoomed the mentioned spectral range for normalization and EW determination as it is to see in the Figure above, in order to show, how this process has to be carried out in your low-resoluted spectra. The EW in your spectrum is - 7.5 A, but it should be -9.8 A compared with the value of the campaign monitoring.
In addition you can see that the Halpha emission shows a strange and unrealistic profile which is based in the too low spectral resolution. With other words: seriously EW determinations in such a sort of spectra are less to not meaningful.

Best wishes,
Ernst

[Christof Wiedemair]

Thank you Ernst,
it happened that I sent our worst spectrum. We had an issue with our spectrograph, it failed to focus the whole spectrum at once. So it was replaced with a new one. The following spectra should be of much better quality. I know your time is limited, still it would be so kind of you if you looked at one of the other two spectra.

We collected spectra on 11 nights during the last fall and winter. On 8 nights we even collected data between the 4800A and 5900A. The data features the decrease of the H-beta emission lines.
We could share all our data with you, if you desire.

About the EW, how did you sum up the data? Did you just sum up the values out of the 1d.fits, or did you use a numeric integration method?

This is our first big project with spectroscopy, and we are still learning. We are glad for every constructive criticism and advice we can get.

Sincerely,

Patrick Franzinelli.

[Andrew Smith]

Ernst can you explain how you carried out the normalisation? Did you use a polynomial fit or interpolation through certain points and if so how were the points selected?

Regards Andrew

[Ernst Pollmann]

Patrick,
yes I saw your other spectra, but I used the one of 170927 because it has the best signal/noise.
For EW evaluation I used VSpec and its EW computation, but I don´t know which integration method is implemented there (rectangle/tangent/trapezoidal/Simpson rule ?). Perhaps Valerie could give an information.
Thank you very much for the offer to make the other spectra available, but for the current Halpha campaign we don´t consider spectra with such a low resolution. Hope you will understand that.
Concerning learning:
I invite you and recommend to take part at our next spectroscopy seminar at the CFG observatory in the city of Wuppertal (Germany) from 20.-22. April this year. For more information see:
http://astrospectroscopy.de/media/files ... r-2018.pdf

Best wishes,
Ernst

[Ernst Pollmann]

Andrew,
in this case I used a smoothed spline function in VSpec after supress of the Halpha profil.

Ernst