On 18 February 2011 11:49, Ciara Quinn <[log in to unmask]> wrote:
> Hi all.
>
> I am trying to manipulate some MOPRA data using kappa, by converting to sdf
> with FITS2NDF and then converting back to a FITS file after the
> manipulation. In the final FITS file I have lost the astrometry (whether I
> open it in KVIS or Gaia).
>
> I get the following error if I try and open the final FITS file in Gaia.
>
> ! astRead(FitsChan): Cannot implement spectral algorithm code '-LSR'
> specified in FITS keyword 'CTYPE3A'.
> ! astRead(FitsChan): Unknown algorithm code or unusable parameter values.
> !! AST: Error at line 176 in file ./generic/GaiaFITS.C.
> ! astRead(FitsChan): Cannot implement spectral algorithm code '-LSR'
> specified in FITS keyword 'CTYPE3A'.
> ! astRead(FitsChan): Unknown algorithm code or unusable parameter values.
>
> The SDF file, made using FITS2NDF doesn't have the keyword CTYPE3A in its
> FITS header.
>
> Does anyone know how I can fix this?
Hi Ciara,
It was caused by a bug in the AST library, now fixed.
This bug prevented the AIPS non-standard CTYPE3 value "VELO-LSR" from
being translated into the equivalent FITS standard keywords
(CTYPE3="VRAD" and SPECSYS="LSR").
You can get the fixed version by copying the current development
version of Starlink from jac (see
http://starlink.jach.hawaii.edu/starlink/rsyncStarlink).
Alternatively, a temporary fix to get round this issue is to delete
the AXIS arrays from the NDF before running ndf2fits:
% kappa
% erase BHR4212CO_MEAN.axis ok
% ndf2fits BHR4212CO_MEAN
Regarding the fact that the NDF created by fits2ndf did not have a
CTYPE3A in its FITS extension, this keyword is added not by fits2ndf,
but by ndf2fits, when the final fits file is created. That's the one
to check.
David
|