alberto.micol at eso.org
Wed Jun 24 02:53:58 PDT 2009
Douglas Tody wrote:
> On Tue, 23 Jun 2009, Alberto Micol wrote:
>> As demonstrated by this long discussion, it is clear that all the above
>> requires clarification in the SpectrumDM text. This kind of confusion
>> (in this and other standards) is killing all our efforts. We have to
>> be precise, and to provide clear and unambiguous (almost mathematical)
>> definitions, accompanied by a good set of real (as opposed to virtual)
>> examples (tutorials). Revision of the IVOA documents should take into
>> account the confusion that certain definitins might introduce in the
>> (even vo-aware) data providers. Crystal-clear documents are an art;
>> we are good, but apparently not in Arts...
> What we most need now are *implementations*, with real data, of both
> SSA services and client applications which begin to use the information
> in the SpectrumDM. After there have been a few more of these we will
> have enough experience to update the SSA/SpectrumDM specifications
> and address most of these subtle issues. Adding a user guide could
> also be very useful.
Agreed, as I think you agree that implementers need crystal clear specs.
I'd suggest a little revision of that SpectrumDM section where the
flux calibration is described.
> If applications do not present this information well to the user
> that should be addressed as the applications are updated to be
> more VO-aware. This won't happen until we have more, better, more
> compliant SSA services.
That's the key, and the question is: why we do not have yet those?
One example that comes to mind is that ground based spectra are very often
published using a 1D image FITS format, where the 1 axis is the wavelength
(using CDELT1, CRVAL1, CPIX1 to specify the wavelength wcs).
That format is NOT supported by the SSA, and that might cause indeed
problems to the ground based data providers. Sure, they could translate
FITS files on the fly, but we shall not forget that data providers have
things to do than try to twist their data into the VO. Just looking at
the number of space-based SSA services greatly outnumber the
At the level of about 90 vs 10 (or worse). The one I stated is certainly
not the only reason
for this (calibrating a ground-based spectrum is much more difficult),
but the format problem
exists (and I can tell, now that I work for a grond base archive!)
>> Did the FITS community ever tried to describe this? Why isn't there
>> a BCALIB = 'RELATIVE' / or ABSOLUTE or UNCALIBRATED or RECTIFIED
>> FITS keyword?
>> Shall we reccomend one? (Haven't I just done that? ;-))
> Note the SSA/SpectrumDM data model spreadsheet suggests FITS keywords
> for many of the data model attributes (the HST/MAS SSA services for
> example are already using these):
> For the FLUX calibration attribute we have FLUX_CAL (also TIME_CAL
Yap, thanks, I forgot that, but I was referring to the fact that the
(not the VO one) never introduced such keyword, as if the problem never
before. I find that puzzling...
> - Doug
More information about the dm