EFOSC2

ESO Faint Object Spectrograph and Camera

Fringes in Spectroscopic mode

25/7/2001

Background: The observer was using the grism 10 with 1" slit and 2x2 binning. He took some spectroscopic dome flats in the afternoon. He found that, using the domeflat to flatfield the science frame still results in residual fringes of the level of about 2% in the red part in the science spectrum. This is not acceptable to his requirement,
since his objects are very faint.

We had a discussion whether the residual fringes were due to:

there is a slight wavelength mismatch between the dome flats and the science frame, causing a small shift in the fringe patterns in the 2 frames, which results in an imperfect flatfielding.
this may be caused by the different colours between the domeflat lamp and the sky.

We did the following experiment: we took an an internal lamp exposure straight after the science exposure. This is then used to flatfield the science exposure. The result is that there is no noticable residual fringe pattern. Hence, scenario 1 above is responsible.

Hence,

When an observer is doing spectroscopy in the near infrared, to get good fringe removal one should take internal lamps at the same telescope position as the object. This only takes a 2 minutes but makes a real improvement.
The flexture of the telescope does cause some shift in wavelength (which should be quantified further). For good wavelength calibration (e.g. for accurate radial velocities) HeAr arcs should also taken at the same telescope location as the object.
We should offer a new template offering spectroscopy with offsets along slit (akin to the templates on FORS and EMMI) which allows the observer to nod the telescope and so be able to median combine science frames to contruct a "super sky (fringe)" pattern.

The below demonstrates the difference in fringe-removal between flatfielding using a spectroscopic dome flat, and an internal lamp flat taken at the location of the science object.

	This is an extracted spectrum of an object, normalised and sky-subtracted, but not flat-fielded yet. This shows the very pronounced fringes in the red part of the spectrum. The grism is no. 10 and slit width is 1", binnig is 2x2.
	This is the previous spectrum, now flat-fielded by the spectroscopic dome flat. Note the residual fringe pattern of 2%
	The object spectrum, flat-fielded by an internal lamp "flat" taken at the same telescope location as the object. The residual fringe pattern is not present at a significant level.

Send comments to : ls-spectro

Last modified: Fri Jan 18 02:54:59 CLST 2002