Defaults in STC

[Arnold Rots]

Defaults are a nasty issue.  I have some sympathy for your argument,
but don't agree.

Most HI data repository keepers DO know what conventions were used.
The problem lies in the large variety and it would be highly
user-unfriendly to put the burden on the client.

If we allow data providers to specify "default", they will use it,
because the conventions are so obvious to them, in their own context.
In many cases it might not matter, but people will have forgotten what
they were doing and not realize that there are datasets where they do
need to be more specific.

For instance, there is the infamous example of the HI absorption line
that had disappeared in a WSRT observation in the early 70s - until it
was realized that the velocity that had been specified was optical and
the online frequency calculation assumed radio: as a result, the line
was completely out of the band.

More recently, I ran into a problem with some CO data of M51.  Things
didn't match - until I noticed the fine print that said that the
velocities were LSR, from which I inferred that they were also likely
to be radio definition; totally counter-intuitive.

What I am trying to say is that it is usually trivial for the data
providers to be explicit and that allowing defaults in cases where it
does not matter will lead to their being used in situations where it
does matter.  Local defaults are usually known - why not insert them
explicitly in the STC elements?

A note on references: these are IDREFs, meant to either being able to
use the same element multiple times in a single XML document, or refer
to a table FIELD.

  - Arnold


Anita Richards wrote:
> 
> 
> STC states that there are no global defaults, which is reasonable if
> this means that the schema requires every element to have a value.  In
> general data providers should be encouraged to describe their data
> rigorously. However for some data it may not be practical for a number
> of reasons, possibly because the data provider thinks something is
> irrelevant and is too busy, or because some details of the data source
> are lost in the mists of time.  Hence we might end up rejecting highly
> desirable data, for example a great deal of spectral data such as HI
> data does not say what reference frame or Doppler conventions it has
> used and in many cases the spectral or velocity uncertainties are
> greater than the differences between LSR, Heliocentric, and between
> optical, radio or relativistic.
> 
> There are several solutions and these could be applied as local
> defaults ('local' meaning either the data provider or the local VO
> data centre etc.).  In the above instance allowing 'Not Available' as
> a reference position location might work, but might break conversions
> between e.g. wavelength and velocity.  Alternatively someone with
> knowledge of the probable observing procedure used might say 'call it
> all heliocentric, optical'.  Bob Hanisch proposed that metadata
> accuracy should have a flag which could reflect this - or the data
> provider could increase all relevant uncertainties by an appropriate
> amount.
> 
> Section 2 of STCv1.0 (p. 4) suggests that such context-dependent
> defaults might be acceptable. Section 6.2.4 describes how
> observational data uses two STC objects; the observatory data
> (location etc.) is used for some operations on observational data.
> Section 7.1 says that all elements can contain actual values or refer
> to other elements. Is it possible to use these sructures to supply
> defaults where these involve making educated guesses about the
> observattion procedure, or even other elements? In some cases
> (e.g. antenna positions) there can also be legitimate improvements
> when more accurate data becomes available.  If not, local defaults
> could simply be hardwired.
> 
> 
> Summary:
> 
> Encourage complete descriptions but rather than reject data:
> Allow Not Available
> Explain how to use references/local defaults
> Use metadata accuracy flag system to warn when this has been done
> 
> thanks
> 
> Anita
> 
> - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
> Dr. Anita M. S. Richards, AVO Astronomer
> MERLIN/VLBI National Facility, University of Manchester,
> Jodrell Bank Observatory, Macclesfield, Cheshire SK11 9DL, U.K.
> tel +44 (0)1477 572683 (direct); 571321 (switchboard); 571618 (fax).
> 
--------------------------------------------------------------------------
Arnold H. Rots                                Chandra X-ray Science Center
Smithsonian Astrophysical Observatory                tel:  +1 617 496 7701
60 Garden Street, MS 67                              fax:  +1 617 495 7356
Cambridge, MA 02138                             arots at head.cfa.harvard.edu
USA                                     http://hea-www.harvard.edu/~arots/
--------------------------------------------------------------------------