[fitsbits] EXPOSURE or OBSTIME?

[Steve Allen]

On Mon 2000-06-05T10:36:35 -0400, Eric Greisen hath writ:

>    Perhaps this is a case for me to recommend the use of a 1-pixel
> axis to specify time, with the axis increment being the integration
> time.

This is a cool idea, but it does not address the complexities of
optical exposures where the cloudcover is variable.  In such temporal
axis cases it is important to know the start time, the end time, and
the exposure-weighted mean time (if possible).

The teams who are spectroscopically discovering extrasolar planets
need exactly this kind of information.  For the sake of expediency,
and because there is only one temporal axis, the current set of
instruments simply report these numbers as locally-defined FITS
keywords.

In the spatial case of optical CCDs it is also the case that the
photon-sensitivity of a CCD pixel is not uniform over each rectangle
of the rectanglar grid.  For any single CCD it depends on f-ratio,
color, and various factors of vignetting.

So for the more general case, this harks back to the sampling theory
question.  As Eric points out, there is currently no means for
communicating whether the sampling over any FITS axis represents
a sample at a point or an integral over a bin.

I wouldn't want to hang up the WCS effort by suggesting that it should
contain a solution to these questions, but the issue of what FITS
means by sampling would be good to discuss for the sake of future
standardization.

--
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