re OBJECT keyword

[Frank Valdes]

I am also in agreement with Steve Allen that the OBJECT keyword has a long
history of use by astronomers and acquisition systems as a human entered
label for the data.  As he noted, this label is also used by many analysis
systems for short, one-line, listings of a collection of data where it
is desired by the observer to get the most information in one line.
Thus including exposure time, filters, coordinates, personal finder
desginations, etc is common practice.

I don't think the original OBJECT keyword was abused but that the original
minimal set of recommend keywords was (appropriately) limited though,
unfortunately, there was nothing defined for an observation label.  So it is
only natural that people used this keyword in preference to the worse
solution of new keywords which other software would not even see.  It is
actually a tribute that many diverse software systems label displays,
graphs, listings, etc reasonably (without great user outcry) using
information in the OBJECT keyword from many diverse sources.

In recognition of this the NOAO FITS keyword dictionary has defined the
OBJECT keyword as the observation label and new keywords for the archival
name following the IAU naming standards.  While it may be some time before NOAO
data taking systems actually make use of this, the proposed names in our
dictionary are OBJNAME/OBJnnnnn (where the latter is when there are
multiple objects in the observation such as with MOS systems).  For an
archive there are also some other things which would be good to define.  In
particular OBJTYPE/OBJTnnnn which give some standard object type such as
star, galaxy, etc.  The dictionary is still a design document which is
quite thorough in trying to define useful information for ground-based
optical astronomical data.

Returning to the question of what happens at the telescope (for the
traditional ground-based scenerio) the acquisition system asks the observer
for a label for the observation.  It is important to minimize how much is
required for the user to enter so a single line is typical.  This
information is added to the data and is generally mapped to the FITS
"standard" in the OBJECT keyword.  This is so entrenched and any changes to
the meaning of OBJECT will not affect past data that I don't think it
should be changed.  What should happen (in new systems) is that the
software systems should take the OBJECT keyword, coordinates, telescope
caches, etc and use name resolvers to automatically create an archival
keyword, such as OBJNAME.  The user could be asked to confirm this which is
an easier requirement during observing than to require a standardized
entry.  Of course, archival ingestion systems can also do this.

I don't think the concern of Arnold Rots is a serious problem.  Any
software can use a precedence and fallback scheme.  The NOAO dictionary
uses this.  An archive system that needs to produce a designation can look
for the more precise keyword, say OBJNAME or IAUDESIG, and if it does
not find it falls back to OBJECT.

There are many ways to address this concern: education, name resolvers, new
FITS standards, queue and proposal entry systems, new systems at new
telescopes, smart archives.  All of these will occur.  But if people are
anxious to do something now that current systems under development might
take into account, then adopting a new keyword is the approach I would
advocate as opposed to trying to change/tighten the original OBJECT keyword
usage.

Frank Valdes
NOAO/IRAF Group