iraf.net

[b:aed5a995d4]From: [/b:aed5a995d4]Ranjan Gupta <rag@iucaa.ernet.in ([email]rag@iucaa.ernet.in[/email])>
[b:aed5a995d4]Date: [/b:aed5a995d4]May 27, 2006 10:52:23 PM MST
[b:aed5a995d4]To: [/b:aed5a995d4]"Frank F. Valdes" <valdes@noao.edu ([email]valdes@noao.edu[/email])>
[b:aed5a995d4]Subject: [/b:aed5a995d4][b:aed5a995d4]Fringe removal in red part[/b:aed5a995d4]
Hi Frank,
Greetings from India! 
Our new 2 meter telescope has started and we have been trying
observations with its Imaging Spectrograph to observe many of our
CFLIB stars. While reducing the spectroscopic data for the red part
(one grism has range 600-850nm) and we see a lot of
fringes (since its a thin CCD). Can you suggest an IRAF task which
can effectively remove the fringes -- (you had done it for our
CFLIB spectra)?
I have attached a .ms.fites file of HR5501 IRAF std star which shows the effect of fringes in the red side.
Looking forward to your reply,
Regs
Ranjan
-- 
Prof. Ranjan Gupta,      Phone: +91-20-25604111 (direct office)
IUCAA, Post Bag 4,       or via operator +91-20-25604100 ask for extension 111
Ganeshkhind,             Res:+91-20-25604411/25604471 (direct home)
Pune-411007, India       or via operator +91-20-25604100 ask for ext 411 or 471
FAX: +91-20-25604699
URL: http://www.iucaa.ernet.in/~rag
Cell: +91-94235-82545
-- 
[b:aed5a995d4]From: [/b:aed5a995d4]Frank Valdes <valdes@noao.edu ([email]valdes@noao.edu[/email])>
[b:aed5a995d4]Date: [/b:aed5a995d4]May 30, 2006 10:11:52 AM MST
[b:aed5a995d4]To: [/b:aed5a995d4]Ranjan Gupta <rag@iucaa.ernet.in ([email]rag@iucaa.ernet.in[/email])>
[b:aed5a995d4]Subject: [/b:aed5a995d4][b:aed5a995d4]Re: Fringe removal in red part[/b:aed5a995d4]
Hi Ranjan,
The usual steps for handling fringing is to remove it during flat fielding.  So the flat field is first processed by CCDPROC and then a "response" flat is created using the task RESPONSE.  The idea in this task is to fit a smooth function of wavelength for the average flat (automatically generated from the 2D flat) and then divide each pixel in the 2D flat with the appropriate function value.  The key is to use a fitting order which is high enough to remove the general shape of the flat field lamp but not so high as to fit the fringing.  By dividing by the low order fit the general shape of the flat field will not be applied to the science exposures but the higher frequency fringing will be divided out of the science data.  The response flat you generate is then applied using CCDPROC with the response flat specified as the flat instead of the original flat.
Note that if you are going to do flux calibration then it really doesn't matter if the shape of the flat field spectrum is introduced to the science data since the flux calibration step will handle that.  So in that case you can either just skip RESPONSE and directly divide by the flat field (in CCDPROC) or use a very low order function like a constant to just normalize the flat field.
There is no other regular way to handle the fringing other than dividing by a flat field.  It would be possible to divide by the extracted flat field.  In other words you could do the flat fielding and fringe removal in 1D instead of in 2D.
Cheers,
Frank