Version 1.9.9 of UCD definition

[Clive Page]

On Tue, 21 Oct 2003, Anita Richards wrote:

> Thus, to keep the 'decades', this seems more suitable:
> XRAY 0.12 - 12 keV (as per UCD2)
> GAMMARAY > 12 keV
> but maybe a high energy astronomer can advise....

I think that boundary between the two is reasonable: some past instruments
called 'X-ray' have some coverage up to maybe 15 keV, but most modern
telescopes depend on grazing-incidence reflection, which practically stops
working above about 10 or 12 keV.

> OVERLAPPING DATA

> How will the UCDs be used? As I understand it, in order to use Vizier or
> the Aladin SED tool to search for e.g. radio observations between 1.3 and
> 1.7 GHz (radio L-band), the software would look for UCDs
> em.radio.750-1500MHz and em.radio.1.5-3GHz.

I think the way it has to work is this: the UCD defined for some dataset
should be as specific as possible, so if the data fall practically within
a single band (say within 750-1500 MHz) then you declare the UCD as
"em.radio.750-1500MHz".  If not then you have to fall back on a less
specific UCD of "em.radio".  A user wanting radio measurments at some
frequency should be prepared to search for both _the_ most specific UCD
and the less specific UCD, i.e. "em.radio".  This may bring up some false
positives, but so will any scheme that we can devise.

The alternative, as you say, is to apply a two or more UCDs to a dataset.
That might be better in principle, but difficult in practice.

> * not use the fine divisions in UCDs, but treat frequency like position
>   and treat every query as a 1D cone search, ie a linear segment search.

Well that's similar to a proposal I made some time ago (maybe in a Data
Models context).  Datasets should specify the range of frequencies they
cover (which requires two numbers for the min frequency and max frequency,
which doesn't imply that coverage between the limits is continuous, gaps
should be ignored); then users also specify the range of interest.  It's
then a trivial exercise for a computer to compare the two intervals and
work out which resources overlap the range of interest.  All this without
us having to dream up any artificial divisions between wavebands.  But
that doesn't fit in with UCDs as devised up to now, and I don't see any
easy way to make it fit.


-- 
Clive Page
Dept of Physics & Astronomy,
University of Leicester,    Tel +44 116 252 3551
Leicester, LE1 7RH,  U.K.   Fax +44 116 252 3311