[fitsbits] exposure time - resolution

[Steve Allen]

On Mon 2000-06-05T17:43:35 -0400, Arnold Rots hath writ:

> 2. In situations where the PSF varies considerably over the field of
> view (X-ray mirrors come to mind; but it's also true in most other
> parts of the spectru, even radio data) there is not much point in
> giving a single value for the resolution.

Of course, variations in a multi-dimensional gaussian describing FWHM
would usually be known at a much lower spatial resolution than the
primary image, and these could be included along with the primary
image using a compressed representation very much like that of the WCS
"Pixel Regularization Image" in Greisen and Calabretta paper I.
For a primary image with NAXIS = n this scheme would require a
"gaussian sampling kernel" image with NAXIS = n*(n+1)/2.

And describing a multi-dimensional gaussian is half as hard by
comparison to the general case of sampling.  In that we would
effectively wish to transmit a model of the full n-dimensional
convolution kernel as sampled at various points within the entire
primary image.  This could be done using the scheme outlined above,
and the "sampling kernel" image for a primary of NAXIS = n would
require NAXIS = n*n dimensions.

Either way this would begin to demonstrate a usage for a significant
number of the 999 FITS axes.  I'm not sure I see any fair way of
doing either of these solely with a set of keywords.

--
Steve Allen          UCO/Lick Observatory       Santa Cruz, CA 95064
sla at ucolick.org      Voice: +1 831 459 3046     http://www.ucolick.org/~sla
PGP: 1024/E46978C5   F6 78 D1 10 62 94 8F 2E    49 89 0E FE 26 B4 14 93