comments concerning SSA and spectral data model

[Igor V. Chilingarian]

Hi all,

Today in Pune we had a short discussion with Jonathan McDowell
and Philipe Prugniel concerning SSA and spectral data model.

I'd like to post some notes as a result of our discussion.

1/
As far many spectra exist, especially in the optical band, represented as
1D-FITS images with a wavelength scale specified as a WCS and this way of
representation doesn't fit the spectral data model, we propose to include
to the metadata some optional keywords describing wavelength sampling in
several simple and widely-used cases in addition to the "wl-flux" table.

Actually, it is not trivial to recover the type of the wavelength sampling
from the individual points, because it is necessary to check different
possibilities: linear sampling, logarithmic sampling etc. So, we propose
to add parameters for
	1) wavelength sampling type: linear or logarithmical
	2) starting wavelength
	3) wavelength step
	4?) unit

The issue is a unit. It is supposed to be already contained in the
metadata, but i.e. for logarithmically rebinned spectra that are widely
used in the extragalactic studies, the wavelength and step are specified
in the logarithmic units that are not connected directly to Angstroms, but
more to km/s. This is a point TBD

Adding this keywords will allow to easyly use existing tools for spectral
data processing and analysis without major modifications.

2/
In the future, it would be nice to have a possibility to describe not only
the spectral resolution in each point of the spectrum (that is a case for
probably v.2), but also a shape of the line-spread function, because for
some studies related to kinematics and stellar population of galaxies this
is a crucial point. I rised this question on the DAL session yesterday.

With best regards,
						Igor