[fitsbits] start of Public Comment Period on the INHERIT convention

[Rob Seaman]

I make it a practice never to disagree with Frank :-)

Regarding the cubing of data, many might suggest that all axes should be coherent independent variables, e.g., spatial X & Y plus wavelength.  Stacking different tiles from a focal plane is as non-physical a representation as anything out there.

In general MEFs are combined into output 2-D mosaics and the software that does so is known as a “pipeline” (with complex pixel remapping, etc.) ;-)  Whether the individual pixel arrays are *input* as separate files, as MEFs, as cubes or anything else is an implementation detail.  The goal is to recreate a snapshot of the focal plane which is itself a representation of the sky.  As Steve says it is commonplace for issues of optical, electrical and digital engineering to seap through to the data engineering.

Doing physics with FITS should trump FITS metaphysics.

Rob
—

> On Jul 1, 2015, at 11:18 AM, Steve Allen <sla at ucolick.org> wrote:
> 
> On Wed 2015-07-01T14:07:37 -0400, Arnold Rots hath writ:
>> As an aside, I wonder how many of the multi-image FITS files could be
>> combined into an image cube.
> 
> Some could be combined, others could not.
> 
> The optical CCD instruments at Keck are not constrained to have the
> same readout geometry for every IMAGE extension in their
> multi-extension FITS files.
> 
> See the tartan/patchwork image example here.
> 
> http://www.ucolick.org/~sla/fits/mosaic/panes/phduNoted.html