Merging echelle orders

Frank Valdes wrote on Jun 26, 1998

Q: I am reducing and analysing multi-order spectra, obtained using the
fiber-fed Echelle spectrograph (Elodie) located in Observatoire
Haute-Provence, France. I have reduced and extracted data covering the
range 3900A to 6800A using Iraf. This data extends over 67 orders. It
is necessary for me to measure features (Diffuse Interstellar Bands)
which extend over two or more orders. Is there anyway that one can
merge two or more orders together so that I would be able to measure
the FWHM etc. of these features using for example 'splot'? If it is
possible to do this, is it possible specify a wavelength range with no
discontinuities or is one limited to merging whole orders together?


A:  Merging echelle orders is one of the hardest steps in echelle
reductions.  The problem is dealing with the blaze function in such a
way that the overlaps match.  One way to deal with this is to continuum
normalize the orders first.  The makes matching the orders much
easier.  The task for continuum normalization is CONTINUUM. This is
probably the approach you should take.  However, when you have broad
features, especial if they span more than one order, this can be hard
or impossible to do.

The usual tool for merging orders is SCOMBINE.  In the overlap you can
choose to average or do some other combining method.  But as I said, if
the blaze function is not handled well you will get a very rippled
result.  In SCOMBINE you specify the final wavelength range and
wavelength steps.  So, in a way, this is what you asked about for
specifying a specific region of interest.

Since you are only interested in a few orders, it may be easier if you
copied out only the orders of interest with SCOPY before you do
anything else.  Some other hints are that you can use DISPCOR to
resample several orders to a common wavelength interval.  You can also
use DISPCOR on a single order (extracted with SCOPY) to any starting,
ending, and interval in wavelength.  SCOPY can be used to extract a
piece of a spectrum and the "rebin" parameter allows resampling like
DISPCOR.  But DISPCOR is more general since SCOPY does not allow you to
change the wavelength interval.  By using SCOPY and DISPCOR you could
get specific pieces which don't overlap but match together in
wavelength; i.e. the last pixel of one order comes just before the
first pixel of the next order.  You would still use SCOMBINE to piece
things together.

---

Last post on Jun 26, 1998