Further thoughts on UCDs
tam at lheapop.gsfc.nasa.gov
Wed Oct 29 10:38:28 PST 2003
One of my problems in understanding how or whether we should
change the UCD strings is the indefiniteness of the current UCD2
proposals which do not specify the full list of UCDs. I've just
spent a few minutes looking at the PHOT hierarchy. Probably
Sebastien and Roy have done this with far greater care, but
I've gone through the old UCD tree and tried to see what UCDs
are suggested by what's in the PHOT hierarchy, where I have
explicitly left out any band information from the UCDs assuming
that to be supplied by qualifiers in the em hierarchy.
There are only about 25 distinct phot words here versus just
under 500 in the original UCD1 tree. Most of this savings is at the cost of
having an extensive em tree describing the bands, but I think
that's helpful since we want to be able to combine
fluxes from different bands. I think it makes the photometry tree
much more accessible. One thought expressed here is that we should
not distinguish between flux, magnitudes and counts in the UCDs, but
rather do that with the units. I'm not sure if this is
a good idea but it seems to me that they are more 'similar'
to each other than each is to say a fluence or a color.
There is a question as to whether we specify colors by providing
special keywords for each color, or by using a pair of em qualifiers.
In the latter case, then the order of the em qualifiers may be
significant. If we go this route (and I think it makes more sense)
then my proposal for well-formed UCDs would have to be modified to
say something like: 'where the order of qualifiers is not significant,
they should be given in alphabetical order'.
… distinctions between counts, magnitudes, fluxes, etc should be carried in the
… units not the UCD.
phot.color (This is really a ratio, but we use color as the traditional
phot.color.diff (A difference in colors, i.e., a kind of second derivative)
em.bolometric (Bolometric measurements are after all just specifying a very
math/arith.ratio (for the phot_sd/b-bright, phot_tot-Bright/b-bright UCDs)
phot.system (is this different from phot.class?)
phot.parameter (some parameter of the photometric system
phot.spectrum (vector valued fields)
phot.timeseries (vector valued fields)
phot.image (vector valued fields)
This was just a quick exercise and doubtless there are better specific choices,
but the phot hierarchy is one area where I imagine that our software may
be very UCD aware. If we really can simplify it to this extent, then I think
this alone is a major impetus for going to UCD2.
More information about the ucd