Obscore Subtleties

[Markus Demleitner]

Hi DAL,

On Wed, Mar 07, 2018 at 11:47:30AM -0500, Arnold Rots wrote:
> The first issue is that it would be helpful for unsuspecting users
> to know that this was a calibration observation. Our database
> knows that, but Obscore does not have a parameter that allows
> communicating the type of observation. In addition to calibration,
> one could think of observing modes, like pointed or scanning,
> objective prism, grating, ...

I have to say that I've missed *some* way to mark up this kind of
information as well.  I have a couple of objective prism exposures,
and so far I've largely kept them out of Obscore (mostly because I
don't really have good calibration for them, but also to keep
unsuspecting users for becoming confused).  And then there's this
kind of thing:

http://dc.zah.uni-heidelberg.de/flare_survey/q/tsdl/static/photos/ESO040_009768.jpg

-- where people were looking for flares by doing multiple exposures
on the same plate.  For these guys, I've put only the "calibration"
(i.e., direct exposure) frames into obscore, and the timed exposures
are in datalink with an improvised semantics.  But of course, that
way obscore queries against time will miss the multiple exposures...

Then there are odd cases from ages ago, where they did a multiple
exposures of the pole sequence and a field of interest.  And sure,
there are the frames that followed an asteroid (I'm leaving these out
of obscore, so far, though one might argue that I should be putting
in the target name and leave out the astrometry).  And more oddities
-- some of my plate collections are a zoo.

> Should Obscore be extended with information of this kind?
> If so, how and to what extent?

Hard to say.  I'd say we need to define discovery cases for them, and
then see how much plain, normal discovery is impacted when they go
in.  We might find that dataproduct_subtype and a taxonomy of fun
people have with imaging is a good answer.

> The second issue concerns the coordinates, and is a question
> of ambiguity.
> The standard says "center of the observation".
> Is that the pointing position (i.e., the tangent point) or the center
> of the field of view? Both have their merits.
> In the case at hand the tangent point fell well outside the FOV.

I'd say FOV -- that's what almost everyone wants to discover on.

           -- Markus