spectro-aras.com

Hi peter,

Yes this is not the best accurate method but it is still a way to correct the greater part of the horizontal shifting.
Anyway, have you tested the pipelines to see how great is the accuracy?

An improvement that is easly to do is to manage all your calibration spectra to build a non linear shifting model (ie polynomial here). This will a be a way to be closer to he experimental shifting.
But sudden jumps will need a deeper coding and processing will the be heavier.

What is the method used by SpectroCalc you have experimented?

Benji

Hi Benji,

Currently I'm evaluating 2 software (spcAudace and SpectroCalc2) for use, so far only by reading the specs and maybe doing quick experiments (more time only at weekends).
I don't rush, and I'm a slow learner.

The SpectroCalc2.pdf manual tells about exactly my need, so easy to understand:
"Prefacing
I am working with a Shelyak LHires III spectrograph with 2400 lines/mm.
...
With my equipment, I got the experience that the wavelength can move more than 10 pixels over the night, caused by temperature shift. For that reason, I take always before and after each image of the spectrum a reference lamp image (except I decide to calibrate with water lines)."

At this moment, I do not know whether any polynomial is used in SpectroCalc.
Currently I'd be happy with simple average of 2 calibrations for each exposure.

Enhancements could be:
- after the 1st (full) calibration, we need only a shift (a single number!) in case of the LHires III, not the full calibration - so that measuring a skew should require less calibration time, that is a big win (by practice, I could spend e.g. 2-3 seconds rather than typical 15 seconds between each exposures).
- error estimate is important (in ISIS currently I can see resolution and calibration error numbers, sometimes very useful)
- polynomial solution (to deal with any shift-noise) would be highly appreciated, but with a manual control (e.g. see a serie of error numebrs between polynomial and the simple averaged)

Since you currently don't have this solution (and won't ask favour in a rush just for me), I'd suggest that I continue exploration with SpectroCalc2 for now (maybe asking author by mail what is really happening), and will write my experience here.
In case you happen to come up with a solution in spcAudace meanwhile, I will be happy to try.

Peter

Hello Peter,

Great that you've founded a beginning of solution.
I'm currious about quantitave values returned by SpectroCalc for such shifting correction.
And a comparaison with the one done by Spcaudace on your spectra. They are stored in fits header keyword SPC_REGH.

Cheers,

Benji

To: Peter,

RE:"James,
Thank you for sharing your method, good to see you also care with the shifts - at least the in case of high-res H-alpha region!
If I'm right, you don't have any tellurics for the CaII region (not even for H-beta), so you can't check for any shift.
In case with fiber fed eShels, you won't meet this problem (although I saw professionals still taking bracketed spectra with fiber-eShels, but only for an error check).
I'm still continuing to search for the best possible processing method, such that can be used in general."

With the help of some of the BassSpectro at io.groups, I was able to construct a Hb-atm file to use around H-beta. However its usefulness is much less, being I was only able to strictly identify 4 lines and only one (possible) unambiguous line at 4848.41A; see:


The CaK region, from my location, has one Hg line at 4046.51A which maybe useful for shifting the spectrum; thank goodness for light pollution!

James

James, interstellar Ca H and K lines can be also used - if present (close to galaxy plane) and if not severely blended - as a fixed reference (but not necessarily absolute!). I'd add that there is actually plenty of interstellar lines in various regions of optical and near-UV spectra, though not all of them are very strong.
See e.g. https://www.eso.org/sci/meetings/santia ... wski-2.ppt

Best wishes,
Tõnis

Hello,

Sorry for not answering sooner, been busy with more people contacting in private and discussing stuff.
To establish a result-based decision, I've found one of my earlier observation session suitable: I've made a full night observation of Gam Cas around He I 4471 to repeat an earlier known phenomena about puslation effect on absorptions (this is known indeed, I just wanted to repeat it myself). That was a shorter summer night: 21:22 - 02:23 UTC (5 hours total!).

I was using the 0th calibration (40 seconds each, 1x1 bin) to calibrate ALL frames shown here (6 prominent lines + 6 lower snr lines) both calibrations (40 sec) and target spectra (350-250 sec) between each other.
I processed calibrations (49 of them) measuring RV using ISIS -> Misc/Radial Velocity (other name: CCF or cross correlation) method.

First shot of example, to imagine how Gam Cas looks in this region (Y range is small): Shifts during 5 hour is 40 km/sec in total: It is non-linear. Bigger shift in the first 1 hour.

To confirm the shift is real (not just my internal Ar/Ne lamp misbehaving), measured RVs against 1st real spectrum: It has some bigger noise, it's why I don't like using the spectrum (here at least 3 features changed, and I took only 1 flat later).

Summary:
- my shift is roughly 8 km/s per hour on average, as 5 x 8 = 40 (went up to roughly 16 km/s per hour here at the beginning)
- sometimes non-linear (see graph), temperature drop suspected (typically beginning 1/3 of my nights)
- target spectrum is not useable for shifts corrections, here used only as a check

Because of my typical use (1-3 hours per spectrum), I prefer staying with ISIS, and transit sum header via different software (BaSS is going support summed fits header transition soon).
Thank you all for reading all of this!

Peter

Note: I have special camera-adapter with many screws added, and a 2" LHires-to-scope adapter, only manual focusing. Others may have very different physical setup!
Note 2: quickly looked at spcAudace/archive_pipe-2a.zip, amazed by the shift within just 30 minute... but because of non-linearity and longer exposure time projects, I prefer staying with ISIS for now.
Note 3: SpectroCalc2 is going to add demo files with youtube later, however it requires darks matching with exposure times (bigger restriction), and no cosmetic processing + cosmic rays facility, and not going to be added.

Peter, that is a very nice analysis!

Best wishes,
Tõnis