[fitswcs] Units along spectral axes

[David Berry]

> I've been working on incorporating Paper III in my IDL software, and came to
> realize that it's vitally important that the program be able to parse the units
> specification in order to implement the non-linear projections.  For example, if
> the axis keywords include
>
> 	CTYPEka  = 'WAVE-V2W'
> 	CUNITka  = 'Angstrom'
> 	RESTWAVa = 3.04E-8	/[m] Rest wavelength
>
> the program has to know how to convert Angstroms into meters in order to compare
> it with the rest wavelength.  I've written a units parser routine based on Paper
> I, but it could be easily confused by deviations (e.g. 'Ang' or 'angstroms'
> instead of 'Angstrom').
>
> The current version of the paper includes the statement
>
> > The IAU-standard prefixes for scaling the units are described in Paper I and
> > should be used with all coordinate types, except that the dimensionless ones
> > are not scaled.
>
> I suggest that this be strengthened to point out the need for strict adherence
> to the system outlined in Paper I.

There was quite a bit of discussion of the handling of units on the IVOA
lists a few months ago. The practical problem is that there is an awful
lot of existing data out there which uses "nearly but not quite FITS"
units syntax, and most software systems will need to be able to
interpret these units variations. There were three basic approaches
disccused on the IVOA lists:

1) Get everyone to change all their existing data archives to
correct non-standard units strings to FITS-WCS paper I syntax. Not
practical!

2) Leave the non-standard units specification unchanged, but add a new
standardised representation of the units (a "dimensional analysis"). This
still requires lots of data to be changed.

3) Make the units handling software clever enought to recognise and
translate the bulk of the common units syntax variations.

I favour the third solution. The Starlink AST library includes such a unit
parser. There's an active demo of it under "Units Facilities" at

http://www.starlink.ac.uk/ast/

David