[Issues] Bad RegriddedFrame astrometry

John P. McFarland mcfarland at astro.rug.nl
Tue Nov 20 14:31:52 CET 2007

Hi All,

A bug was just discovered and fixed related to astrometry in RegriddedFrames 
created on or after 11 Oct 2007.  If you have created any RegriddedFrames 
(or data products derived from them) since this date, you should recreate 
them and mark the previous versions as invalid (i.e., is_valid=0 or 
SuperFlag in CalTS).

Normally, a RegriddedFrame is made by SWarp by supplying input image
filenames and specifying the proper extensions for optional weight frames
and separate header files containing the updated astrometry.  This assumes
the weight and header files have the same base filename (e.g., file.fits,
file.weight.fits, file.head, respectively).  When CoaddedRegriddedFrames are
made, the same procedure is used, but no header files are needed (the
proper astrometry is already in the RegriddedFrame), and symbolic links are
made and later removed to prevent long command-line length issues
(e.g., 0+.fits, 0+.weight.fits, 1+.fits, 1+.weight.fits, etc.).

Approximately 6 weeks ago, the functionality to link input image and weight 
frame filenames to shorter names was moved from astro/main/RegriddedFrame.py 
to astro/external/Swarp.py to optimize the process and to consolidate 
functionality into appropriate locations.  This had the unfortunate 
side-effect of breaking the image-weight-header file commonality.  The 
header file containing the updated astrometry for the to-be-created 
RegriddedFrame no longer had the same base name and it was not applied to 
the output RegriddedFrame because SWarp did not find it.

This has now been fixed.  Please update your check-outs, if applicable, 
before (re)creating any RegriddedFrames.



More information about the Issues mailing list