Utypes for Light Curves?

[Doug Tody]

On Tue, 15 Jun 2010, Roy Williams wrote:

>>  Have a look at Bruno's
>>  http://www.ivoa.net/internal/IVOA/InterOpNov2009DM/sample.html
>
> Alberto
> This looks like an interesting document. Finally the Utypes are brought out 
> into daylight where we can see them.

See also http://www.ivoa.net/internal/IVOA/SSA_v1_1/ssap-keywords-rc2.xls
which is the SSA/spectrum data model in spreadsheet form.  This is 
essentially the same as in Bruno's web page rendition, but is more
complete were it comes to describing an actual spectrum or time series.
In particular it includes the fields for the actual data values.

> Following this, the Time Axis data model and Flux Axis data models look quite 
> relevant. I am imagining a representation of a Light Curve that has some 
> parameters and a table wrapped in a VOTable. The parameters and fields will 
> include these things from the above, for example
> <FIELD utype="FluxAxis.Accuracy.StatError"/>  for error in flux
> <FIELD utype="TimeAxis.Coverage.Location.Value"/> for time of observation
>
> Is that correct?

Yes, this is part of the characterization of the data, as opposed to
the actual data.

> If so we need a few more Utypes:
> -- The FluxAxis data model does not seem to have an actual measured value of 
> Flux. What the Utype for the FIELD that has the flux value?

FluxAxis is part of Char, hence it does not give the actual data
samples.  In Spectrum this is given by Spectrum.Data.FluxAxis.Value.
For a time series one would presumably have something like
TimeSeries.Data.FluxAxis.Value.

> -- How do we represent the filter used for the observations? Don't we need 
> upper and lower limits for the filter, central wavelength, name of the 
> filter, that sort of thing? The Spectral DM lookslike it has some useful 
> Utypes, but a Spectrum is not the same a Filter.

This would be provided by the Photometry data model, which goes beyond
what we had when we did SSA/Spectrum.  Ultimately we just add a few more
attributes to the overall model.

> -- How does the TimeAxis model work with STC? It says "Midpoint of exposure 
> on MJD scale". What time scale is this, where is the observer, is it 
> barycentric, etc etc. How does this sort of thing integrate with the Utypes?

Basically we just take whatever global frame information we need from
STC and include it in the model in the COORDSYS class.  Hence there
is a CoordSys.TimeFrame.Name to define the time scale.  The spatial
reference position is CoordSys.SpaceFrame.RefPos.  And so forth.
Possibly for time series data more detail is needed, but it should
be pretty close already.

It would be interesting to do a use case for some actual time series
datasets to see what may be missing.

 	- Doug