Hi folks,
I took a recent spectrum of Gamma Cas H alpha and for some reason the EW turned out to be way way too low. This was after a good spectrum was taken just a week or so earlier. I am scratching my head trying to figure out what I could have done during the processing to cause it. First time I have ever seen this happen. It's a very easy target, only a second magnitude star and the signal to noise is great, resolution (which I understand does not affect the EW) great. I tried retaking my darks and bias frames, reprocessing with and without instrument response, recalculating instrument response, using a different comparison star, trying different reference regions of the continuum and even trying different versions of the processing program. No changes noted thus far. So while I am waiting for better weather to try again, I was hoping someone could tell me potentially what processing steps if gone awry could affect the EW measurement adversely. I attached a file showing the calculated difference (black arrow) between the good spectrum and the recent one that was way off. The difference should be negligible which it clearly is not. Thanks, Dave Doctor
Perhaps the difference is real. Gamma Cas H alpha is variable. eg
http://www.spectro-aras.com/forum/viewtopic.php?f=14&t=714
How do your EW compare with other spectra in BeSS ?
http://basebe.obspm.fr/basebe/
Cheers
Robin
If it is not real, then either the continuum is too high or the emission line is too low. Here are some other things you can check:-
Saturated pixels in the emission line ?
Spectrum binning zone does not includes full width of spectrum
Sky background binning zones are contaminated
Cheers
Robin
Also check how your response using your comparison star looks.(Look for residual contamination from H alpha in reference star in the response) If you used a flat, the response should be almost a flat line over such a small range at this wavelength. It should also be almost unchanged over a long period provided you are using the same setup and flat lamp. Try without response correction to check if the difference is still there.
Cheers
Robin
Yes I checked with and without response and no difference meaning the "difference" was still there. There could be an issue with saturated pixels in the emission. A rough estimate looking at image stats of a subframe shows that the max is extremely close to saturation. So certainly exposure could be reduced next time.
Dave
One final item to add to the check list is the cosmic ray filter if you are using one. If the threshold is set too aggressive, it can "heal" pixels containing real data in high contrast areas, particularly if the spectrum is close to being undersampled. If you are using ISIS, check the intermediate @map images which show which pixels are being healed.
Cheers
Robin
Thanks Robin. The filter is set to 15 which I thought was pretty low. Yes, in ISIS I have not previously saved the intermediate files but will look at them. Thanks again!
Dave
Thanks Robin. I was told the difference is not real. BeSS looked at it and concluded the same thing ,that despite the variable nature it isn't that unstable to exhibit such a large change in that short a period. The comparison spectra there did not show it. Also those actively researching this system currently have told me my spectrum is a definite outlier and not consistent.
Regards
Dave
Dave,
the BeSS hint you got is totaly correct. Our current monitoring in different ways provides absolutely no hint of such a strange EW variability whithin such a short time.
So, it is reasonable to assume that there are instrumental influence which lead to the too low EW. In deed, if the emission has been taken near the limit of saturation would be a plausible explanation.
I am interested to get the raw spectra in order to check that issue.
Ernst Pollmann
Thanks Ernst. Here is the download link to the raw files. I did test it but let me know if it doesn't work.
Regards
Dave
https://www.dropbox.com/s/c7rqxrasxa0xhcb/Gam%20Cass%203_3.zip?dl=0
Hi Dave,
I can see lots of saturated 65535 value pixels in the cores of the emission line in the ISIS image preview I am afraid.
(By the way, how did you attach that image in your original post? When I asked admin how it could be done in the new forum format, I was told it had been disabled)
Cheers
Robin
I ran a quick experiment on your images and if it is a critical observation it looks like it should be possible to salvage the data by just binning the unsaturated rows at the edges of the spectrum image
Robin
Interesting! Not sure how critical this particular observation is though. It is a fairly easy observation to do and can easily be repeated with the proper exposure times. I would be interested myself to know how you would accomplish this in ISIS perhaps for future reference?
Thanks!
Dave
Well it will be a bit fiddly and probably not worth it in this case as you say.
It is a bit difficult to describe without pictures but as an experiment I identified the rows in the spectrum image with saturated pixels, effectively the rows down the middle of the width of the spectrum. I then set the binning zone just above these so it just included the unsaturated rows along the top the spectrum image and generated a profile. I then did the same with the binning zone covering the lower unsaturated rows.
The two profiles were almost identical giving confidence in the method but you must make sure the tilt in the spectrum is corrected accurately otherwise you can easily get ramps in the spectrum doing this. For the final result I would average the two unsaturated spectra. (Note this probably should not be attempted where there is any sign of change in focus or chromatism in the spectrum which might cause the spectrum to vary across the cross section of the spectrum image but that is unlikely to be a problem here and as an emergency last resort might prove useful)
Cheers
Robin
Hi Robin,
That's great! I really appreciate it. I understand the gist of what you are doing there.
Regards,
Dave
HI Robin,
Yes thanks that's exactly what I was thinking once I looked at it. I was a little overzealous with exposure trying to get the max SNR but it's a very bright star! Hopefully that's it and reducing exposure time should do the trick. However I would be interested to know where in ISIS you are able to see it? I had to look at it in another program.
Not sure how I was able to attach that image. I thought something was wrong when the button was greyed out and maybe reloaded the page somehow to get it to come up but I can't remember specifically.
Thanks again
Regards
Dave
Hi Dave,
If you view an image in ISIS (either via the "image" or "calibration" tabs), you can hover your mouse over an image and read off X Y coordinates and I value of the pixel at the mouse position, shown at the bottom of the right hand panel. A large swathe of identical values is a good sign that the pixels are saturated.
Cheers
Robin
Excellent! Thanks Robin.
Regards
Dave
Hi Dave,
the attached screen shot of a spectrum scan with MAXIM DL shows very clear: the spectrum is over-saturated.
/sites/default/files/users/user63093/gamcas.png
Ernst
Excellent! Thanks very much Ernst. Yes that is very clear and exactly what I would have expected to see in the raw line profile if oversaturated .This is all good news as it is an easy fix :)
Thanks Ernst and Robin for your help!
Regards
Dave