[Fwd: VOTimeSeries]

[Doug Tody]

Hi All -

A couple of points here for the record:

    o	The general spectrophotometric model upon which SSA is based,
	and which we have always intended to use for time series as
	well, uses STC (the data model not the XML serialization) for
	all reference frames, including time.  Hence any time frame (or
	spatial frame) defined by STC can potentially be used for spectral
	or time series data.  Of course, the scope of a model such as
	Spectrum is broader, and can express things which STC cannot.
	Models such as Spectrum and STC are not necessarily competing with
	each other; you don't necessarily have to choose one or the other.

    o	A general abstract data model such as STC, which could be used
	for many things, needs to be fairly complete in terms of what it
	can describe.  When we apply such a general model to a specific
	case however, for example time series, or a data discovery query,
	or instrumental calibration, etc., it may be sensible to restrict
	the model to allow only certain frames, units, and so forth.
	In other words, just because a general model defines something,
	does not need one has to use in an application of the model.

Regarding flexibility in terms of specifying frames and units, this
issue comes up frequently.  In the past we have always concluded that
we can restrict the choice of things such as frames and units in some
circumstances (e.g., dataset characterization, services interfaces) but
not so easily in others (representing actual observational datasets).

It is not just time which is the problem - photometric systems and units
for example are also quite complex, probably more of a problem than time.
For many actual observations, even time series, time does not have to
be known to great precision.

	- Doug