[QUANTITY] Plea for pragmatism

[Doug Tody]

> Maybe we are not on the same wavelength here, let's try to clarify this
> please. Let's take the case of "flux bandpass":
> 
> >     flux bandpass
> >         range of flux values in data
> >         hiValue, loValue, units, fillFactor (no refValue)
> >         loValue is also known as the limiting flux or magnitude
> >         hiValue is saturation limit or maximum flux
> 
> Here loValue is the limiting flux, and hiValue is the saturation limit.
> those two values have nothing to do with the photometric accuracy,
> eg, the error on the zeropoint. Where do you describe that ?
> 
> Similarly the sky coverage (be it a circle or a polygon) should
> have an associated astrometric accuracy (any vertex position is knwown
> within a certain precision).

Ok, I see what you are getting at.  To describe this we probably do need
to add a new attribute to each model.  I agree this is important to fully
characterize the dataset.  This will potentially duplicate information
in any actual full calibration included in the dataset, e.g., a WCS,
a photometric solution, etc., but that is what the summary data models do.

 
> Also, for the spatial bandpass I thought you meant 
> loValue to be the resolution, and hiValue the field of view; 
> did I missinterpreted that ?
>
> > All of these models are assumed to be observational models calibrated to
> > the sky, not instrumental models.  Hence the spectral bandpass is not just
> > for a filter, but also for the telescope, instrument, atmosphere, etc.
> > The spatial bandpass would reflect both instrumental resolution and,
> > at the low end, seeing effects (note also that for instruments such as
> > interferometers the spatial frequency response and filling factor will
> > in general not be flat as it is for O/IR instruments).
> 
> Agreed, but the pixel scale is still an important piece of information
> and it is useful to know whether an image is oversampled or not.
> And, yes, the PSF might vary within the field of view.

Agreed.  Perhaps for the spatial bandpass we should have:

    refValue    Spatial resolution. (characteristic value)

    loValue	Instrumental resolution.  This is the theoretical minimum
		resolution (projected back to the sky) as determined from
		the instrument, e.g., the pixel size on a CCD as mounted
		in the specific telescope used, the limiting resolution as
		determined from the maximum baseline of an interferometer,
		and so forth.

    hiValue	Maximum spatial frequency in the data.	This is the field
		of view for an imaging detector, the maximum field of view
		for an interferometer determined by some TBD heuristic,
		and so forth.

    fillFactor	For the spatial bandpass this provides a measure of
		whether all spatial frequences between the loValue and
		hiValue are present in the data.  For an imaging detector
		this would be 1.0.

I wonder if there is a useful refValue for the flux bandpass, since we
have found one for everything else?