[fitswcs] RADESYS = 'GAPPT'

[Rob Seaman]

But via a coherent mapping relative to the CCD array?  And the array relative to the bore sight of the telescope?  And the optics relative to the drive encoders?  At some point the utility of a top-down pointing model gives way to the mask preparation process (and the catalogs Steve mentions).  Similarly, at some point a utility threshold of mapping the telescope pointing across to wide-field imaging wcs folds.  Metadata doesn't always win out over data.  And data logistics aren't in thrall to observing logistics.

A modern instrument has hundreds of engineering parameters that always generate a discussion about which to include in the FITS headers visible to the astronomers.  It seems to this observer that these discussions are often backwards.  Rather than might this-or-that keyword be handy to keep around, ask what science metadata requirements flow from the science use cases.  Such an inventory may well then imply that the data handling jump out of the well worn path of previous instrumentation and include data-dependent components like astrometry.net rather than metadata-derived computations from an earlier stage of the empirical process.

Which is to say that I've had surreal conversations too many times about the "meaning" of RA versus TELRA versus the WCS keywords versus "reference pixels" versus...

Rob
--
On Mar 16, 2011, at 11:52 AM, <patrick.wallace at stfc.ac.uk> wrote:

> Rob Seaman wrote:
> 
>> For other instrumentation such as multi-object spectrographs, the
>> notion of telescope pointing is quite a bit different.  What does
>> it mean for a bench spectrograph to be "on-target"?
> 
> That the light illuminating the chosen piece(s) of slit is emanating
> from the desired celestial coordinates.  In practical terms, there
> will probably be an array of fibres in the focal plane, and "on-
> target" means that the sky is being imaged onto the array as
> intended.
> 
> Patrick Wallace