[fitswcs] Paper III

[Mark Calabretta]

On Tue 2004/08/10 14:38:41 CST, Eric Greisen wrote
in a message to: fitswcs at nrao.edu

>to see if there are any problems.  Typos to major suggestions will be
>considered and, the former anyway, gratefully received.

(I don't mean to be contrary - this just happened to arise in the normal
course of events!)

Consider setting up a radio spectral observation; if the observer wants
to measure HI in 3C273 there's no point tuning to 1.420GHz because the
line will probably be Doppler shifted out of the band.  So you compute
the required Doppler shift and tune to the Doppler-shifted frequency.

While the reference frame for the spectral axis is still
SPECSYS = 'TOPOCENT', it would clearly be desirable to record the source
recession velocity in the FITS header and ZSOURCE (or VSOURCE) seems
like the obvious place.

However, Paper III says that ZSOURCE (VSOURCE) is recorded as "the
apparent radial velocity with respect to the observer", i.e. 
topocentric.  This may be reasonable in defining the source reference
frame, SPECSYS = 'SOURCE', but not so if you simply want to record the
source redshift.  In either case, it would be more convenient if ZSOURCE
could be recorded in some standard reference frame, e.g. barycentric or
LSRK.

For this reason, I propose introducing a new keyword SSYSSRC, to record
the reference frame in which ZSOURCE (VSOURCE) was measured.

While on this subject, it seems to me that VSOURCE is not a useful
keyword.  It is meant to record the "apparent radial velocity" of the
source, i.e. the relativistic velocity derived from the observed
redshift, but this is never used in practice - either the optical or
radio velocities are used.  Hence I suggest either abandoning VSOURCE
altogether leaving only ZSOURCE, or augmenting it with new keywords
VOPTSRC and VRADSRC (which of course suggests that ZSOURCE and VSOURCE
should really be ZOPTSRC and VELOSRC).

Mark Calabretta
ATNF