>>>> 2.3:
>>>>
>>>> How can you have metadata on virtual data? How should we anticipate all
>>>> possible ways a user may ask for spectral cutouts, extraction etc. to be
>>>> prepared to answer?
>>>
>>> This is what on-demand data generation and virtual data are all about.
>>> The service describes the metadata of the virtual data product it would
>>> generate.
>>
>> How do you know the SNR ratio a priori for all possible spectral cutouts?
>
> You wouldn't know it without computing it from the data for the cutout
> region, but the interface assumes that the SNR may not be known for spectra
> in general, hence it is optional. Most metadata can however be computed
> for virtual data given the overall archival dataset values.
One could also argue that for this case (subset of a larger observation) one should just return the SNR of the full dataset. If for example we have a spectral observation with an overall SNR=10, and we see nothing in a cutout region (where presumably a line might be expected), that tells us more than if we see nothing and have computed SNR=1.0 for the cutout region.
I agree BTW, that exactly what SNR means for a spectrum is not well defined. This is similar to the absolute sensitivity of any observation, which should ideally be a measure of the minimum absolute flux which can be detected (although this can be wavelength dependent). Char should deal with this, but I don't think we understand it well enough yet to specify it. Perhaps SNR for an overall spectrum is similar to the limiting flux for an image through a given bandpass? In that case the SNR for a cutout of a spectrum may be approximated by the overall SNR (or limiting flux), although due to variations in sensitivity across the spectrum there could be significant deviations for spectra with a wide spectral coverage.